EphA2 on urinary extracellular vesicles as a novel biomarker for bladder cancer diagnosis and its effect on the invasiveness of bladder cancer

Abstract

Background

Urinary extracellular vesicles (uEVs) secreted from bladder cancer contain cancer-specific proteins that are potential diagnostic biomarkers. We identified and evaluated a uEV-based protein biomarker for bladder cancer diagnosis and analysed its functions.

Methods

Biomarker candidates, selected by shotgun proteomics, were validated using targeted proteomics of uEVs obtained from 49 patients with and 48 individuals without bladder cancer, including patients with non-malignant haematuria. We developed an enzyme-linked immunosorbent assay (ELISA) for quantifying the uEV protein biomarker without ultracentrifugation and evaluated urine samples from 36 patients with and 36 patients without bladder cancer.

Results

Thirteen membrane proteins were significantly upregulated in the uEVs from patients with bladder cancer in shotgun proteomics. Among them, eight proteins were validated by target proteomics, and Ephrin type-A receptor 2 (EphA2) was the only protein significantly upregulated in the uEVs of patients with bladder cancer, compared with that of patients with non-malignant haematuria. The EV-EphA2-CD9 ELISA demonstrated good diagnostic performance (sensitivity: 61.1%, specificity: 97.2%). We showed that EphA2 promotes proliferation, invasion and migration and EV-EphA2 promotes the invasion and migration of bladder cancer cells.

Conclusions

We established EV-EphA2-CD9 ELISA for uEV-EphA2 detection for the non-invasive early clinical diagnosis of bladder cancer.

Introduction

Bladder cancer is one of the most common cancers and significant causes of cancer-related deaths worldwide [1]. Patients with early-stage bladder cancer can undergo curative treatment such as transurethral resection. However, muscle-invasive bladder cancer (MIBCa) is clinically destructive, progresses and metastasises rapidly, and is usually lethal [2]. Therefore, early detection of bladder cancer is important for improving prognosis.

Cystoscopy offers high diagnostic accuracy but is unsuitable as a screening method owing to its high cost and invasiveness. The current standard of screening for bladder cancer is a urine cytology, but it exhibits poor sensitivity for low-grade or early-stage bladder cancer, and the accuracy of interpretation depends on the expertise of the pathologist [3]. Therefore, urine cytology is not an efficient screening test.

To date, several urinary biomarkers have been approved by the US Food and Drug Administration but are not widely adopted owing to low specificity or high data heterogeneity [4, 5]. Thus, non-invasive and more accurate biomarkers are needed for bladder cancer.

Extracellular vesicles (EVs), including exosomes and microvesicles, are lipid bilayer vesicles secreted by many cell types into bodily fluids. EVs play an important role in intercellular communication and harbour various bioactive molecules, including nucleic acids, proteins and lipids, characteristic of the host cells [6]. In particular, cancer-derived EVs reportedly contain cancer-specific components capable of promoting cancer progression [7]. Therefore, there is increasing interest in cancer-derived EVs as a promising source of diagnostic biomarkers and therapeutic targets in oncology [8, 9]. Given that urine comes in direct contact with the urogenital system, it is expected to contain abundant EVs from bladder cancer tissues [10, 11]. We previously demonstrated that urine is a suitable biofluid for detecting bladder cancer-derived EV proteins [12]; thus, proteomic analysis of urinary EVs (uEVs) is a powerful approach to determine potential bladder cancer biomarkers [13–16].

The aim of this study was to identify a uEV-based protein biomarker for the diagnosis of bladder cancer and develop an enzyme-linked immunosorbent assay (ELISA) for analysing the biomarker protein on uEVs. In addition, we evaluated whether the protein identified on the EVs exerts a cancer-promoting effect on bladder cancer.

Materials and methods

Patients and biological sample collection

In the discovery phase for tandem mass tag-labelling liquid chromatography-tandem mass spectrometry (LC-MS/MS) analysis, voided urine samples (>50 mL each) were collected from seven patients with bladder cancer [non-muscle-invasive bladder cancer (NMIBCa)], n = 3; MIBCa, n = 4] and four healthy individuals, as previously reported [12].

In the validation phase for selected reaction monitoring/multiple reaction monitoring (SRM/MRM) analysis, urine samples were collected from 49 patients with bladder cancer (NMIBCa, n = 24; MIBCa, n = 25) and 48 individuals without bladder cancer (healthy controls, n = 36; patients with non-malignant haematuria, including cystitis, n = 12) at Osaka University Hospital, Osaka Police Hospital and Osaka General Medical Centre, Japan.

For ELISA, we collected urine samples from 36 patients with bladder cancer (NMIBCa, n = 28; MIBCa, n = 8) and 36 individuals without bladder cancer (healthy controls, n = 26; patients with non-malignant haematuria, n = 10), at Osaka University Hospital. Clinical and pathological information of patients used in shotgun proteomics (tandem mass tag-labelling LC-MS/MS), target proteomics (SRM/MRM) and ELISA are shown in Table 1.

Table 1.

Clinical and pathological information of patients in proteomics and ELISA.

