Human neuroblastoma tissue arrays were obtained from US Biomax, Inc. (Rockville, MD, USA). The array consisted of 61 cases of neuroblastoma cancer and normal tissues with stage and grade information. Tissue arrays were processed for immunohistochemistry as per previously described standard protocol (20). Briefly, slides were deparaffinized and endogenous peroxidases activity was blocked at room temperature by 5–10 min incubation in 0.3% H₂O₂ in PBS (pH 7.7). The slides were then rinsed in PBS for 5 min, followed by antigen retrieval by heating at 95°C in citrate buffer (0.01 mol/l sodium citrate buffer, pH 6.0) for 10 min. The tissues were blocked using normal goat serum for 20 min at room temperature and incubated in primary antibody (anti-VEGF or SPARC) for 60 min at room temperature. The slides were then rinsed in PBS followed by the addition of a horseradish peroxidase (HRP)-conjugated secondary antibody for 20 min at room temperature. HRP substrate DAB (3,3′-diaminobenzidine tetrahydrochloride) at 1 mg/ml in 50 mmol/l Tris, pH 7.2, and 0.3% H₂O₂ was then added for the development of DAB substrate followed by haematoxylin staining. Next, the slides were dehydrated and mounted with coverslips.

Neuroblastoma cell line SK-N-BE(2) [obtained from the American Type Culture Collection (ATCC; Manassas, VA, USA)] was obtained, characterized and authenticated by the ATCC, low passage cells were used for experiments in all the cases (less than 6 months of passages) and NB1691 (provided by Peter Houghton, St. Jude Children's Research Hospital, Memphis, TN, USA) were characterized and authenticated by the providers institution (less than 6 months of passages) were cultured in Opti-MEM medium with 5% fetal bovine serum (FBS) and 1% penicillin/streptomycin in a humidified 5% atmospheric CO₂ at 37°C. Endothelial cells were cultured using the endothelial cell growth Medium-200 (Thermo Fisher Scientific, Waltham, MA, USA) with 10 ng/ml EGF, 10 ng/ml FGF, 10 ng/ml heparin, and 1 μg/ml of hydrocortisone, 2% FBS and 1% penicillin/streptomycin in a humidified 5% CO₂ atmosphere at 37°C with Calcein AM Dye (Thermo Fisher Scientific) as per manufacturer's instructions.

SK-N-BE(2) and NB1691 cells were transfected with SPARC overexpression plasmid RC209964 (Origene Technologies, Inc., Rockville, MD, USA) as per standard protocols using Lipofectamine with 10 μg of plasmid on 60% confluent plates. SPARC overexpression was also validated in vitro using SPARC Adenovirus (Human) from Applied Biological Materials Inc., (Richmond, BC, Canada) on 60% confluent plates at 50 MOI infection. Empty adenovirus was used as control. SPARC expression was validated by western blot analysis.

Antibodies were obtained from the following sources: SPARC (Aviva Systems Biology Corp., San Diego, CA, USA), VEGF-A and β-actin (Santa Cruz Biotechnology, Dallas, TX, USA).

SK-N-BE(2) and NB1691 cells were transfected with SPARC overexpression plasmid. After 24 h, cells were treated with or without 5 Gy of ionizing radiation and incubated for another 24 h. Cells were collected and total protein extracted by using M-PER mammalian protein extraction reagent (Thermo Fisher Scientific, Waltham, MA, USA) and protein concentrations measured using Pierce 660 nm protein assay reagent (Thermo Fisher Scientific). Equal amounts of protein (10 μg/lane) were electrophoresed under reducing conditions on 4–16% gradient polyacrylamide gels. After SDS-PAGE, the separated proteins were transferred on to a polyvinylidene difluoride membrane (Bio-Rad Laboratories, Hercules, CA, USA). Membranes were blocked with TBS-T containing 5% non-fat skim milk for 1 h. Subsequently, membranes were immunoprobed with primary antibody and appropriate horseradish peroxidase-labelled secondary antibody. Specific protein bands were visualized using enhanced chemiluminescence detection reagents (Life Technologies, Carlsbad, CA, USA). A similar protocol was used for total protein isolated from subcutaneous tumors of nude mice described in section below (see method for neuroblastoma subcutaneous tumor model).

