Bladder cancer extracellular vesicles drive tumorigenesis by inducing the unfolded protein response in endoplasmic reticulum of nonmalignant cells

The field cancerization effect has been proposed to explain bladder cancer's multifocal and recurrent nature, yet the mechanisms of this effect remain unknown. In this work, using cell biology, flow cytometry, and qPCR analyses, along with a xenograft mouse tumor model, we show that chronic exposure to tumor-derived extracellular vesicles (TEVs) results in the neoplastic transformation of nonmalignant human SV-HUC urothelial cells. Inhibition of EV uptake prevented this transformation. Transformed cells not only possessed several oncogenic properties, such as increased genome instability, loss of cell–cell contact inhibition, and invasiveness, but also displayed altered morphology and cell structures, such as an enlarged cytoplasm with disrupted endoplasmic reticulum (ER) alignment and the accumulation of smaller mitochondria. Exposure of SV-HUC cells to TEVs provoked the unfolded protein response in the endoplasmic reticulum (UPRER). Prolonged induction of UPRER signaling activated the survival branch of the UPRER pathway, in which cells had elevated expression of inositol-requiring enzyme 1 (IRE1), NF-κB, and the inflammatory cytokine leptin, and incurred loss of the pro-apoptotic protein C/EBP homologous protein (CHOP). More importantly, inhibition of ER stress by docosahexaenoic acid prevented TEV-induced transformation. We propose that TEVs promote malignant transformation of predisposed cells by inhibiting pro-apoptotic signals and activating tumor-promoting ER stress–induced unfolded protein response and inflammation. This study provides detailed insight into the mechanisms underlying the bladder cancer field effect and tumor recurrence.

Urothelial carcinoma of the urinary bladder is the seventh most common cancer in the world, with 81,190 cases estimated to have occurred in the United States in 2018 (1). This malignancy often occurs as multifocal tumors within the urothelium, and there is a high risk of local recurrence after treatment. Synchronous or metachronous multicentric tumors may or may not share similar histology, thus making it more difficult to predict the course of the disease and posing challenges for clinical management. A principal unresolved issue is whether multifocal tumors spread from a monoclonal origin or arise independently. In 1953, Slaughter et al. (2) proposed the field cancerization hypothesis to describe situations in which some carcinogenic influence creates an area of preconditioned epithelium where genetically distinct tumor clones later arise, and there is evidence for a field cancerization effect in bladder cancer. Bladder tumors are often accompanied by abnormalities in the nearby urothelium such as genetic instability (3), dysplasia, and carcinoma in situ (4), suggesting the presence of a common factor affecting the field. Genetic and epigenetic changes in the epithelium have been implicated and are likely determinants of the cancer field process; however, the underlying mechanism is still poorly understood.
Extracellular vesicles (EVs) 2 are now recognized as intriguing entities in cancer pathogenesis. Cancer cells secrete substantially greater numbers of EVs than nonmalignant cells, and the cargo they carry has been shown to sustain tumors in various ways (5) such as promoting cell survival (6), angiogenesis (7), and immune suppression (8,9). Indeed, a number of studies have shown that cancer EVs are capable of initiating carcinogenesis in normal recipient cells (10,11). Our curiosity about the potential role of bladder cancer EVs in the cancerization of the adjacent urothelial field prompted us to examine the cancer-related pathological alterations on recipient human urothelial cells after long-term cancer EV exposure.
The unfolded protein response of the endoplasmic reticulum (UPR ER ) is essential for maintaining normal cell physiology. When unfolded proteins accumulate in the ER lumen, stress signals are transduced by the dissociation of binding immunoglobulin protein (BiP), an ER molecular chaperone, from three ER-resident transmembrane sensors: inositol-required enzyme 1 (IRE1), protein kinase-like ER kinase (PERK), and activating transcription factor 6. These activate signal cascades that halt the translation of new proteins and promote increased protein folding and proteasome activity to restore ER homeostasis. However, the sustained overactivation of UPR ER produces pathological alterations and potentially leads to diseases including cancer (12). The rapid proliferation of cancer cells requires increased rates of protein folding, assembly, and transport which may increase ER stress, and the elevated expression of UPR ER signaling network proteins like BiP, IRE1, or PERK is characteristic of many cancers (13,14). Interference with the UPR ER pathway has been shown to suppress tumor growth in many preclinical models (14), and small molecules that block the activity of UPR components can serve as anti-tumor agents (15)(16)(17)(18)(19). Because tumor cell EVs differ from normal cell EVs in protein and RNA content (20 -23), including several different classes of foreign proteins (oncoproteins, signaling molecules, membrane trafficking proteins, adhesion molecules, and immunomodulatory molecules) (24), horizontal transfer of TEV cargo would likely provoke a stress response such as UPR ER in the recipient cells.
In the current study we demonstrate that chronic exposure to TEVs was sufficient to induce malignant transformation of the immortalized human urothelial cell line SV-HUC. The transformed cells were in a state of ER stress and inflammation that may explain the acquisition of persistent survival advantage. Inhibition of EV internalization and ER stress in the urothelial cells prevented malignant transformation. For the first time, we have revealed a novel pro-tumorigenic pathway mediated by cancer EVs that might explain the multifocal and heterogeneous nature of bladder carcinoma and also suggest new therapeutic strategies.

