Integrin-αvβ3 is a Therapeutically Targetable Fundamental Factor in Medulloblastoma Tumorigenicity and Radioresistance

Brain tumors represent the second most common childhood tumors, following hematologic cancers (1). Medulloblastoma is among the most common brain tumor in the pediatric population, accounting for approximately 20% of all childhood brain tumors (2). Medulloblastomas typically originate in the posterior fossa, primarily within the cerebellum, often manifesting early in life (3). The current standard of care for medulloblastomas includes a maximal tumor resection followed by craniospinal irradiation and chemotherapy (4). Recent advances in the molecular understanding of medulloblastoma pathophysiology have delineated four distinct prognosis molecular subgroups: wingless (WNT), sonic hedgehog (SHH), group 3, and group 4. Nevertheless, the clinical translation of these observations is just starting to emerge. Therefore, the treatment strategies employed for medulloblastomas are currently hinged upon categorizing patients into standard- or high-risk groups according to the presence of metastases, age, and extent of postoperative residual disease.

Standard-risk patients (constituting ∼70% of medulloblastomas) are typically children ages over 3 years with no evidence of disseminated disease and postoperative residual tumor of <1.5 cm². In such instances, the 5-year postdiagnosis overall survival (OS) stands as 85%. If at least one of these criteria is unmet, the patient is considered high-risk, and the 5-year OS drops to 60% (5). Despite the recent identification of four core molecular subgroups (SHH, WNT, group 3, and group 4), clinical stratification of the patients still mainly relies on the aforementioned clinical features. In addition, the tumors of about 30% of the patients originally included in the standard-risk group eventually relapse. Consequently, the accurate identification of high-risk medulloblastomas and the effective management of recurrent medulloblastomas remain substantial challenges. These challenges are further accentuated by the absence of robust prognostic markers and limited therapeutic avenues for cases of recurrence. These issues highlight the need to refine medulloblastoma diagnostic criteria with new prognostic and predictive markers and therapeutic options designed explicitly for recurring disease.

Over the past decades, our understanding of the molecular pathogenesis of malignant brain tumors such as gliomas has yielded insights into novel regulators of key processes in glioblastoma (GBM) cell growth, invasion, and angiogenesis (6). Among these factors, integrins have emerged as pivotal players and promising therapeutic targets for GBM. Integrins are a large family comprising noncovalent heterodimeric complexes involving 18 α and 8 β subunits, possibly forming at least 24 different receptor types (7). Integrins are membrane-bound receptors specific to interactions between cells and extracellular matrix (ECM) components such as fibronectin, laminin, and collagen. Upon binding to their ligands, integrins activate downstream signaling pathways that regulate many cellular effects under physiologic and pathologic conditions. While integrin-encoding genes are rarely mutated in cancer, dysregulation of their expression level or integrin signaling is common, especially in brain tumors (8).

Among these integrin complexes, integrin-αvβ3 was the first to be found aberrantly expressed in high-grade brain tumors. Integrin-αvβ3 belongs to the integrin subtypes that recognize the tripeptide Arg-Gly-Asp (RGD) sequence found in many ECM proteins, including fibronectin and vitronectin. Many studies have highlighted the central role of αvβ3 in sustaining the proliferation and invasive properties of brain tumors through the activation of focal adhesion kinase (FAK), leading to the activation of downstream pathways, including the PI3K/protein kinase B (Akt) and ERK/MAPK pathways (6). In addition to these features, αvβ3 was found necessarily and sufficiently to reprogram tumor cells toward a cancer stem cells phenotype (tumor initiation, self-renewal) in several solid tumors (9). These significant insights have been bolstered by clinical investigations, where the expression of αvβ3 has been linked to unfavorable prognosis and reduced time-to-progression (10).

Given their involvement in tumor progression, integrins have garnered significant attention as attractive therapeutic targets for solid tumors. Therefore, several strategies have been explored to target integrin-αvβ3, including the use of small molecules. They include a cyclic pentapeptide blocking the RGD binding site (cilengitide; cyclo-Arg-Gly-Asp-DPhe-NMe-Val) that acts as a selective αvβ3 inhibitor. Cilengitide was found to impair GBM tumorigenesis by hampering angiogenesis, cell proliferation, and invasion. This agent was subsequently evaluated in clinical studies (11, 12). Cilengitide had moderate antitumor efficacy in patients with GBM, with antitumor effects restricted to high-αvβ3–expressing tumors, highlighting the potential need to stratify patients based on αvβ3 expression in their tumors (13). In this context, the advent of nuclear medicine is currently one of the greatest components of theranostics concepts. As a molecular probe, radiolabeled RGD was developed for αvβ3-based imaging and therapy of solid tumors, including GBM (8). Nevertheless, such studies remain to be conducted for medulloblastoma.

In this novel study, we aimed to evaluate the role and the relevance of integrin-αvβ3 in medulloblastoma. First, the integrin-αvβ3 expression was determined in patient-derived samples and naïve and radioresistant medulloblastoma cell lines. Then, integrin-αvβ3’s role was explored using genetic depletion of the β3-subunit or cilengitide in vitro and in vivo in orthotopic medulloblastoma mouse models. Finally, molecular imaging with ^99mTc-radiolabeled-RGD peptides was used to noninvasively measure integrin-αvβ3 expression in medulloblastomas. By integrating these diverse methodologies, our study aimed to provide a comprehensive understanding of the role and relevance of integrin-αvβ3 in the complex landscape of medulloblastoma.

