Glioma-derived extracellular vesicles promote tumor progression by conveying WT1

Abstract

Glioma persists as one of the most aggressive primary tumors of the central nervous system. Glioma cells are known to communicate with tumor-associated macrophages/microglia via various cytokines to establish the tumor microenvironment. However, how extracellular vesicles (EVs), emerging regulators of cell–cell communication networks, function in this process is still elusive. We report here that glioma-derived EVs promote tumor progression by affecting microglial gene expression in an intracranial implantation glioma model mouse. The gene expression of thrombospondin-1 (Thbs1), a negative regulator of angiogenesis, was commonly downregulated in microglia after the addition of EVs isolated from different glioma cell lines, which endogenously expressed Wilms tumor-1 (WT1). Conversely, WT1-deficiency in the glioma-derived EVs significantly attenuated the Thbs1 downregulation and suppressed the tumor progression. WT1 was highly expressed in EVs obtained from the cerebrospinal fluid of human patients with malignant glioma. Our findings establish a novel model of tumor progression via EV-mediated WT1–Thbs1 intercellular regulatory pathway, which may be a future diagnostic or therapeutic target.

Introduction

Glioma is the most common primary tumor of the central nervous system and persists as one of the most aggressive and lethal tumors due to its high invasion rate and frequent development of drug resistance (1). Among all gliomas, the most prevalent form with the most dismal prognosis is the World Health Organization (WHO) grade IV glioblastoma multiforme (GBM). Current therapies against GBM, such as surgery, radiotherapy and chemotherapy, frequently lead to recurrence, which universally leads to inevitable progression to death with a median survival of only 14.6 months (2). In addition to the rapid growth, high invasiveness and genetic heterogeneity of GBM, the poor survival of GBM patients originates from the poor understanding of the precise mechanism governing disease progression.

Extracellular vesicles (EVs), particularly exosomes, emerge as important regulators of tumor progression, such as in brain tumors, by mediating multiple biological effects (3). EVs are complex structures composed of a lipid bilayer that deliver various proteins, nucleic acids and lipids to their recipient cells, working as vehicles of intercellular communication (4). Tumor cells secrete EVs containing molecular effectors of various processes underlying tumor progression, such as growth, invasion, drug resistance, angiogenesis, metastasis and immune suppression. The uptake of these effectors may change the properties of recipient cells and impact the tumor microenvironment. The roles of EVs in glioma progression are still unclear. We, therefore, examined the roles of glioma-derived EVs in tumor progression.

Materials and methods

Cells, antibodies and human samples

U87, U251 and T98 human glioma cell lines (purchased from ATCC in 2009), and an MG6 mouse microglial cell line (purchased from RIKEN BRC, Japan in 2012) were cultured as described previously (5,6). Normal human astrocytes (NHAs) (purchased from Lonza, Switzerland in 2017) were maintained according to the manufacturer’s instructions. These cell lines were characterized at each resource institute by short tandem repeat profile analysis. They were expanded by culturing them for fewer than two passages and stored as aliquots in liquid nitrogen. Low-passage cells were used for experiments within a period of 6 months after resuscitation. All cells were negative for mycoplasma contamination. Monoclonal antibodies against CD9 (HI9a, BioLegend), CD63 (H5C6, BioLegend), WT1 (D8I7F, CST), TSG101 (4A10, Abcam), GAPDH (3H12, MBL, Japan), β-actin (AC74, MilliporeSigma) and polyclonal antibodies against Iba1 (Wako, Japan) were used. Brain samples and cerebrospinal fluid (CSF) from glioma and control patients were obtained with informed consent under a protocol approved by the Human Research Ethics Committee of Kanazawa University.