Experimental model of bladder cancer field effect
The carcinogenic transformation of cells requires the gradual accumulation of multiple genetic changes (25). We hypothesized that the chronic influence of EVs secreted by high-grade urothelial carcinoma cells might be sufficient to induce such changes in nearby recipient cells. In one scenario, cancer EVs might cancerize a field of normal urothelium; in another, they might drive cells in a previously cancerized field (partially transformed after exposure to tobacco carcinogens, for example) further along the pathway to tumorigenic transformation. To model the second scenario, we used SV-HUC, a human urothelial cell line that was transformed by simian virus 40 to yield cells that can survive extended in vitro culture but are not yet tumorigenic (26). We reasoned that this could serve as a model of clinically indolent urothelium in a cancerized field and allow us to investigate the potential of bladder cancer EVs to induce complete neoplastic transformation. In our experimental model of the bladder cancer field effect, we included groups of SV-HUC cells treated with SV-HUC EVs or DPBS vehicle and subjected to the same serial passaging as cancer EV-treated cells to control for the possibility that the observed alterations were produced by extended culture. Because of the heterogeneous nature of EV donor cells of patient-derived EVs, we chose to use the cancer EVs from the high-grade human urothelial carcinoma lines TCCSUP and T24. TEVs were purified via two ultracentrifuge spins, and purified EV quality was analyzed; our group had reported this previously (27). Here, we characterized purified EVs for purity by immunostaining with EV marker TSG101 and by nanoparticle tracking analysis (Fig. S1, A and  B). Structural and functional integrity of EVs were examined by transmission EM and confirmation of EV internalization. Interestingly, internalization of TCCSUP EVs and SV-HUC EVs did not influence the growth of recipient SV-HUC cells (Fig. S1, C-E). Recipient cells were treated with EVs for up to 13 weeks. Following the withdrawal of EVs, the cells were maintained in normal culture for at least 5 weeks and then assayed to observe persistent cellular alterations

Cancer EVs induce malignant transformation of nonmalignant SV-HUC cells
One of the hallmarks of cancer is the loss of cell-cell contact inhibition of proliferation (28). Cancer cells in vitro continue to proliferate and do not experience growth arrest when they are confluent. To determine the oncogenic property of the chronic cancer EV-exposed SV-SHC cells, we seeded the cells at high density and analyzed their cell cycle profiles by flow cytometry. At day 1 after seeding, both parental cells and 13-week TCCSUP EV-treated cells displayed normal cell cycle profiles. At confluence on day 6, cell cycle analysis revealed that parental SV-HUC cells were accumulating in the G 1 phase, whereas fewer cells were seen in the G 2 /M phase, indicating that they had stopped proliferating (Fig. 1A). In contrast, confluent 13-week TCCSUP EV-treated cells continued to grow and had a cell cycle distribution similar to day 1, indicating that the cancer EV-treated cells had acquired resistance to cell-cell contact inhibition.
SV-HUC cells at early passage do not form tumors in athymic nude mice but are capable of anchorage-independent growth (29). In an assay to test whether chronic exposure to cancer EVs enhanced colony formation in vitro, we found that cells cultured with TCCSUP EVs for 13 weeks showed a significant induction of colony number and size as compared with 8-week treated cells or controls (Fig. 1, B and C). Treatment with T24 EVs for 13 weeks also induced colony-formation ability but to a lesser extent (Fig. S2). We next engrafted 8-week and 13-week TCCSUP EV-treated cells subcutaneously into athymic nude mice to test whether they were sufficiently transformed to be tumorigenic in vivo. 12 weeks after engraftment, the 13-week, but not 8-week, cancer EVtreated cells had formed tumors with an average weight of 20 mg (Fig. 1, D and E). Histological examination of tissue sections confirmed that the masses were composed of disorganized tumor cells (Fig. 1F). Taken together, the in vitro and in vivo data provide direct evidence that chronic exposure to TCCSUP EVs is sufficient to drive the neoplastic transformation of SV-HUC urothelial cells.

