Assessment of TSPAN Expression Profile and Their Role in the VSCC Prognosis

The role and prognostic value of tetraspanins (TSPANs) in vulvar squamous cell carcinoma (VSCC) remain poorly understood. We sought to primarily determine, at both the molecular and tissue level, the expression profile of the TSPANs CD9, CD63, CD81, and CD82 in archived VSCC samples (n = 117) and further investigate their clinical relevance as prognostic markers. Our studies led us to identify CD63 as the most highly expressed TSPAN, at the gene and protein levels. Multicomparison studies also revealed that the expression of CD9 was associated with tumor size, whereas CD63 upregulation was associated with histological diagnosis and vascular invasion. Moreover, low expression of CD81 and CD82 was associated with worse prognosis. To determine the role of TSPANs in VSCC at the cellular level, we assessed the mRNA levels of CD63 and CD82 in established metastatic (SW962) and non-metastatic (SW954) VSCC human cell lines. CD82 was found to be downregulated in SW962 cells, thus supporting its metastasis suppressor role. However, CD63 was significantly upregulated in both cell lines. Silencing of CD63 by siRNA led to a significant decrease in proliferation of both SW954 and SW962. Furthermore, in SW962 particularly, CD63-siRNA also remarkably inhibited cell migration. Altogether, our data suggest that the differential expression of TSPANs represents an important feature for prognosis of VSCC patients and indicates that CD63 and CD82 are likely potential therapeutic targets in VSCC.


Introduction
Vulvar cancer, albeit rare, represents about 4% of all gynecological malignancies. The American Cancer Society (ACS) estimated that, in 2020, more than 6000 new cases of vulvar cancer were diagnosed in the USA alone, with over 1300 patients dying from the disease due to cancer progression [1,2]. Among all types of vulvar malignancies, vulvar squamous cell carcinoma (VSCC) is the most frequent histological subtype (~90%) [1,3,4]. VSCC may During data analysis, we found that 13% of the patients (n = 10) out of 79 medical records available received neoadjuvant chemotherapy as part of their care (Table 1). To further evaluate the impact of neoadjuvant chemotherapy on clinicopathologic variables, we performed a series of univariate and multivariate analyses (Table S1). For instance, our analyses indicate that, within the patients receiving neoadjuvant therapy, 5 patients were reported having perineal commitment against 11 patients in the non-treated arm (p = 0.015). Still within the non-treated group (n = 68), 75% of them showed tumors larger than 6 mm 3 (p = 0.014), with about 86% of those (n = 66) presenting deeper lesions when compared to neoadjuvant treated patients (p = 0.050). Additionally, nearly 61% of patients (n = 40) in the non-treated group had dermis commitment compared to other lesion types within the same group. The frequency of tumor relapse was also higher in patients that did not receive neoadjuvant chemotherapy (78%) as compared to treated individuals (22%) (p = 0.010). Surprisingly, 100% of non-treated patients showed infection by high-risk HPV compared to patients receiving chemotherapy (p = 0.021).
Furthermore, we found that 60.8% of patients (n = 31) within the 51 VSCC cases with clinical annotated status on HPV-infection were positive for the virus (Table 1). Considering HPV positive patients, 80.6% of them (n = 25) exhibited high-risk HPV and 9.7% (n = 3) showed a low-risk type of HPV. Considering only patients with high-risk HPV, 28% (7 out of 25) showed coinfection with two or three virus isotypes. Additionally, 9.7% (3 out of 31) of HPV positive samples had coinfection with both high and low-risk types (Table S2). Additionally, the HPV 16 type was found predominantly in 58% of HPV-infected patients (18 out of 31), followed by the HPV 33 type, which affected 23% of patients, as shown in Table S3.