Plasmids and transfection

To establish glioma cells deficient in TSG101 or WT1, CRISPR/Cas9 pX330 vectors targeting human TSG101 exon-4 (TAATAGTCATTGAACTAGTAGG) and human WT1 exon-1 (AGCCAGGCGTCATCCGGCCAGG) were constructed (PAM sequences are underlined). Each plasmid was introduced into the cells with ViaFect Transfection Reagent (Promega) to induce frameshift mutations in both alleles. Knockout clones were isolated by limiting dilution and confirmed by DNA sequence and western blot analyses. Control cells were established by transfecting an empty vector. To overexpress Wilms tumor-1 (WT1) in glioma cells, a DNA fragment for a full-length coding sequence of human WT1 was prepared using reverse transcription–polymerase chain reaction with the following primers: WT1-Fw (5′-gatcctggacttcctcttgATGCAG-3′) and WT1-Rv (5′-CTCAAAGCGCCAGCTGGAGTTTGGTC-3′), and was inserted into pMXs retrovirus vector. The plasmid was transfected into PLAT-A packaging cells to produce retrovirus, which was used to infect glioma cells to establish stable transformants.

EV quantification and isolation

To prepare EV samples, the culture medium of glioma cells was changed to the fresh culture medium without fetal bovine serum (FBS) and incubated for 24 h. The collected conditioned culture media were centrifuged at 300g for 10 min, at 1000g for 15 min and at 12 000g for 30 min to remove cell debris and large EVs (mainly microvesicles), and supernatants containing EVs (mainly exosomes) were prepared. To quantify the relative amounts of EVs, the supernatants from glioma cells (5 × 10⁴ cells) were subjected to enzyme-linked immunosorbent assay (ELISA) using PS Capture Exosome ELISA Kit (Wako, Japan) with anti-CD9 or anti-CD63 antibody. To isolate EVs, two methods were performed. For the ultracentrifugation method, the supernatants were further centrifuged at 100 000g for 90 min. The pellets were washed with phosphate-buffered saline (PBS) and recentrifuged at 100 000g for 90 min, again. The pellets were suspended in PBS and stored at −80°C until use. To obtain highly purified EVs for RNA-Seq and quantitative PCR (qPCR) analyses, the MagCapture Exosome Isolation Kit PS (Wako, Japan) (7) was used as an alternative method. The concentration and size distribution of EVs were determined by nanoparticle tracking analysis using NanoSight (Malvern, UK).

EV uptake assay

EV uptake assay was performed as described previously (6). In brief, U87 cell-derived EVs were labeled with PKH26 (MilliporeSigma). MG6 cells (1 × 10⁵ cells) were cultured in eight wells with fresh Dulbecco’s modified Eagle’s medium with 3% EV-free FBS and were incubated with PKH26-labeled EVs (1 × 10¹⁰ particles) for 3 h. The cells were fixed with 4% paraformaldehyde and were observed by BZ-X710 (KEYENCE, Japan) or analyzed by BD FACSCanto II.

In vitro scratch wound healing assay

Glioma cells were seeded in a 12-well plate until confluency, and monolayers of cells were scratched with a 200-µl pipette tip to induce a wound. Cells were washed twice with PBS to remove detached cells and smooth the edge of the wound. After that, the cells were cultured in FBS-free medium. The wounded edges were imaged, and the area of the wound was measured at each time point, at 0, 12, 24 and 36 h, using BZ-X710. The percentage of the difference between the area of the wound at the 0 h time point and area at the time point being analyzed was calculated.

Mouse model of glioma

U87 cells (1 × 10⁵ cells) were injected into the brain parenchyma 3 mm down from the cerebral seam of 8–10-week-old BALB/c nude mice (Charles River Laboratories, Japan). Implantation of either TSG101 KO1 or control U87 cells were randomized and blinded. After 3 weeks, all mice were euthanized. In some experiments, TSG101 KO1 cells mixed either with PBS or EVs (1 × 10¹⁰ particles) collected from control U87 cells were injected, and in other experiments, a mixture of U87 cells (1.5 × 10⁵ cells) and T98 cells (1.5 × 10⁵ cells) was injected. In these experiments, the mice were euthanized after 4 weeks. Brain tissues were dissected, fixed with 4% paraformaldehyde, embedded in paraffin, and then cut into 4-µm serial coronal sections. The tissue sections were stained using a standard hematoxylin and eosin staining technique. The surface included by the tumor contour of the region of interest in the coronal section was calculated as the maximal area of each tumor. To evaluate angiogenesis, microvessel density was quantified as described previously (8). In brief, we quantified the number of CD34⁺ microvessels in ten random areas of tumor microenvironment within 1 µm from a tumor. All animal experiments followed the Guidelines for the Care and Use of Laboratory Animals at Kanazawa University that covers the national guidelines.