	Shotgun proteomics		Target proteomics			EV-EphA2-CD9 ELISA
	(n = 11)		(n = 97)			(n = 72)
	Non-bladder cancer	Bladder cancer	Non-bladder cancer		Bladder cancer	Non-bladder cancer		Bladder cancer
	HC	(n = 7)	HC	Haematuria	(n = 49)	HC	Haematuria	(n = 36)
	(n = 4)		(n = 36)	(n = 12)		(n = 26)	(n = 10)
Age (y), median (range)	67.5 (42–79)	73 (66–83)	57 (41–73)	57 (20–82)	71 (31–86)	61.0 (34–81)	67.5 (45–85)	73.5 (59–93)
Sex, n (%)
Male	3 (75.0)	6 (85.7)	22 (61.1)	5 (41.7)	32 (70.0)	11 (42.3)	6 (60.0)	25 (69.4)
Female	1 (25.0)	1 (14.3)	14 (38.9)	7 (58.3)	17 (30.0)	15 (57.7)	4 (40.0)	11 (30.6)
Urine cytology, n (%)
Negative	4 (100)	2 (28.6)	36 (100)	12 (100)	24 (49.0)	26 (100)	10 (100)	18 (50.0)
Positive	0 (0.0)	5 (71.4)	0 (0.0)	0 (0.0)	24 (49.0)	0 (0.0)	0 (0.0)	18 (50.0)
Unknown	0 (0.0)	0 (0.0)	0 (0.0)	0 (0.0)	1 (2.0)	0 (0.0)	0 (0.0)	0 (0.0)
Pathological T stage n, (%)
Ta	—	3 (42.9)	—		24 (49.0)	—		22 (61.1)
Tis	—	0 (0.0)	—		0 (0.0)	—		3 (8.3)
T1	—	0 (0.0)	—		0 (0.0)	—		3 (8.3)
pT2≤	—	4 (57.1)	—		25 (51.0)	—		8 (22.2)
Pathological grade, n (%)
Low grade	—	2 (28.6)	—		14 (28.6)	—		18 (50.0)
High grade	—	5 (7.14)	—		35 (71.4)	—		18 (50.0)

ELISA enzyme-linked immunosorbent assay, HC healthy controls.

Approval was obtained from the Institutional Review Board (Osaka university hospital Institutional Review Board, Protocol Number: 13397-11) before initiating the study, and all patients provided written informed consent. The study was performed in accordance with the Declaration of Helsinki.

All patients with bladder cancer were histologically diagnosed by experienced senior pathologists based on the observation of standard haematoxylin and eosin-stained sections. Tumours were staged according to the 7th AJCC TNM staging system and graded according to the 2016 World Health Organization criteria. Urine cytology was also evaluated by specialists according to the Paris system; positive urine cytology is categorised as Suspicious for High-Grade Urothelial Carcinoma, Low-Grade Urothelial Carcinoma, and High-Grade Urothelial Neoplasia.

The collected urine samples were stored at 4 °C for up to 6 h before processing, followed by centrifugation at 2000 × g for 30 min. The supernatants were separated and stored at −80 °C until further processing.

Cell lines and cell culture

T24 cells (lot no. 02052018, Apr. 2019) were purchased from the Japanese Collection of Research Bioresources Cell Bank (Osaka, Japan); J82 cells (lot no. 58307736, Oct. 2011), UM-UC-3 cells (lot no. 61729357, Aug. 2014), TCC-SUP cells (lot no. 57938022, Feb. 2012), RT4 cells (lot no. 58078661, Jan. 2010), SV-HUC-1 cells (lot no. 3836614, Sept. 2007) and HEK293 cells (lot no. 08202010, Sept. 2014) were purchased from the American Type Culture Collection (VA, USA); U-BLC1 cells (lot no. 18E092, Jan. 2021) were purchased from the European Collection of Authenticated Cell Cultures (Wiltshire, UK); 5637 cells [lot no. I-5051 (N6-2-7, 43), Aug. 2009] were obtained from Cell Resource Centre for Biomedical Research Institute of Development, Aging and Cancer, Tohoku University (Miyagi, Japan). All cell lines were routinely tested for Mycoplasma using CycleavePCR^TM Mycoplasma Detection Kit (Takara Bio, Shiga, Japan) and used for experiments within 12 weeks of thawing. Cell culture details are available as Supporting Information.

EV isolation

Isolation of uEVs was performed by ultracentrifugation on a 30% sucrose/D₂O cushion, as previously described [17]. We also isolated uEVs using the MagCapture^TM Exosome Isolation Kit PS (FUJIFILM Wako, Osaka, Japan), according to the manufacturer’s instructions [17, 18].

For EV isolation from cell lines, sub-confluent cells, in 15 cm dishes, were washed twice with phosphate-buffered saline and grown in a culture medium supplemented with 10% exosome-depleted foetal bovine serum for 48 h. The conditioned medium was collected, and EVs were isolated by differential centrifugation. The final pellet was resuspended in 50 μL of phosphate-buffered saline. The size and number of EVs obtained were determined using a qNano nanoparticle characterisation system (Izon Science, Christchurch, New Zealand), and the protein concentration of the EVs was measured using a Micro BCA protein assay kit (Thermo Fisher Scientific, MA, USA). The details of EV isolation from urine samples and cell lines are available as Supporting Information.

Selection of biomarker candidate proteins and target peptides for SRM/MRM analysis

Candidate biomarkers were selected from 1960 uEV proteins previously identified by shotgun proteomics [12] based on the following criteria: (a) proteins expressed at high levels in uEVs from patients with bladder cancer (fold change > 2.0; P < 0.05), and (b) membrane proteins reported in the UniProt Knowledgebase (UniProtKB). To verify candidate proteins as bladder cancer biomarkers, target peptides of the proteins were selected for SRM/MRM analysis. Details of the criteria for target peptide selection are available as Supporting Information.

Target proteomics

SRM/MRM analyses were performed as previously described [12]. The quantification values obtained for each uEV sample were normalised according to the values of CD9, for deviations in EV collection from urine. The details of target proteomics are available as Supporting Information.