TSPANs Expression in Patients' Samples
To determine the expression profile of TSPANs in VSCC at the molecular level, we first performed a quantitative real-time PCR (qRT-PCR) analysis for CD9, CD63, CD81, and CD82 in 16 archived frozen tumors and matched normal adjacent tissues randomly selected from our 117 patients. As depicted in Figure 1a, our analyses revealed that the overall relative expression (mean ± SD) of CD9 was significantly down-regulated by nearly 50% in tumor samples compared to normal tissues (p = 0.0006). In contrast, VSCC samples had a significant increase of CD63 mRNA levels by~1.0-fold (p < 0.005), as compared to the control. Notably, no statistical differences were observed when CD81 and CD82 were compared to normal tissues. Upon further investigation of the expression frequency of each TSPAN in our VSCC cohort, we found that CD63 was upregulated in 75% (n = 12) of the samples (fold regulation (FR) >1), as compared to its counterparts (CD81 at 25%, CD82 at 20%, and CD9 at 6%, respectively; Figure 1b).
Next, to expand on our initial findings and determine whether the protein expression ( Figure 1c) and frequency ( Figure 1d and Table 2) of each TSPAN were consistent with the gene expression results (Figure 1a,b), a series of immunohistochemical analyses were performed on pre-determined FPPE samples (n = 117, Table 1). To ensure reproducibility and consistency in data acquisition and semi-quantitative protein analysis, tonsils were used as a positive control reference (Figures S1 and S2). At the tissue level, CD9 was found to be predominantly localized at the cell membrane, whereas CD63, CD81, and CD82 were primarily found in the cytoplasm of the tumor samples ( Figure 2). Our analysis also showed a heterogeneous staining pattern of all four TSPANs, which were classified from a negative to strong expression, regardless of their cell compartment localization, according to the H-score method (0-12). Considering only samples with a H-score result ≥3 (moderated and strong staining-Figures 1c and 2), we found that CD81 was poorly expressed in VSCC samples, with 20% of the tissue showing a dull staining pattern (H-score of~1.5). CD82 and CD9 were both found to be moderately expressed in most samples with a frequency of 51% and 41%, respectively. Lastly, and similar to the gene expression results, CD63 was found to be more highly expressed (H-score ≥ 4, Figures 1d and 2) and the most frequent (positive in 83% of the samples, Figure 1d and Table 2) TSPAN in all VSCC samples evaluated. Primary antibodies specificity and sensitivity can be seen in Supplementary Figure S3     Statistically, the comparison analysis within the protein expression of each TSPAN showed significant differences between CD63 and CD9 (p < 0.01), as well as between CD9 and CD81 (p < 0.0001, Figure 1c). Robust statistical differences were also found among the expressions of CD63, CD81, and CD82 (p < 0.0001) and between CD81 and CD82 (p < 0.001). Moreover, co-expression analysis among TSPANs showed a moderate positive correlation between CD82 and CD9 (r = 0.597, p < 0.0001) and four less robust, yet positive correlations between CD9/CD63 (r = 0.325, p = 0.001), CD9/CD81 (r = 0.423, p < 0.0001), CD63/CD82 (r = 0.422, p < 0.0001), and CD81/CD82 (r = 0.356, p < 0.0001) ( Table 3), thus suggesting that TSPANs may interact with each other in VSCC. Statistically, the comparison analysis within the protein expression of each TSPAN showed significant differences between CD63 and CD9 (p < 0.01), as well as between CD9 and CD81 (p < 0.0001, Figure 1c). Robust statistical differences were also found among the expressions of CD63, CD81, and CD82 (p < 0.0001) and between CD81 and CD82 (p < 0.001). Moreover, co-expression analysis among TSPANs showed a moderate positive correlation between CD82 and CD9 (r = 0.597, p < 0.0001) and four less robust, yet positive correlations between CD9/CD63 (r = 0.325, p = 0.001), CD9/CD81 (r = 0.423, p < 0.0001), CD63/CD82 (r = 0.422, p < 0.0001), and CD81/CD82 (r = 0.356, p < 0.0001) ( Table 3), thus suggesting that TSPANs may interact with each other in VSCC.