RNA-Seq and qPCR analyses

EVs (1 × 10¹⁰ particles) were added to MG6 cells (1 × 10⁵ cells) that were previously cultured in 24-well plates with Dulbecco’s modified Eagle’s medium and 3% EV-free FBS and incubated for 12 h. Total RNA was collected using RNeasy Plus Mini Kits (QIAGEN). RNA-Seq was performed using HiSeq4000 (Illumina), and gene expression difference analysis was performed using CLC Genomics Workbench (QIAGEN). For qPCR analyses, RNA was reverse transcribed using ReverTra Ace qPCR master mix (TOYOBO, Japan), and complementary DNA products were amplified using a LightCycler 96 (Roche) with Universal SYBR Select master mix (ThermoFisher). Data were analyzed using the delta Ct method and normalized to GAPDH RNA expression in each sample. The primers for real-time PCR were as follows: Gm20405-Fw (5′-ACACACACACCGAGAGGATG-3′) and Gm20405-Rv (5′-TTGTCAGGGACAGCCAGAC-3′); Nr4a-Fw (5′-CTGTCCGCTCTGGTCCTC-3′) and Nr4a-Rv (5′-AATGCGATTCTGCAGCTCTT-3′); Thbs1-Fw (5′-CACCTCTCCGGGTTACTGAG-3′) and Thbs1-Rv (5′-GCAACAGGAACAGGACACCTA-3′).

Statistical analyses

Statistical analyses were performed with SPSS 23.0 (IBM). The data obtained from triplicate samples were expressed as the mean ± SD. Log-rank tests and Student’s t-tests were used to determine P values. Continuous variables in figures are shown as the mean ± SEM. For all measurements, two-tailed tests were performed, and P < 0.05 was considered statistically significant. All experiments were performed at least three times.

Results

Glioma-derived EVs promote tumor progression

As the potential roles of EVs in brain tumors have been described (9), we examined the effect of TSG101-deficiency on brain tumor cells, which is one of the major molecules involved in EV production (10,11). We introduced a CRISPR-Cas9 targeting vector against TSG101 or an empty vector into U87 human glioma cells and established two TSG101^−/− clones (TSG101 KO1 and KO2) and control cells (Figure 1A). The EVs secreted by these TSG101 KO cells decreased to half those secreted by control cells as determined by ELISA using antibodies against CD9 or CD63 EV markers and nanoparticle tracking analysis (Figure 1B), consistent with previous findings (10). To analyze the effect of TSG101-deficiency on cell growth ability, we performed an in vitro cell proliferation assay and found no decrease between TSG101 KO1 or KO2 and the corresponding control cells (Figure 1C). We further performed an in vitro scratch wound healing assay and found no difference between TSG101 KO1 and their control cells, while TSG101 KO2 cells showed a slightly lower ability of wound healing (Figure 1D). These data suggest that TSG101-mediated EV secretion has a minimal effect on in vitro cell growth and migration abilities under these assay conditions.

Figure 1.