ELISA

Frozen urine samples were thawed in a 37 °C water bath and vortexed thoroughly. The samples were centrifuged at 10,000 × g for 30 min at room temperature, and the supernatant was filtered through a 0.22 μm filter. uEVs were isolated from 5 mL of filtered urine using the MagCaptureTM Exosome Isolation Kit PS (FUJIFILM Wako). A 96-well ELISA plate (BioLegend, CA, USA) was coated with an anti-human EphA2 antibody (#356802, BioLegend). uEVs were added to the plates and incubated at room temperature for 2 h. After three washes, biotinylated anti-human CD9 antibody (1 K, FUJIFILM Wako) was added, and the plate was incubated at room temperature for 1 h. After three washes, the plate was incubated with horseradish peroxidase-conjugated streptavidin at room temperature for 2 h. After five final washes, the plate was incubated with 3,3′,5,5′-tetramethylbenzidine reagent at room temperature for 30 min, followed by the addition of a stop solution (2 M H₂SO₄). The absorbance was measured at 450 nm using a microplate reader (Bio-Rad, CA, USA), with the correction wavelength set at 620 nm. A standard curve was prepared using EVs purified from the cell culture supernatant of COLO201 cells (COLO201 exosomes, FUJIFILM Wako).

Immunohistochemistry (IHC)

Immunohistochemical staining was performed using primary antibodies against EphA2 (1:200; PA5-14574; Thermo Fisher Scientific) and CK5/6 (1:100; D5/16B4; Sigma–Aldrich, MO, USA), as previously reported [19]. The details are available as Supporting Information. The IHC score of each field was assigned from 0 to 3+ (0, no staining; 1+, weak; 2+, moderate and 3+, strong), based on the intensity (0–3+) and extent, using the following criteria:

If no staining was observed: 0 (negative).

If the field stained 1+ or less (except all negative): 1+ (weak).

If more than half of the field was stained 1+ and the rest (less than half) was stained 2+ or 3+: 2+ (moderate).

If more than half of the field was stained 2+ and the rest (less than half) was stained 3+: 3+ (strong).

If more than half of the field was stained 3+ and the rest (less than half) was stained 1+ or 2+: 3+ (strong).

The average IHC score of five randomly selected fields (400×) was calculated in each case as the final result, and a positive case for EphA2 staining was defined as the average IHC score of 1+ or higher. In addition, we categorised tumours as the basal subtype with a cut-off of 20% positivity for CK5/6 (cytoplasmic staining), as this cut-off best categorises tumours based on their molecular subtypes [20, 21].

Establishment of EphA2-knockout (KO) 5637 cells and EphA2-overexpressing (OE) HEK293 cells

EphA2-KO 5637 cells were established using the CRISPR/Cas9 system [22]. A single guide RNA (sgRNA, CTACTATGCCGAGTCGGACC) targeting the EphA2 gene was cloned into the pSpCas9(BB)-2A-Puro (PX459) V2.0 plasmid (Addgene, MA, USA).

Stable EphA2-OE HEK293 cells were then established. The EphA2-coding sequence was cloned into the pcDNA3.1 + C-DYK plasmid (GenScript, NJ, USA). Cloning was confirmed by restriction digestion and DNA sequencing. EphA2 knockout and overexpression were confirmed by western blot analysis. The details are available as Supporting Information.

Matrigel invasion assay to assess the effect of EV-EphA2 on the cell invasion potential of bladder cancer cells

U-BLC1 cells (4.0 × 10⁴ cells/well) were seeded into the upper chamber of a Corning® BioCoat™ Matrigel Invasion Chamber (pore size—8 µm) (Corning, AZ, USA) in serum-free media, with the base plate containing culture media supplemented with 10% exosome-depleted FBS, and EVs or phosphate-buffered saline (50 μL/well) were added to the upper chamber.

For the experiments of pre-conjugating Ephrin-A1 protein to EV-EphA2, U-BLC1 cells (4.0 × 10⁴ cells/well) in the upper chamber were co-cultured with 5637 cell-derived EVs or EphA2-OE HEK293 cell-derived EVs with or without a 10 min pre-incubation with recombinant human Ephrin-A1 Fc chimera protein (R&D Systems, MN, USA), which was derived from NS0 mouse myeloma cells. After incubation for 72 h at 37 °C with 5% CO₂, cells that had penetrated the Matrigel were fixed and stained using Diff-Quik stain (Sysmex, Hyogo, Japan) and counted. The analysis for each experimental group was repeated three times.

Wound-healing assay to assess the effect of EV-EphA2 on the cell migration potential of bladder cancer cells

A confluent layer of U-BLC1 cells in a 24-well plate was scratched using a sterile 1 mL pipette tip, and the detached cells were removed by washing with PBS. The cells were then incubated in a fresh culture medium supplemented with 10% exosome-depleted FBS, and EVs or PBS were added to each well (50 μL/well). The cells were incubated at 37 °C with 5% CO₂. Cell migration was captured using a fluorescence microscope BZ-X710 (KEYENCE, Osaka, Japan) at 0 and 16 h after scratching, and quantified by measuring the size of the recovered area using ImageJ 1.53e. The assay for each experimental group was repeated three times.

Statistical analyses

Statistical analyses were performed using JMP Pro (v.16.0.0; SAS Institute, NC, USA), and visualisation quantification was performed using the GraphPad Prism software (v.7.05; GraphPad Software, CA, USA). Univariate analysis was performed using Welch’s t-test, two-tailed Student’s t-test and Mann–Whitney U test. The Tukey–Kramer method was used for multiple comparisons, and Dunnett’s test was used to compare several treatments with a single control. Differences were considered significant at P < 0.05. The optimal cut-off value for diagnosis was determined from the receiver operating characteristic curve using the Youden index, and the sensitivity and specificity for diagnosis were calculated according to each optical cut-off value.