Human medulloblastoma cell lines were used in this study. DAOY (ATCC: HTB-186; SHH group) and CHLA-01-Med (ATCC: CRL-3021) were purchased from the ATCC, HD-MB03 [Deutsche Sammlung von Mikroorganismen und Zellkulturen (DMSZ): ACC 740; group 3] from DMSZ. D458 (Cellosaurus: CVCL_1161; group 3) was from Dr. Celio Pouponnot (Institut Curie, Paris, France). ONS-76 (Cellosaurus: CVCL_1624, SHH group) was obtained from Dr. Gilles Pagès [Institute for Research on Cancer and Aging-Nice, Nice, France, originally obtained from Health Science Research Resources Bank (IFO50355)]. All cell lines were tested negative for Mycoplasma during routine surveillance [PlasmoTest—Mycoplasma Detection Kit (Mycostrip, Invivogen)] and reauthentication was not performed. DAOY, D458, and ONS-76 cells were cultured in DMEM supplemented with 1 mmol/L sodium pyruvate, 2 mmol/L Glutamax, and 7.5% FBS. CHLA-01-Med cells were cultured in DMEM/F12 supplemented with 2% B-27, 20 ng/mL EGF, and 20 ng/mL basic FGF. HD-MB03 cells were cultured in RPMI medium supplemented with 7.5% FBS. All cell lines were cultured less than 10 passages from thawed vials and maintained at 37°C in 5% CO₂ humidified incubator, and mostly subcultured twice a week.

DAOY wildtype (WT) cells were transfected using jetPEI (Polyplus, 101000053) according to the manufacturer's instructions with pSpCas9(BB)-2A-GFP (PX458) plasmid (a gift from Feng Zhang; Addgene plasmid #48138; http://n2t.net/addgene:48138; research resource identifier: Addgene_48138) containing clustered regularly interspaced short palindromic repeats (CRISPR)-CRISPR-associated protein 9 (Cas9) targeting regions for the first [guide RNA (gRNA): 5′-GAGGCGGACGAGATGCGAGCG-3′] and 10th (gRNA 5′-AGACGGGCTGACCCTCCCGG-3′) exons of the β3-integrin gene (ITGB3). The ITGB3 sequence (pcDNA3.1-beta-3 was a gift from Timothy Springer; Addgene plasmid # 27289) was subcloned into pLenti6.3/TO/V5-Blasti (A11144, Thermo Fisher Scientific) to create β3-integrin expression plasmids. Lentiviruses were produced by triple transfection of HEK-293T cells with the lentiviral transfer vector pLenti6.3/TO/V5-Blasti and the packaging plasmids psPAX2 (Addgene plasmid #12260) and pMD2.G (Addgene plasmid #12259), a kind gift from Didier Trono, at a ratio of 0.3:0.3:0.1. Transfection was performed using jetPEI (101000053, Polyplus). The viral supernatant was collected 48 hours after transfection, filtered through a 0.45 µm filter, and added to the target cells.

For in vivo tumorigenesis assays, DAOY-derived and HD-MB03–derived cells were transduced with lentiviral particles containing RFP-Luc (Lenti-One RFP-Luc, GEG Tech) at a multiplicity of infection of 0.3 in the presence of hexadimethrine bromide (4 µg/mL).

Real-time qRT-PCR analyses were performed using human cerebellar mRNA (Biochain) and medulloblastoma cell lines. mRNAs were prepared using a Nucleospin RNA Kit (Macherey-Nagel), and cDNA synthesis was performed using a Maxima First Strand cDNA Synthesis Kit for qRT-PCR with dsDNase (Thermo Fisher Scientific). Quantitative PCR analyses were performed on an Applied Biosystems StepOnePlus System using TB Green Premix Ex Ta (Tli RNase H Plus; Takara Bio) reagents. The primers used are listed in Supplementary Table S1. Relative expression levels were determined using the ΔCt method and normalized to the reference gene 36B4. Results are expressed relative to the normal cerebellum.

Cells were lysed in 1.5 × Laemmli buffer, and protein concentrations were determined using the Pierce bicinchoninic acid Protein Assay (Thermo Fisher Scientific). Protein extracts (40 µg) were separated via electrophoresis on 10% SDS-polyacrylamide gels and transferred to polyvinylidene difluoride membranes (Millipore). Membranes were blocked in 2% milk PBS and incubated with the following anti-human antibodies: rabbit β3-integrin [1:1,000; 13166, Cell Signaling Technology (CST)], rabbit p-FAK (1:1,000; 8566, CST), rabbit FAK (1:1,000; 71433, CST), mouse p-Akt (1:1,000; 4051, CST), rabbit Akt (1:1,000; 9272, CST), rabbit p-ERK1/2 (1:1,000; 4370, CST), rabbit ERK1/2 (1:1,000; 4695, CST), and rabbit PARP (1:1,000, 9542; CST). Actin was used as the protein loading control (1:5,000; MA5-15739, Thermo Fisher Scientific). Immunoreactive bands were detected with horseradish peroxidase-coupled anti-mouse or anti-rabbit antibodies (CST) using the ECL System (Merck Millipore; WBKLS0500). Immunoblot analysis was performed using the LI-COR Odyssey Imaging System.