Glioma progression in vivo is dependent on EV secretion. (A) Whole-cell lysates of TSG101^−/− U87 cells (KO1 and KO2 cells) and control U87 cells were immunoblotted with anti-TSG101 or anti-β-actin antibody. (B) EV secretion was analyzed by ELISA with anti-CD9 or anti-CD63 antibody, and nanoparticle tracking analysis. *P < 0.01, versus control cells, Student’s t-test. (C) Control (red squares), TSG101 KO1 (blue circles) and KO2 (green circles) cells were analyzed by an in vitro proliferation assay. Proliferation was determined as the average of the number in three wells. (D) The migratory ability of these cells was analyzed by an in vitro scratch wound healing assay. The percentage of the difference between the area of the wound at the 0 h time point and the area at the time point being analyzed was calculated. *P < 0.05, Student’s t-test. (E) Mouse models of glioma were established by implanting control or TSG101 KO1 cells (1 × 10⁵ cells) into mouse brains. After breeding the mice for 3 weeks, their brains were dissected and fixed with 4% paraformaldehyde. Thinly sliced sections of paraffin-embedded blocks were subjected to hematoxylin and eosin (HE) staining. A primary tumor that grew near an injection site is shown with black arrowheads, while a secondary tumor that metastasized from the primary tumor is shown with white arrowheads. (F) The areas of tumors in coronal sections showing the maximal area of each tumor (primary and secondary tumors) were measured (three mice for each group). *P < 0.05, Student’s t-test. (G) Mice injected with each cell type (control or TSG101 KO1 cells) were bred for up to 5 weeks, and their survival was checked every day (nine mice for each group). *P < 0.01, log-rank test. (H) TSG101 KO1 cells (1 × 10⁵ cells) were implanted into mouse brains either with PBS or EVs (1 × 10¹⁰ particles) collected from control U87 cells. Four weeks later, their brains were dissected and stained with HE. Sections with the maximal area of each primary tumor were shown, and the sizes were plotted in a graph (two mice for each group).

We next examined whether the decrease in EV secretion affects tumor progression in vivo. We implanted either TSG101 KO1 or control cells into mouse brains following a previous protocol (12) and compared the size of ‘the primary tumor’ around the injection site and ‘the secondary tumors’ that metastasized from the primary tumor (Figure 1E). We found that the primary tumors derived from TSG101 KO1 cells were 7.35-fold smaller than the tumors derived from control cells (control: 2.06 ± 0.56 mm², KO1: 0.28 ± 0.25 mm², P < 0.05) and that the secondary tumors derived from TSG101 KO1 cells were 13.3-fold smaller than the tumors derived from control cells (control: 1.06 ± 0.47 mm², KO1: 0.08 ± 0.03 mm², P < 0.05) (Figure 1F). Accordingly, the survival of mice implanted with TSG101 KO1 cells was 6 days longer than those implanted with control cells (control: 21.9 ± 0.8 days, KO1: 27.3 ± 0.7 days, P < 0.01) (Figure 1G). Furthermore, when TSG101 KO1 cells were coimplanted with EVs collected from control cells, the primary tumors became 5.28-fold larger than those with PBS (EV: 6.44 mm², PBS: 1.22 mm²) (Figure 1H), indicating that glioma-derived EVs enhance tumor progression in vivo.

Glioma-derived EVs promote microglial recruitment and angiogenesis

To analyze the effect of tumor cell implantation in the brain, number and the sizes of microglia in the tumor regions were compared. It revealed an increase in the expression of microglia around the implantation site of U87 cells (Figure 2A), which was not induced by sham injection (Supplementary Figure S1, available at Carcinogenesis Online). On the other hand, implantation of TSG101 KO1 cells caused fewer microglial recruitment, smaller microglial sizes and less angiogenesis than control cells in size-matched tumors (Figure 2B and C), but coimplantation with EVs collected from control cells rescued the decrease (Figure 2D and E), suggesting that brain tumors can affect the surrounding microenvironment via microglia through EVs. Taken together, these data suggest that the decrease in EV production does not directly affect tumor growth per se but rather affects the surrounding microenvironment in brain tumors. In other words, glioma-derived EVs are likely involved in the growth and progression of brain tumors by affecting microglia.

Figure 2.

Modulation of tumor microenvironment by glioma-derived EVs. (A) Brain samples of glioma model mice injected with U87 cells were immunostained with an anti-Iba1 antibody (green) and counter-stained with 4′,6-diamidino-2-phenylindole (blue). Scale bar, 100 µm. (B) After implantation of each cell type (control or TSG101 KO1 cells), number and the sizes of microglia in the tumor regions were compared by immunostaining with an anti-Iba1 antibody (green). Scale bar, 50 µm. (C) Microvessel density was determined by the number of CD34⁺ microvessels in ten random areas of each tumor microenvironment. *P < 0.01, Student’s t-test. (D) After implantation of TSG101 KO1 cells either with PBS or EVs from control U87 cells, number and the sizes of microglia in the tumor regions were compared. Scale bar, 100 µm. (E) Microvessel densities in ten random areas of each tumor microenvironment were plotted in a graph. *P < 0.01, Student’s t-test.

