Moesin (MSN) as a Novel Proteome-Based Diagnostic Marker for Early Detection of Invasive Bladder Urothelial Carcinoma in Liquid-Based Cytology.

Bladder urothelial carcinoma (BUC) is the most lethal malignancy of the urinary tract. Treatment for the disease highly depends on the invasiveness of cancer cells. Therefore, a predictive biomarker needs to be identified for invasive BUC. In this study, we employed proteomics methods on urine liquid-based cytology (LBC) samples and a BUC cell line library to determine a novel predictive biomarker for invasive BUC. Furthermore, an in vitro three-dimensional (3D) invasion study for biological significance and diagnostic validation through immunocytochemistry (ICC) were also performed. The proteomic analysis suggested moesin (MSN) as a potential biomarker to predict the invasiveness of BUC. The in vitro 3D invasion study showed that inhibition of MSN significantly decreased invasiveness in BUC cell lines. Further validation using ICC ultimately confirmed moesin (MSN) as a potential biomarker to predict the invasiveness of BUC (p = 0.023). In conclusion, we suggest moesin as a potential diagnostic marker for early detection of BUC with invasion in LBC and as a potential therapeutic target.


Introduction
Bladder urothelial carcinoma (BUC) is the most common and lethal malignancy of the urinary tract [1]. The evaluation of cancer cell invasion beyond the subepithelial layer is crucial due to the
Subsequently, a gene ontology analysis on biological process revealed enrichment in cytoskeleton organization, cell migration, and cell motility, which implicated significant alterations in the cytoskeletal architecture and invasion process ( Figure 1C, Table S4). Especially, DEPs involved in cell motility and invasion were mostly upregulated in MIBUC compared to NIBUC. A further comparison of stromal-invasive BUC (SIBUC) and NIBUC revealed that biological processes with ribonucleoprotein complex biogenesis and antigen processing/presentation of peptide antigen were significantly enriched comparison of stromal-invasive BUC (SIBUC) and NIBUC revealed that biological processes with ribonucleoprotein complex biogenesis and antigen processing/presentation of peptide antigen were significantly enriched in SIBUC by upregulated and downregulated DEPs, respectively ( Figure S2, Table S4-S6). Molecular functions with UFM1 activating enzyme activity and oxidoreductase activity were enriched while comparing MIBUC and SIBUC groups.

Proteomic Library of BUC Cell Lines Identified Candidate Biomarkers
For the discovery of candidate biomarkers related to invasion, we performed a tandem mass tag (TMT) proteomic analysis and constructed a BUC cell line proteomic library ( Figure 2, Table S7). First, we assessed the invasion and migration ability of eight BUC cell lines to categorize them into invasive BUC cell line (IBUC_CL) and non-invasive BUC cell line (NIBUC_CL). Among the BUC cell lines, T24, J82, and 253J-BV (IBUC_CL) revealed the most invasive and proliferative capacity, while RT4, HT1376, and HT1197 showed the least aggressive ability (NIBUC_CL) (Figure 2A,B). Next, we conducted a proteomic analysis between IBUC_CL and NIBUC_CL for the discovery of candidate biomarkers related to cancer invasion and identified 677 DEPs and aforementioned proteins in LBC proteomics, including ATP1B1, CLTC, GRHL2, KPNA3, LDHB, LLGL2, MSN, MVP, NCAM2, PARP4, PPA2, and VAPA ( Figure 2C, Table S8).

Proteomic Library of BUC Cell Lines Identified Candidate Biomarkers
For the discovery of candidate biomarkers related to invasion, we performed a tandem mass tag (TMT) proteomic analysis and constructed a BUC cell line proteomic library ( Figure 2, Table S7). First, we assessed the invasion and migration ability of eight BUC cell lines to categorize them into invasive BUC cell line (IBUC_CL) and non-invasive BUC cell line (NIBUC_CL). Among the BUC cell lines, T24, J82, and 253J-BV (IBUC_CL) revealed the most invasive and proliferative capacity, while RT4, HT1376, and HT1197 showed the least aggressive ability (NIBUC_CL) (Figure 2A,B). Next, we conducted a proteomic analysis between IBUC_CL and NIBUC_CL for the discovery of candidate biomarkers related to cancer invasion and identified 677 DEPs and aforementioned proteins in LBC proteomics, including ATP1B1, CLTC, GRHL2, KPNA3, LDHB, LLGL2, MSN, MVP, NCAM2, PARP4, PPA2, and VAPA ( Figure 2C, Table S8).