DAOY-derived and HD-MB03–derived cells were seeded into 6-well dishes for 24 hours. Cells were rinsed with PBS, detached with accutase (00-4555-56, Thermo Fisher Scientific), and collected. Cells (250,000) were incubated with a mouse anti-αvβ3 antibodies (1 µg/mL; ab190147, Abcam) for 1 hour in PBS/1% BSA on ice. Next, cells were centrifuged and washed twice with PBS. Then, cells were incubated with a goat anti-mouse IgG secondary antibody coupled to AlexaFluor488 (1 µg/mL; A-11029, Invitrogen). All experiments were performed in triplicate. Ten thousand events were analyzed per sample using a BD FACS Melody cytometer (BD Biosciences). Data were analyzed using the FlowJo software.

Cell adhesion to ECM proteins was performed using a CytoSelect 48-well Cell Adhesion Assay Kit (Cell Biolabs) according to the manufacturer's protocol. Briefly, 100,000 DAOY-derived and 250,000 HD-MB03–derived cells were collected in 200 µL of serum-free medium and added to the prewarmed ECM adhesion plate. The plates were incubated at 37°C for 90 minutes in a CO₂ incubator. The wells were washed three times with PBS and stained with 200 µL of cell staining solution for 10 minutes at room temperature. After washing and drying, the stained cells were extracted with extraction solution and incubated in an orbital shaker for 10 minutes. Then, 150 µL of each sample was transferred to a 96-well plate and quantified by measuring absorbance at 560 nm.

To determine cilengitide's IC₅₀, 1 µg of fibronectin was plated onto 96-well plates for 60 minutes at 37°C in a CO₂ incubator. Serum-starved cells were added to the fibronectin-coated wells (10,000 cells for DAOY and 30,000 for HD-MB03) in the presence of serial cilengitide dilutions (0–200 µmol/L), and the plates were incubated for 90 minutes at 37°C with 5% CO₂. Then, cells were washed three times with PBS and stained for 20 minutes with a 1% Crystal Violet solution at room temperature. After washing with PBS, adherent cells were solubilized with DMSO and quantified by measuring absorbance at 590 nm. The IC₅₀ was calculated using the nonlinear regression method in the Prism 8 software (Graphpad Software Inc.).

Cells (25,000 and 50,000 DAOY-derived and HD-MB03–derived cells, respectively) were seeded onto 6-well plates in triplicate per cell line and per condition. Cell proliferation was assessed by daily trypsinization and counting (Coulter Z1; Beckman) for 4 days (96 hours). Both adherent and floating cells were counted. The cell proliferation index was calculated as the percentage of day 0 by standardizing each measurement to the cell number obtained 24 hours after seeding (day 0) or after initiation of cilengitide (0.5 or 2.5 µmol/L) or LM609 (1 or 10 µg/mL) treatment.

Cells (3,000 and 10,000 for DAOY and HD-MB03 cells, respectively) were seeded in 96-well plates (Corning Inc.) in 100 µL of medium per well. Cilengitide concentrations from 0 to 500 µmol/L were tested. Its effect was determined using the 3-(4,5-dimethylthiazol-2yl)-diphenyltetrazolium bromide (MTT) colorimetric assay (Sigma-Aldrich) according to the manufacturer's instructions. When applicable, the EC₅₀ was calculated using the nonlinear regression method in the Prism 5 software (Graphpad Software Inc.).

In vitro migration and invasion assays were performed in Boyden chambers (Corning) with porous membranes coated or not with 50 µg/mL Matrigel for 2 hours at 37°C to assess invasion or migration, respectively. DAOY (25,000 cells) and HD-MB03 (75,000 cells) cells were placed in serum-free DMEM in the upper compartment [with or without cilengitide (0.5 or 2.5 µmol/L) or LM609 (1 or 10 µg/mL) treatment] and DMEM in the lower compartment of the chambers. After 12 and 24 hours incubation for DAOY and HD-MB03, respectively, cells were stained with Giemsa for 30 minutes. Invasive cells were counted under the microscope. Results are expressed as % of the control.

Cells were seeded in 12-well plates (50,000–150,000 cells per well, triplicate per condition) at 37°C/5% CO₂ in their respective media. Cells were treated with cilengitide (20 µmol/L) for 48 hours. Both floating and adherent cells were collected and centrifuged. Then, cell pellets were resuspended in FACS buffer (PBS, 0.2% BSA, and 2 mmol/L ethylenediaminetetraacetic acid) and stained with 2 µg/mL propidium iodide (PI; Invitrogen). PI was added just before the analysis. Experiments were performed at least three times, 10,000 events were analyzed per sample using a BD FACSMelody cytometer, and data were analyzed using the FlowJo software.

