Adenosine receptor 2B activity promotes autonomous growth, migration as well as vascularization of head and neck squamous cell carcinoma cells

,


Introduction
HNSCC is the sixth most common cancer entity within the male population worldwide with an incidence of about 690.000 per year (2015). 1 More than 90% of all head and neck cancers represent HNSCC, originating from different anatomic sites and caused by diverse etiological agents, for example, alcohol, tobacco, but also HPV-infection. 2 Despite increasing treatment options, HNSCC is characterized by poor prognosis with a 5-year overall survival rate of approximately 50%, often associated with loco-regional recurrence of the tumor and distant metastasis. 2 Recently, an immunosuppressive character of the HNSCC microenvironment has been revealed. 3The purine nucleoside adenosine (ADO) was identified among the immunosuppressive factors. 4nside the tumor microenvironment (TME), ADO can be produced by immune cells, stromal and endothelial cells as well as tumor cells, expressing CD39 (gene name: ENTPD1) and CD73 (gene name: NT5E) on their surface.ADO is generated by the sequential conversion of extracellular ATP or ADP to AMP and further to ADO by the ectonucleotidases ENTPD1 and NT5E, respectively. 5ADO causes suppression of antitumor immunity due to inhibitory effects on e.g.cytotoxic CD8 + cells.At the same time ADO induces the differentiation of regulatory T cells (Treg) within the TME, which further suppresses the immune system. 4,6oreover, recent reports indicate that exosomes, derived from tumor cells or suppressive populations inside the TME, express high levels of ENTPD1/ NT5E on their surface and are therefore able to spontaneously generate ADO, 4,7 further contributing to immunosuppression within the TME.
The TME extracellular concentration of ADO is relatively high when compared to healthy tissue. 8ADO activates four different G-protein coupled receptors (GPCR), A1, A2A, A2B and A3. 9 Importantly, tumor cells themselves also express ectonucleotidases as well as ADO receptors.Therefore, besides its suppressive effects on immune cells, ADO directly modulates neoplastic cell activities. 6DO may exert opposite effects on tumor cells, pro-or anti-tumorigenic, depending on the type of ADO receptor predominantly expressed and on the tumor entity itself.Thus, intervention with an ADORA1 antagonist induces apoptosis in breast cancer and colon cancer cells.[10][11][12] Furthermore, the activation of ADORA2A induces apoptosis in Caco-2 human colon cancer and HepG2 hepatoma cells.13,14 In contrast, a protumorigenic effect upon A2A receptor activation was reported in lung and breast cancer cells.15,16 Also, by acting via A3 receptor in different tumor types protumorigenic and antitumorigenic effects were observed.[17][18][19][20] In a wide range of tumor entities, including bladder urothelial carcinoma, 21,22 colon carcinoma, 23 prostate cancer, 24,25 breast cancer, 22,24,26 melanoma 24 and also in tumors of the oral cavity, 27 an A2B-mediated protumorigenic effect was observed.However, the precise role of the adenosinergic system in HNSCC remains poorly understood.In order to elucidate whether targeting of the adenosinergic system could be a future treatment option for patients suffering from HNSCC or in general from tumor entities upregulating ADORA2B expression, the present study aims to characterize the biological consequences of the expression and function of the adenosinergic axis for HNSCC cancerogenesis.For this, in vitro and in vivo studies were applied using HNSCCderived cell lines as well as ADORA2B expressing prostate cancer cell lines, for which a link between ADORA2B activity and cell proliferation was previously suggested.24,25

Detection of ADORA-transcripts
The transcripts of ADORA1, ADORA2A, ADORA2B, ADORA3 and GAPDH were amplified by end point-PCR from tumor cells and positive control samples (ADORA1: CD19-positive B-cells isolated from human tonsil tissue; ADORA2A: Jurkat T cells; ADORA2B: prostate cancer cell lines; ADORA3: Glioblastoma) using specific primers 25 (Supporting Information Table S1).Products were separated on 1.8% agarose gels and visualized by Midori Green Advance DNA Stain (NIPPON Genetics Europe GmbH) using the ChemiDoc MP Imaging System (Bio-Rad Laboratories GmbH, Munich, Germany).Three independent experiments were performed.