Glioma-derived EVs modulate microglial gene expression

As tumor-associated macrophages/microglia (TAMs) are known to be key regulators of the tumor microenvironment (13), we examined how glioma-derived EVs affect microglial functions. First, we modeled the transfer of EVs from glioma to microglial cells in vitro. When EVs from U87 cells were labeled with PKH26 dye and cocultured with MG6 mouse microglial cells, 71.4% of microglial cells engulfed the glioma-derived EVs (Figure 3A and B). Next, to investigate the effect of glioma-derived EVs on microglia, we performed genome-wide analyses of gene expression in microglia cells with or without glioma-derived EVs. As a comparison, other human glioma cells, such as U251 and T98, were also used as a source of EVs, and commonalities were investigated. We found that the expression of several genes was commonly downregulated when microglia were treated with different glioma-derived EVs (Figure 3C and Supplementary Table S1, available at Carcinogenesis Online). To narrow down candidates for microglial effectors, we picked up three genes: Gm20425 and Nuclear Receptor Subfamily 4 Group A (Nr4a) because they were downregulated by multiple sources of EVs, and thrombospondin-1 (Thbs1) because thrombospondin-1 (TSP-1) protein is known as a negative regulator of angiogenesis (14) and TAMs are major inducers of tumor angiogenesis (15). To confirm gene expression, we performed qPCR analyses and found that the expression of the three genes was actually downregulated in microglia, which had been incubated with any kind of glioma-derived EVs (Figure 3D). Notably, expression of the Thbs1 gene in microglia was strikingly decreased; furthermore, we found that the decreasing effect was glioma-specific and not observed when microglial cells were incubated with EVs derived from NHAs (Figure 3E), although microglia engulfed NHA-derived EVs as efficiently as glioma-derived EVs (Supplementary Figure S2, available at Carcinogenesis Online). Our results suggest that glioma-derived EVs promote tumor progression by facilitating angiogenesis via the downregulation of Thbs1 gene expression in microglia.

Figure 3.

Microglial genes downregulated by the uptake of glioma-derived EVs. (A) Glioma-derived EVs (1 × 10¹⁰ particles) were labeled with PKH26 (red) and added to cultured MG6 cells (1 × 10⁵ cells). After 3 h, the cells were fixed with 4% paraformaldehyde and counter-stained with 4′,6-diamidino-2-phenylindole (blue). Scale bar, 40 µm. (B) The proportion of MG6 cells that engulfed PKH26-labeled EVs was quantified by fluorescence-activated cell sorting analysis. (C) EVs from U87, U251 and T98 cells (1 × 10¹⁰ particles) were added to cultured MG6 cells (1 × 10⁵ cells) on a 24-well plate and incubated for 12 h. Total RNA was collected and analyzed by RNA sequencing. The numbers of genes commonly downregulated in cells treated with EVs from different glioma cell lines are shown. (D) EVs isolated from each glioma cell line (U87, U251 and T98) were added to MG6 cells. After 12 h of incubation, total RNA was collected from the MG6 cells and reverse transcribed. qPCR analyses were performed to examine the expression of genes indicated, which was normalized to that of GAPDH. MG6 cells without addition of EVs were used as a control of the relative expression. *P < 0.01, **P < 0.05, versus control sample, Student’s t-test. (E) EVs from NHAs or U87 cells were added to MG6 cells, and the relative gene expression of the Thbs1 against control was examined. *P < 0.01; n.s., not significant, versus control sample, Student’s t-test.