Clinical Implication of TSPANs Expression for VSCC Patients
Based on the unique molecular and tissue signature of CD9, CD63, CD81, and CD82, we sought to investigate whether their expression could be clinically useful as markers for diagnosis and/or prognosis or used to predict patterns of treatment response and outcome. Therefore, their expression profiles were evaluated in tandem with the patients' clinical features and available follow-up medical information.

Clinical Implication of TSPANs Expression for VSCC Patients
Based on the unique molecular and tissue signature of CD9, CD63, CD81, and CD82, we sought to investigate whether their expression could be clinically useful as markers for diagnosis and/or prognosis or used to predict patterns of treatment response and outcome. Therefore, their expression profiles were evaluated in tandem with the patients' clinical features and available follow-up medical information.
CD81 down-expression was mainly associated with tumor location, treatment, and death. Specifically, 73% of the cases with labia majora commitment were attributed to patients with low CD81 expression, compared to 27% of similar cases in CD81 positive samples (p = 0.029). In total, 85% of patients with CD81 negative tumors did not undergo neoadjuvant chemotherapy or radiotherapy (p = 0.023), and low expression of CD81 was associated with increasing mortality as compared to CD81 positive patients (p = 0.039, Table S7).
Lastly, we found that the majority of large tumors (>6 mm 3 ) were negative for CD82, while in contrast, CD82 positive tumors represented about 61% of smallest tumor samples (<5 mm 3 , p = 0.027). A large number of CD82 positive samples was also found to be associated with patients who underwent adjuvant therapy (p = 0.004). Moreover, 60% of the CD82 positive tumors show HPV coinfection (p = 0.051, Table S8).
No other clinicopathological feature showed relevant statistical significance associated with the expression of TSPANs at the gene or protein levels. A complete list of the variables analyzed is shown in Tables S2-S6.

Role of TSPANs in Metastatic and Non-Metastatic Cell Lines
To mechanistically investigate the role of TSPANs in VSCC, we selected the tetraspanins CD63 and CD82 to study, as they are well characterized as having opposite roles in tumor development and prognosis in the literature.
First, we assessed the basal expression profiles of CD63 and CD82 by qRT-PCR in both SW954 (non-metastatic) and SW962 (metastatic) human cell lines. Figure 4a shows that the expression of CD63 was 4.5 times higher in SW962 compared to SW954 cells (p < 0.05), whereas the expression of CD82 was decreased by nearly 2 times in the metastatic cells compared to SW954 (p < 0.05).
Next, we transiently silenced CD63 expression by transfecting both cell lines with a small interfering RNA (siRNA). By qRT-PCR, we found that SW954 and SW962 cells transfected with CD63-siRNA had a significant decrease of CD63 mRNA by 70% and 90% (p < 0.05), respectively, as compared to the non-transfected cells (CC) and siRNA control (Ct siRNA) (Figure 4b,c). Consistently, at the protein level, the intensity of CD63 staining in immunocytochemistry (ICC) was found to be reduced by 80% and 90% in SW954 and SW962 CD63-siRNA transfected cells, respectively, as compared to the controls (Figure 4d). Further semi-quantitative analysis in both silenced cell lines showed a significant statistical difference in downregulation of CD63 expression (using the score of intensity and frequency of stained cells) in both cell lines when compared to the controls (p = 0.0005, Table 5).  Next, we transiently silenced CD63 expression by transfecting both cell lines with a small interfering RNA (siRNA). By qRT-PCR, we found that SW954 and SW962 cells transfected with CD63-siRNA had a significant decrease of CD63 mRNA by 70% and 90% (p < 0.05), respectively, as compared to the non-transfected cells (CC) and siRNA control (Ct siRNA) (Figure 4b,c). Consistently, at the protein level, the intensity of CD63 staining in immunocytochemistry (ICC) was found to be reduced by 80% and 90% in SW954 and SW962 CD63-siRNA transfected cells, respectively, as compared to the controls (Figure At the cellular level, silencing of CD63 in both cell lines led to a significant reduction in cell viability (proliferation assay) as early as 24 h and up to 96 h after CD63-siRNA transfection. Based on the siRNA efficiency data, we assessed the CD63 silencing effects 48 h after transfection. Significant differences in the cell proliferation were observed for CD63-siRNA SW954 cells after 72 h (~50% of inhibition, p < 0.05) and 96 h (~70% of inhibition, p < 0.005) (Figure 5a), while CD63-siRNA SW962 cells showed higher inhibition rates after 96 h (~20%, p < 0.05) (Figure 5a). CD63-dependent cell migration was also investigated in both siRNA transfected and controls cells using the wound healing (scratch) assay 48 h after CD63 silencing. As depicted in Figure 5b, the ability migration of SW954 cells was compromised after CD63 silencing. However, a significant effect was observed 6 h after the scratch (CC = 40%, siRNA C = 55%, CD63-siRNA = 20%; p < 0.05, Figure 5b), and after 12 h, the effect was sustained but without statistical significance. Similarly, a robust and pronounced migration inhibition was also observed in CD63-silenced SW962 cells 24 h after scratch (CC = 50%, siRNA C = 50%, CD63-siRNA = 25%; p < 0.05, Figure 5c), without wound closure even after 48 h (CC = 90%, siRNA C = 90%, CD63-siRNA = 50%; p > 0.05). The migration analysis of the cells has continued until total wound healing in the control cells group. Based on our findings, we propose, as schematically represented in Figure 6, that the differential expression of TSPANs, in particular CD9, CD63, CD81, and CD82, might contribute to VSCC development and poor prognosis. Our results indicate that these alterations, along with high risk-HPV infections, are associated mainly with larger tumor size, Based on our findings, we propose, as schematically represented in Figure 6, that the differential expression of TSPANs, in particular CD9, CD63, CD81, and CD82, might contribute to VSCC development and poor prognosis. Our results indicate that these alterations, along with high risk-HPV infections, are associated mainly with larger tumor size, lymph node metastasis, and mortality. In vitro, we showed that the expression of CD63 is associated with increased cell survival and migration in the VSCC cells, highlighting the potential clinical benefit of targeting CD63 as a novel therapeutic modality for patients with VSCC.