All experiments were evaluated objectively, with groups blinded at the time of analysis.

More comprehensive details of the Materials and Methods are available as Supporting Information.

Results

Proteomic analysis of urine identified uEV-EphA2 as a bladder cancer biomarker

An overview of biomarker protein identification for bladder cancer diagnosis is shown in Fig. 1a. Among the 1960 uEV proteins previously identified by shotgun proteomics (tandem mass tag-labelling LC-MS/MS) [12], uEV proteins from patients with bladder cancer were compared with those from individuals without bladder cancer. We focused on 30 proteins upregulated in uEVs obtained from patients with bladder cancer (fold change > 2.0; P < 0.05; Welch’s t-test). Considering the development of EV-based clinical assays, we selected 13 membrane proteins from the 30 proteins as potential biomarker candidates (Table S1). To validate candidate proteins, target proteomics (SRM/MRM) was performed on uEVs isolated from 97 individuals, including 49 patients with bladder cancer and 48 individuals without bladder cancer (36 healthy controls and 12 patients with non-malignant haematuria, including six with cystitis)]. The clinical and pathological information of the patients is summarised in Table 1. Among the 13 candidate biomarkers, eight were detected by SRM/MRM, all of which were more significantly upregulated in patients with bladder cancer than in individuals without bladder cancer, including patients with non-malignant haematuria (P < 0.05; Mann–Whitney U test) (Table 2). However, among them, only EphA2 showed a significant difference in expression between patients with bladder cancer and patients with non-malignant haematuria (P = 0.01). The levels of other candidate proteins were more significantly elevated in patients with bladder cancer than in individuals without bladder cancer, but the differences reduced upon comparison with patients with non-malignant haematuria (Fig. S1). The relative quantification data of uEV-EphA2 in shotgun and target proteomics are summarised in Fig. 1b; the other seven uEV proteins included gamma-enolase, complement decay-accelerating factor, myristoylated alanine-rich C-kinase substrate, tight junction protein ZO-2, protein diaphanous homolog 1, myristoylated alanine-rich C-kinase substrate-related protein and epidermal growth factor receptor (Fig. S1). The list and levels of the 11 peptides for the eight proteins in target proteomics are shown in Tables S2 and S3.

Fig. 1. Identification and validation of urinary EV-EphA2 as a biomarker for bladder cancer diagnosis.

a Strategy for bladder cancer biomarker identification in urinary EVs. The selection method at each step and the number of shortlisted candidate proteins are indicated. b Relative quantification data of urinary EV-EphA2 between patients without bladder cancer [healthy control (HC)] and patients with bladder cancer in shotgun proteomics (left), between patients without bladder cancer (HC and non-malignant haematuria) and patients with bladder cancer (centre) and between patients with haematuria and patients with bladder cancer (right) in target proteomics. Receiver operating characteristic (ROC) curve analysis of urinary EV-EphA2 for bladder cancer diagnosis in target proteomics is also shown. Data are expressed as median with 95% confidence interval (Mann–Whitney U test; ***P < 0.01; *P < 0.05, n.s. not significant). The area under the curve (AUC) is indicated on the ROC curve.

Table 2.

List of eight proteins verified in target proteomics.

	Bladder cancer (n = 49)		Bladder cancer (n = 49)
	vs		vs
	Non-bladder cancer (HC + haematuria) (n = 48)		Haematuria (n = 12)
Protein name	Fold change	P-value	Fold change	P-value
Ephrin type-A receptor 2 (EphA2)	2.51	<0.0001	1.95	0.01
Gamma-enolase (ENO2)	1.95	0.0008	1.79	0.132
Complement decay-accelerating factor (CD55)	1.73	0.0265	0.91	0.16
MARCKS-related protein	2.76	<0.0001	1.95	0.165
Tight junction protein ZO-2 (TJP2)	2.86	<0.0001	1.78	0.28
Protein diaphanous homolog 1 (DIAPH1)	1.72	<0.0001	1.05	0.751
Myristoylated alanine-rich C-kinase substrate (MARCKS)	2.29	0.0002	1.52	0.849
Epidermal growth factor receptor (EGFR)	1.69	0.006	1.25	0.849

HC healthy controls.

In target proteomics, receiver operating characteristic curve analysis indicated that the area under the curve (AUC) value of uEV-EphA2 for bladder cancer diagnosis was 0.79 in bladder cancer vs non-bladder cancer (healthy controls and patients with non-malignant haematuria) (sensitivity: 77.6%; specificity: 72.9%) and 0.74 in bladder cancer vs non-malignant haematuria (without healthy control) (sensitivity: 87.8%; specificity: 58.3%) (Fig. 1b). EphA2 is a single-pass type I membrane protein; hence, it is expected to be a potential EV surface biomarker. Therefore, we focused on EphA2 for further analyses.

The development of ELISA for EV-EphA2 quantification

To determine the clinical application of urinary EV-EphA2 as an early diagnostic marker for bladder cancer, we designed an ELISA system to measure EV-EphA2 without ultracentrifugation and mass spectrometry. We employed the phosphatidylserine affinity method, which can isolate EVs in a short time without ultracentrifugation [18], and selected CD9, which is a common EV marker, as the detection target for EV-EphA2 in our assay. A schematic of the assay is shown in Fig. 2a. EVs isolated from 5 mL of urine using the phosphatidylserine affinity method were first captured with the anti-EphA2 antibody immobilised on a 96-well plate and detected with an anti-CD9 antibody; thus, bladder cancer-derived EVs were quantified by recognition of EphA2/CD9 double-positive EVs.