Quantitative expression analysis of ADORA2B, ENTPD1 and NT5E
Relative mRNA expression was quantified by SYBR Green PCR assay.Specific primers (Supporting Information Table S2) were obtained from biomers.netGmbH.Gene-specific PCR products were measured with the LightCycler 96 Instrument (Hoffmann-La Roche AG).For relative quantification the 2 −ΔΔCT -method was used. 31tection and quantification of ENTPD1 and NT5E on cell surfaces Cells were stained using monoclonal anti-CD39PE-Cyanine seven coupled antibodies, eFluor 450 coupled anti-CD73-antibodies and corresponding murine IgG1 kappa isotype control antibodies (eBioscience, San Diego, CA).Flow cytometry was performed using a Gallios Flow Cytometer (Beckmann Coulter Life Sciences, San Jose, CA).The mean fluorescence intensity was determined.

Proliferation assay
Cell proliferation was assessed by MTT assay using the CellTiter 96 nonradioactive cell proliferation assay (Promega Corporation, Fitchberg, WI).Then, 1 × 10 3 cells per well were seeded on 96-well plates and grown at 37 C and 5% CO 2 in complete DMEM (10% FBS).After 24 hr, fresh medium was given (1% FBS) containing DMSO or PSB-603 (1-10 μM; Sigma Aldrich).The absorbance was measured with a Tecan Infinite M200 Pro microplate reader (Tecan Group AG).Three experiments were performed in quadruples.The proliferative capacity of shRNAexpressing cells was assessed by the xCELLigence real time cell analysis system (OMNI Life Science).Electrodes were coated with 1% fibronectin for 30 min at RT and 5 × 10 3 cells/well were seeded in triplicates in complete DMEM (10% FBS).Proliferation was monitored over time.

Scratch assay/wound closure assay
A total of 3 × 10 4 cells per cm 2 were seeded on 12-well plates in DMEM supplemented with 10% FBS and cultivated at 37 C and 5% CO 2 overnight.An artificial wound was scratched into the confluent cell monolayer using a pipette tip of a 1,000 μl pipette.Photomicrograph images (Axio Observer D1 fluorescence microscopy, Carl Zeiss AG) were immediately captured (time 0 hr), and the cells were subsequently incubated in DMEM (1% FBS) supplemented with DMSO, NECA, NECA +PSB or PSB (each 1 μM; Sigma Aldrich).In case of coincubation (NECA+PSB), PSB was added 20 min after NECA.The migration of the cells and the closing of the scratch were observed by taking microphotographs.Three experiments were performed in triplicate.

Migration assay
Two-chamber migration apparatus (pore size 8 μm) was used (Falcon Corning Inc, Corning, NY) to perform cell migration assays.Then, 1 × 10 6 cells were seeded in the upper chamber in serum free medium supplemented with 0.1% BSA.Medium supplemented with 10% FBS was added to the lower chamber as the chemoattractant.In both chambers DMSO, NECA, NECA+PSB-603 or PSB-603 (each 1 μM) were added.In case of co-incubation (NECA+PSB-603), PSB-603 was added 20 min after NECA.After 16 hr incubation at 37 C and 5% CO 2 , cells were carefully removed from the upper chamber using a cotton swab and the cells that had traversed the membrane were fixed with 3.7% paraformaldehyde and stained with DAPI.Cells fluorescence was monitored with an Axio Observer D1 microscope (Carl Zeiss AG).Images were captured using the software Axio Vision 40 (Carl Zeiss Imaging Solutions GmbH).The cell nuclei in the captured images were counted for each experimental condition.Three experiments were performed in duplicate.

VEGFA-ELISA
HNSCCs were seeded with a density of 2.5 × 10 4 cells per cm 2 on 12-well plates and grown at 37 C and 5% CO 2 in complete DMEM (10% FBS) overnight.Medium was replaced (0 hr time) with DMEM (1% FBS) supplemented with either DMSO or PSB-603 (1 μM).Supernatants were collected and subjected to detection of VEGFA release using a human VEGFA ELISA Kit (Life Technologies) following the manufacturer instruction manual.Extinction was analyzed at 450 nm using Infinite M200 Pro microplate reader (Tecan Group AG) and Magellan V7.1 software (Tecan Group AG).The experiment was performed three times in duplicate.

cAMP measurement
Intracellular cAMP levels were determined using the cAMP parameter assay kit from R&D systems according to the manufacturer's instructions using Infinite M200 Pro microplate reader (Tecan Group AG) and Magellan V7.1 software (Tecan Group AG).

Cell-cycle analysis and apoptosis assays
Cell cycle analysis was performed as described earlier. 25Flow cytometry analyses of Annexin V (Abcam) and 7AAD (BD Biosciences) were performed to examine apoptosis.Additionally, the TUNEL-Assay-Kit-FITC (Abcam) was applied according to the manufacturer's instructions.

