The Adenosine A3 Receptor Regulates Differentiation of Glioblastoma Stem-Like Cells to Endothelial Cells under Hypoxia

Glioblastoma (GBM) is a neoplasm characterized by an extensive blood vessel network. Hypoxic niches of GBM can induce tumorigenic properties of a small cell subpopulation called Glioblastoma stem-like cells (GSCs) and can also increase extracellular adenosine generation which activates the A3 adenosine receptor (A3AR). Moreover, GSCs potentiates the persistent neovascularization in GBM. The aim of this study was to determine if A3AR blockade can reduce the vasculogenesis mediated by the differentiation of GSCs to Endothelial Cells (ECs) under hypoxia. We evaluated the expression of endothelial cell markers (CD31, CD34, CD144, and vWF) by fluorescence-activated cell sorting (FACS), and vascular endothelial growth factor (VEGF) secretion by ELISA using MRS1220 (A3AR antagonist) under hypoxia. We validate our results using U87MG-GSCs A3AR knockout (GSCsA3-KO). The effect of MRS1220 on blood vessel formation was evaluated in vivo using a subcutaneous GSCs-tumor model. GSCs increased extracellular adenosine production and A3AR expression under hypoxia. Hypoxia also increased the percentage of GSCs positive for endothelial cell markers and VEGF secretion, which was in turn prevented when using MRS1220 and in GSCsA3-KO. Finally, in vivo treatment with MRS1220 reduced tumor size and blood vessel formation. Blockade of A3AR decreases the differentiation of GSCs to ECs under hypoxia and in vivo blood vessel formation.


Introduction
Glioblastoma (GBM) is considered the most common tumor of the central nervous system and one of the most devastating types of cancer. After multimodal therapy, consisting of surgical resection followed by radio-and chemo-therapy with temozolomide, about 99% of cancer cells are eliminated; however, the tumor recurs and, as a result, patients die on average within 15 months [1][2][3]. Therapy failure is mainly attributed to a cell subpopulation called glioblastoma stem-like cells (GSCs) which, like normal neural stem cells (NSCs), have unlimited self-renewal and a multi-lineage differentiation capacity. In addition, GSCs have the potential to form in vivo tumors and exhibit a higher resistance to therapy than differentiated cancer cells [4][5][6][7]. Therefore, strategies are currently aimed at eliminating GSCs or promoting their differentiation to less aggressive phenotypes. A key feature of GBM is the presence of extensive blood vessel networks that support both tumor growth and resistance to treatment [8,9]. Unlike normal vasculature where the endothelium has a low proliferation rate, tumor vasculature is highly proliferative and the formed blood vessels are often disorganized and tortuous [10]. This entails inefficient oxygen supply and as a result, extensive hypoxic areas that promote further vasculogenesis [11]. In fact, the hypoxia inducible factor 1α (HIF-1α) induces the expression of vascular endothelial growth factor (VEGF), which in turn stimulates proliferation and migration of endothelial cells (ECs) [12][13][14][15]. GSCs are enriched in hypoxic areas within the tumor which promote both their stemness and radio-and chemo-resistance [16][17][18][19], therefore, future therapies targeting GSCs should consider this niche. In addition, like GSCs, tumor-derived ECs are found in a greater proportion in internal and hypoxic tumor regions. Importantly, GSCs can differentiate into ECs, a phenomenon where hypoxia appears to be critical [20][21][22]. However, the mechanisms and pathways that control differentiation of GSCs to ECs are not fully understood. Signaling pathways modulated by adenosine are up-regulated under hypoxia and participate in various aspects of cancer, including vasculogenesis [23][24][25][26][27]. This nucleoside is produced mainly by extracellular ATP hydrolysis and signals through its four Adenosine Receptors (ARs): A 1 (A 1 AR), A 2A (A 2A AR), A 2B (A 2B AR), and A 3 (A 3 AR) [26]. We showed that GBM cells have increased extracellular adenosine levels and exhibit high expression of A 3 AR compared to non-tumoral normal cells [28]. Similar results were observed in GSCs, where extracellular adenosine levels and A 3 AR expression are higher than in differentiated GBM cells [29,30]. However, the role of this receptor on GSCs biology is poorly understood. Therefore, our aim was to determine if A 3 AR blockade can reduce the vasculogenesis mediated by the differentiation of GSCs to ECs under hypoxia.

