Hippo pathway suppression reprograms TNFα-primed glioblastoma extracellular vesicles transcripts cargo to drive mesenchymal stem/stromal cells vasculogenic mimicry

Abstract

Background

Glioblastoma (GBM) secrete extracellular vesicles (EVs) which play a pivotal role in brain tumor progression by mediating intercellular communication within the inflamed tumor microenvironment (TME). EVs’ cargo transports biomolecules that promote tumor progression, immune evasion, and resistance to therapies. While Hippo inhibitors play a significant role in mitigating cancer inflammation, their specific impact on EVs cargo remains unknown.

Methods

Human grade IV U87 GBM-derived cells were cultured and EVs isolated from the conditioned media of tumor necrosis factor (TNF)α-primed cells. Total RNA was extracted using TRIzol™, and differential gene expression assessed through gene arrays and validated by RT-qPCR. Protein cell and EVs lysates were used for immunoblotting. 3D mesenchymal stem/stromal cells (MSC) in vitro vasculogenic mimicry (VM) was assessed using Cultrex matrices.

Results

Our study shows that U87 cells are responsive to pro-inflammatory stimulation by TNFα as the phosphorylation status of ERK, IκB, and NFκB increased. Among the Hippo pathway inhibitors tested, VT107 inhibited both the TNFα-induced phosphorylation, induction of the downstream Hippo pathway CYR61, and cargo of secreted EVs as assessed upon gene array screens. Pro-inflammatory genes that were reduced by VT107 in EVs included, among others, COX2, IL6, IL1B, and several members of the CCL, CXCL, and Interleukin/Interleukin receptors family. EVs isolated from VT107-treated TNFα-primed U87 cells had decreased paracrine regulation of MSC in vitro VM.

Conclusions

By inhibiting the Hippo pathway and TNFα-induced pro-inflammatory cargo of GBM-derived EVs, our data support VT107 as a potential candidate to inhibit tumor-promoting processes involved in therapy resistance such as paracrine induction of MSC-mediated VM.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12964-025-02401-x.

Background

Glioblastoma (GBM) represents a highly malignant and therapeutically resistant primary brain tumor, characterized by rapid proliferation, extensive infiltration, and adaptive mechanisms that limit treatment efficacy [1]. Extensive infiltrative growth further enables malignant cells to disseminate throughout the brain parenchyma, thereby rendering complete surgical resection exceedingly challenging [2]. GBM tumors are also highly heterogeneous, leading different regions of the tumor to respond differently to treatments, reducing the effectiveness of standard therapies [3]. While GBM resistance to chemotherapy and radiation requires activation of several survival pathways, the blood-brain barrier further appears to limit drug delivery, preventing many therapeutic agents from reaching the tumor site effectively [4–6]. Additionally, immunotherapy efforts are hindered by the tumor’s immunosuppressive environment, preventing immune cells from effectively targeting cancer cells [7]. Despite advances in research, personalized treatments remain limited due to the tumor’s unpredictable genetic alterations, and recurrence leads to poor prognosis. Overall, GBM represents a complex therapeutic challenge requiring new innovative and multimodal approaches for better patient outcomes.

The Hippo signaling pathway plays a crucial role in GBM chemoresistance by regulating tumor cell survival, proliferation, and adaptation to therapy [8, 9]. When dysregulated, this pathway promotes the activity of yes-associated protein (YAP) and transcriptional coactivator with a PDZ-binding domain (TAZ), that drive oncogenic gene expression and support drug resistance mechanisms [10]. GBM cells exploit Hippo pathway alterations to evade apoptosis, allowing them to persist despite chemotherapy-induced stress [8]. Additionally, Hippo signaling influences the tumor microenvironment (TME), enhancing interactions with stromal cells that shield GBM from therapeutic agents. This pathway also modulates autophagy, a survival strategy that tumor cells use to withstand chemotherapy [11, 12]. Hippo pathway modulation can also disrupt vasculogenic mimicry (VM), a process where tumor cells form blood vessel-like structures to sustain growth [13–15]. This can weaken the tumor’s ability to maintain its nutrient supply and limit its invasive potential. Recently, transcriptional regulation of CYR61 and CTGF by LM98, a synthetic YAP-TEAD inhibitor, was found to alter in vitro VM in GBM cells [14, 16]. Furthermore, Hippo-related signaling contributes to metabolic reprogramming, enabling GBM cells to sustain energy demands even under drug pressure [8]. Immune evasion is another consequence, as Hippo pathway dysregulation helps suppress anti-tumor immune responses [17]. Combined, these factors make targeting Hippo signaling a promising strategy for overcoming GBM chemoresistance and improving treatment outcomes.