Discussion
In the present study, 117 FFPE archived samples of VSCC (from 1978 to 2009) and associated medical records were selected from a larger cohort of 300 samples previously and exhaustively characterized by several studies [14,[44][45][46][47][48][49]. A review study published by Maia et al. in 2012 highlighted the uniqueness of the Brazilian's VSCC registry and concluded that the main prognostic factors for VSCC are mainly the patient's age, treatment, recurrence, and survival, corroborated with the global literature at the time. Nearly a decade later, the main prognostic factors for VSCC remain virtually unchanged and features such as the patient's age, HPV/16 status, VINs occurrence, tumor stage and grade, tumor size, depth of invasion, lymph node metastasis, stromal change, and margin status continue to be relevant [1,7,9,10,16,51,52]. As such, our analysis revealed that the patient's age (~70 years old), therapy status (~80% of patients did not receive any therapy), VSCC stage (~87% of the tumors were VSCC I/II), FIGO stage (~25% of patients were stage IIIB), tumor recurrence (~50% of patients had recurrence), LN metastasis (~40% of patients did present LN involvement), and mortality rate (~54% of the patients died) were adequate prognostic features for selection of the study population and further correlative studies.
Although prognostic factors, such as those discussed above, are clinically important for patient management and decision making, the lack of associated molecular markers to further predict patient outcome (e, OS, DFS, and objective response rate) or to anticipate VSCC progression remain an unmet clinical need. It is well stablished in the literature that alterations in TSPAN expression are directly correlated with prognosis in many cancer types [1]. However, in VSCC, their expression profile and molecular role remain poorly elucidated. Here, we found that, among all four TSPANs analyzed, CD63 was predominantly expressed and more frequent in VSCC samples at the molecular (qRT-PCR, 75%) and tissue level (IHC, 83%) than CD9, CD81, and CD82. Curiously, no significant differences at OS were found when the expression of CD63 or the other three TSPANs were evaluated in conjunction with the patient's clinical record.
CD9 is ubiquitously expressed in various normal and cancer tissues [53,54], and consequently, its expression as a molecular marker is often ambiguous and cancerdependent [36,[55][56][57][58][59][60][61][62]. Higher positive expression of CD9 has served as a good prognostic marker in breast, lung, colon, and pancreatic carcinomas [55,63,64], as it is thought to have a metastatic suppressor role by inhibiting tumor cell motility, at least in vitro [65,66]. However, in a previous study of tumor progression in penile squamous cell carcinomas [67], we observed that invasive tumors have a heterogeneous pattern of CD9 expression, ranging from strong expression in 80% of the cells to a marked loss of CD9 expression in the same case. Similarly, Hori et al. (2004) and Soyuer et al. (2010) described that higher CD9 expression was associated with more aggressive tumors and increased number of lymph node metastases in gastric cancers, therefore representing a marker of poor prognosis [60,61]. In VSCC, although a direct correlation between CD9 expression and the potential of lymph node metastasis and survival was not statistically significant, we anticipate that patients that are CD9-low or CD9-negative will, more likely, bear tumors that are phenotypically more aggressive and, therefore, have a worse prognosis.