Fig. 2. Development of EV-EphA2-CD9 ELISA and evaluation of bladder cancer diagnostic performance.

a Schematic overview of EV-EphA2-CD9 ELISA. b Correlation of the OD (450–620 nm) value with the COLO201 exosome concentration in EV-EphA2-CD9 ELISA. c Relative quantitation data and d receiver operating characteristic (ROC) curve analysis of urinary EV-EphA2-CD9 for differentiating between patients without bladder cancer (including ten patients with non-malignant haematuria) and patients with bladder cancer in EV-EphA2-CD9 ELISA. Data are expressed as median with 95% confidence interval (Mann–Whitney U test; ***P < 0.001). The area under the curve (AUC), sensitivity and specificity are shown on each graph.

To examine the quantitative reactivity of this ELISA (EV-EphA2-CD9 ELISA), we measured EV-EphA2-CD9 using commercially available COLO201 exosomes. EphA2 expression on COLO201 exosomes was confirmed using a concentration gradient (Fig. S2). The standard curve exhibited concentration-dependent linearity (Fig. 2b).

EV-EphA2-CD9 ELISA was then used to analyse 72 urine samples from 36 patients with bladder cancer and 36 individuals without bladder cancer (26 healthy controls and 10 patients with non-malignant haematuria). The clinical and pathological information of the patients is summarised in Table 1. EV-EphA2-CD9 expression was significantly higher in uEVs from patients with bladder cancer than in those from individuals without bladder cancer (P < 0.0001; Mann–Whitney U test) (Fig. 2c). In addition, both patients with low-grade and high-grade bladder cancer had significantly higher uEV-EphA2-CD9 expression than individuals without bladder cancer in multiple analyses (Bonferroni corrected Mann–Whitney U test) (Fig. S3).

Receiver operating characteristic curve analysis indicated that the AUC value for bladder cancer diagnosis was 0.78 (95% CI: 0.67–0.89), with a sensitivity of 61.1% (22/36) and specificity of 97.2% (35/36) (Fig. 2d). Additionally, the sensitivity and specificity of urinary cytological detection in patients with bladder cancer were 50% (18/36) and 100%, respectively. When EV-EphA2-CD9 detection and urinary cytology results were combined, the sensitivity and specificity for identifying patients with bladder cancer were 80.6% (29/36) and 97.2% (35/36), respectively (Table S4).

EphA2 exerted cancer-promoting effects in bladder cancer cells and was widely expressed in bladder cancer tissues

We examined EphA2 expression in five bladder cancer cell lines and a normal uroepithelial cell line (SV-HUC-1), using western blotting. Among bladder cancer cell lines, 5637 cells showed the highest EphA2 expression level; SV-HUC-1 cells also expressed EphA2, although at marginally lower levels than bladder cancer cells (Fig. 3a).

Fig. 3. Functional analysis and immunohistochemical staining of EphA2 in bladder cancer cells.

a Expression of EphA2 protein in five bladder cancer cells and a normal uroepithelial cell line (SV-HUC-1). Five micrograms of cell lysate proteins were applied to each lane. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was used as the loading control. The molecular weight markers are indicated on the right. b Knockdown of EphA2 using siRNA in 5637 and T24 cells. GAPDH was used as the loading control. The molecular weight markers are indicated on the right. c T24 and 5637 cells were transfected with siEphA2 or negative control siRNA for 24 h and reseeded in a 96-well plate, incubated for the indicated time durations, and examined by MTS cell proliferation assay. Data are expressed as mean ± standard deviation (SD) (Student’s t-test; *P < 0.05). d T24 and 5637 cells were transfected with siEphA2 or negative control siRNA for 72 h and were reseeded into the upper chamber of Matrigel-coated Transwell membrane inserts (5637 cells, 4.0 × 10⁴ cells; T24 cells, 2.0 × 10⁴ cells); the lower chamber was filled with the complete medium and then cultured for 72 h (5637 cells) or 48 h (T24 cells). Invasive cells that had penetrated the Matrigel membrane were fixed and stained. The statistical results of three independent experiments are summarised in the left panel. Data are expressed as mean ± SD (Student’s t-test; ***P < 0.001; **P < 0.01). e T24 and 5637 cells were transfected with siEphA2 or negative control siRNA for 72 h, and cell migration was measured 16 h after wound creation. Representative results of cell migration in the scratch wound-healing assay are shown. The statistical results of three independent experiments are summarised in the left panel. Data are expressed as mean ± SD (Student’s t-test; ***P < 0.001; **P < 0.01). f EphA2 expression in bladder cancer tissues and normal uroepithelium was assessed in immunohistochemical analysis. (a–c) Moderate-strong expression of EphA2 in bladder cancer tissues. (d–f) Weak expression of EphA2 in normal uroepithelium. Scale bars: 50 μm. g Comparison of EphA2 IHC scores between normal uroepithelium and bladder cancer tissues. Data are expressed as median with 95% confidence interval (Mann–Whitney U test, **P < 0.01).