WT1 in EVs downregulates microglial Thbs1 gene expression

To elucidate the signaling pathway mediated by brain tumor-derived EVs, we next investigated which molecules in glioma-derived EVs are responsible for the downregulation of Thbs1 gene expression. Several molecules including transcription factors have been reported that tightly regulate the Thbs1 gene expression (16). We, therefore, examined the protein expression of these candidates in EVs and found that WT1 was expressed in EVs from U87, U251 and T98 human glioma cells but not in EVs from NHAs (Figure 4A). In contrast, we could not detect other Thbs1 regulators, including ATF1, c-Jun and YY1, in the glioma-derived EVs (data not shown). As the downregulation of Thbs1 gene expression in microglia was most strongly induced by T98 cell-derived EVs (Figure 3D), we established WT1^−/− T98 cells by introducing a CRISPR-Cas9 targeting vector against WT1 (Figure 4B) and confirmed the loss of WT1 expression in the EVs (Supplementary Figure S3A, available at Carcinogenesis Online). We also established cell transformants in which WT1 was overexpressed in WT1^+/+ T98 cells or re-expressed in WT1^−/− T98 cells (Figure 4C) and confirmed the overexpression of WT1 in the EVs (Supplementary Figure S3B, available at Carcinogenesis Online). When cell proliferation was compared among these cells, the deficiency or overexpression of WT1 did not affect cell growth in vitro (Figure 4D). We next collected EVs from these cells and examined whether they downregulated Thbs1 gene expression in microglia. As mentioned above, EVs from WT1^+/+ T98 cells suppressed the expression of the Thbs1 gene, but overexpression of WT1 did not enhance the suppression, suggesting that endogenous expression of WT1 was sufficient for this effect (Figure 4E). In contrast, the downregulation of Thbs1 gene expression was partially attenuated by EVs from WT1^−/− T98 cells, implicating that there might be other factors in EVs which also regulate Thbs1 expression. However, re-expression of WT1 in WT1^−/− T98 cells rescued the suppressive effects of EVs. By using these cells, we next investigated the effect of WT1 in EVs on tumor progression. We noticed that T98 cells could not form tumors when implanted into mice (data not shown), which was rather advantageous for using these cells as a source of EVs to examine their effects on the progression of other tumor cells such as U87 cells. In addition, T98 cells were found to secret nine times more EVs than U87 cells in vitro (Supplementary Figure S3C, available at Carcinogenesis Online). We, therefore, mixed U87 cells and T98 cells at a ratio of 1:1 and examined how the progression of U87 cell-derived tumors can be regulated by T98 cell-derived EVs. As shown in Figure 4F, WT1-deficiency in T98-derived EVs significantly suppressed the progression of U87 cell-derived tumors, which was reversed by the re-expression of WT1. Furthermore, angiogenesis in the surrounding areas of these tumors was correlated to the expression levels of WT1 in EVs (Figure 4G). On the other hand, neither the deficiency nor the overexpression of WT1 in EVs dramatically changed the number and the sizes of microglia (Supplemental Figure S3D, available at Carcinogenesis Online), indicating that other factors might be involved in these processes. Taken together, these data suggest that WT1 in glioma-derived EVs promotes tumor progression by inhibiting Thbs1 gene expression in microglia, which can affect the tumor microenvironment.

Figure 4.

WT1 in glioma-derived EVs represses Thbs1 gene expression in microglia. (A) The expression of the WT1 protein in EVs from each cell line was analyzed by western blotting with anti-WT1, anti-CD9 or anti-CD81 antibody. A total of 1 × 10¹⁰ EV particles were added to each lane. (B) WT1^−/− T98 cells were established with the CRISPR/Cas9 system, and the loss of WT1 expression in the cells was confirmed by western blotting. (C) T98 transformants (WT1^+/+ or WT1^−/−) that stably overexpress exogenous WT1 were established and shown as OE. The expression of WT1 in these cells was compared by western blotting. (D) In vitro proliferation of each T98 cell line (WT1^+/+ (red), WT1^−/− (green), OE-WT1^+/+ (blue) or OE-WT1^−/− (yellow)) was compared. (E) EVs (1 × 10¹⁰ particles) from each T98 cell line (WT1^+/+, WT1^−/−, OE-WT1^+/+ or OE-WT1^−/−) were added to the cultured MG6 cells (1 × 10⁵ cells) and incubated for 12 h. And quantitative reverse transcription PCR analyses of Thbs1 gene expression in the MG6 cells were performed. MG6 cells without addition of EVs were used as a control of the relative expression. *P < 0.01, **P < 0.05; n.s., not significant, Student’s t-test. (F) U87 cells (1.5 × 10⁵ cells) were mixed with an equal number of each T98 cell line (WT1^+/+, WT1^−/− or OE-WT1^−/−), and implanted into mouse brain (five mice for each group). After breeding the mice for 4 weeks, the size of U87-derived primary tumors was compared. *P < 0.05; n.s., not significant, Student’s t-test. (G) CD34⁺ microvessel density in the tumor microenvironment of mouse brain implanted with each T98 cell line (WT1^+/+, WT1^−/− or OE-WT1^−/−) was compared (three mice for each group). *P < 0.01; n.s., not significant, Student’s t-test.