Multi-Omic Platforms Selected Moesin (MSN) as a Potential Biomarker for Invasive BUC
For the discovery of potential biomarkers related to invasion, we performed a stepwise analysis on a multilayer platform ( Figure 3A, Table S9). First, we compared DEPs from one-way ANOVA and paired t-test of LBC samples. Proteomic analysis of LBC revealed 182 DEPs and 188 DEPs in one-way ANOVA and paired t-test, respectively. Cross-validation with these two platforms showed 139 common DEPs in LBC proteomics. Next, we employed an extra platform of BUC cell line proteomics to determine a predictive biomarker for bladder cancer invasion. We used the 677 proteins for BUC cell line proteomics for a comparative analysis with DEPs derived from human LBC proteomics. Consequently, the cross-validation of DEPs from LBC and BUC cell lines revealed 12 invasion-

Multi-Omic Platforms Selected Moesin (MSN) as a Potential Biomarker for Invasive BUC
For the discovery of potential biomarkers related to invasion, we performed a stepwise analysis on a multilayer platform ( Figure 3A, Table S9). First, we compared DEPs from one-way ANOVA and paired t-test of LBC samples. Proteomic analysis of LBC revealed 182 DEPs and 188 DEPs in one-way ANOVA and paired t-test, respectively. Cross-validation with these two platforms showed 139 common DEPs in LBC proteomics. Next, we employed an extra platform of BUC cell line proteomics to determine a predictive biomarker for bladder cancer invasion. We used the 677 proteins for BUC cell line To shortlist the optimal candidates, we evaluated the change in proteomic intensities among groups based on cancer invasion in all the proteomic data, including that of label-free LBC and the cell line TMT ( Figure 3B, Table S9). Seven out of the 12 candidates showed a random alteration of protein intensity regardless of the advanced tumor stage in BUC patients and of invasive phenotype in cell lines, which were eventually excluded for further validation tests. The remaining five candidate biomarkers, namely, GRHL2, LLGL2, MSN, NCAM2, and VAPA, demonstrated a gradual increase of protein intensity in groups with more invasive phenotypes, which guided us to select them as the final candidates for further validation.
In gene ontology analysis, the five final candidate biomarkers were associated with cellular transport (LLGL2 and VAPA), immune response (VAPA), invasion process (LLGL2 and MSN), and cellular organization (GRHL2 and MSN) ( Figure 3C). All four candidates, except MSN, were downregulated in MIBUC. GRHL2 affects cell morphogenesis and epithelial-mesenchymal transition (EMT) and acts as a tumor suppressor in various tumors [22,45]. LLGL2 and VAPA are involved in cellular transport and affect invasion process. These genes show a tumor-suppressive role in various tumors [21,24]. On the other hand, MSN expression was upregulated in MIBUC and consistent with its oncogenic role in invasion process [16,18,19].

The Inhibitory Effect of Moesin (MSN) Depletion on Cancer Invasion in BUC
Next, we performed a two-dimensional (2D) invasion and migration assay with T24 and J82 BUC cell lines to evaluate how the five selected candidates modulated the invasion ability of BUC cells ( Figure 4A). The invasion and migration assay showed that BUC cells were significantly reduced in both the MSN-depleted cell lines as opposed to other candidate biomarkers including LLGL2, NCAM2, and VAPA that all failed to prove significant alteration of invasion ability ( Figure S3). A further 3D invasion assay showed concordant findings that MSN knockdown T24 and J82 BUC cells exhibited a remarkable reduction in cell invasion ( Figure 4B,C). A further pre-ranked gene enrichment analysis utilizing gene ontology term-defined gene sets linked MSN to proteins involved in actin dynamics (CORO1A, FLNA, and LCP1), formin (FMNL1), integrin signaling (FERMT3, ITGAM, ITGA6, and ITGB4), extracellular matrix (ECM) remodeling (MMP9), EMT phenotype (VIM), small GTPase activator (ARHGEF2), and mitogen-activated protein kinase (MAPK) pathway (MAP2K1) that were upregulated in the MIBUC group ( Figure 5). A further co-expression analysis using TCGA data revealed strong correlation of MSN expression with proteins involved in actin dynamics (FLNA), integrin signaling (ITGAM), and EMT phenotype (VIM) and suggested a co-operative role of MSN with signaling pathways associated with cell motility ( Figure S4). Cancers 2020, 12, x 6 of 17