Transformed SV-HUC cells express genome instability and gain invasiveness
First, we examined whether cancer EVs create DNA damage in recipients. Increased genomic instability is a general phenomenon of most cancers and is also found in otherwise Extracellular vesicles drive neoplastic transformation normal-appearing urothelium associated with bladder tumor (3). In previous studies, cancer EVs were found to increase reactive oxygen species (ROS) levels in nonmalignant recipient cells and thus possibly induce the DNA damage response (DDR) (30). We performed quantitative realtime PCR (qPCR) to screen for alterations of several key genes involved in anti-oxidative stress and DDR and found that the anti-oxidative stress gene SOD and the DDR genes GADD45A, GADD45B, GADD45G, and RAD50 were up-regulated in the cancer EV-transformed cells ( Fig. 2A). Corresponding to the gene profile alteration, we assessed cellular ROS status and found a significantly higher ROS level in 13-week TCCSUP EV-treated cells as compared with untreated cells (Fig. 2B). We then assessed DNA damage by measuring histone H2AX phosphorylation in response to DNA double-strand breaks (31). As shown in Fig. 2C, the nuclei of cells treated with TCCSUP EVs for 13 weeks had significantly more ␥H2AX foci as compared with untreated and 8-week treated cells. Together, these results suggest that cancer EV-treated cells have elevated oxidative stress, DNA damage, and impaired DDR resulting in genome instability.
Motility and invasiveness are hallmarks of progression to advanced cancer status. To measure the invasive potential of EV-transformed cells we used a Transwell invasion assay. As shown in Fig. 2D, the TCCSUP EV-transformed cells exhibited a striking increase in invasiveness, with significant numbers of cells transmigrating through basement membrane extract in contrast to untreated cells. We next measured the mRNA expression of two key factors involved in cell adhesion and mobility by qPCR. In concordance with the invasive phenotype, the gene coding for the invasion promoter N-cadherin (CDH2) was significantly up-regulated, and the gene coding for the invasion suppressor E-cadherin (CDH1) was down-regulated in the EV-transformed cells versus untreated cells (Fig. 2E).

Extracellular vesicles drive neoplastic transformation Suppression of EV uptake inhibits EV-induced transformation
EVs have multiple mechanisms of action and can exert influence in the microenvironment by various means, including receptor-ligand interactions at the recipient cell surface or the release of EV cargo into the intracellular space (32,33). To determine whether the transformation of SV-HUC cells required internalization of the cancer EVs, we used two EV inhibitors: Dynasore inhibits dynamin which is required for EV internalization (34,35), and NSC23766 blocks the GTPase activity of Rac1 that controls actin modulation and micropinocytosis (34,36) and processes in EV secretion and uptake, respectively. First, we determined the nontoxic doses of each inhibitor (Fig. S3). To determine whether these two com-pounds could inhibit EV uptake alone or in combination, we performed an EV internalization assay and found that at doses at which Dynasore (10 M) or NSC23766 (5 M) alone had little or no impact, a combination of both compounds significantly reduced EV internalization (Fig. 3A). After 8 weeks of TCCSUP EV exposure, we began adding the EV inhibitor combination or vehicle for the remaining 5 weeks of EV treatment. Cells treated with the EV inhibitor combination or vehicle in the absence of cancer EVs produced few colonies in the assay of anchorage independent growth in vitro. Continuation of TCCSUP EV exposure during the final 5 weeks with the addition of vehicle produced cells with significantly increased colony-forming potential, and addition

Extracellular vesicles drive neoplastic transformation
of the EV inhibitor combination reduced the number of colonies by more than half (Fig. 3B).