Expression of WT1 in CSF-derived EVs of patients with glioma

To investigate whether WT1 is actually expressed in EVs from human samples, we obtained CSF of patients with glioma by lumbar puncture and analyzed them. We found the expression of WT1 in CSF-derived EVs collected from patients with various types of glioma (primary GBM, secondary GBM and diffuse intrinsic pontine glioma) but not in CSF-derived EVs from a patient with brain tumors that had metastasized from breast cancer (Figure 5A). We further confirmed the expression of WT1 in the tumor lesion of a GBM patient but not in the non-tumor lesion or the brain of a control patient (Figure 5B), suggesting that WT1 can be specifically expressed in primary glioma and promotes tumor progression by being transferred to microglia via EVs.

Figure 5.

Specific expression of WT1 in CSF-derived EVs from glioma patients. (A) Expression of WT1 in EVs isolated from patient CSF (5 ml) was analyzed by western blotting with anti-WT1 or anti-CD9 antibody. WT1 was expressed only in EVs from glioma patient (Pt. 1: secondary GBM, Pt. 2: primary GBM, Pt. 3: metastatic brain tumor of breast cancer, Pt. 4: diffuse intrinsic pontine glioma). (B) Cell lysates of brain samples from patients were analyzed by western blotting with anti-WT1 or anti-GAPDH antibody. WT1 was highly expressed in the tumor lesion of a GBM patient (Pt. 2) (lane 3) but not in the non-tumor lesion of the same GBM patient (lane 2). WT1 was not expressed in the normal brain of a control patient (lane 1). (C) The overview of our study. In glioma cells, WT1 is highly expressed and can be incorporated into intraluminal vesicles that are formed inside multivesicular endosomes. Glioma cells secrete EVs that contain the WT1 protein, and these EVs are taken up by tumor-associated microglia. In the microglia, the WT1 represses the Thbs1 gene expression, which may lead to enhance angiogenesis in glioma.

Discussion

In this study, we propose a novel mechanism by which glioma cells establish their microenvironment through EV-mediated transfer of a transcription factor WT1. The WT1 gene was originally considered to mainly function as a tumor suppressor gene in Wilms’ tumor of kidneys (17). However, it was revealed that the WT1 gene acts as an oncogene in many cancers. Although WT1 is predominantly a nuclear protein in normal cells and tissues, it is mostly expressed in the cytoplasm in the majority of WT1-expressing tumors (18). Therefore, WT1 can be encapsulated in intraluminal vesicles inside multivesicular endosomes and be secreted as an EV component. WT1 in glioma-derived EVs serves as a potent promoter of tumor progression by inhibiting Thbs1 gene expression in microglia, thus working as an intercellular messenger that regulates tumor microenvironment (Figure 5C).

It has been reported that glioma-derived EVs can affect endothelial cells to enhance angiogenesis by delivering proangiogenic factor VEGF-A or long non-coding RNA (19–21). On the other hand, the main effector cells in our model are microglia. The glioma–microglia crosstalk has been extensively studied (22,23), and glioma cells are known to secrete various chemokines (e.g. CCL2, CX3CL1 and SDF-1) and cytokines (e.g. GM-CSF, CSF-1 and TGF-β) to attract microglia and turn them into a protumorigenic phenotype, which plays various roles in glioma progression including proliferation, survival, motility and immunosuppression. It would be intriguing to examine whether these chemokines and cytokines stimulate the microglial uptake of glioma-derived EVs, which might cause synergistic effects.