Extracellular Adenosine Concentration and A 3 Adenosine Receptor Expression Increase under Hypoxia
U87MG GSCs were cultured under 0.5% O 2 in order to evaluate hypoxia effect on extracellular adenosine generation and A 3 AR expression. We found that U87MG GSCs had increased extracellular adenosine production (~7 fold) after 24 h of hypoxia ( Figure 1A). A 3 AR expression increased in U87MG GSCs under hypoxia ( Figure 1B,C). Similarly, the percentage of A 3 AR-positive GSCs increased under hypoxic conditions ( Figure 1D). These results suggest that the high levels of extracellular adenosine in U87MG GSCs culture could activate the A 3 AR under hypoxia. (A) Extracellular adenosine concentration in U87MG glioblastoma stem-like cells (GSCs) under hypoxia. U87MG GSCs were exposed to hypoxia for 24 h. Adenosine concentrations (nM) were normalized to total protein concentration (µg); (B) Western blot of HIF-1α and A 3 adenosine receptor (A 3 AR) expression in U87MG GSCs under normoxia and hypoxia for 24 h; (C) Flow Cytometry analysis of the mean fluorescence intensity (M.F.I.) of A 3 AR expression in U87MG GSCs under normoxia and hypoxia for 24 h; (D) Flow Cytometry graph of A 3 AR-positive U87MG GSCs (left panel) and a representative Flow Cytometry histogram (right panel) under normoxia and hypoxia for 24 h. Graphs represent the mean ± standard deviation (S.D.). * p < 0.05; ** p < 0.01; *** p < 0.001 normoxia versus hypoxia (24 h). n = 3.

Differentiation of Glioblastoma Stem-Like Cells to Endothelial Cells Increases under Hypoxia
To evaluate the effect of hypoxia on the differentiation of GSCs to ECs we evaluated the expression of endothelial cell markers (CD31, CD34, CD144, and vWF) and VEGF secretion. No differences were observed in the expression of endothelial markers through Flow Cytometry between U87MG GSCs ( Figure 2A). However, the percentage of positive cells for CD34 and vWF increased after 24 h of hypoxia ( Figure 2B,C). To evaluate VEGF secretion in GSCs under hypoxia, we evaluated the presence of VEGF-165 in U87MG GSCs medium during 72 h of hypoxia. We observed an increase in VEGF-165 secretion at 48 (~2 fold) and 72 (~2.7 fold) hours under hypoxia ( Figure 2D). These results suggest that U87MG GSCs could differentiate into ECs, especially under hypoxia. These results propose that hypoxia promotes the expression of endothelial cell markers and the secretion of VEGF in GSCs.

A 3 AR Blockade Decreases Differentiation of Glioblastoma Stem-Like Cells to Endothelial Cells under Hypoxia
We explored the effect of A 3 AR blockade on the differentiation of GSCs to ECs under hypoxia. Cells were treated with MRS1220, a selective A 3 AR antagonist, under hypoxia and then the expression of endothelial cell markers and VEGF secretion were analyzed. A 3 AR blockade did not change the expression of endothelial markers ( Figure 3A), nevertheless, decreased the percentage of CD31, CD144, and vWF positive GSCs after 24 h under hypoxic conditions ( Figure 3B,C). VEGF secretion in U87MG GSCs decreased~25% with MRS1220 after 72 h of hypoxia ( Figure 3E). To validate the effect of MRS1220 in U87MG GSCs differentiation to ECs, we used an A 3 AR knockout cell line (GSCs A3-KO ) to evaluate its intrinsic differentiation ability to ECs under hypoxia. Similarly, we observed a decreased percentage of CD31, CD144, and vWF positive cells ( Figure 3B,D), and an almost total decrease in VEGF secretion ( Figure 3E) in GSCs A3-KO under hypoxia. These results suggest that the ability of U87MG GSCs to differentiate into ECs could be regulated by A 3 AR activation under hypoxia.