Likewise, the low expression of CD82 in the tissue samples analyzed was associated with larger tumors, as well as increased stromal invasion. Such findings are consistent with previous studies, where downregulation of CD82 expression was associated with both advanced cancer stages (e.g., in breast, bladder, ovary, and prostate carcinomas and also in oral/esophageal/laryngeal/penile squamous cell carcinomas) and the transition to a metastatic phenotype [37,59,[68][69][70]. For example, Yang et al. (2001) reported that the overexpression of CD82 in breast cancer cells resulted in the suppression of in vitro invasion and in vivo metastasis, suggesting that CD82 does function as a tumor suppressor [71]. While the tumor suppressing ability of CD82 was also shown in bladder, oral cancer, prostate, and ovary cancers, where cancer cell migration and invasion were drastically reduced upon its upregulation [72][73][74][75][76], in penile cancer, loss of CD82 led to an increase in tumor metastasis and higher HPV16 infection rates [24]. In this context, our analysis did reveal that a large majority of CD82 negative tumors (61% of 30 patients) were associated with HPV16, thus further indicating that the lack of CD82 expression may contribute to a poor prognosis in vulvar carcinomas. The etiological role of HPV in the development of vulvar cancer has been well recognized [77], and further elucidating the role of TSPANs in this process may be a key point for better understanding VSCC development.
On the other hand, both CD63 and CD81 were found to be overexpressed among the VSCC samples analyzed. Our results showed that the strong CD63 tissue staining was particularly associated with well-differentiated VSCC1 + VSCC2 samples, while lost or downregulated in more aggressive tumors. A similar correlation between high expression of CD63 and larger tumor size, as well as increased vascular and stromal invasion, and higher mortality was observed in our analysis. We found that these results are well aligned with recent publications, suggesting that increased expression of CD63 in certain cancer is correlated with increased activity of pro-tumorigenic cell signaling pathways, such as activation of β-catenin, phosphoinositide 3-kinases (PI3K), and extracellular signalregulated kinases (ERK) [5]. Furthermore, its increased expression was also reported as an indicator of poor disease specific survival (DSS) in solid tumors [3,68], thus corroborating our clinical findings and suggesting that CD63 functions as an oncogene in VSCC. Indeed, our in vitro analyses showed that silencing of CD63 reduced the proliferation rate of both VSCC cell lines and the migration capacity of the metastatic cells.
Lastly, the expression of CD81 in VSCC samples was robustly correlated with tumor relapse, especially in patients who underwent neoadjuvant therapy. In fact, in the latter, we found a staggering tumor relapse median that was three times higher than untreated patients. Such observation can be attributed to the fact the that CD81 has been shown to regulate cell migration and invasion and, therefore, is implicated in cancer progression and chemoresistance. As such, its overexpression or downregulation has served as a prognostic marker in several tumor types, including breast, lung, prostate, melanoma, brain cancer, and lymphoma [78][79][80]. In our analysis, despite the low frequency of CD81 positive cases, the results indicate that CD81 contributes to a worse DFS and unfavorable prognosis, further corroborating with the literature [81,82].
Although more comprehensive analyses are necessary to better understand the role of TSPAN in VSCC, our data indicate that these molecules are important benchmarks of VSCC development and progression. Furthermore, to the best of our knowledge, this was the first study of its kind shedding some light onto the relationship between TSPANs and VSCC, and as such, we believe this work will pave the way for the development of novel treatment modalities, as well as new perspectives for disease management options.