TSP-1 is a multifunctional extracellular matrix protein, and the antiangiogenic properties of TSP-1 are well established (14). TSP-1 inhibits the proliferation and migration of endothelial cells by upregulating proapoptotic proteins such as Fas ligand (24). Although TSP-1 levels are normally low in the adult brain, activated microglia and reactive astrocytes express this protein to promote neurite growth and regeneration (25,26). In GBM, the number of TAMs can be very high and constitute up to 30% of the tumor mass (27). Therefore, glioma cells might deliver WT1 to TAMs to prevent the production of large amounts of TSP-1. WT1 is known to repress the Thbs1 transcription by binding to -210 region of its promoter (28). Although TSP-1 has been shown to reduce GBM growth and vascularity (29), the roles of TSP-1 are complex. Notably, a recent study has been demonstrated that TSP-1 secreted by GBM promotes tumor cell invasion and growth by interacting with CD47 on the tumor cells (30). Therefore, upregulating the TSP-1 expression specifically in microglia by blocking EV-mediated convey of WT1 from GBM to inhibit angiogenesis would be a good anti-GBM therapeutic strategy.

The expression of WT1 in high-grade glioma has been validated in a clinical study (31), and a phase I trial of dendritic cell-based immunotherapy targeting WT1 in patients with recurrent malignant glioma has been reported (32). The WT1 gene expresses at least 11 isoforms and encodes various proteins that can activate or inhibit each target gene, leading to proliferation, differentiation and apoptosis (33); however, the difference among WT1 isoforms has rarely been investigated in clinical studies in glioma. We found that only long-form of WT1 (70 kDa) was expressed in EVs from CSF in glioma patients, although it is not highly expressed within the cells compared with short-form (55 kDa) (Figure 5A and B), which may suggest that only the long-form of WT1 is selectively incorporated into EVs to provide tumor-promoting effects for tumor-associated cells. In any case, as brain tumorigenesis was found to be regulated through the EV-mediated WT1-Thbs1 regulatory pathway, WT1 in EVs from CSF can be a relatively non-invasive diagnostic marker for glioma and be used as a target for the treatment of patients with glioma.

Supplementary material

Supplementary data are available at Carcinogenesis online.

Supplementary Table S1. A list of microglial genes downregulated by the uptake of glioma-derived EVs

Supplementary Figure S1. After sham injection, the brain samples were HE stained or immunostained with an anti-Iba1 antibody (green) and DAPI (blue). Scale bar, 100 µm.

Supplementary Figure S2. NHA-derived EVs (1 × 10¹⁰ particles) were labeled with PKH26 (red) and added to cultured MG6 cells (1 × 10⁵ cells) for 3 h. Scale bar, 40 µm.

Supplementary Figure S3. (A and B) WT1 expression in EVs or the amount of EVs from the T98 cell line (WT1^+/+, WT1^−/−, OE-WT1^+/+ or OE-WT1^−/−) was compared by western blotting with anti-WT1 or anti-ALIX antibody, respectively. (C) The concentration of EVs secreted from U87 or T98 cells was compared by nanoparticle tracking analysis. *P < 0.05, Student’s t-test. (D) Brain samples of glioma model mice injected with each T98 cell line (WT1^+/+, WT1^−/− or OE-WT1^−/−) were immunostained with an anti-Iba1 antibody (green). Scale bar, 200 µm.

Abbreviations

Abbreviations
 
• CSF

  cerebrospinal fluid

 
• EV

  extracellular vesicle

 
• FBS

  fetal bovine serum

 
• GBM

  glioblastoma multiforme

 
• PBS

  phosphate-buffered saline

 
• WT1

  Wilms tumor-1

Funding

This work was supported by Core Research for Evolutional Science and Technology (CREST) from Japan Science and Technology Agency (JST) (No. JPMJCR18H4 to R.H.), Grants-in-Aid for Scientific Research (KAKENHI) from the Ministry of Education, Culture, Sports, Science and Technology (MEXT) (No. 25713020 and 19K22578 to R.H., No. 18H02910 to M.N., No. 15K06710 to H.K. and No. 17K16635 to T.T.).

Acknowledgements

We thank Y. Okayasu for secretarial assistance. We are grateful to Dr S. Horike at the Advanced Science Research Center, Kanazawa University for technical support.

Conflict of Interest Statement: None declared.