Chorionallantoic membrane assay and IHC analysis
Chorionallantoic membrane (CAM) assays were performed as described. 32Briefly, 1.5 or 2.0 × 10 6 cancer cells as indicated were applied within 5 mm silicon rings on the surface of the chicken CAM, 8 days after egg fertilization.Xenografts were either treated with vehicle control or treated with different concentrations of the ADORA2B inverse agonist/antagonist PSB-603 (1, 2.5 or 5 μM) 24 and 48 hr after tumor xenograft.Four days after implantation tumors were harvested and photographed.Tumor area was quantified using ImageJ software. 33umors were fixed in formalin, paraffin embedded and further subjected to IHC.The 5-μm sections were analyzed for pancytokeratin (pan-KRT; 1:1; EMD Millipore Corp.; clone AE1/AE3), desmin (DES; 1:150; Dako; clone D33) and Ki-67 (MKI67) 1:50 or 1:75; Santa Cruz; clone MIB-1).To determine the number of Ki-67-positive cells, at least 300 cells in three distinct, randomly chosen areas within the tumor (according the cytokeratin staining) were examined.Desmin reactivity was quantified by subtracting background staining from the specific desmin staining using ImageJ software. 33

Statistical analysis
Data are expressed as the mean AE standard deviation (SD).Between groups, statistical differences were evaluated with Student's t-test.Statistical significance was marked with * for p < 0.05, ** for p < 0.01 and *** for p < 0.005.

Data availability
The data that support the findings of our study are available from the corresponding author upon reasonable request.

Human HNSCC cell line express CD73 (NT5E) and ADORA2B constitutively on their surface
To analyze the presence of players of the adenosinergic system in HNSCC cells, quantitative RT-PCR (Supporting Information Fig. S1a; Fig. 1a) and flow cytometry analyses (Supporting Information Fig. S1b and S1c) detecting ENTPD1 and NT5E were performed involving 10 human HNSCC-derived cell lines.As internal controls, three ADORA2B-expressing prostate carcinoma cell lines (DU145, PC-3, LnCap) 25 as well as human lymph node derived CD19 + B cells, expressing ENTPD1 and NT5E but not ADORA2B at their surface, 34 were included into the study.Additionally, the Burkitt lymphoma cell line Namalwa was analyzed.While we detected a substantial expression of ENTPD1 on CD19+ B cells, we were unable to identify any ENTPD1 mRNA expression in prostate cancer cell lines, HNSCC cell lines or Namalwa B cells (Supporting Information Fig. S1a).In contrast, we observed a considerable expression of NT5E in all HNSCC-derived cell lines as well as in the analyzed prostate cancer cells.Interestingly, NT5E showed higher expression in cancer cells as compared to purified CD19+ B cells (Fig. 1a).The absence of ENTPD1 and the presence of NT5E expression at protein level was confirmed by flow cytometry (Supporting Information Fig. S1b and S2c).To define the   exact ADO-sensing receptors expressed in HNSCC cell lines, endpoint RT-PCR experiments were performed.These revealed that all analyzed HNSCC cell lines expressed exclusively ADORA2B (Supporting Information Fig. S2).In line with published data, 25 prostate cancer cell lines DU145, PC-3 and LnCap also express ADORA2B, with the highest expression detected in PC-3 cells.In contrast, expression of A1, A2A and A3 receptors was not observed in any of the analyzed HNSCC, prostate carcinoma or Burkitt lymphoma cell lines (Supporting Information Fig. S2).
The expression of ADORA2B was further evaluated by quantitative RT-PCR.To exclude a potential deregulation of housekeeping genes, three different genes, RPL13, B2M and HPRT1, were used for normalization, resulting in different ADORA2B expression levels estimations (Fig. 1b).Immunoblotting analysis confirmed ADORA2B expression (Fig. 1c) in all HNSCC cell lines tested, albeit at different extent.