Cancer EVs induce UPR ER and the release of inflammatory cytokines
Cancer cells constitutively release high numbers of EVs, and those TEVs contain a variety of oncogenic cargoes consisting of proteins and nucleotides. Through the differential LC-MS/MS profiling EV proteins from TCCSUP and SV-HUC cells, we had reported on the identification of 110 unique proteins (37), and miRNAs that present only in the EVs derived from TCCSUP. Thus, horizontal transferring TCC-SUP EV cargo molecules to SV-HUC recipient cells would alter protein homeostasis and possibly provoke ER stress response in recipient cells. To test this possibility, we performed Western blotting to examine the UPR ER signals on SV-HUC cells that were treated with TCCSUP EVs and SV-HUC EVs. As shown in Fig. 4A, we found that TCCSUP EVs, but not SV-HUC EVs, dose-dependently increase the expression of total and phospho-PERK and that tunicamycin, an ER stress inducer, also significantly elevated phosho-IRE1 levels. These data illustrate that treatment with TCCSUP EVs provokes an immediate UPR ER to restore protein homeostasis. If UPR ER fails to alleviate ER stress under prolonged or severe ER stress, EV chaperones accumulate in the ER lumen, and cells activate apoptosis pathways to eliminate damaged cells. Our in vitro cell line model was designed to explore the pathological consequences of clinically indolent cells receiving EVs in the vicinity of tumor over an extended period of time. Thus, we also examined UPR ER signals on cancer EVtransformed cells and control cells (Fig. 4B). We found that the expression of ER chaperone protein BiP and the sensor protein IRE1 were up-regulated in the cancer EV-treated groups. As expected, we found that PERK signaling was activated in the 8-week treated cells and activated the pro-apoptotic signal C/EBP homologous protein (CHOP). Interestingly, 13-week treated cells had shifted this program toward the IRE1-mediated prosurvival pathway, in which IRE1 expression remained high, PERK was reduced, and CHOP was now absent. NF-B, a central transcriptional regulator of inflammatory signals downstream of IRE1 and PERK activation, was substantially upregulated in the cancer EV-transformed cells. Several ER stressinduced protein folding molecules (PDI, Ero1-L␣, and calnexin) were either slightly increased or remained constant. These data suggest that chronic cancer EV exposure possibly promotesmalignanttransformationthroughsuppressionofapoptosis and activation of inflammatory response.
Morphologically, EV-transformed cells were dramatically larger than untreated cells, as seen in phase-contrast microscopy (Fig. 4C, top panels). In analyzing the ultrastructure of EV-transformed cells by transmission EM, we found that the alignment of endoplasmic reticulum (ER) with mitochondria is

Extracellular vesicles drive neoplastic transformation
disrupted, and greater numbers of small-sized mitochondria are present (Fig. 4C, bottom right panel). These morphological abnormalities also support that transformed cells might activate ER stress response to cope with genotoxic stress induced by cancer EVs. Inflammation is one of the pro-cancer properties that can be induced by activation of UPR ER signals. To confirm the activation of inflammatory signals in TEV-transformed cells, we examined several ER stress-associated inflammatory cytokines by ELISA and found significant up-regulation of leptin and CCL2 and TGF␤ but to a lesser degree in transformed cells as compared with control cells (Fig. 4D). Finally, to show clinical relevance of the tumorigenesis pathway mediated by TER-UPR ER activation, we cross-checked the expression profiles of those cancer EV-altered UPR ER molecules from a published mRNA microarray dataset of bladder cancer patient tissues (38). We found that pro-apoptotic protein CHOP, which is abolished in the cancer EV-transformed cells, is also down-regulated in superficial and invasive bladder cancer relative to normal bladder tissue (Fig. 4E). We also examined the expression of ATF4, the transcription factor that regulates CHOP expression, and found that ATF4 is significantly reduced in superficial bladder cancer specimens versus normal controls as analyzed by Blaveri et al. (39) (Fig. S4C). Next, we tested whether TEV-transformed SV-HUC cells are resistant to ER stress-induced apoptosis; we applied bortezomib, a proteasome inhibitor, to induce ER stress. As expected, the TEVtransformed SV-HUC cells were significantly more resistant to bortezomib-induced growth inhibition as determined by MTT

Extracellular vesicles drive neoplastic transformation
and flow cytometry analyses (Fig. S4, A and B). Combined with the cellular alterations, these data suggest that cancer EV exposure might reprogram cells to switch to the survival branch of the UPR ER pathway and inhibit CHOP-mediated apoptosis, thereby enhancing cell fitness and survival.