Slide-Based Moesin Immunocytochemical Test Predicts Invasive Urothelial Carcinoma on Urine Liquid-Based Cytology
We further verified the diagnostic role of moesin to predict BUC invasion and its clinical application through ICC based on an independent urine LBC cohort, which was composed of NIBUC, SIBUC, and MIBUC. The proportion of moesin immunoreactivity significantly increased with BUC invasion-38.5%, 80.0%, and 85.7% in NIBUC, SIBUC, and MIBUC, respectively (p-value = 0.046; Figure 6, Table 1). The predictive ability was more powerful in a dichotomous comparison between the BUC group without invasion and the other group with invasion (p-value = 0.023, moesin immunoreactive rates, 38.5% vs. 82.4%, respectively).

Slide-Based Moesin Immunocytochemical Test Predicts Invasive Urothelial Carcinoma on Urine Liquid-Based Cytology
We further verified the diagnostic role of moesin to predict BUC invasion and its clinical application through ICC based on an independent urine LBC cohort, which was composed of NIBUC, SIBUC, and MIBUC. The proportion of moesin immunoreactivity significantly increased with BUC invasion-38.5%, 80.0%, and 85.7% in NIBUC, SIBUC, and MIBUC, respectively (p-value = 0.046; Figure 6, Table 1). The predictive ability was more powerful in a dichotomous comparison between the BUC group without invasion and the other group with invasion (p-value = 0.023, moesin immunoreactive rates, 38.5% vs. 82.4%, respectively).

Discussion
In this study, we prioritized moesin (MSN) as a protein biomarker for early detection of BUC invasion using a liquid-based cytologic test, which is the most widely used clinical screening method for monitoring bladder cancer progression. Moesin showed predictive ability for invasion of BUC in the independent ICC cohort. In situ immunoreactivities of moesin on LBC slide-based tests revealed statistical discriminative power when more than one cell was immunostained in LBC slides, which

Discussion
In this study, we prioritized moesin (MSN) as a protein biomarker for early detection of BUC invasion using a liquid-based cytologic test, which is the most widely used clinical screening method for monitoring bladder cancer progression. Moesin showed predictive ability for invasion of BUC in the independent ICC cohort. In situ immunoreactivities of moesin on LBC slide-based tests revealed statistical discriminative power when more than one cell was immunostained in LBC slides, which was concordant with the previous study where positive immunostaining in any cancer cell was significantly associated with poor overall survival in BUC [17]. In the TCGA public dataset, the higher expression of MSN transcript was also marginally associated with unfavorable clinical outcomes (p-value = 0.061; Figure S5). The higher expression of MSN was also associated with advanced American Joint Committee on Cancer (AJCC) staging (p-value = 0.001) and angiolymphatic invasion (p-value = 0.050) (Table S10).
Growing evidence has shown that moesin (MSN) plays a crucial role in invasion by cytoskeletal reorganization and EMT in various malignant tumors [16,18,19]. Although the functional relevance of MSN has not been fully revealed in urothelial carcinoma [9,17], our in vitro 3D spheroid invasion assay along with the 2D invasion assay confirmed significantly decreased invasion ability in MSN-depleted BUC cell lines. The 3D tumor spheroid invasion assay has advantages, for example, tumor spheroids mimic a more physiologic tissue-like morphology and recapitulate tumor cells and microenvironment [46,47] Our proteomic data demonstrated that moesin (MSN) upregulation is one of the major factors for BUC invasion, which can be more critical as the previous proteomic analysis of urine extracellular vesicles revealed moesin as one of the candidate biomarkers for bladder cancer diagnosis [9]. In a further network analysis with protein-protein interactions, we confirmed a tight clustering of MSN with several key proteins, including ITGA6, ITGB4, FERMT3, FLNA, LCP1, CORO1A, FMNL1, ARHGEF2, and MMP9, all of which modulate membrane ruffling, lamellipodia and filopodia formation, cell-ECM interaction, and ECM remodeling that play a crucial role in cancer cell invasion [48] (Figure S6). Moesin binds to phosphatidylinositol 4,5-bisphosphate (PI(4,5)P 2 ), CD44, and Na + /H + exchanger 1 (NHE-1), which are all key factors for cytoskeletal reorganization by modulating integrin signaling and integrin complex formation [16]. The complex that consists of integrin subunit α6 (ITGA6) and subunit β4 (ITGB4) and sequentially interacts with laminin [49] and kindlin-3 (FERMT3) is involved in tumorigenesis by modulating tumor cell-ECM interaction [50]. Filamin A (FLNA), plastin-2 (LCP1), coronin-1A (CORO1A), and formin-like-1 (FMNL1) also affect cytoskeletal dynamics by modulating actin filaments which eventually prompt cancer mobility and invasion [51,52]. Additional molecular studies need to be carried out, focusing on the above-selected proteomic markers. The results and how they can be interpreted in perspective of previous studies and working hypotheses should be discussed. The findings and their implications should be discussed in the broadest context possible. Future research directions may also be highlighted.