Further, to investigate the biological function of EphA2 in bladder cancer cells, we used siRNA to knockdown EphA2 expression in 5637 and T24 cells. As expected, the protein levels of EphA2 were successfully downregulated by the siRNA (Fig. 3b). The knockdown of EphA2 significantly inhibited the proliferation of 5637 and T24 cells on day 4 in the MTS cell proliferation assay (Fig. 3c). In the Matrigel invasion assay, EphA2 knockdown significantly attenuated the invasion of 5637 and T24 cells (Fig. 3d). Furthermore, cell migration was significantly inhibited by siEphA2 in both cell lines (Fig. 3e). These results indicated that EphA2 exerts cancer-promoting effects in bladder cancer cells. We further conducted an immunohistochemical analysis of EphA2 expression in bladder cancer tissues (n = 34) and normal uroepithelium (n = 26). The clinical and pathological information of patients evaluated in the immunohistochemical analysis is shown in Table S5. Typical patterns of EphA2 immunohistochemical expression in bladder cancer tissues and normal uroepithelium are shown in Fig. 3f.

EphA2 expression was observed in 88.2% (30/34) of the tumours (IHC score median, 1.4; range, 0.2–2.6). In addition, EphA2 expression was observed in 73.1% (19/26) of the normal uroepithelial tissues (IHC score median, 1.0; range, 0–2.2); however, the IHC score of the normal uroepithelium was statistically lower than that of bladder cancer tissues (P = 0.01, Mann–Whitney U test) (Fig. 3g).

Thus, EphA2 was expressed in most bladder cancer tissues and was also weakly expressed in the normal uroepithelium.

Next, we analysed the associations of the IHC scores of EphA2 with tumour grade and stage. Notably, the top three cases with the highest EphA2 IHC score in the NMIBCa group were the three cases of BCG-refractory NMIBCa. However, there was no statistical difference between low-grade (n = 6) and high-grade tumours (n = 28) (P = 0.79, Mann–Whitney U test) (Fig. S4A) and between the NMIBCa (n = 20) and MIBCa groups (n = 14) (P = 0. 37, Mann–Whitney U test) (Fig. S4B). We also stained muscle-invasive bladder cancer tissues for CK5/6, which is used as a marker for the basal subtype of bladder cancer. We found that all the MIBCa cases in our present study were classified as the basal subtype, with a cut-off of 20% positivity for CK5/6, and observed a negative correlation between the IHC scores of CK5/6 and EphA2 in this group (Fig. S5A). In contrast, the IHC score of EphA2 was not correlated with the value of urinary EV-EphA2 in target proteomics (SRM/MRM) or EV-EphA2-CD9 ELISA in the present study (Fig. S5B and S5C).

EV-EphA2 promoted the invasiveness and migration of bladder cancer cells

We performed western blotting (0.5 μg/well) to identify the bladder cancer cell line with the highest EphA2 expression on EVs (EV-EphA2). As shown in Fig. 4a, EV-EphA2 expression was the highest in 5637 cells. Additionally, we found that SV-HUC-1 cells secreted EV-EphA2 at a lower level than bladder cancer cell lines.

Fig. 4. The effect of EV-EphA2 on the invasion and migration potential of bladder cancer cells.

a Expression of EV-EphA2 isolated from the cell culture media of five bladder cancer cell lines and one normal uroepithelial cell line (SV-HUC-1). EV protein (0.5 µg) was applied to each lane. β-actin was used as the loading control. The molecular weight markers are indicated on the right. b Distribution of particle sizes and number of EVs isolated from the culture medium of 5637 cells seeded in three 15 cm dishes. c Western blotting using the CRISPR/Cas9 system showed that EphA2 was knocked out in the lysate of 5637 cells (left) and 5637-EVs (right). GAPDH (left) and CD9 (right) were used as the loading controls. The molecular weight markers are indicated on the right. d 5637 cell-derived EVs (5637-EVs) significantly increased the invasiveness of U-BLC1 cells, whereas EphA2 knockout disabled the invasion-promoting effect. e Western blotting showed that EphA2 was overexpressed in the lysate of HEK293 cells (left) and HEK293-EVs (right). GAPDH (left) and CD9 (right) were used as the loading controls. The molecular weight markers are indicated on the right. f EphA2-OE HEK293 cell-derived EVs (EphA2-OE HEK293-EVs) significantly promoted the invasion potential of U-BLC1 cells compared with phosphate-buffered saline (PBS) and mock HEK293-EVs. g 5637-EVs significantly promoted the migration of U-BLC1 cells, whereas EphA2-KO 5637-EVs did not exert a migration-promoting effect. h EphA2-OE HEK293-EVs significantly promoted the migration potential of U-BLC1 cells compared with PBS and mock HEK293-EVs. The statistical results of three independent experiments are summarised. Data are expressed as mean ± standard deviation (SD) (Tukey–Kramer method; ***P < 0.001; **P < 0.01; *P < 0.05; n.s. not significant). i Schematic of a blocking experiment in which recombinant Ephrin-A1 protein was pre-conjugated to EV-EphA2 to inhibit the binding of EphA2 to Ephrin-A1 ligands on cancer cells. j U-BLC1 cells were cultured with PBS (control), 5637-EVs or EphA2-OE HEK293-EVs with or without pre-incubation with 2 μg of recombinant Ephrin-A1. 5637-EVs and EphA2-OE HEK293-EVs, both of which express EphA2 at high levels, as shown in the bottom image, promoted the invasiveness of U-BLC1, but pre-incubation with Ephrin-A1 reduced the invasion-promoting effect of 5637-EVs and EphA2-OE HEK293-EVs. The statistical results of three independent experiments are summarised. Data are expressed as mean ± SD (Dunnett’s test; *P  <  0.05; n.s. not significant).

Next, we investigated whether EVs from 5637 cells, which secreted the highest levels of EV-EphA2 among the five bladder cancer cell lines, promoted the invasiveness of bladder cancer cells. The concentration of EVs isolated from 5637 cells was 6.7 × 10⁸ particles/50 µL, and almost all the particles were less than 200 nm in size (Fig. 4b). We then established EphA2-KO 5637 cells using the CRISPR/Cas9 system and collected EVs with or without EphA2 (Fig. 4c).