ADORA2B activity affects proliferation, migration and VEGFA secretion of HNSCC cells
To characterize the oncogenic potential of ADORA2B activity on HNSCC, cells were treated with different concentrations of ADO and monitored over time.Surprisingly, we did not observe any effect on cell proliferation induced by ADO (Fig. 2a).Only the use of high concentrations of ADO (10 mM) was associated with reduced proliferation.However, this observation was rather due to an ADORA2B-independent, toxic effect, since proliferation of Namalwa cells, which do not express ADORA2B or any other ADO receptor, was also affected at this high ADO concentrations (Figs.1b and 1c and Supporting Information Fig. S2).This hypothesis was confirmed by AnnexinV/7AAD staining as well as by TUNEL assays, revealing a significant increase of apoptotic cells upon treatment with 10 mM ADO (Supporting Information Figs.S3a and S3b).In contrast, the use of increasing concentrations of ADORA2B inverse agonist/antagonist PSB-603 (below just called ADORA2B antagonist) resulted in a clear concentration-and time-dependent decrease in proliferation for all HNSCC cell lines analyzed (Fig. 2b).Internal controls were performed with ADORA2B expressing PC-3 cells ( 25 and Figs.1b  and 1c) and ADORA2B negative Namalwa cells (Figs. 1b and  1c).Of note, ADORA2B-negative Namalwa cell proliferation remained unaffected by PSB-603 treatment ranging between 1 and 5 μM.However, at concentrations of 10 μM PSB-603 Namalwa cell proliferation was severely diminished (Fig. 2b) suggesting an ADORA2B-independent, direct toxic effect of PSB-603.In order to exclude ADORA2B-independent PSB-603 effects, a 10 μM PSB-603 concentration was no longer used in subsequent experiments.We further confirmed the PSB-603-mediated inhibitory effect on HNSCC cell proliferation in scratch assays, in which PSB-603 significantly delayed wound closure (Figs.2c and 2d).In contrast, for some (UDSCC1, 3, 5 and PC-3) but not all tested cell lines we observed a faster wound closure when cells were cultured in the presence of the nonselective ADORA2B agonist NECA (Figs. 2c and 2d).Interestingly, co-incubation of cancer cells with NECA together with PSB-603 abrogated the positive effect of NECA (Figs. 2c and  2d).Together, these results indicate that HNSCC cell proliferation depends on ADORA2B activity.
To investigate which ADORA2B-mediated signaling pathways are involved in the regulation of proliferation, inhibitor studies were conducted.Similar to PSB-603 treatment, we observed a concentration-dependent inhibition of HNSCC-derived cell line proliferation when the adenylate cyclase inhibitor MDL12330A was involved, leading to comparable reduced cAMP levels as observed with PSB-603 (Figs. 2e and 2f ).Therefore, our data suggest an involvement of Gα s coupled proteins responsible for elevated cAMP levels associated with ADORA2B-dependent proliferation of HNSCC.
ADORA2B expression is associated with progression of oral cavity tumors 27 and also with increased invasiveness and metastasis of triple negative breast cancer. 24Since increased migratory activity is a prerequisite for invasiveness and metastasis, we next analyzed whether migration of HNSCC-derived cell lines depends on ADORA2B activity.We observed a substantial migratory capability of analyzed HNSCC as well as prostate cancer cell lines, which was found significantly reduced in the presence of PSB-603 in all analyzed ADORA2B-expressing cancer cells (Fig. 2g).Of note, treatment with NECA did not significantly increase migration of tested cell lines.
Since VEGFA is a key molecule involved in the induction of tumor vascularization, we tested whether ADORA2B activity is also necessary for VEGFA secretion by HNSCC cells.Abrogation of ADORA2B activity resulted in significant decrease in VEGFA secretion by HNSCC and prostate cancer cells (Fig. 2h).Interestingly, VEGFA secretion was also detectable in ADORA2B-negative Namalwa B cells, which was consequently unaffected by PSB-603 treatment.Similar to our wound closure experiments, both NECA and ADO treatments increased slightly but significantly the VEGFA secretion at concentration of 5 μM after 96 hr of treatment (Fig. 2i).Again, the NECA induced VEGFA secretion was suppressed by PSB-603.In addition, the phospholipase C inhibitor U73122 was also able to inhibit VEGFA secretion by HNSCC cells (Fig. 2j), indicating a link of ADORA2B to the Gα q subunit.