Patient Selection and Clinicopathologic Review
A total of 16 surgically confirmed LBC samples and an independent BUC cohort of 30 LBC specimens encompassing NIBUC, SIBUC, and MIBUC were employed for quantitative proteomic analysis and verification of diagnostic utility of ICC, respectively ( Table 2). This study was approved by the Institutional Review Board of Seoul National University Hospital (IRB No. H-1602-150-747). Detailed information can be found in the Supplementary Methods.     Figure S7 indicates the key steps in our approach for the proteomic discovery of novel biomarkers. Tumor cells from LBC slides were scraped and the peptide was digested using the filter-aided sample preparation (FASP) procedure as previously described [53]. Each sample was desalted [54] and was followed by LC-MS/MS analysis. For BUC cell lines, a proteomic analysis was performed after eight BUC cell lines, namely, T24, J82, 253J-BV, 253J, 5637, RT4, HT1376, and HT1197 (ATCC; Manassas, VA, USA), were categorized as IBUC_CL and NIBUC_CL based on their invasion and migration capacities. Each sample was labeled by TMT and was followed by LC-MS/MS analysis. The LC-MS/MS analysis was conducted using a Q Exactive Plus Hybrid Quadrupole-Orbitrap mass spectrometer (Thermo Fisher Scientific Inc., Waltham, MA, USA) and an Ultimate 3000 RSLC system (Dionex, Sunnyvale, CA, USA) as previously described [53,55]. MaxQuant version 1.5.3.1 (Max Planck Institute of Biochemistry, Munich, Germany) [56] with the Andromeda search engine [57] and Proteome Discoverer 2.1 software (Thermo Fisher Scientific Inc., Waltham, MA, USA) [58] with the SEQUEST-HT search engine were employed for processing LBC and cell line data, respectively. More detailed information is available in the Supplementary Methods.

Cell Migration and Invasion Assays with Small Interfering RNA (siRNA) Transfection
The T24 and J82 BUC cell lines were selected to evaluate cell migration and invasion abilities. RNA interference siRNAs targeting GRHL2, LLGL2, MSN, NCAM2, and VAPA were employed, followed by transfection to BUC cells. Detailed information is available in the Supplementary Methods.

Tumor Spheroids and 3D Spheroid Invasion Assay
Tumor spheroids were generated for suspension culture. Mixed collagen/Matrigel matrices were constructed as previously described [59]. The dissemination of spheroids was assessed under a phase-contrast microscope. A confocal laser scanning microscope (Leica TCS SP8; Leica microsystems, Wetzlar, Germany) was employed for the detection of stained F-actin. Phalloidin-rhodamine (Thermo Fisher Scientific Inc., Waltham, MA, USA; 1:100 in phosphate-buffered saline (PBS)) was used for visualization of the actin cytoskeleton in 3D spheroid cells. Supplementary Methods contain additional information.