Suppression of ER stress inhibits EV-induced transformation
Our data suggest that SV-HUC recipient cells use UPR ER as a survival strategy in a chronic cancer EV-enriched tumor microenvironment, where persistent ER stress facilitates surviving cells to acquire characteristics that lead to malignant transformation. To determine whether activation of UPR ER by cancer EVs is the driving force for SV-HUC cell transformation, we applied an ER stress inhibitor, docosahexaenoic acid (DHA). First, we examined whether DHA can inhibit ER stress induced by cancer EVs in SV-HUC cells; cells were treated with TCCSUP EVs in the presence/absence of DHA, and we found that TCCSUP EVs activate ER stress where the expression of PERK and IRE1 were up-regulated and adding DHA reversed the cancer EV-induced ER activation (Fig. 5A). Then we tested whether inhibition of ER stress by DHA can alleviate the cancer EV-mediated SV-HUC cell transformation. After 8 weeks of TCCSUP EV exposure, we began treating SV-HUC cell cultures with ER stress inducer tunicamycin alone, or the TCCSUP EVs for an additional 5 weeks in the presence or absence of DHA, cells were then maintained for 5 weeks in normal culture and assayed for colony formation ability. Indeed, tunicamycintreated SV-HUC cells produced a significantly higher number of colonies, suggesting that activation of UPR ER increases tumorigenic potential (Fig. 5B). As expected, continuation of TCCSUP EV exposure during the final 5 weeks produced cells with significantly increased colony-forming potential, and addition of DHA reversed this effect (Fig. 5B). Interestingly, cells treated with DHA alone produced fewer colonies in the assay of anchorage-independent growth in vitro. These data strongly suggest that activation of UPR ER signals by TCCSUP EV in SV-HUC cells promotes malignant transformation.
In summary, we have shown that horizontal transfer of cancer EV cargo following internalization results in malignant transformation of clinically indolent urothelial cells, and inhibition of EV internalization by small inhibiting compounds mitigates the transformative potential of cancer EVs (40). Critically, cells were evaluated following the withdrawal of cancer EVs and after a recovery period in normal culture for at least 5 weeks, indicating that the neoplastic changes were persistent. The proposed underlying mechanism is that cancer EVs perturb ER homeostasis and produce UPR ER in recipient cells. Prolonged induction of UPR ER signals leads to the activation of the survival branch of the UPR ER pathway, in which cells have elevated expression of IRE1, NF-B, and inflammatory cytokines, whereas the pro-apoptotic protein CHOP is silenced, resulting in enhanced cell survival and the expansion of malignant clones. More importantly, inhibition of UPR ER signals by DHA suppresses cancer EVinduced transformation. These data provide a novel mechanism by which cancer EVs in a cancerized urothelial field may promote de novo neoplasms.