To determine in vitro angiogenic inhibition by SPARC overexpression in neuroblastoma cells, we collected conditioned media from controls or radiated NB1691 and SK-NB-E(2) cells with or without SPARC overexpression as per standard protocols. The collected conditioned media (100 μl) was used to culture human endothelial cells over Matrigel. As a positive control for VEGF-A inhibition, we used 50 μg/ml of bevacizumab a recombinant humanized monoclonal antibody that targets VEGF-A in control NB1691 or SK-N-BE(2) conditioned media to compare angiogenic inhibition with SPARC overexpression. The cells were monitored every hour for network formation and after 6 h endothelial cell network was visualized at 490 nm excitation and 520 nm emission using an inverted fluorescent microscope. Angiogenesis was quantified by determining the number of branch points as per standard protocols.

BALB/c nude female mice aged 6–8 weeks were obtained from Harlen Labs, Inc. (Indianapolis, IN, USA) and housed in micro-isolation cages in groups of five animals in ventilated racks at a constant temperature of 20–26°C and humidity of 30–70%. All animal experiments were carried out after obtaining approval from the Institutional Animal Care and Use Committee.

The in vivo angiogenic assay was done as previously described with minor modifications (20). SK-N-BE(2) or NB1691 cells (1×10⁶) treated with SPARC overexpressing plasmid alone or with radiation were loaded into a diffusing chamber. A 2-cm long incision was made horizontally along the edge of the dorsal air sac of the mice and the chambers were placed underneath the skin. The mice were sacrificed 10 days later; and carefully skinned around the implanted chambers. The skin folds covering the chambers were photographed under visible light, and tumor induced vasculature quantified.

The subcutaneous tumor model was done as previously described by us with minor modifications (21). SK-N-BE(2) and NB1691 cells were implanted subcutaneously into nude mice on day 0 (1×10⁵ cells). After 25 days, animals developed detectable tumors. Mice were separated into 4 groups of 5 animals per group. The animals that lost ≥20% of body weight or had trouble ambulating, feeding or grooming were sacrificed. On day 28 and day 32, two groups of animals received 150 μg/animal per day intratumoral injection of SPARC plasmid DNA given at three different locations on the tumor (plasmid concentration of 1.5 μg/μl in PBS, 50 μg per site 120° apart) at a depth of 3–5 mm using a 30-gauge insulin syringe. The overall cumulative dose was 300 μg/animal. Two groups were left as controls. On day 30, one group each of SPARC treated and control animals were exposed to 5 Gy ionizing radiation. Tumor growth was followed, till day 45, after which animals were sacrificed. Tumors were harvested, imaged followed by, protein and total RNA isolation by standard protocols. Tumors were fixed in buffered formaldehyde and paraffin sectioned for immunohistochemical analysis of SPARC and VEGF-A. In vivo tumor size measurement: Subcutaneous tumors volumes were measured using the following formula: π/6 (L × W²) where L is the length and W is the width.

Excised tumors were fixed in 10% buffered formalin and embedded in paraffin. Tissue sections (5 μm thick) collected on slides were deparaffinized and endogenous peroxidases activity was blocked at room temperature by 5–10 min of incubation in 0.3% H₂O₂ in PBS (pH 7.7). The slides were then rinsed in PBS for 5 min, followed by antigen retrieval by heating to 95°C in citrate buffer (0.0 mol/l sodium citrate buffer, pH 6.0) for 10 min. The tissues were blocked using normal goat serum for 20 min at room temperature and incubated in primary antibody (anti-VEGF or SPARC) for 60 min at room temperature. The slides were then rinsed in PBS followed by the addition of a horseradish peroxidase (HRP)-conjugated secondary antibody for 20 min at room temperature. HRP substrate DAB (3,3′-diaminobenzidine tetrahydrochloride) at 1 mg/ml in 50 mmol/l Tris, pH 7.2, and 0.3% H₂O₂ was then added for the development of DAB substrate followed by haematoxylin staining. Next, the slides were dehydrated and mounted with coverslips. Negative control slides were obtained by treating with non-specific IgG. Sections were mounted and analyzed using an inverted microscope. The intensity of VEGF-A or SPARC expression per unit area was measured in arbitrary pixel units using ImageJ software.