ADORA2B is constitutively active in HNSCC cells
Since ADO itself had no effect on HNSCC cell proliferation that in turn was efficiently inhibited using PSB-603, we wondered whether ADORA2B is constitutively active in HNSCC.Indeed, HNSCC cell lines UDSCC1 and UDSCC6 showed considerable high basal levels of cAMP (13-14 pmol/ml per 6 × 10 4 cells initially seeded).This was significantly decreased after treatment of cells with PSB-603 (Figs. 2f, 2k and 2l).Notably, neither ADO nor NECA treatment (2.5 and 5 μM) significantly increased intracellular cAMP levels.In line with previous reports, 35 only high concentrations of ADO (10 μM) promotes a further increase of cAMP levels in HNSCC cell lines taken into the study (Supporting Information Figs.S4a  and S4b).Together with the fact that incubation of HNSCC and PC-3 cancer cells with ADO did not result in increased proliferation (Fig. 2a), our data suggest that the constitutive and ligand-independent activity of ADORA2B promotes the autonomous proliferation, migration and VEGFA secretion of tumors of the head and neck region.Still, we could not exclude a preactivation of ADORA2B by endogenous ADO.
To evaluate this possibility, we repeated our experiments in the presence of ADO deaminase (ADA) that catabolizes ADO to Inosine.ADA did not influence basal nor NECA-induced cAMP production (Figs.2k and 2l) indicating that in our experimental setup an influence of endogenous ADO on ADORA2B activity can be excluded.expense of G0/G1 and S phase in UDSCC1 cells (Supporting Information Fig. S5) upon inhibition of ADORA2B activity.This was associated with reduced cyclin D1 (CCND1) expression as well as with impaired phosphorylation at S780 of retinoblastoma protein (RB1).At the same time, we observed a downregulation of the antiapoptotic protein Bcl-xl (BCL2L1), which was associated with enhanced levels of the proapoptotic protein cleaved caspase 9 (CASP9; Figs.3c-3e).To corroborate these results, HNSCC cell lines were examined by Annexin V/7AAD staining as well as by TUNEL assay (Supporting Information Fig. S3).Both analyses revealed a PSB-603-induced apoptosis.Thus, in UDSCC1 and UDSCC6 cells, PSB-603 provoked cell cycle arrest and induced the programmed cell death pathway in a concentration-dependent manner (Figs.3c-3e; Supporting Information Fig. S3).Similar to wound closure experiments, NECA has a stronger effect on UDSCC1 cell cycle progression than on UDSCC6 cells (Supporting Information Figs.S5a-S5d and data not shown).In UDSCC1 cells, NECA promotes an increase of S-phase, most likely due to enhanced cyclin D1 and pRB1 expression.Again, NECA-induced effects were abrogated in the presence of PSB-603.In contrast, no effect of NECA on protein expression influencing cell cycle progression could be observed in UDSCC6 cells.PSB-603 treatment of cancer cells did not affect the expression of LC3A/B (MAP1LC3A/B) and ATG7 proteins, indicating no effect on autophagy (data not shown).

Inhibition of ADORA2B activity impairs tumor growth as well as angiogenesis in vivo
Having demonstrated the requirement of ADORA2B activity for HNSCC autonomous cell growth in vitro, we next analyzed the effect of ADORA2B inhibition on tumor progression    in vivo using the chicken chorioallantoic membrane (CAM) assay.Inhibition of ADORA2B activity resulted in reduced number of developing tumors (Figs.4a and 4b).Moreover, increasing concentrations of PSB-603 resulted in reduced size of the developing tumors (Figs.4c and 4d), significant reduction of proliferating Ki67 (MKI67)-positive tumor cells and strong reduction of tumor-induced vascularization of xenografts, demonstrated by staining with the pericyte marker desmin (DES; Figs.4d-4f ).

ADORA2B depletion in HNSCC results in reduced tumor growth in vitro and in vivo
To validate our inhibitor data presented in Figure 4, knockdown experiments were performed.HNSCC cells were transduced with lentiviruses, expressing either scrambled shRNA (shScr) or shRNA targeting ADORA2B (shADORA2B).We confirmed the abrogation of ADORA2B expression in UDSCC1 by quantitative RT-PCR (Fig. 5a).Real-time cell analyses revealed a clear dependence of UDSCC1 cells on ADORA2B expression for autonomous cell proliferation (Fig. 5b).To analyze the effect of ADORA2B down-modulation in vivo, scrambled or ADORA2B-specific shRNA infected UDSCC1 were xenografted in two different amounts (1.5 × 10 6 and 2.0 × 10 6 ) on the CAM of chicken embryos and tumor development was assessed.Abrogation of ADORA2B expression resulted in markedly reduced tumor formation capability in vivo (Figs.5c  and 5d).This was associated with significantly reduced tumor size (Fig. 5e), reduced proliferation index, as well as reduced tumor-induced vascularization (Figs.5f-5h), independent of the cell number initially seeded.Together, these data strongly suggest that ADORA2B is an important modulator of HNSCC autonomous cell growth in vitro and in vivo and therefore a potential therapeutic target for HNSCC treatment.