Patient Samples
The present study was approved by the Research Ethics Committee of the Faculdade de Medicina da Universidade de Sao Paulo-FMUSP (Number 1.540.225) and complied with the Helsinki Declaration. Selected samples included 117 formalin-fixed paraffin-embedded (FFPE) tissues with invasive VSCC diagnoses, which were collected between 1979 and 2009 at the Anatomic Pathology Department of the AC Camargo Cancer Center in Sao Paulo, Brazil [15]. A total of 16 frozen tissues were further selected from the 117 cohort, along with 16 donor-matching morphologically normal tissue edges, for transcriptional analysis of the TSPANs.

Cell Lines, Culture Conditions, and Authentication
Non-metastatic-SW954 (American Type Culture Collection [ATCC] HTB-117) and metastatic-SW962 (ATCC HTB-118) VSCC human cell lines were grown in Roswell Park Memorial Institute-1640 (RPMI; Vitrocell, Embriolife, Sao Paulo, SP, Brazil), medium supplemented with 2 g/L of sodium bicarbonate (Sigma, Darmstadt, HE, Germany), 10% (v/v) fetal bovine serum (FBS; Invitrogen, Carlsbad, CA, USA), and antibiotic mix (ampicillin/streptomycin/amphotericin; Gibco, Waltham, MA, USA). Cells were cultured in 75 cm 2 flasks and maintained at 37 • C in a humidified atmosphere containing 5% CO 2 . Cell line authentication was performed using GenePrint 10 System (Promega, Madison, WI, USA), followed by identification and matching of the short tandem repeat profiles to cell line profiles at the ATCC data bank [83]. Total RNA was extracted from frozen tissues using the RNeasy Mini Kit (Qiagen, Hilden, Germany), following the manufacturer's recommendations. RNA isolation from cell lines (SW954 and SW962) was performed with Trizol (Thermo Fisher Scientific, Carlsbad, CA, USA), according to the manufacturer's instructions. For studies using TSPANsilenced cells, 5 × 10 6 SW954 and SW962 cells were seeded in 4-well plates (NUNC™ Delta Treated, Thermo Fisher Scientific, Carlsbad, CA, USA) and transfected with specific TSPANs, targeting siRNA or siRNA control (described in detail below). Total RNA from silenced cells was isolated using the PureLink RNA Mini Kit (Ambion, Austin, TX, USA), following the manufacturer's protocol. All RNA samples were quantified with spectrophotometer NanoDrop 2000 (Thermo Fisher Scientific, Carlsbad, CA, USA) prior to downstream applications. cDNA synthesis (reverse transcription) was carried out using the High-Capacity cDNA Reverse Transcription Kit (Thermo Fisher Scientific, Carlsbad, CA, USA). To evaluate the expression of each TSPAN, TaqMan assays for CD9 (assay ID: Hs00233521_m1), CD63 (assay ID: Hs00156390_m1), CD81 (assay ID: Hs01002167_m1), and CD82 (assay ID: Hs00356310_m1) were used, following the manufacturer's recommendations. B2M (TaqMan assay ID: Hs99999907_m1) and HRPT (TaqMan assay ID: Hs99999909_m1) were predicted as the stable normalizer by the NormFinder algorithm software [84] and used in our studies as housekeeping (HK) genes for data normalization. qRT-PCR amplifications were performed in an ABI 7500 Real-Time PCR instrument (Thermo Fisher Scientific, Carlsbad, CA, USA). As a reference sample, a pool of RNA isolated from tumor-free edge tissues (n = 16) was used. Amplifications were run in duplicate, and averages were obtained after normalization with either B2M (for cells) or HPRT (for tissues). Relative quantification was set up comparing cycle threshold by ∆∆CT method in SDS 3.0 and RQ manager LV2.0.6 software, and the fold regulation (FR) expression cut-off values were established in ≥+2 and ≤−2 to identify differentially expressed genes.