Total RNA was extracted from the tissue samples using mirVana™ miRNA isolation kit (Ambion-Life Technologies, Carlsbad, CA, USA). cDNA was synthesized from RNA using miScript II RT kit (Qiagen, Valencia, CA, USA) according to the manufacturer's instructions. Real-time PCR was performed using miScript SYBR-Green PCR kit in an automated thermal cycler (Bio-Rad Laboratories). miR-410 and SNORD68 primers were obtained from Qiagen. SNORD68 was used as an internal control for normalization. The expression levels of pro-angiogenic factors VEGF-A, -B, -C, FGF2, PDGF, ANGPT1 and ANGPT2 (Table I) were measured as per standard protocol and normalized to HPRT. Relative gene expression was calculated by the 2^−ΔCt method.

MicroRNA transfection reagent, miR-410 mimic (AAUAUAACACAGAUGGCCUGU) and miR-410 inhibitors (UUAUAUUGUGUCUACCGGACA) were obtained from Sigma-Aldrich (St. Louis, MO, USA). SK-N-BE(2) and NB1691 were transfected as per the manufacturer's instructions and conditioned media from these cells were used in CAM assay or for ELISA determination of VEGF-A protein levels. To determine the effectiveness of miR-410 inhibition, miR-410 inhibitor treated cells were assayed for the expression levels of miR-410 using RT-PCR as per standard protocols and normalized to SNORD68 expression.

Fertilized white leghorn chicken eggs were obtained from Charles River Laboratories (Wilmington, MA, USA) and incubated at 37°C with 60% humidity. On day 7, a small window was made on the egg shell under aseptic conditions. Sterile filter paper disks (Whatman, 1–2 mm size) were placed on top of the CAM. Conditioned medium collected from SK-N-BE(2) and NB1691 cells were added on top of the filter disks, gap sealed with porous adhesive tape and the eggs were returned to the incubator. On day 12, CAM surrounding the filter disk was carefully removed, placed on glass slides and visualized for angiogenesis using a stereomicroscope. Angiogenesis was quantified by measuring the length of newly formed capillaries by using imageJ software.

Protein levels of VEGF-A were determined from the conditioned media of SK-N-BE(2) and NB1691 cells overexpressed for SPARC with or without miR-410 inhibitor using the VEGF-A ELISA kit from Thermo Fisher Scientific as per manufacturer's protocols.

All data are expressed as mean ± SD. Statistical analysis was performed using Student's t-test. A P-value of ≤0.05 was considered statistically significant. All experiments were performed in triplicate, or as indicated.

Human neuroblastoma tissue array from US Biomax, Inc., consisting of 61 neuroblastoma cases, was processed for immunohistochemistry for the expression of VEGF-A and SPARC as previously described (20). From the tissue array, we observed that the expression of SPARC was always inversely correlated with the expression of VEGF-A (Fig. 1A). Further semi-quantification of expression of SPARC and VEGF-A in the human tissue array reveled that higher expression of SPARC correlated with retroperitoneal tumors rather than with tumors in the abdominal cavity, mediastinum, or adrenal gland (data not shown).