Adherent cells were irradiated with 8 Gy of X-rays using a 6 MeV Novalis TrueBeam linear accelerator (Novalis TrueBeam STX, 3 Gy/minute) and returned to the incubator afterward. Radioresistant cell lines (DAOY-RR and HD-MB03-RR) were developed from their respective parental cell lines (DAOY and HD-MB03) by weekly exposure to single fractions of radiation. An initial dose of 2 Gy was followed by incremental doses for 12 weeks (from 2 to 8 Gy). During this period, cells received a total of 60 Gy.

Tumor sections (5 µm cryostat sections) were fixed in 4% paraformaldehyde for 10 minutes at room temperature and blocked in 1% horse serum in TBS for 1 hour. The sections were incubated with rat monoclonal anti-mouse platelet and endothelial cell adhesion molecule 1 (PECAM1/CD31; clone MEC 13.3, 1:1,000; BD Pharmingen) and monoclonal anti-mouse α-smooth muscle actin (αSMA; A2547, 1:1,000; Sigma) or anti-Ki67 (ab16667, 1:500; Abcam) antibodies diluted in TBS containing 1% BSA at 1:100 overnight at room temperature, and thereafter washed with TBS containing 0.025% Triton. Afterward, the preparations were incubated with anti-rabbit Alexa488- (#4412, CST) and anti-mouse Alexa-555–coupled secondary antibodies. Next, preparations were washed with TBS containing 0.025% Triton, and the nuclei were counterstained with Hoechst33342 (Thermo Fisher Scientific). Fluorescence images of the cells were taken with a DMI400 (Leica Microsystems) inverted microscope equipped with a 10 × objective (Leica Microsystems) and a Zyla 5.5 camera (Andor Technologies). Preparations were mounted and imaged using a Leica microscope (DMI4000B, Leica) and counted at 10 × (CD31/αSMA) or 40 × (Ki67) magnification with Fiji software (14).

Medulloblastoma tissue microarrays (catalog number BC17012c) were purchased from US Biomax. Integrin-αvβ3 expression was assessed by IHC. Briefly, after deparaffinization, sections were saturated with TBS containing 1% BSA and 1% horse serum for 1 hour at room temperature. Tumor sections were incubated with a mouse anti-human anti-integrin-αvβ3 antibody (1:100; ab7166, Abcam) overnight at 4°C. Then, sections were washed with TBS containing 0.25% Triton and incubated with a secondary biotinylated antibody (Vectastain ABC-HRP Kit, peroxidase, PK-4002, vector), followed by 3,3′-diaminobenzidine stain (Vector).

DAOY-Luc spheroids (eight per animal) or HD-MB03-Luc cell suspensions (5,000 cells per animal) were stereotaxically implanted into the brains of 9-week-old Rj:NMRI-Foxn1 nude (nu/nu) female mice (Janvier Labs). Briefly, DAOY spheroids were generated with 2,500 cells grown for 48 hours in ultralow-adhesion spheroid 96-well plates (Corning). DAOY-Luc spheroids and HDMB-03-Luc cells were implanted into the left cerebellar hemisphere (2 mm posterior, 1.5 mm left of the lambda point, and 2.5 mm deep) using a Hamilton syringe fitted with a needle (Hamilton) and following a previously described procedure (15). Cilengitide was resuspended in 200 µL of an aqueous solution of 5% DMSO in PBS. Mice were administered 300 µg cilengitide three times per week via intraperitoneal injection. Intraperitoneal with vehicle solution served as a control. Survival of mice was evaluated by daily monitoring using neuropathological symptoms, including gait abnormalities and weight loss >10% as endpoints. At least 6 mice per group were selected to achieve sufficient statistical power.

Mice were imaged 2 days after cell implantation, and tumor growth was evaluated by bioluminescent imaging (BLI; IVIS, PerkinElmer). Mice were monitored for up to 170 days (DAOY-derived cells) or 50 days (HD-MB03–derived cells) after intraperitoneal injection of 3.3 mg of d-Luciferin dissolved in 100 µL of PBS. Ten minutes postinjection, mice were anesthetized with isoflurane and imaged with an IVIS Spectrum (field of view: C; binning: medium, f-stop: 1; exposure time: 1 minute). Bioluminescence signals [total flux (photons/second)] were quantified using Living Image 2.0 (Caliper Life Sciences).

The in vivo imaging studies used 12 nine-week-old Rj:NMRI-Foxn1 nude (nu/nu) female mice (Janvier Labs). Acquisitions were made using dedicated MRI and single-photon emission computed tomography (SPECT)/CT MRI systems on 6 mice bearing DAOY tumors and 6 mice bearing HD-MB03 tumors 7 and 4 weeks after tumor implantation, respectively, (Nanoscan PET/MRI3T and Nanoscan SPECT/CT; Mediso). Brain tumors were visualized using T1-weighted coronal images acquired after shimming and 2 minutes after intravenous administration of 100 µL of DOTAREM (0.5 mmol/mL gadoteric acid) using a fast spin-echo MRI sequence (repetition/echo time = 526/27.8 ms, 12 average, 32 slices, and voxel size = 0.34 × 0.34 × 1 mm³). Once tumors were observable on MRI acquisitions, SPECT/CT acquisitions were performed 1 hour after intravenous injection of 63.0 ± 12.6 MBq of ^99mTc-RAFT-RGD. SPECT quantification was performed using volumes of interest located within the tumor and contralateral healthy cerebellum, and the tumor-to-contralateral ratio was determined (VivoQuant, Invicro). Then, the animals were euthanized using CO₂ inhalation, and the brain was harvested. Following ex vivo gamma-well counting, left and right cerebellum samples were frozen, and 20-µm-thick slices were obtained (NX50V, MM France). After overnight exposure to a phosphor-screen (BAS-IP SR 2025 E, Cytiva), autoradiographic images were acquired (Typhoon IP, Cytiva) and quantified using dedicated software (ImageQuantTL). Regions of interest were drawn around tumor lesions and healthy contralateral cerebellum on a minimum of three slices. Results were background corrected and expressed as average tumor-to-contralateral ratios.