Discussion
The adenosinergic system plays an important role in the development of many types of cancer. 4To investigate the importance of ADO-generating and sensing molecules in HNSCC, we first studied the presence of these molecules involved in these processes in HNSCC-derived cell lines in vitro.Our study shows that ENTPD1 is not expressed in HNSCC-derived cells, neither at mRNA nor at protein level.Since hypoxic conditions are responsible for upregulation of ENTPD1 within the TME, 36 cultivation of cells under normoxic settings could explain this observation.In contrast, a considerable expression of NT5E in HNSCC cells was observed.This finding is in line with previous reports demonstrating significantly higher expression of NT5E in epithelial dysplasia and HNSCC compared to normal oral mucosa. 37Moreover, NT5E expression in HNSCC was found to be associated with lymph node metastasis and poor prognosis. 37Contrary to findings of the aforementioned study, 37 we were not able to detect any coexpression of NT5E with ADORA2A or ADORA3.In agreement with the findings of Kasama et al., 27 we detected exclusively ADORA2B in HNSCC cells.Kasama and colleagues showed a substantial upregulation of ADORA2B in oral squamous cell carcinoma (OSCC)-derived cell lines as well as in primary OSCC compared to normal oral keratinocytes or normal oral tissue. 27ADORA2B is one of four ADO sensing receptors exhibiting the lowest ADO affinity. 35hysiological ADO concentrations below 1.0 μM are not sufficient to activate ADORA2B. 38Therefore it was suggested, that ADORA2B may play an important role in tumor conditions associated with hypoxia-induced cell death and subsequent ATP release, in which ADO concentrations can reach 50-100 μM inside the TME. 8The expression of both NT5E and ADORA2B in HNSCC-derived cells suggests the contribution of the adenosinergic system in the development, maintenance and progression of HNSCC.Although ADORA2B was found to be implicated in many types of solid tumors, [21][22][23][24][25][26][27] its precise role in HNSCC carcinogenesis is not well understood.Therefore, the present study focused on the identification of the oncogenic function of ADORA2B in HNSCC-derived tumor cells.
We found that treatment with ADO had no effect on HNSCC cell proliferation.Even high concentrations of ADO (10 μM), associated with elevation of intracellular cAMP levels, failed to promote cell proliferation.In line with previous findings, 20,39 ADO concentrations of 10 mM (at least 100-fold higher than in a hypoxic TME 8 ) induced apoptosis in analyzed cell lines through an ADORA2B-independent mechanism, possibly mediated by intracellular transport of ADO. 40Also, the ADORA2B agonist NECA showed only moderate or no influence on ADORA2B activity measured by intracellular cAMP content, wound closure assay, protein expression of cell cycle associated genes and VEGFA secretion.We assume that these observations are caused by the constitutive ADORA2B activity revealed in HNSCC-derived cells, as considerable basal cAMP levels were detected in our study, even in the presence of ADA abolishing the influence of endogenous ADO on ADORA2B activity.
2][43] Furthermore, elevated levels of intracellular cAMP with increasing expression levels of transfected GPCR were observed. 44In line with these observations, the upregulation of ADORA2B in HNSCC shown in the present study and by others 27 could be responsible for the constitutive ADORA2B activity observed in HNSCC cells.Moreover, Vecchio et al. 43 reported a ligand-independent ADORA2B activity in native cells derived from prostate cancer.Thus, the fact that similar ADORA2B characteristics are observed in different tumor entities, such as prostate cancer-and HNSCC-derived cells, suggests that constitutive ADORA2B activity might be a general phenomenon among various cancer entities.
In contrast, the ADORA2B antagonist PSB-603 efficiently inhibited proliferation of HNSCC cells in a concentrationdependent manner, which was associated with decreased concentration of intracellular cAMP.Moreover, we showed that PSB-603 induced cell cycle arrest as well as apoptosis.Notably, in colorectal cancer cells, PSB-603 induces alteration in cell metabolism changing the redox state in an ADORA2Bindependent manner. 40Thus, the observed reduction of Namalwa cell proliferation in the MTT assay under 10 μM PSB-603 may be caused by ROS-mediated induction of apoptosis.Also, we cannot exclude a PSB-603-induced ROS production in HNSCC-derived cells that may further contribute to ADORA2Bindependent induction of apoptosis even at lower concentrations.However, shRNA-mediated inhibition of ADORA2B activity in HNSCC cells reduced tumor formation, proliferation and vascularization in in vivo xenograft experiments.Therefore, an A2Bdependent effect on HNSCC tumorigenesis was confirmed, and a PSB-603-induced ADORA2B-dependent effect on tumor growth and vascularization can be concluded.
6][47][48][49][50] Similar to PSB-603 treatment, inhibition of the phospholipase C results in the reduction of VEGF secretion and the abrogation of adenylate cyclase activity leads to the inhibition of HNSCC cell proliferation.Therefore, after ligation of ADORA2B, an involvement of G q -as well as G αs -mediated signaling can be suggested.However, the A2B-induced downstream signaling pathways in HNSCC-derived tumor cells remain to be elucidated.In contrast to the present study describing an A2B-mediated protumorigenic effect on HNSCC, activation of ADORA2A (also utilizing G αs and therefore elevated cAMP levels for intracellular signal transduction 46 ) induces antitumorigenic effects in colon cancer and hepatoma cells. 13,14Notably, in melanoma cells ADO-concentration dependent pro-proliferative and antiproliferative effects through ADORA2A activation were observed. 20A2A-mediated protumorigenic effects were also found in lung and breast cancer cells. 15,16The opposite effects mediated by ADORA2A and ADORA2B may depend on (i) the basal receptor activation status as we observed a constitutive, ligandindependent ADORA2B activity in HNSCC cells, which was not described for ADORA2A, (ii) the concentration of extracellular ligand, (iii) the ADO-receptor-independent agonist or antagonist action, (iv) the expression of additional ADO-sensing receptors, (v) the ADO-receptor-mediated intracellular signaling cascade, and on (vi) the tumor entity.Thus, further investigation on ADOreceptor function and signaling is needed to better understand their precise role in the development and maintenance of cancer.
Together, our data suggest ADORA2B being an important biomarker and an interesting therapeutic target for treatment of HNSCC as well as other ADORA2B expressing solid tumor entities hitting two important cancer characteristics: cell viability and angiogenesis.