The role of SPARC in angiogenesis has been demonstrated by us and other researchers (22,23). The addition of radiation treatment was included because radiation induced vascular injury can promote the development of leaky vasculature that can contribute to local hypoxic regions promoting an angiogenic response (24–26). Therefore, targeting angiogenesis accompanied with radiation would greatly impact the success of the consolidation therapy phase (6). As we have observed that VEGF-A is a significant player in angiogenesis in neuroblastomas and that expression of SPARC is always inversely correlated with the expression of VEGF-A (Fig. 1A), overexpression of SPARC could be a strategy for controlling radiation induced angiogenesis that is mediated by VEGF-A. To determine whether overexpression of SPARC suppressed angiogenesis, we performed the dorsal skin fold angiogenesis assay using nude mice as previously described (20). We observed that control mice showed increased tumor induced vasculature (TV) which was clearly differentiated from pre-existing vasculature (PV). Neuroblastoma tumor that overexpressed SPARC showed a significantly reduced amount of tumor induced vasculature (20%) in both NB1691 cells (P<0.001) and in SK-N-BE(2) (50%) cells (P<0.05). Addition of radiation treatment along with SPARC further suppressed tumor cell induced vasculature by 94% in NB1691 cells (P<0.001) and by 85% in SK-N-BE(2) cells (P<0.001) (Fig. 1B and C). Notably, addition of radiation alone, did increase tumor cell induced vasculature to 110% (P=0.07) in NB1691 and to 128% (P=0.01) in SK-N-BE(2) of controls in the dorsal sac chamber nude mouse models.

Based on the results from the in vivo angiogenic assay, we attempted to validate the involvement of VEGF-A as a significant angiogenic contributor in SK-N-BE(2) and NB1691 neuroblastoma cell lines. We performed RT-PCR to determine the expression levels of known pro-angiogenic molecules VEGF-A, -B, -C, FGF2, PDGF, ANGPT1 and ANGPT2 (Table I) (27) and observed that expression of VEGF-A was significantly higher than the expression levels of VEGF-B, -C, FGF2, PDGF, ANGPT1 or ANGPT2 in both SK-N-BE(2) and NB1691 cells (Fig. 1D). Since the expression of VEGF-A is the prominent contributor of angiogenesis; we further investigated whether the suppression of angiogenesis by SPARC is mediated by modulation of VEGF-A expression.

As we determined that VEGF-A is a major contributor to angiogenesis in both SK-N-BE(2) and NB1691 cells (Fig. 1D), and in continuation with this finding and our previous findings (22), we attempted to determine whether SPARC expression can suppress tumor growth and angiogenesis by modulating the expression of VEGF-A. We performed immunohistochemical analysis for VEGF-A and SPARC on neuroblastoma subcutaneous tumors developed in mice from SK-N-BE(2) or NB1691 neuroblastoma cells (Fig. 2A and B). As described, SK-N-BE(2) and NB1691 cells were implanted subcutaneously into nude mice followed by intratumoral injection of SPARC plasmid DNA (150 μg/animal) and/or 5 Gy ionizing radiation. Tumors were harvested after 45 days, and we observed that mice radiated with 5 Gy showed slightly larger tumors when compared to non-radiated controls. SPARC overexpression plasmid treated mice showed decreased tumor size in NB1691 tumors but not so much in SK-N-BE(2) tumors. Furthermore, mice treated with both radiation and SPARC overexpression showed the greatest decrease in tumor growth with two of the five mice showing undetectable tumors in both NB1691 and SK-N-BE(2) implanted mice (Fig. 2B and C). Furthermore, we observed that the expression of SPARC and VEGF-A were inversely correlated (Fig. 2D) and was similar to the human tissue array expression of SPARC and VEGF-A (Fig. 1A), where we observed that the expression of SPARC and VEGF-A was inversely correlated. Graphical representation of VEGF-A expression obtained by determining the pixel intensity of HRP-DAB reaction showed that SPARC overexpression strongly inhibited VEGF-A expression in both SK-N-BE(2) (P<0.05) and NB1691 (P<0.05) neuroblastoma subcutaneous tumors (Fig. 3A). Addition of radiation further reduced VEGF-A levels in SK-N-BE(2) (P<0.001) and NB1691 (P<0.001) derived tumors (Fig. 3A). Furthermore, mice injected with SPARC overexpressing plasmids did show increased expression of SPARC (P<0.001) (Fig. 3B). To validate the immunohistochemistry data, proteins were extracted from these tumors and expression levels of SPARC and VEGF-A determined by western blot analysis (Fig. 3C). This also correlated well with the human tissue array data (Fig. 1A) and the immunohistochemistry data (Fig. 2D). To further validate these findings, we performed western blots for VEGF-A and SPARC on NB1691 and SK-N-BE(2) cells infected with adenovirus expressing SPARC at 50 MOI with and without radiation. We observed that upregulation of SPARC resulted in significant reduction of VEGF-A expression in both NB1691 and SK-N-BE(2) cells (Fig. 3D). To determine whether the suppression of angiogenesis was similar to bevacizumab, we treated human endothelial cells with conditioned media from NB1691 or SK-N-BE(2) cells with or without SPARC overexpression and with or without radiation. We used bevacizumab, a known VEGF-A inhibitor as a positive control. We observed that angiogenic inhibition by overexpression of SPARC was comparable to cells treated with bevacizumab both with and without radiation (Fig. 3E and F).