All animal experiments were conducted in strict accordance with the recommendations of the Guide for the Care and Use of Laboratory Animals. All animal studies were approved in advance by the local animal care committee (Veterinary Service and Direction of Sanitary and Social Action of Monaco; APAFIS # 19480-2019022616164184v4).

Data are presented as mean ± SEM. Data were compared between groups using nonparametric Mann–Whitney (two groups) or two-way ANOVA corrected for multiple comparisons using Sidak test (>2 groups). Tumor growth was compared between groups using two-way ANOVA corrected for multiple comparisons using Sidak test. Survival was compared between groups using the log-rank test (Mantel–Cox). A P < 0.05 was statistically significant.

The data generated in this study are available upon request from the corresponding author.

Integrin-αvβ3 expression was first assessed in a tissue array of medulloblastoma-derived samples (Fig. 1A). Among the 20 patients analyzed, 20% (4/20) expressed significant integrin-αvβ3 levels. Because several integrins are upregulated in brain tumors, their expression was examined in several medulloblastoma cell lines (ONS-76, DAOY, D458, CHLA-01Med, and HD-MB03). Most integrin genes were overexpressed in DAOY and ONS-76 (SHH group) cell lines but showed low expression in HD-MB03, D458, and CHLA-01 (groups 3 and 4) cell lines (Fig. 1B). Among them, high ITGB3 expression was detected in DAOY and ONS-76. This observation was confirmed at the protein level with higher expression in DAOY cells (Fig. 1C). To investigate the role of αvβ3 in medulloblastoma tumorigenesis, ITGB3 was knocked out (KO) in DAOY cells using CRISPR/Cas9 technology, and β3 was overexpressed in DAOY-KO and HD-MB03 cells (Fig. 1D and E; Supplementary Fig. S1).

Because a major function of integrin-αvβ3 is cell adhesion to the ECM, an adhesion assay was first performed on various ECM proteins (Supplementary Fig. S2). Genetic ablation of the β3-subunit in both KO clones #22 and #36 resulted in significantly decreased adhesion of DAOY cells to fibronectin and fibrinogen. Overexpression of the β3-subunit in DAOY-KO clones or HD-MB03 cells restored cell adhesion to these ECM proteins (Supplementary Fig. S2). The proliferation of DAOY-KO cells was significantly slower than control and β3-overexpressing cells (Fig. 1F). The same results were observed with HD-MB03 cells (Fig. 1G). Analysis of downstream-integrin-αvβ3 signaling revealed decreased pFAK, pAkt, and pERK1/2 levels in DAOY-KO and HD-MB03_WT cells (Supplementary Fig. S3). As a key player in cell adhesion, the role of integrin-αvβ3 in migration and invasion processes was further investigated. Genetic ablation of the β3-subunit decreased cell migration of both KO clones by 80% (Fig. 1H), while its overexpression in HD-MB03 cells increased cell migration by 600% (Fig. 1I). Similar observations were made in invasion assays (Fig. 1J and K). Restitution of β3-integrin in DAOY-KO cells completely abolished the anti-migratory and anti-invasive effects of β3-depletion. Therefore, integrin-αvβ3 promotes medulloblastoma cell proliferation, migration, and invasion.

The fundamental role of integrin-αvβ3 in tumorigenesis features prompted us to investigate its role in vivo in orthotopic medulloblastoma models (Fig. 2). β3-depletion in DAOY cells impaired orthotopic tumor growth (Fig. 2A–C), increasing the median survival of mice bearing DAOY-KO_#22 and KO_#36 tumors by 95% and 54%, respectively, compared with control (P < 0.001 vs. DAOY_control; Fig. 2E). In HD-MB03 xenografts, β3 expression increased tumor growth rate and decreased mouse survival by 20% compared with control tumors (Fig. 2B, D, and F). Consistent with these observations, ex vivo analyses show an increase in Ki67-positive nuclei in β3-expressing DAOY or HD-MB03 tumors (Fig. 2G and H). Altogether, these results highlight the protumorigenic role of integrin-αvβ3 in medulloblastoma.