Recent evidence has revealed important crosstalk between tumor necrosis factor-alpha (TNFα), a well-characterized pro-inflammatory cytokine, and the Hippo signaling pathway [18]. Mechanistically, TNFα influences the activity of core Hippo components, including MST1/2 and LATS1/2 kinases, through intermediates such as the Ras-association domain family (RASSF) proteins, which link TNFα receptor signaling to Hippo activation [19], and which shapes the metabolic and inflammatory landscape of GBM [20]. Additionally, TNFα’s clinical relevance, given its involvement in multiple disease contexts and its status as a target of anti-inflammatory therapies, offers potential translational implications for Hippo-modulating agents. From a practical standpoint, TNFα is a robust and reproducible ligand for in vitro assays, allowing for controlled activation and precise pathway analysis. While other molecules such as lysophosphatidic acid, epidermal growth factor, and mechanical stimuli can also influence Hippo activity, TNFα was prioritized in this exploratory study due to its strong biological relevance and feasibility as an experimental tool.

Targeting the Hippo pathway offers a promising avenue to also influence extracellular vesicle (EVs) communication, altering tumor-stroma interactions and preventing the spread of pro-survival signals [21, 22]. Whether Hippo pathway inhibitors may become valuable components of multimodal GBM therapy remains to be more documented. Targeting EVs cargo in GBM is crucial because these play a significant role in tumor progression, resistance, and communication within the TME [22, 23].

As EVs carry bioactive content such as proteins, RNA, lipids, and even organelles such as mitochondria that influence immune evasion, enhance tumor cell survival, and promote therapy resistance, this work questions whether by modifying or inhibiting specific cargo components, one can potentially disrupt the pathways that GBM cells rely on to proliferate and evade treatment [24]. Moreover, EVs also facilitate intercellular communication, increasing cell survival while spreading pro-tumor signals to surrounding cells, which contributes to angiogenesis and metastasis [25–28]. We therefore questioned whether regulating EVs cargo could help prevent paracrine regulation of chemoresistance associated biological process such as VM. Since EVs influence both local and distant tumor environments, manipulating their cargo could help suppress GBM progression and improve patient outcomes.

Methods

Materials

Sodium dodecyl sulfate (SDS) and bovine serum albumin (BSA) were purchased from Sigma-Aldrich Corp (St. Louis, MO, USA). Cell culture media Eagle’s Minimum Essential Medium (EMEM) was from Wisent (320-005 CL). Electrophoresis reagents were purchased from Bio-Rad Laboratories (Hercules, CA, USA). The HyGLO™ Chemiluminescent HRP (horseradish peroxidase) Antibody Detection Reagents were from Denville Scientific Inc (Metuchen, NJ, USA). Micro bicinchoninic acid (BCA) protein assay reagents were from Pierce (Micro BCA™ Protein Assay Kit; Thermo Fisher Scientific, Waltham, MA, USA). Antibodies against Src (#2101), phosphor-Src (Tyr416, #2101), STAT3 (79D7 – CAT# 4904 S), phospho-STAT3 (D3A7 – CAT# 9145 S), GAPDH (D4C6R – CAT# 97166 S), IκBα (4814 S), phospho-IκBα Ser32/36 (9246 S), p44/42 MAPK (ERK1/2), and phospho-p44/42 MAPK (ERK1/2) were from Cell Signaling Technology (Danvers, MA, USA). Antibodies against CD63 (10628D) and CD9 (Cat: 10626D) were from Invitrogen (Burlington, ON, Canada). HRP-conjugated donkey anti-rabbit and anti-mouse immunoglobulin (Ig) G secondary antibodies were from Jackson ImmunoResearch Laboratories (West Grove, PA, USA).

Cell culture

Cells and culture media were from the American Type Culture Collection (ATCC; Manassas, VA, USA). The human U87 GBM-derived cells were cultured in 10% fetal bovine serum (FBS). Cells were kept subconfluent and expanded by a 1:4 bi-weekly split, in a humidified incubator at 37 °C with 5% CO₂.

Total RNA extraction, cDNA synthesis, and real-time quantitative PCR

Total RNA was extracted from U87 cell monolayers and EVs using TRIzol, following the manufacturer’s recommendations (Life Technologies, Gaithersburg, MD, USA). Total RNA concentration was measured using a NanoPhotometer P330 (Implen) and 2 µg were reverse-transcribed into cDNA using a high-capacity cDNA reverse transcription kit (Applied Biosystems; Foster City, CA, USA). Samples were prepared with primer sets and SsoFast EvaGreen Supermix (Bio-Rad; 1725204). The primer sets used were all purchased from QIAGEN: The following QuantiTect primer sets were provided by QIAGEN: YAP1 (Hs_YAP1_1_SG, QT00080822), TEAD1 (Hs_TEAD1_1_SG, QT00000721), CTGF (Hs_CTGF_1_SG, QT00052899), CYR61 (Hs_CYR61_1_SG, QT00003451), AXL (Hs_AXL_1_SG, QT00067725), NF2 (Hs_NF2_va.1_SG, QT01004934), GAPDH (Hs_GAPDH_1_SG, QT00079247) and PPIA (Hs_PPIA_4_SG, QT01866137). Gene expression was quantified by real-time quantitative PCR (RT-qPCR) on a CFX Connect (Bio-Rad) with the Bio-Rad CFX manager Software version 3.0. The relative RNA quantities of each target gene were normalized against two housekeeping genes, GAPDH and PPIA, using the standard 2^−ΔΔCq method.