DNA Isolation and HPV Genotyping
Total DNA was extracted from up to eight tissue sections (10-µm-thick). Sections were pretreated with 350 µL ATL lysis buffer from the DNA FFPE kit (Qiagen, Valencia, CA, USA) and added directly to the paraffin sections, followed by incubation of the tightly closed tubes at 120 • C for 20 min. All procedures followed the supplier's specification. DNA was then quantified at the NanoDrop ND-1000 spectrophotometer (Wilmington, DE, USA) and analyzed on 1% agarose gel.

CD63 Silencing-siRNA Transfection
Silencing of CD63 was performed using MISSION siRNA CD63 (EHU032781-Sigma-Aldrich, MO, USA). Cells were cultured in Opti-MEM I Reduced Serum Medium (Thermo Fisher Scientific, Carlsbad, CA, USA), and the siRNA transfection was carried out with Lipofectamine RNAiMAX Reagent (Invitrogen, CA, USA) without antibiotic, according to manufacturer's instructions. The siRNA eGFP (EHUEGFP-Sigma-Aldrich, Saint-Louis, MO, USA) was used as a negative control. Cell seeding density and the concentration of each reagent were standardized for each functional assay, according to the manufacturer's recommendations.

Immunocytochemistry (ICC)
Prior to ICC studies, the specificity of the anti-CD63 antibody (MA1-19281, Thermo Fisher Scientific, Carlsbad, CA, USA, Figure S3) was validated and confirmed by western blot analysis ( Figure S3). Because western blot analysis was not feasible due to the small number of cells recovered after siRNA transfection, ICC was the method of choice to assess the expression of CD63. All reactions were carried out in 4-well plates (NUNC™ Delta Treated, Thermo Fisher Scientific, GE, USA), with 10 5 cells/well. Plates were washed with phosphate buffered saline (PBS), and 500 µL of 3.7% formaldehyde (Merck, Darmstadt, HE, Germany) was added for cell fixation. Hydrogen peroxide (H 2 O 2 ; VIC pharma, Sao Paulo, BR) was applied to block the endogenous peroxidase and nonspecific proteins. Next, cells were incubated with PBS and 5% bovine serum albumin (BSA) for 20 min at room temperature. For permeabilization, 0.1% Triton X 100 (Merck, Darmstadt, HE, Germany) in PBS was added for 20 min. Anti-CD63 antibody was diluted 1:100 in PBS supplemented with 0.1% Triton X 100 and incubated at room temperature for 1 h. Cells were submitted to three washing steps and incubated with biotinylated secondary antibody for 20 min at room temperature. After several washes, polymer HRP-peroxidase was added and extensively washed. Samples were stained with DAB (1:50, Merck, Darmstadt, HE, Germany) and counterstained with hematoxylin for 10 min at room temperature. Microphotographs and analysis were carried out with optical microscope Nikon Fase Contrast 0.90 Dry-Eclipse Ni (Nikon, Tokyo, Japan), equipped with a photo camera Nikon DS-Ri1 (Nikon, Tokyo, Japan).