In Vivo Antagonization of A 3 AR Decreases Tumor Size and Blood Vessel Formation
To ensure tumor growth, the formation of blood vessels that supply oxygen and nutrients to neoplastic cells is crucial [27]. Increased tumor volume is linked to a larger network of blood vessels, which is why in recent years the generation of new anti-angiogenic therapies has been sought [31]. To evaluate the in vivo effect of A 3 AR antagonization, we generated an allogeneic rat subcutaneous tumor using GSCs from the rat C6 glioma cell line. At day ten post-inoculation with C6 GSCs, we treated animals with MRS1220 for fifteen days. We observed a reduction close to 80% and 90% in tumor volume compared to the vehicle-treated group at day ten and fifteen post-treatment, respectively ( Figure 4A). The histopathological analysis showed extensive necrotic areas with blood vessel formation, which was reverted after pharmacological blockade of A 3 AR. The number of blood vessels per field was reduced by three times with MRS1220 ( Figure 4B), indicating a strong in vivo anti-angiogenic effect. Original magnification ×20 (H&E); Arrows indicate the location of blood vessels. Counting the amount of blood vessels per field in vehicle and MRS1220 treated groups are represented (right panel). Graphs represent the mean ± S.D. * p < 0.05; vehicle versus MRS1220. n = 3.

Discussion
The prognosis for GBM treatment is worsened by the presence of GSCs due to their self-renewal and cell differentiation properties, for example to endothelial cells (ECs), promoting angiogenesis and neovascularization [32]. Since GSCs are highly chemo-and radio-resistant, they are maintained in tumor niches even after treatment. This sustained maintenance and increased neovascularization is linked to the high extracellular adenosine concentrations found in tumors and even higher concentrations in hypoxic niches [27]. AR subtypes have different affinities to adenosine; A 1 and A 2A are high affinity receptors and A 2B and A 3 are low affinity receptors; therefore, in pathological conditions, such as cancer, extracellular adenosine levels are increased and mainly activate A 2B AR and A 3 AR subtypes enhancing several signaling pathways, such as PI3K/AKT, MAPK, among others [27,29,33].
Knockout of ectonucleotidase CD73, which is important for extracellular adenosine production from AMP, decreased angiogenesis in melanoma models: a process that is reverted when using different AR agonists [34]. The overall effect was greater when inhibiting adenosine production compared to AR blockade, concluding that the AR subtypes have a summatory effect on angiogenesis. Several studies have confirmed that A 3 AR is important to angiogenesis in different tumors, specifically in the generation of blood vessels and neovascularization [34][35][36]. In this study, we proposed that adenosine regulates the differentiation of GSCs to ECs in vitro through A 3 AR activation, and that this promotes the formation of new blood vessels in an in vivo GBM tumor model. GSCs produce more adenosine than differentiated cells [29], and their production is enhanced under hypoxia in other tumor models [27]. In this study we showed that GSCs not only enhance adenosine production but also increase A 3 AR expression, suggesting a loop of positive regulation between the AR and its ligand; probably through expression of the transcription factor HIF-1α, which increases during early hypoxia [32]. Hypoxia-promoted cell differentiation and subsequent expression of HIF-1α have been described in different models [37], however this is poorly understood in GSCs. In this study, we showed for the first time that hypoxia increased A 3 AR expression, and its blockade decreases the cell population positive to several endothelial cell markers, such as CD34, CD144, and vWF, suggesting that GSCs could be differentiated into ECs, possibly through a mechanism dependent on extracellular adenosine-A 3 AR axis. In addition, hypoxic conditions increased VEGF secretion, which was previously observed in other tumor models but not in GSCs [38,39]. VEGF-165 secretion, which is the most abundant and potent VEGF isoform in GBM [40], was specifically evaluated. The results suggested that differentiation of GSCs into ECs could promote neovascularization and angiogenesis. These processes are highly relevant to the progression and prognosis of GBM as they support tumor growth and infiltration into surrounding healthy tissue.
In this study we used MRS1220, an A 3 AR pharmacological antagonist, to demonstrate that this receptor is involved in the differentiation of GSCs to ECs. In addition, our research group previously produced the U87MG A 3 AR KO cell line, with the ability to differentiate into GSCs, demonstrating their role in chemoresistance [29]. The A 3 AR antagonist and the KO model showed a decrease in the CD31, CD144, and vWF positive cell population, suggesting that the expression of these markers depends on AR activation; probably due to increased adenosine production in GSCs under hypoxic conditions. Expression of in vitro markers does not necessarily reflect phenotypic changes in vivo, however, the decrease in blood vessel production in GBM models correlates with the low levels of some markers, such as CD31 [41]. To corroborate whether cell differentiation is directly linked to in vivo neovascularization, a previously validated murine model and an MRS1220 antagonist were used [29], producing a decrease in tumor size and blood vessel formation. The high concentrations of adenosine in the tumor, specifically in hypoxic niches, promotes the expression and over-activation of the A 3 AR, facilitating neovascularization. This process surely depends on HIF-1α activation; however, the possible signaling pathways involved must still be studied in depth. These results provide adenosine and its signaling with a new and important role in GBM.