Discussion
Many studies have shown that cancer EVs are capable of inducing tumorigenesis by a variety of mechanisms including transfer of miRNAs that regulate recipient cell gene expression (22, 40 -42), mRNAs that can be translated into functional proteins (41,42), and active proteins that directly influence recipient cell physiology (14,42). Recipient cells import functionally active cargoes that inhibit apoptosis, promote tumor survival and growth, and suppress the immune surveillance response that seeks out and destroys transformed cells (43). One concern is that such EV-mediated horizontal transfer of oncogenic material produces transient changes that fade following the withdrawal of EV influence (44). In this study, we provide the first evidence that cancer EVs are capable of inducing lasting changes in the recipient cell processes governing ER homeostasis and UPR. We found that prolonged exposure to cancer EVs resulted in induction of the prosurvival IRE1 branch of the UPR pathway, production of inflammatory cytokines, activated DDR, and ultimately malignant transformation. Therefore we propose cancer EV-driven UPR ER as a novel mechanism of tumorigenesis.
It is now widely accepted that EVs can deliver functional cargo to recipient cells, and although the mechanisms of cargo processing that occur following EV internalization remain mostly obscure, EV interactions with the ER have been examined. Heusermann et al. (45) found that internalized EVs were transported to the ER. Approximately 90% of EVs remained in close association with the ER membrane for up to 20 min, raising the possibility of cargo release. This prompts us to hypothesize that EVs might deliver an abundance of nonspecific cargo, such as mRNA transcripts or misfolded proteins, directly to the ER that overwhelms its functional capacity and induce ER stress. Alternatively, cancer EVs may deliver specific molecules that influence ER stress pathways. Javeed et al. (46) showed that EVs derived from pancreatic cancer cells cause ␤-cell dysfunction and up-regulated ER stress genes, indicating that these EVs may disrupt ER homeostasis. Our previously published MS data TCCSUP EVs contain many proteins that can increase angiogenesis, oxidative stress, and cell metabolism that would alter ER protein homeostasis and cause ER stress in recipient SV-HUC cells. Among them, one protein, PDI might be directly involved in TCCSUP EV-induced ER stress. PDI is an ER-resident protein induced during ER stress that is responsible for formation of disulfide bonds for proper protein folding (47). An in-depth loss-of-function study would be helpful to dissect their involvement. Moreover, the transformed cells in the current study displayed an abnormal accumulation of small-sized mitochondria with disrupted ER alignment. We suspect that Mfn2, a factor involved in mitochondrial fusion and tethering to ER (48), is affected in the transformed cells. Mfn2 downregulation has been shown to activate the three signaling branches of the UPR ER (49,50); therefore, it might be of interest to investigate a possible link between cancer EVs and Mfn2.
For the first time, we provide direct evidence that inducing UPR response either by tunicamycin or cancer EVs promotes tumorigenesis. Transformed cells use UPR ER pathways as a survival strategy, and interference with UPR ER pathways has been shown to suppress tumor growth in many preclinical models (15)(16)(17)(18). UPR ER signaling network proteins are often found elevated in cancers; thus small molecules that block the activity of UPR components can serve as anti-tumor agents (14,19). Nevertheless, activation of UPR ER signaling has both prosurvival and anti-survival effects on cells, and caution is necessary in designing therapies that target UPR components. UPR ER -induced inflammation is a double-edged sword: the stressed cells release inflammatory cytokines that function as alarm signals to the environment to maintain tissue integrity. However, chronic inflammation can lead to production of tumor-promoting cytokines, such as leptin, CCL2, and TGF␤ in our case, and over time it contributes to all stages of cancer development and progression through multiple mechanisms, such as genome instability, promoting cell proliferation, survival and invasion, and inducing angiogenesis (51). Inflammation signals play key roles in the etiology of bladder cancer. For instance, the loss of SPARC (secreted protein acidic and rich in cysteine), a tumor suppressor gene, promotes an inflammation phenotype that promotes bladder cancer development (52).
In the current study, we used DHA to inhibit ER stress, thereby inhibiting tumorigenesis as a proof of principle to demonstrate that activation of ER stress is the driving force behind cancer EV-induced tumorigenesis. DHA is a fatty acid that can sustain calcium homeostasis in the smooth ER by inhibiting the IP3 signaling pathway (53, 54). However, DHA is a versatile molecule that also interferes in multiple pro-tumor properties, such as inflammation, ROS, and mitogenic signaling pathways (protein kinase C and extracellular signal-regulated kinase 1/2) (53, 55). DHA has been linked to the alleviation of cancer; for instance, administration of omega-3 fatty acids, including both DHA and eicopentaenoic acid, significantly reduces carcinogen-induced bladder tumor in rats (56). We speculate that DHA may also regulate EV internalization and biogenesis through maintaining sodium/potassium pump activity and modifying plasma membrane lipid raft structure (57,58). Thus, in addition to suppressing ER stress, DHA might target multiple pro-tumor pathways including alleviating inflammation, oxidative stress, pro-tumorigenic signaling, and EV kinetics in treated cells. This might explain why DHA treatment suppressed colony formation ability significantly not only in the TCCSUP EV-treated cells, but also in PBS-treated controlled cells shown in Fig. 5.
Unlike other mechanisms of EV-driven tumorigenesis, the induction of chronic ER stress would presumably require a sustained assault by cancer EVs over a prolonged period of time. Cancer cells release elevated numbers of EVs (44) and would exert continuous long-term influence over normal recipient cells at local and distant sites prior to clinical detection. Because adjacent cells would be subject to the highest EV concentrations, transformation via induction of ER stress might be expected relatively close to the primary tumor, and field cancerization in this vicinity might represent incomplete transformation. In light of this, it may be that an initial bladder cancer diagnosis should be followed by immediate intervention with an EV suppression strategy aiming to stop further cancerization of the nearby field and reduce the potential for disease recurrence. Different therapeutic approaches are available to poten-

Extracellular vesicles drive neoplastic transformation
tially alleviate EV-induced ER stress and prevent future bladder cancer recurrence. Suppression of EV release by cancer cells has been achieved in the clinic using drugs like amiloride (59), which blocks both biogenesis and macropinocytosis of EVs via mechanisms involving ion channel-mediated endosome maturation and Rac1/Cdc42-related endocytosis and intracellular trafficking (60).