Wound Healing Assay
SW954 (2 × 10 5 /well) and SW962 (1 × 10 5 /well) cells were seeded in 24-well plates and allowed to reach complete confluence in growth media containing no antibiotics. To make the wound, a plastic P200 pipette tip (Fisher) was used to scratch the cell monolayer to create a cleared area. The wounded cell layer was washed once with 1X PBS to remove loose cells and then transfected with CD63-siRNA and refed with fresh growth media containing 1% FBS. The wounds were observed using phase contrast microscopy on an inverted microscope Zeiss (AxioCam ERc 5c, Wetzlar, HE, Germany). Images were taken at regular intervals over the course of 0 to 48 h and then analyzed by ImageJ (v1.50i, Bethesda, MD, USA).

Statistical Analysis
Distribution of continuous data was analyzed by the Shapiro-Wilk normality test. For comparison between TSPAN expression profiles in SW954 and SW962 cells, the Mann-Whitney U test was used. The Kruskal-Wallis with Dunn's post-test was used to compare the expression levels of different TSPANs. To evaluate the statistical differences among frequencies of each variable, the Chi-square test or Fisher's exact test was used. The mean and SD (mean ± SD) were used to represent the patient's age, as previously described [87]. The Kaplan-Meier method evaluated patient outcome, and the differences between each group were analyzed using the log-rank test (Mantel-Cox) and multivariate analysis with Cox's regression test by the Cox proportional hazard model. Survival rates were calculated based on months. OS time was set up between the date of the patient's surgery and either death date or the last information date/follow-up. DFS was determined from the date of surgery to the date of relapse or the last follow-up. Spearman's rank correlation coefficient was used to measure the strength of the linear relationship between two non-parametric variables, as described elsewhere [87,88]. Statistical analysis was performed using the SPSS (v21, IBM Corp., Armonk, NY, USA) and Graph Pad Prism 5.0 (v3, GraphPad Software Inc., San Diego, CA, USA) software [87]. All differences were considered statistically significant when p < 0.05.

Conclusions
We saw that the imbalance in the TSPAN expression is associated with poor prognosis in VSCC patients. CD63, the most frequent protein found in the samples, seems to have an important and possibly stage-specific role in those tumors. Additionally, CD82 and CD63 might be potential targets for therapy in VSCC.

Supplementary Materials:
The following are available online at https://www.mdpi.com/article/ 10.3390/ijms22095015/s1. Figure S1: Tetraspanins protein expression in patient tissues compared to normal tonsil (positive control); Figure S2: Staining pattern for CD9, CD63, CD81, and CD82 in positive control tissues (Tonsils); Figure S3: Specificity of anti-TSPANs by western blot, (a) Detection of CD9, CD63, CD81, CD82, and Actin-B (loading control; C) in total cell lysate from the human breast cancer cell line, MCF7, (b) reactivity of anti-CD63 in total cell lysate from VSCC non-metastatic (SW954) and metastatic (SW962) cell lines. Protein sizes are indicated; Table S1: Comparison between clinical and pathological features from patients, considering neoadjuvant chemotherapy treatment; Table S2: Comparison of HPV types and co-infection in patients with VSCC; Table S3: Prevalence of HPV, considering patients with and without coinfection (n = 31); Table S4: VSCC clinicopathological  features and CD9 protein expression; Table S5: VSCC clinicopathological features and CD63 protein expression using two categories (negative and positive expression); Table S6: Comparison between clinical and pathological features and CD63 protein expression using three categories (negative, moderate, and strong expression); Table S7: VSCC clinicopathological features and CD81 protein expression; Table S8: VSCC clinicopathological features and CD82 protein expression.
Author Contributions: K.P.F., B.C.d.A., and L.G.d.A., experiments, data analysis, and manuscript preparation; G.B., patient s information data bank construction; F.A.S. and R.M.R., cases selection and review; E.C.B. and A.S.D., manuscript review, intellectual and technical support; K.C.C., idea conception, experiments, and data and manuscript preparation. All authors have read and agreed to the published version of the manuscript. Informed Consent Statement: Informed consent was obtained from all subjects alive and in the follow up involved in the study. For those who died, the waiver of the informed consent was requested. Written informed consent was obtained from the patient(s) to publish this paper.
Data Availability Statement: All data generated or analyzed during this study are included in this article or Supplementary Material.