Figure 1 .
Figure 1.Ectonucleotidase NT5E (CD73) as well as ADORA2B are constitutively expressed in HNSCC-derived cell lines.(a) Quantitative RT-PCR analyses of NT5E mRNA expression in 10 HNSCC-derived cell lines relative to three different housekeeping genes (RPL13, B2M, HPRT1).In parallel, three prostate-carcinoma (Ca) cell lines (DU145, PC-3 and LNCap) were included into the analyses.Additionally, the Burkitt lymphoma cell line Namalwa as well as CD19+ tonsillar B cells were analyzed.NT5E mRNA expression levels of tumor cell lines are depicted relative to the NT5E transcript level of CD19 + B cells.(b) Quantitative RT-PCR analyses of ADORA2B mRNA expression in 10 HNSCC-derived cell lines relative to three different housekeeping genes (RPL13, B2M, HPRT1).In parallel, three prostate-carcinoma (Ca) cell lines (DU145, PC-3 and LNCap) were included into the analyses, also, the Burkitt lymphoma cell line Namalwa as well as CD19+ tonsillar B cells.ADORA2B mRNA expression levels of tumor cell lines are depicted relative to the ADORA2B transcript level of the DU145 prostate-carcinoma (Ca) cell line, for which ADORA2B expression was described previously.25 (a and b) Each individual experiment was performed in duplicates.Three independent experiments were performed.Presented are the mean values AE SD.(c) Evaluation of ADORA2B protein expression by Western Blotting.The detection of AKTB (β-Actin) served as loading control.

Figure 2 .Figure 2 .
Figure2.Constitutive ADORA2B activity drives autonomous proliferation, migration as well as VEGFA secretion of HNSCC-derived cells lines.(a) Adenosine does not induce proliferation of HNSCC-derived cell lines.Cell lines as indicated were seeded, treated with different amounts of adenosine (ADO) and proliferation was assessed over a time period of 72 hr.(b) HNSCC-derived cell lines as indicated as well as the prostate-ca cell line PC-3 (highly expressing ADORA2B; positive control) and the B lymphoma cell line Namalwa (not expressing ADORA2B; negative control) were seeded and treated with different amounts of the ADORA2B antagonist PSB-603.The proliferation was assessed over time.(c) Wound closure capacities were analyzed for HNSCC-derived cell lines as well as for the prostate-ca cell line PC-3, which were either left untreated or treated with the NECA, with the PSB-603 or with a combination of both.Wound closure was monitored microscopically (exemplary depicted for UDSCC5) and (d) evaluated taking the measurements of microphotograph images.(e, f ) UDSCC1 and UDSCC6 cells were seeded and either left untreated or cultivated in the presence of PSB-603 or different concentration of the adenylate cyclase inhibitor MDL12330A.Proliferation (e) as well as intracellular cAMP levels (f ) were assessed.(g) In transwell assays, the migratory capabilities of HNSCC-derived cell lines as well as of the prostate-ca cell line PC-3 were assessed in the presence of NECA, PSB-603 or in the presence of both.The percentage of migration relative to the DMSO-treated cells (control) is depicted.(h) Cell lines UDSCC1, UDSCC5, PC-3 and Namalwa were either left untreated or treated with PSB-603.After time points indicated the supernatant was taken and analyzed for the amount of secreted VEGFA.(i) UDSCC5 cells were either left untreated or treated with PSB-603, NECA, with equimolar concentrations of both compounds together or with ADO.After time points indicated the supernatant was taken and analyzed for the amount of secreted VEGFA.( j) UDSCC5 cells were either left untreated or treated with the ADORA2B antagonist PSB-603 or increasing concentrations of the phospholipase C inhibitor U73122.After time points indicated the supernatant was taken and analyzed for the amount of secreted VEGFA.(k and l) The UDSCC1 and UDSCC6 cells were either left untreated or treated with the PSB-603, NECA, with equimolar concentrations of both compounds together or with ADO at concentrations indicated, either in the absence or presence of 1 U/ml ADA.The changes in intracellular cAMP concentrations are depicted relative to the determined basal cAMP level, which was set as 1.Each individual experiment was performed in triplicates.None less than three independent experiments were performed.Presented are the mean values AE SD.Significant levels were determined relative to the DMSO control values.*p < 0.05, **p < 0.01, ***p < 0.005.Significant levels were also determined relative to samples that were treated with NECA together with PSB-603 (NECA+PSB): ##p < 0.01.[Color figure can be viewed at wileyonlinelibrary.com]