To mimic β3-depletion, we explored the pharmacologic impact of cilengitide, an RGD-derived compound, on medulloblastoma cell lines in vitro (Fig. 3). The IC₅₀ of cilengitide was determined using MTT or adhesion assays. In both assays, the effect of cilengitide was restricted to β3-expressing medulloblastoma cell lines with IC₅₀s of <2 and <10 µmol/L in DAOY-WT and HD-MB03-β3+, respectively (Supplementary Fig. S4). Consistent with the genetic approach, cilengitide impaired β3-downstream signaling in DAOY cells in a dose-dependent manner (Fig. 3A). This effect resulted in antiproliferative and anti-invasive effects at 0.5 and 2.5 µmol/L (Fig. 3C–E). Moreover, 48 hours treatment with cilengitide induced cell apoptosis in DAOY-WT cells (Supplementary Fig. S5).

To substantiate this observed phenotype, we conducted parallel investigations employing an anti-integrin-αvβ3 specific antibody (LM609 clone). LM609 treatment decreased impaired β3-downstream signaling in DAOY cells (Fig. 3B). This led to decrease in cell adhesion, invasion, and proliferation of DAOY_WT cells with no impact on DAOY_KO#22 (Fig. 3D and E, Supplementary Fig. S4–S6).

Subsequently, we assessed the efficacy of cilengitide in orthotopic models of DAOY-WT, DAOY-KO, HD-MB03_WT, and HD-MB03_β3+ tumors to explore its relevance for integrin-αvβ3–expressing medulloblastomas (Fig. 4). Cilengitide treatment resulted in a delay in tumor growth only in mice with DAOY-WT or HD-MB03_β3+ tumors (Fig. 4A and B; Supplementary Fig. S7). This antitumor effect translated into a significant increase in mouse survival, with an 81% and 46% extension of survival observed in the DAOY-WT and HD-MB03_β3+ models, respectively (Fig. 4C, left; Supplementary Fig. S7C). Conversely, no significant effects were observed in the DAOY-KO and HD-MB03 tumors, underscoring the specificity of cilengitide for tumoral integrin-αvβ3 (Fig. 4A–C).

Because of its antiangiogenic function, cilengitide administration resulted in a 60% decrease in intratumoral blood vessels in each model (Fig. 4E). However, a decrease in Ki67 nuclei was observed only in DAOY-WT tumors, consistent with cilengitide's antitumor effect (Fig. 4D, left).

Cilengitide's restricted efficacy for αvβ3-positive tumors highlights the need for patient stratification. As a key component of personalized medicine, nuclear imaging is a relevant strategy to answer this issue. Therefore, we investigated the ability of ^99mTc-RAFT-RGD, a radioligand targeting integrin-αvβ3, to noninvasively evaluate its expression in DAOY and HD-MB03 orthotopic tumors (Fig. 5). The blood–brain barrier permeability was found to be permeable in both tumor models as demonstrated by gadolinium induced enhancement in T1-weighted fast spin-echo images. SPECT-MRI acquisitions showed that ^99mTc-RAFT-RGD uptake was readily observable in DAOY tumors, while no signal was found in HD-MB03 tumors (Fig. 5A). Tumor-to-cerebellum ratios were significantly higher in DAOY tumors than in HD-MB03 tumors (P < 0.010; Fig. 5B). Autoradiography of tumor slices was consistent with SPECT imaging quantification, with higher tumor-to-cerebellum ratios in DAOY tumors (Fig. 5C and D). These results suggest that radiotracers targeting αvβ3, such as ^99mTc-RAFT-RGD, could be used as companion markers to identify responders before cilengitide administration.

Because radioresistance is an important issue in brain tumor therapy, αvβ3-integrin expression was examined in radioresistant DAOY and HD-MB03 cells generated after multiple X-ray irradiations (Fig. 6). Cell survival was higher for radioresistant DAOY and HD-MB03 (DAOY-RR/HD-MB03-RR) than for naïve cells after a single irradiation (Supplementary Fig. S8). The mRNA analysis performed in these cells showed a significant increase in ITGB3 expression in the radioresistant cells (Fig. 6A). This observation was confirmed at the protein level with higher expression in radioresistant cells (Fig. 6B).

The effect of cilengitide was further investigated in vivo in both models. It decreased cell proliferation of both naïve and radioresistant DAOY cells. However, its effect was limited to radioresistant HD-MB03 cells expressing integrin-αvβ3 (Fig. 6C and D). Invasion assays performed with these cells showed a significantly higher invasion potential for radioresistant cells than for naïve cells (Fig. 6E and F). Cilengitide decreased the number of invasive cells in the DAOY WT and radioresistant populations but only in the HD-MB03 radioresistant population. Cilengitide induced cell apoptosis in these cells, as shown by PARP cleavage after 24 hours of treatment (Fig. 6G and H). Altogether, these results suggest that αvβ3-integrin is a potential therapeutic target for radioresistant medulloblastoma.

Medulloblastoma stands as the most frequently diagnosed primary brain tumor among children (16, 17). The standard treatment protocol comprises maximal safe resection followed by craniospinal radiation and adjuvant therapy. However, approximately 30% of children with medulloblastoma experience disease relapse, which is almost always fatal (18). Therefore, recurrent medulloblastoma is a major therapeutic challenge. Given the apparent need to improve therapeutic strategies, considerable efforts have been made to understand the biology of brain tumors. The low survival rates of brain tumors, including medulloblastoma and GBM, are at least partly due to extensive brain tissue invasion.