Cell culture and EV isolation
Cell lines were obtained from the American Type Culture Collection (Manassas, VA) and maintained according to instructions in a humidified chamber at 37°C and 5% CO 2 . For EV collection, the cells were cultured in medium containing EV-depleted fetal bovine serum (FBS; Thermo Fisher Scientific) as described previously (37,61). Cell culture supernatants were processed immediately after collection by serial centrifugation at 400 ϫ g for 10 min and 15,500 ϫ g for 30 min to remove cells and debris and then stored at Ϫ80°C. EVs were isolated from thawed samples by ultracentrifugation performed twice at 200,000 ϫ g for 70 min at 4°C, and the resulting pellets were resuspended in a small volume of DPBS. Aggregates were removed from the samples by another 15,500 ϫ g centrifugation for 5 min. Final total protein concentrations of the samples were measured by Micro BCA assay (catalog no. 23235, Thermo Fisher Scientific), and samples were stored at Ϫ80°C.

SV-HUC transformation
SV-HUC cells were seeded 2 ϫ 10 4 cells/well in 24-well plates and maintained in Ham's F-12K (Kaighn's) medium (Life Technologies) supplemented with 2% EV-depleted FBS. The cells were treated twice per week with PBS vehicle, 20 g/ml of EVs derived from TCCSUP, T24, or SV-HUC lines, DHA (catalog no. 90310, Caymen), or tunicamycin (catalog no. 11445, Caymen) as described. The cells were passaged approximately weekly as needed and replated to the starting density. The cells were harvested at 8 and 13 weeks and maintained in normal culture for at least 5 weeks before further analysis.

Colony-forming assay
2 ml/well of 0.8% noble agar in serum-free F-12K were added to 6-well plates and allowed to solidify. Cancer EV-treated and control cells were suspended at a density of 1 ϫ 10 5 cells/ml in medium containing 0.4% Noble agar (catalog no. A5431, Sigma-Aldrich) and then placed on top of the 0.8% agarose underlayer. Medium was refreshed twice per week for 6 weeks. The colonies were photographed and counted using ImageJ image analysis software.

EV internalization assay
EVs were labeled with green fluorescent dye by suspending 160 g/ml EVs in PKH67 dye solution (4 ϫ 10 Ϫ6 M; PKH67GL, Sigma-Aldrich) and incubating for 3 min. The staining reaction was stopped by adding 50% EV-depleted FBS in F-12K. EVs were then washed twice in a large volume of DPBS and repelleted by ultracentrifugation at 100,000 ϫ g for 1 h to remove unbound dye. EV pellets were then resuspended in F-12K growth medium and incubated with recipient SV-HUC cells in the presence of EV inhibitors or DMSO vehicle control. Inhibitors Dynasore (10 M) and NSC23766 (5uM) were used alone and in combination. After 6 h, the wells were thoroughly washed twice with DPBS, and then the cells were stained with DAPI and photographed using a Leica DM5000B fluorescence microscope. EV internalization was determined by measuring the integrated density of the PKH67 signal with ImageJ software and then normalizing by the cell number in the field as determined by manually counting the DAPI-stained nuclei.

Xenograft mouse tumor model
The study was approved by the University of Rochester Committee on Animal Resources, and the mice were kept in a specific pathogen-free environment. Young adult female athymic NCr-nu/nu mice (NCI-Frederick) at 8 -10 weeks of age were subcutaneously injected with 1 ϫ 10 6 cancer EV-treated cells into the dorsal flanks. At 12 weeks, the tumors were collected, weighed, and processed for hematoxylin and eosin staining.