We performed the Matrigel invasion assay and found that EVs derived from 5637 cells significantly increased the invasiveness of U-BLC1 cells, whereas EphA2 knockout inhibited the invasion-promoting effect (Fig. 4d). To further determine the invasion-promoting effect of EV-EphA2 on bladder cancer cells, EphA2-OE HEK293 cells were established, and EphA2-OE EVs were collected (Fig. 4e). The invasion assay showed that EVs from EphA2-OE HEK293 cells significantly promoted the invasive potential of U-BLC1 cells more than did the control EVs from HEK293 cells (Fig. 4f).

Next, we attempted to inhibit the ability of EV-EphA2 to bind to its major ligand, Ephrin-A1, on cells. Pre-incubation of Ephrin-A1 with 5637 cell-derived EVs or EphA2-OE HEK293 cell-derived EVs reduced their ability to promote the invasion of U-BLC1 cells (Fig. 4g, h).

Additionally, we knocked down Ephrin-A1 in U-BLC1 cells and investigated the impact on the invasion-promoting effect of EV-EphA2 (Fig. S6A). Although the transfection of negative control siRNA to U-BLC1 cells did not alter the invasion-promoting effect of EV-EphA2 (Fig. S6B), the knockdown of Ephrin-A1 in U-BLC1 cells attenuated the invasion-promoting effect of EV-EphA2 (Fig. S6C).

Collectively, these findings suggest that EphA2 on EVs exerts invasion-promoting effects on bladder cancer cells through the ligand Ephrin-A1 on cancer cells.

Furthermore, we found that EVs derived from 5637 cells significantly promoted the migration potential of U-BLC1 cells, whereas EVs from EphA2-KO 5637 cells did not promote the migration potential (Fig. 4g), and EVs from EphA2-OE HEK293 cells significantly promoted the migration potential of U-BLC1 cells more than did the control EVs from HEK293 cells (Fig. 4h). Taken together, these results indicate that EV-EphA2 promotes the migration and invasion potential of U-BLC1 cells.

We also determined if EV-EphA2 promoted the proliferation of bladder cancer, U-BLC1 and RT4 cells; the proliferation-promoting effect of EVs was found to be independent of EphA2 expression (Fig. S7).

Discussion

Cancer-derived EVs are of increasing interest as diagnostic biomarkers and therapeutic targets. Previously, we focused on EV proteins that were identified both in uEVs and tissue-exudative EVs, which were extracted directly from cultures of surgically resected viable cancerous tissues [12], but in this study, we re-analysed the shotgun proteomics data and performed a comparative proteomic analysis of uEVs between bladder cancer and non-bladder cancer patients, and identified uEV-EphA2 as a novel biomarker for bladder cancer diagnosis. In addition, we established an EV-EphA2-CD9 ELISA to quantify uEV-EphA2 for the diagnosis of bladder cancer. We further demonstrated that EphA2 and EV-EphA2 promote the progression of bladder cancer.

To date, several studies have focused on uEV proteins as bladder cancer biomarkers and have undertaken proteomic analyses [13–16]. Lee et al. collected uEVs from patients with bladder cancer and healthy individuals and identified mucin-1, carcinoembryonic antigen, epidermal growth factor receptor kinase substrate 8-like protein 2 and moesin as bladder cancer biomarkers [13]. Using the same approach, other studies have identified potential protein biomarkers [14–16].

However, the protein biomarkers identified in these studies were not validated in patients with non-malignant haematuria, such as patients with cystitis, which is expected in clinical practice. In addition, uEVs obtained in these reports were isolated using ultracentrifugation alone, and it is known that performing only ultracentrifugation tends to lead to the co-isolation of non-EV-associated proteins, such as Tamm-Horsfall proteins, at high levels [23–25]. It has been reported that the proteomic profile differs significantly between uEVs isolated using ultracentrifugation alone and uEVs isolated using density fractionation, and contaminating proteins could cause the false-negative identification of cancer-specific biomarkers [25]. Therefore, in this study, we adopted an additional purification step using a sucrose/D₂O cushion to isolate uEVs with minimal contamination by Tamm-Horsfall protein and other soluble proteins, as previously reported [12]. This isolation method and validation with patients with non-malignant haematuria increased the reliability and integrity of the proteomic data for biomarker discovery. Resultantly, we considered uEV-EphA2 as a reliable biomarker for bladder cancer diagnosis, which can be used to distinguish between patients with bladder cancer and non-bladder cancer patients with haematuria.

Recently, blood EV-EphA2 was reported as a potential diagnostic biomarker of pancreatic cancer [26, 27]; however, to the best of our knowledge, there are no reports on EV-EphA2 as a diagnostic biomarker of bladder cancer. This study showed, for the first time, that uEV-EphA2 could be a potential biomarker for bladder cancer.

In general, EV-associated transmembrane proteins can exist in either an ‘outside-out (localised to the outer surface of both the plasma and EV membranes)’ or ‘inside-out (localised to the inner surface of the plasma membrane and moving to the outer surface of the EV membrane)’ configuration. EphA2 is known as a transmembrane protein in the plasma membrane, and anti-human EphA2 antibodies (BioLegend) were shown to bind to the extra-plasma membrane region of EphA2 protein because the EphA2 antibody can be used to detect cell surface EphA2 in flow cytometry analysis [28]. In addition, the same EphA2 antibody can capture EV-EphA2 on the ELISA plate, suggesting that the antibody-binding sequence is present in the extra-EV membrane region of EphA2. Therefore, EphA2 is assumed to be an ‘outside-out’ EV-associated transmembrane protein.