Invasion is finely regulated and is mainly controlled by cancer and stromal cell interactions (19, 20). In addition to invasive features, GBM and medulloblastoma show marked tumor cell proliferation and increased angiogenesis. All these processes have mainly been studied in GBM, where integrins have been shown to be widely expressed and to play fundamental roles (21–23). Therefore, integrins have emerged as compelling targets for therapeutic intervention in GBM. Among them, the proangiogenic αvβ3 was the first to be found to be abundantly expressed in high-grade brain tumors (23).

Integrin-αvβ3 belongs to the integrin subtypes that recognize the tripeptide RGD sequence found in many ECM proteins, including fibronectin or vitronectin. Many studies have highlighted the role of αvβ3 in sustaining GBM's high proliferative, migrative, and invasive properties and promoting angiogenesis (24–26). Nevertheless, the expression and the role of integrin-αvβ3 in medulloblastoma remain relatively unexplored. In 2005, Lim and colleagues reported moderated integrin-αvβ3 expression in 5 patients with medulloblastoma (27). In our study, we detected a substantial integrin-αvβ3 expression in 20% of patient-derived samples, underscoring its significance for a subpopulation of patients with medulloblastoma. The restricted integrin-αvβ3 expression found in medulloblastoma-derived samples was corroborated in medulloblastoma cell lines, where only two of five exhibited ITGB3 overexpression.

Within the SHH subgroups, DAOY and ONS-76 exhibited integrin expression, while D458 and HD-MB03 (group 3) as well as CHLA (group 4) did not display such expression. These findings imply the potential restriction of integrin-αvβ3 to a specific molecular subset of medulloblastomas. To validate this observation, further comprehensive studies are warranted. Afterward, the role of integrin-αvβ3 in medulloblastoma was explored through genetic manipulation: the β3-subunit was depleted in DAOY cells, and overexpression was induced via lentiviral transduction in DAOY-KO and HD-MB03 cells. While integrin-αvβ3 was conventionally associated with neovessels expression, IHC analysis of GBM samples revealed significant expression by tumor cells rather than angiogenic cells, thereby promoting GBM cell proliferation, migration, and invasion (23). In both cell lines, integrin-αvβ3 was a major player in these three processes. As orthotopic experiments showed, depleting the β3-subunit in two independent DAOY clones strongly reduced their tumorigenic potential. To a lesser extent, similar results were observed in HD-MB03 cells. These results support the original role of integrin-αvβ3 in medulloblastoma tumorigenesis. These results are in agreement with those obtained by Franovic and colleagues in orthotopic models of GBM using shRNA against β3-subunit (28).

Since its initial discovery, the core integrin-binding domain RGD in fibronectin has attracted considerable attention in the field of anticancer therapies. Among the various RGD-containing peptides developed to impair tumorigenesis, cilengitide was identified as a selective αvβ3 and αvβ5 inhibitor (29–31). This compound demonstrated antiangiogenic, cytotoxic, and anti-invasive activities in preclinical GBM models. Despite this promise, subsequent phase III CENTRIC and phase II CORE clinical trials showed no significant effects on OS in newly diagnosed GBM (12, 13). Nevertheless, in the CORE study, high αvβ3 levels in tumor cells were associated with improved OS. This finding suggests that cilengitide's efficiency was mediated by its action on αvβ3-positive tumor cells rather than angiogenic cells. Nonetheless, the mere presence of αvβ3-positive tumor cells might not prove adequate to render tumors receptive to integrin inhibition. For instance, recent findings by Cosset and colleagues, unveiled that the effectiveness of αvβ3 antagonists (such as cilengitide or LM609) in GBM models is confined to tumors reliant on GLUT3 (glucose transporter 3)-dependent glucose uptake (32).

In this study, we conducted an evaluation of cilengitide within our medulloblastoma models. The MTT assay showed that cilengitide did not affect HD-MB03 cells, while it strongly decreased the proliferation and viability of DAOY and HD-MB03-β3–overexpressing cells. The DAOY-KO cells were found to be only slightly affected by cilengitide treatment, with an IC₅₀ >20-fold higher than that of DAOY-WT cells. This difference could be due to αvβ5 expression in DAOY-WT and -KO cells. Significantly, cilengitide exhibited a dose-dependent impairment of downstream integrin-αvβ3 signaling, evidenced by diminished phosphorylation of FAK, Akt, and ERK1/2. These molecular repercussions translated into noteworthy anti-proliferative, anti-migratory, and anti-invasive effects observed in DAOY.

Cilengitide was subsequently evaluated in vivo in orthotopic models of DAOY-WT, DAOY-KO, HD-MB03, and HD-MB03_β3+ models. As anticipated, cilengitide treatment led to a notable reduction in blood vessel density across our models. This aligns with cilengitide's well-documented potent antiangiogenic properties. However, the antitumor effect of cilengitide was observed exclusively in tumors with high integrin-αvβ3 expression (DAOY, and HD-MB03_β3+), leading to a significant increase in median survival in both models. This discovery underscores the critical importance of patient stratification before initiating cilengitide therapy, echoing findings from the CORE study conducted with patients with GBM. Further investigations are required to ascertain the mere presence of αvβ3 is adequate to confer tumor sensitivity to cilengitide. Given the intriguing observations made by Cosset and colleagues regarding the correlation between αvβ3-antagonist efficacy and GLUT3 addiction in GBM, it would be prudent to explore the transposition of these findings to the context of medulloblastoma. An option is to perform GLUT3 invalidation experiments to assess cilengitide effects in medulloblastoma models with suppressed GLUT3 expression. This could reveal whether the observed cilengitide sensitivity linked to αvβ3 presence is influenced by GLUT3 status in medulloblastoma.