Cell-cell contact inhibition assay
To create confluent culture, 2.33 ϫ 10 6 parental SV-HUC cells or 13-week TCCSUP EV-treated SV-HUC cells were seeded in 35-mm dishes. Culture medium was changed to serum-free F-12K medium at day 3 to synchronize cells. At days 1 and 6, both groups were trypsin-dissociated and counted. 2 ϫ 10 6 cells were fixed with ice-cold 70% ethanol in Ϫ20°C overnight and then treated with 0.2 mg/ml RNase in 0.1% Triton X-100 at 37°C for 30 min. The cells were stained with 20 g/ml propidium iodide (PI) (P3596, Life Technologies) in 0.1% Triton X-100. PI intensity was quantified by using flow cytometry, and the cell cycle was analyzed using FlowJo software.

Transwell invasion assay
Transwell polycarbonate membrane inserts with an 8-m pore size (catalog no. 3422, Corning Life Sciences) were coated with growth factor-reduced basement membrane extract (catalog no. 3433-005-02, Trevigen) and incubated for 4 h according to the instructions. Parental SV-HUC cells and cancer EVtreated SV-HUC cells were serum-starved for 18 h and added to each insert at a concentration of 1 ϫ 10 6 /ml. F-12K growth medium with 10% FBS was placed in the bottom well as a chemoattractant. After 16 h, inserts were harvested, fixed in methanol, stained with 1% toluidine blue, and photographed with a Leica MZ9.5 microscope. ImageJ was used to measure the total area of toluidine blue-stained cells on the bottom of each insert.

Extracellular vesicles drive neoplastic transformation Total RNA extraction and qPCR
Total RNA was collected from cells using acid guanidinium thiocyanate-phenol-chloroform extraction (TRIzol) (catalog no. 1596018, Life Technologies) and quantified using spectrophotometry (NanoDrop, Thermo Fisher Scientific). First strand cDNA was synthesized using 1 g of total RNA in a 20-l reaction using the iScript cDNA synthesis kit instructions (Bio-Rad). cDNA levels were measured in triplicate by iQ SYBR Green (Bio-Rad), and relative target expression was normalized to RPLP0. Primer sequences will be provided upon request.

EM
Glass chamber slides with cultured cells were fixed in 0.1 M sodium cacodylate-buffered 2.5% glutaraldehyde for 24 h and then post-fixed in 1.0% buffered osmium tetroxide for 30 min. The plastic chambers were removed from the glass slides for dehydration through a graded series of ethanol to 100% (ϫ3) and infiltration with Spurr's epoxy resin overnight. The next day, several size 3 BEEM capsules filled with fresh resin were inverted and placed over the cells on the glass slide surface and polymerized at 60°C overnight. The polymerized BEEM capsules were "popped off" (62) the glass slide by dipping several times into liquid nitrogen and then gently wiggling them to break the surface tension between the glass surface and the polymerized epoxy capsule. The capsules containing the entrapped cells were trimmed with a razor blade to a small trapezoid face and thin sectioned at 70 nm using a diamond knife and an ultramicrotome. The thin sections were placed onto carbon-coated slot grids and stained with aqueous uranyl acetate and lead citrate. The grids were examined using a Hitachi 7650 transmission EM and an attached Gatan Erlangshen 11 megapixel digital camera for photography.

Proliferation assay
2000 cells/well were seeded in 96-well plates (day 0), rested for 24 h, and then treated with serial dilutions of Dynasore hydrate (D7693, Sigma-Aldrich) and NSC23766 (catalog no. 2161, Tocris) as indicated every other day. Cell growth rate was determined in triplicate using the MTT (M6494, Invitrogen) proliferation assay. 10 l of 12 mM MTT solution were added to each well. After incubation for 2 h at 37°C, culture medium was removed, and the formazan crystals were dissolved in 100 l DMSO. The optical density in each well was measured at 540 nm using a Synergy H1 microplate reader (BioTek Instruments, Inc.).

Nanoparticle tracking analysis
Particle size distribution and concentration in EV isolates were measured using a NanoSight NS300 (Malvern Instruments). Each sample was diluted 1:1000 in DPBS with negligible background signal and recorded into five video files of 30 s each.

Reactive oxygen species level analysis
1 ϫ 10 6 cells of interest were seeded in 6-well plates 24 h before analysis. Cells dissociated by trypsin-EDTA were incubated with 20 M of DCFDA solution (ab13851, Abcam) following protocol instruction. DCFDA fluorescence intensity was detected by using flow cytometry.