VEGF-A is a major contributor of angiogenesis and is a recognized therapeutic target (28–30). Our data from miRanda analysis presented shows that miR-410 is a significant modulator of VEGF-A expression (Table II). To validate this in a biological setting, we investigated the relevance of miR-410 as an angiogenic regulator using a chicken chorioallantoic membrane (CAM) assay (Fig. 4A), which is an accepted model for in vivo angiogenesis (31). We observed that NB1691 and SK-N-BE(2) cells treated with SPARC reduced angiogenesis when compared to controls in both NB1691 (P<0.001) and SK-N-BE(2) cells (P<0.001) (Fig. 4B and C). Interestingly, we also found that miR-410 mimic (AAUAUAACACAGAU GGCCUGU) alone showed the same angiogenic inhibitory effect in both NB1691 (P<0.001) and SK-N-BE(2) cells (P<0.001) (Fig. 4B and C). To study if this inhibition of angiogenesis is specifically controlled by miR-410, we inhibited miR-410 using a specific inhibitor (UUAUAUUGUGUCU ACCGGACA) in SPARC overexpressing cells. Inhibition of miR-410 reversed SPARC induced reduction of angiogenesis in both NB1691 (P<0.001) and SK-N-BE(2) cells (P<0.001) (Fig. 4B and C). This reversal of angiogenic vessel formation was also consistent in the presence of radiation. These findings clearly show that SPARC downregulates VEGF-A mediated angiogenesis specifically by upregulating the expression of miR-410. The P-values represent change in vessel length in 5 replicates. Similar to the in vivo dorsal skin assay where we observed that radiation increased tumor cell induced vasculature, in the CAM assay also we observed that radiation treated cells induced an angiogenic response greater than non-radiated cells. This radiation induced angiogenesis in the CAM ex vivo model was, however, not as pronounced as seen in the in vivo model.

To test whether SPARC regulates VEGF-A expression via microRNAs, we first performed a target prediction analysis using the miRanda algorithm (32) and determined that miR-410 can be a potential mediator for VEGF-A suppression. Of several miRNAs that had potential target regions in the VEGF-A mRNA, miR-410 targeted 4 specific regions of VEGF-A mRNA (Fig. 4D). To study if SPARC was associated with miR-410 expression, we measured miR-410 levels in tumor samples collected from mice treated with SPARC overexpressing plasmid. Our results show that SPARC overexpression significantly increased miR-410 levels in NB1691 cells (P<0.05) and SK-N-BE(2) cells (P<0.001) and addition of miR-410 inhibitor showed suppression of SPARC induced miR-410 expression in both SK-N-BE(2) and NB1691 cells (Fig. 4E).

To further validate our findings, we looked for secreted levels of VEGF-A using ELISA in conditioned media used for the CAM (Fig. 5A). We observed that SPARC overexpression alone decreased VEGF-A protein levels by 40% in NB1691 cells (P<0.001) and 42% in SK-N-BE(2) cells (P<0.05) (Fig. 5A). Addition of miR-410 inhibitor to SPARC overexpressed cells significantly rescued VEGF-A to 77% of controls (P=0.05) in NB1691, whereas in SK-N-BE(2) cells no such reversal was observed; P-value represents change in VEGF-A protein levels in culture media compared to controls in 3 replicates.