Noninvasive molecular imaging presents a promising avenue for confirming target existence and monitoring tumor response through tracking its expression. Over recent decades, several αvβ3-targeting radiotracers have been developed and investigated for potential clinical application (8). Most are based on the tripeptide RGD due to its high affinity and specificity for integrin-αvβ3. Preclinical and clinical studies have found no physiologic brain uptake of different RGD peptides. However, clinical studies showed significant uptake in patients with medulloblastoma (8). In our study, we used a ^99mTc-radiolabeled tetrameric RGD-based peptide (RAFT-RGD) previously validated for imaging and internal vectorized therapy in preclinical GBM models (33). In alignment with our findings regarding cilengitide activity, our results revealed higher uptake of ^99mTc-RAFT-RGD in DAOY tumors compared with HD-MB03 tumors.

The combination of SPECT and MRI in these studies paves the way for nuclear imaging use in patients with medulloblastoma. Our results with radioresistant medulloblastoma suggest that integrin-αvβ3 is expressed in radioresistant tumors (34). Therefore, SPECT/MRI in recurrent medulloblastoma could be considered to determine eligibility for cilengitide treatment. Indeed, accumulating evidence supports the role of integrin-αvβ3 in resistance to conventional therapies. Other studies have reported integrin-αvβ3 expression in radioresistant gliomas and prostate cancer tumors (34, 35). If integrin-αvβ3 expression is an important determinant of cilengitide efficacy, several factors, such as redundancy or compensatory mechanisms, could explain its limited efficacy in the clinic.

The limited efficacy of cilengitide in clinical application suggests the significance of integrating integrin-αvβ3 imaging as a prerequisite before initiating treatment. Furthermore, it emphasizes the need to explore alternative strategies beyond pharmacologic inhibition. The remarkable advances in nuclear medicine, especially targeted therapies, may provide new tools for treating recurrent medulloblastoma. Most therapeutic radiopharmaceuticals are labeled with β-emitting isotopes. These particles penetrate only a few millimeters into tissue and allow irradiation of cells within a limited radius, resulting in tumor cell death while sparing surrounding healthy tissue. Commonly used β-emitters include lutetium-177 (¹⁷⁷Lu), which has a half-life of 6.7 days and relatively high β-emission (0.497 MeV). These characteristics offer the advantage of delivering a high dose and achieving prolonged tumor irradiation. In addition, a unique feature of radionuclides is that they can exert a “cross-fire” effect that destroys neighboring cells. Several ¹⁷⁷Lu-radiolabeled RGDs have been studied in preclinical GBM models. One, the ¹⁷⁷Lu-NOTA-EB-RGD, was found to completely eradicate tumor growth from αvβ3-expressing patient xenografts (36). Although all these compounds require further clinical investigation, they may have great potential for integration into combined therapies for brain tumors, including medulloblastomas. Future perspectives of our study include the evaluation of ¹⁷⁷Lu-RAFT-RGD in mouse medulloblastoma models, which could be the beginning of the era of theranostics in these tumors.

C. Ortholan reports personal fees from Brainlab outside the submitted work. V. Picco reports grants from GEMLUC and Fondation Flavien during the conduct of the study. No disclosures were reported by the other authors.

W. Echavidre: Data curation, formal analysis, validation, investigation, visualization, writing-original draft. J. Durivault: Resources, data curation, formal analysis, validation, investigation, visualization, methodology. C. Gotorbe: Data curation, validation, investigation. T. Blanchard: Data curation, formal analysis, validation, investigation, visualization. M. Pagnuzzi: Data curation, formal analysis, validation, investigation. V. Vial: Resources, data curation, formal analysis, validation, investigation. F. Raes: Data curation, validation, investigation. A. Broisat: Data curation, validation, investigation. R. Villeneuve: Resources, data curation, formal analysis, investigation, visualization, methodology. R. Amblard: Resources, data curation, formal analysis, investigation, visualization, methodology. N. Garnier: Resources, data curation, formal analysis, investigation, visualization, methodology. C. Ortholan: Resources, data curation, formal analysis, investigation, visualization, methodology. M. Faraggi: Data curation, formal analysis, validation. B. Serrano: Resources, data curation, formal analysis, investigation, visualization, methodology. V. Picco: Conceptualization, resources, data curation, formal analysis, supervision, funding acquisition, validation, investigation, visualization, methodology, project administration. C. Montemagno: Conceptualization, resources, data curation, software, formal analysis, supervision, funding acquisition, validation, investigation, visualization, methodology, project administration.

The Government of the Principality of Monaco, Fondation Flavien, AS Monaco FC, and Groupement des Entreprises Monégasques dans la Lutte contre le Cancer (GEMLUC). This work was partly supported by France Life Imaging (grant no. ANR-11-INBS-0006).