Immunocytochemical Analysis
Immunocytochemical staining was conducted on LBC slides. Immunostaining of moesin was performed using Benchmark XT (Ventana Medical System, Inc., Tucson, AZ, USA). A monoclonal mouse anti-moesin antibody (Santa Cruz Biotechnology, Dallas, TX, USA) was diluted to 1:500. The binding of the primary antibody was identified using an Optiview universal DAB kit (Ventana Medical Systems, Inc., Tucson, AZ, USA) according to the manufacturer's protocol. ICC analysis defined negative expression for tumor cells with no moesin expression as opposed to positive expression when at least more than one tumor cell expressed moesin [17]. We also assessed the intensity and proportion of positive BUC cells for H-score evaluation [60].

Statistical Analyses
All proteomic datasets were submitted to the ProteomeXchange Consortium (http://proteomecentral. proteomechange.org) (project ID: PXD016437) [61]. ToppGene Suite resources (https://toppgene.cchmc. org/) [62] and String [63] were used for gene ontology annotation and interaction network model construction, respectively. Cytoscape version 3.7.1 (Institute for Systems Biology, Seattle, WA, USA) [64] was used for the illustration of the network model. Statistical analyses were conducted using the Perseus software (Max Planck Institute of Biochemistry, Munich, Germany) [65] for proteomic data. The H-score for ICC validation was analyzed by utilizing the Kruskal-Wallis test and Mann-Whitney U test for the comparison of BUC groups with the GraphPad Prism 8.0 program (GraphPad Software, Inc., CA, USA). The cross-tabulation analysis was conducted by Pearson's χ 2 test and Fisher exact test with IBM SPSS Statistics version 20 (IBM Corp., Armonk, NY, USA). Detailed information is available in Supplementary Methods.

Conclusions
Taking advantage of advanced proteomic techniques, the present study identified a novel promising diagnostic biomarker that can be applied as a new ancillary test for the prediction of BUC with invasion using voided urine LBC samples, the most frequently used diagnostic sample in routine practice. We successfully demonstrated that the immunoreactivity of moesin can be utilized as a diagnostic marker for early surgical intervention. Further investigation will be necessary for our future studies to validate the predictive ability of moesin in a larger cohort.  Table S1: Total list of identified protein groups in bladder urothelial carcinoma (BUC) (liquid-based cytology (LBC) samples), Table S2: Total list of differentially expressed proteins (DEPs) by ANOVA in bladder urothelial carcinoma (BUC) (liquid-based cytology (LBC) samples), Table S3: Total list of differentially expressed proteins (DEPs) by pairwise t-test in bladder urothelial carcinoma (BUC) (liquid-based cytology (LBC) samples), Table S4: Gene ontology (GO) analysis of differentially expressed proteins (DEPs) in bladder urothelial carcinoma (BUC) (liquid-based cytology (LBC) samples), Table S5: Gene ontology (GO) analysis of differentially expressed proteins (DEPs) with upregulation in bladder urothelial carcinoma (BUC), Table S6: Gene ontology (GO) analysis of differentially expressed proteins (DEPs) with downregulation in bladder urothelial carcinoma (BUC) (liquid-based cytology (LBC) samples), Table S7: Total list of identified protein groups in tandem mass tag (TMT)-based proteomics of bladder urothelial carcinoma (BUC) cell lines, Table S8: Total list of differentially expressed proteins (DEPs) in tandem mass tag (TMT)-based proteomics of bladder urothelial carcinoma (BUC) cell lines, Table  S9: Comparison of differentially expressed proteins (DEPs) between proteomics of liquid-based cytology (LBC) samples of bladder urothelial carcinoma (BUC) and tandem mass tag (TMT)-based proteomics of BUC cell lines, Table S10: Correlation between MSN RNA expression and AJCC staging and angiolymphatic invasion in bladder urothelial carcinoma (BUC) in The Cancer Genome Atlas (TCGA) public data.

Conflicts of Interest:
The authors declare no conflicts of interest.