High Performance Liquid Chromatography (HPLC)
Quantification of adenosine production of U87MG GSCs was performed by HPLC using the protocol described by Torres et al., [29]. Briefly, GSCs were incubated in 1 mL of Tyrode's buffer for 1 h at 37 • C. 200 µL of incubation medium was mixed with 100 µL of citrate buffer (pH 6). Adenosine, AMP, ADP, and ATP contents were quantified with 2-chloroacetaldehyde derivatizations by HPLC fractionation in a Chromolith Performance RP-18 column (Merck, Darmstadt, Germany) and by fluorescent detection [42]. Adenosine concentration (nM) was normalized to the total protein concentration (µg).

Flow Cytometry
To measure endothelial cell marker expression, cells were analyzed by flow cytometry (FACS Jazz; BD Biosciences, Franklin Lakes, NJ, USA). Cells were previously fixed with PFA 3.7% for 15 min at room temperature. Cells were then blocked for 45 min (1xPBS-BSA 0.5% at room temperature) and marked with anti-CD31 (58068), anti-CD34 (562577), or anti-CD144 (561569) antibodies (BD Biosciences, Franklin Lakes). For vWF detection, cells were incubated with anti-vWF (555849) (BD Biosciences, Franklin Lakes) followed by an anti-mouse Alexa 488 (Life Technologies). Lastly, events were acquired through the FL1 filter of the cytometer.

Enzyme-Linked ImmunoSorbent Assay
VEGF in culture medium was quantified using Human VEGF ELISA Kit (KHG0111, Life Technologies) [40]. For kinetic analysis, a cell density of 10 4 GSCs/well [40] were incubated for 24, 48, and 72 h under hypoxia conditions. Treatments with 10 µM MRS1220 were carried out for 72 h under hypoxia conditions with the same cell density. VEGF levels (ng) were measured according to the manufacturer's instructions and normalized to total protein content (mg).

In Vivo Studies and Histopathological Analysis
A total of eight, 8 week-old male Sprague-Dawley rats were maintained under standard laboratory conditions, approved by the Ethics Committee of Animal Experiments at the Universidad Austral de Chile (Permit Number: 248-2016; date: 23 March, 2016). A density of 2 × 10 6 GSCs C6 cells were inoculated by subcutaneous injection in previously anesthetized rats (ketamine (100 mg/kg)/xylazine (10 mg/kg) intraperitoneal). At day ten post-inoculation, animals were divided for the following treatments (i) 1xPBS-0.001% DMSO (Vehicle) (Merck), and (ii) MRS1220 (0.15 mg/kg/72 h) administered by intraperitoneal inoculation. Tumor size was measured each five days until 25 days post-inoculation when rats were euthanized by intraperitoneal administration of Sodium Thiopental (120 mg/kg). Subcutaneous tumors were removed, fixed in 3.7% paraformaldehyde, dewaxed with xylol, and rehydrated using alcohols in decreasing concentration. The samples were immersed in hematoxylin and eosin (H&E) for 5 min and finally passed through an ascending alcohol concentration followed by xylol and then mounted (Histomount, Thermo Fisher Scientific Inc.). Tumor preparations were analyzed and the blood vessel count was performed by dividing each sample into ten quadrants where the number of blood vessels was quantified and averaged using ImageJ software (NIH, Bethesda, MD, USA) [29].

Statistical Analysis and Artwork
Values are expressed as the mean ± Standard Deviation (S.D.), where n indicates number of independent experiments. Statistical analyses were performed using ANOVA, Student's t-test (unpaired data), and Tukey-test. P values ≤ 0.05 were considered statistically significant. GraphPad Prism 6 (La Jolla, CA, USA) software was used to create all graphs and statistical analyses.

Conclusions
In this study, we conclude that A 3 AR promotes GSCs differentiation to ECs under hypoxia. Expression of endothelial cells markers, such as CD144, CD31, and vWF and VEGF secretion are regulated by adenosine and A 3 AR activation. Our data suggest that hypoxic niches and the adenosine axis are responsible for neovascularization; proposing GSCs and the adenosine axis as plausible therapeutic targets for GBM ( Figure 5).