In the present study, to develop a simple assay for EV-EphA2 measurement with clinical value, we employed the MagCapture^TM Exosome Isolation Kit PS (FUJIFILM Wako) to isolate EVs. MagCapture^TM is a useful method for easily isolating pure EVs without ultracentrifugation, which is time-consuming and requires expensive equipment.

In the validation of ELISA with the test cohort, EV-EphA2-CD9 ELISA demonstrated good diagnostic performance, particularly when it was combined with urine cytology; the combination of EV-EphA2-CD9 ELISA with urine cytology might reduce the use of cystoscopy at bladder cancer diagnosis and during follow up.

EphA2 is a member of the erythropoietin-producing hepatocellular tyrosine kinase receptor family.

EphA2 signalling is mediated through interaction with the cell surface-anchored ligand Ephrin-A1 via direct cell–cell interactions, resulting in the bidirectional activation of signals in the corresponding cells [29, 30]. EphA2 signalling is highly regulated in normal tissues but is deregulated during carcinogenesis, owing to the loss of cell contact. As a result, EphA2 is overexpressed, leading to a ligand-independent increase in oncogenic signal transduction. EphA2 overexpression has been reported in various solid tumours, including colorectal cancer [31], non-small cell lung cancer [32], breast cancer [33, 34], gastric cancer [35], oesophageal cancer [36], pancreatic carcinoma [37], prostate cancer [38], head and neck cancer [39], glioblastoma [40], ovarian cancer [41] and bladder cancer [42, 43]. Abraham et al. reported that the staining intensity of EphA2 was less in normal uroepithelium but higher in advancing stages of bladder cancer [42]. However, Komoun et al. demonstrated that EphA2 positivity was higher in NMIBCa than in MIBCa, and hence, a later stage did not correspond to greater EphA2 expression [43]. These findings suggested that EphA2 expression in bladder cancer tissues could also vary according to the molecular subtype, and indeed, in our present study, in some cases, the MIBCa tissues (especially in patients with a high CK5/6 expression among basal-subtype tumours) showed low EphA2 expression in the immunohistochemical analysis. However, EphA2 expression in bladder cancer tissues did not correlate with the uEV-EphA2 level. We assumed that this could be attributed to the different levels of uEV-EphA2 secreted by bladder cancer tissues. Focusing on the functional aspect, EphA2 is reportedly associated with aggressive features in tumours, mediating tumorigenic and metastatic functions in non-small cell lung cancer [32], breast cancer [33, 34] and melanoma [44, 45]. However, it is unclear whether EphA2 exerts cancer-promoting effects in bladder cancer. In addition, EV-EphA2 is reported to mediate cancer cell–cell communications and promote malignant transformation of several cancers, such as pancreatic [26] and breast cancer [27, 46], through the Ephrin-A1-dependent reverse pathway without direct cell–cell interaction; however, to the best of our knowledge, no report has investigated the effects of EV-EphA2 on bladder cancer. In the present study, we showed, for the first time, that EphA2 enhances the proliferation, migration and invasion-promoting potential of bladder cancer cells, and that EV-EphA2 promotes the invasion and migration of bladder cancer cells. Thus, EV-EphA2 derived from bladder cancer cells is not only a potential diagnostic biomarker but also a potential therapeutic target for bladder cancer.

Our study, however, has several limitations. First, we demonstrated the diagnostic ability of EV-EphA2-CD9 ELISA as a proof of concept; the number of samples was limited. Thus, the clinical value of EV-EphA2-CD9 ELISA should be tested in a larger prospective study. Second, the effects of EphA2 knockdown on proliferation were statistically positive but considerably modest compared to the effects on invasion or migration. Combined with the finding that EV-EphA2 did not exert a proliferation-promoting effect, EphA2 expression in bladder cancer cells might be a secondary effect of the invasion- and migration-promoting effects. Third, the functional analysis of EphA2 and EV-EphA2 was conducted only in vitro in this study; thus, it would be desirable to verify whether EV-EphA2 enhances the invasive behaviour of bladder cancer cells in vivo.

In conclusion, we identified uEV-EphA2 as a novel diagnostic biomarker for bladder cancer by proteomic analysis and demonstrated a proof-of-concept for EV-EphA2-CD9 ELISA, which showed potential for good diagnostic performance in bladder cancer. Large-scale studies are necessary to validate the diagnostic potential of uEV-EphA2 in bladder cancer.

Reporting summary

Further information on research design is available in the Nature Research Reporting Summary linked to this article.

Author contributions

Study conception and design: ET and KF; data acquisition: ET, KM and RN; data analysis: ET, KM, RN, AY, TU, GY, YK, MM and YH; provision of resources: ET, KM, YH, MH, EB, RU, TT and ST; drafting the manuscript and figures: ET and KF; reviewing the manuscript: TK, KH, AK, MU, HU, JA, TT and NN. All authors have read and approved the final draft for submission.

Funding

This study was supported by the Japan Society for the Promotion of Science under KAKENHI (grant numbers: 17K16788 and 20K18140) and the Japan Agency for Medical Research and Development under Translational Research (grant number: A102).

Data availability

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

Competing interests

The authors declare no competing interests.

Ethics approval and consent to participate

Approval was obtained from the Institutional Review Board (Osaka University Hospital Institutional Review Board, Protocol Number: 13397-11) before initiating the study, and all patients provided written informed consent. The study was performed in accordance with the Declaration of Helsinki.

Consent for publication

Not applicable.

Supplementary information

The online version contains supplementary material available at 10.1038/s41416-022-01860-0.