Figure 3 .
Figure 3.The ADORA2B antagonist PSB-603 stops cell cycle progression and induces the apoptotic pathway.(a) Cell cycle analyses were performed using UDSCC6 cells that were either left untreated (DMSO control) or were treated with PSB-603 at different concentrations as indicated for 24 hr.Depicted are the mean values AE SD of three independent experiments.(b) Shown are representative graphs of cell cycle analyses described in (a) using ModFit LT software.(c) UDSCC1 and (d) UDSCC6 cells were either left untreated (DMSO control) or treated with different concentrations of PSB-603 as indicated.After 48 hr of incubation, lysates were prepared and subjected to Western blot analysis detecting proteins involved in cell cycle regulation CCND1 and pRB1 (Ser780), or in regulation of pro-and anti-apoptotic events, cleaved CASP9 and BCL2L1.Presented are the mean values of protein expression relative to the housekeeping gene AKTB AE SD determined by densitometry analyses of three independent experiments.Bands with identical AKTB controls are derived from the same blot.Significant levels were determined relative to the DMSO control values.*p < 0.05, **p < 0.01, ***p < 0.005.(e) Representative Western blot images of experiments described in (c) and (d) are shown.

Figure 4 .
Figure 4. Inhibition of ADORA2B activity impairs tumor growth and tumor blood vessel formation on CAM.(a) UDSCC6 cells were applied to CAM.After 24 and 48 hr tumors were treated with concentrations as indicated of PSB-603.Four days after implantation tumors were harvested and photographed.(b) Evaluation of tumor formation.(c) Quantification of tumor area is presented.(d) IHC using antibodies directed against desmin (DES), Ki-67 (MKI67), and cytokeratin (pan-KRT) is depicted.(e) Desmin immunoreactivity was quantified by subtracting background staining from specific desmin signal using ImageJ software.(f ) Quantification of Ki-67 positive cells is shown.Presented are the mean values AE SD.Significant levels were determined relative to the DMSO control values.*p < 0.05, **p < 0.01, ***p < 0.005.

Figure 5 .
Figure 5. Down-regulation of ADORA2B expression impairs proliferation as well as tumor formation in vitro as well as in vivo.(a) UDSCC1 cell were lentiviral transduced with vectors expressing either scrambled (shScr) or ADORA2B shRNA (shADORA2B).The efficiency of ADORA2B down-modulation was estimated using quantitative RT-PCR by calculating ADORA2B expression relative to the expression determined in shScr expressing cells levels and two housekeeping genes, HPRT1 and B2M.(b) shScr or shADORA2B expressing cells were analyzed for their proliferating rates over time using a real time cell analyzer.The calculated cell index is presented mean AE SD.(c) shScr or shADORA2B expressing cells were applied to CAM in different amounts, either 1.5 × 10 6 or 2.0 × 10 6 cells.Four days after implantation tumors were harvested and photographs were taken.(d) Evaluation of tumor formation.(e) Quantification of tumor area is presented.(f ) IHC using antibodies directed against desmin (DES), Ki-67 (MK67), and cytokeratin (KRT) is depicted.(g) Desmin immunoreactivity was quantified by subtracting background staining from specific desmin signal using ImageJ software.(h) Quantification of Ki-67 positive cells is shown.Presented are the mean values AE SD. *p < 0.05, **p < 0.01, ***p < 0.005.