Human inflammation profiler PCR arrays

Premade RT² Profiler PCR arrays for Human Inflammatory Cytokines and Receptors (PAHS-011ZD) were purchased from QIAGEN and used following the manufacturer’s instructions. Briefly, the genomic DNA was removed before 0.5 µg of total RNA was reverse transcribed via the RT² First Strand Kit (QIAGEN, 330404). Each plate was used to assess one cDNA sample prepared with RT² SYBR Green qPCR Mastermix (QIAGEN, 330502). The relative expression analysis of 84 genes and controls was performed through the GeneGlobe analysis center, a website provided by QIAGEN (https://geneglobe.qiagen.com/us/analyze), via the standard fold change 2^−ΔΔCq method. Based on the overall number of genes and modulation profile, the fold regulation used in the figures for upregulated genes was cutoff > 2, and for downregulated genes cutoff < −2.

Protein-to-protein interactions

The database STRING v11 (https://www.string-db.org/) was used to identify and build protein-to-protein interaction networks [29], with a confidence score of 0.4. The maximum number of interactions shown was set to 10, to help with the readability of the predictions.

Western blot

Total cell lysis was performed in a buffer containing 1 mM each of sodium fluoride (NaF) and sodium orthovanadate (Na₃VO₄). Proteins were heat-denatured (5 min at 95 °C) except those from VEs. Proteins (8–20 µg) were then separated by SDS-polyacrylamide gel electrophoresis (PAGE). Next, proteins were electro-transferred to low-fluorescence polyvinylidene difluoride membranes and blocked for 1 h at room temperature with 5% nonfat dry milk in Tris-buffered saline (150 mM NaCl, 20 mM Tris-HCl, pH 7.5) containing 0.3% Tween-20 (TBST; Bioshop, TWN510-500). Membranes were washed in TBST and incubated overnight with the appropriate primary antibodies (1/1,000 dilution) in TBST containing 3% BSA and 0.1% sodium azide (Sigma-Aldrich) at 4 °C and on a shaker. After three washes with TBST, the membranes were incubated for 1 h with horseradish peroxidase-conjugated anti-rabbit or anti-mouse IgG at 1/2,500 dilutions in TBST containing 5% nonfat dry milk. Immunoreactive material was visualized by ECL.

Capillary-like structure formation assay (in vitro VM)

Human bone marrow-derived mesenchymal stem/stromal cells (MSC) were purchased from the ATCC. Cell culture media was from Life Technologies Corp (Carlsbad, CA, USA). Cells were plated in high glucose αMEM supplemented with 10% FBS and 50 units/mL penicillin/streptomycin and cultured in a humidified incubator at 37 °C with 5% CO₂. MSC were kept subconfluent and expanded in number over 10 passages by a 1:2 split on a weekly basis. VM was assessed in vitro using Cultrex (3432-010-01, R&D Systems) to monitor capillary-like structures formation [30, 31]. In brief, each well of a 96-well plate was pre-coated with 50 µl of Cultrex. MSC suspension in culture media (10⁴ cells/100 µl) was then seeded on top of polymerized Cultrex. Tested compounds were added to the cell culture media and incubated at 37 °C in a CO₂ incubator. Pictures were taken over time using a digital camera coupled to a phase-contrast inverted microscope. The number of loops upon tube branching formed by the cells were quantified using the Wimasis analysis software (https://www.wimasis.com; Cordoba, Spain).

Extracellular vesicles isolation

Twenty mL of U87 cell culture conditioned media were collected and subjected to sequential centrifugation steps. Initially, cell debris and apoptotic bodies were pelleted down at 4,000 x g for 10 min at 4 °C. The supernatant (S1) was then further centrifuged at 16,500 x g for 20 min at 4 °C. The resulting supernatant (S2) was subjected to ultracentrifugation at 100,000 x g for 24 h at 4 °C to produce a pellet fraction containing EVs. For ultracentrifugation, polycarbonate tubes were used in SW32 Ti Rotor, Swinging Bucket with Beckman Coulter Optima XE Ultracentrifuge, speed set at 24,100 rpm to reach RCF = 100,000 x g; maximum acceleration and deceleration; k-factor = [2.533 × 10⁵ × ln (rmax/rmin)/(RPM/1000)²] = 200,4, adjusted k Factor (k adj) = k-factor× (max rpm/actual rpm)² = 353.3). The final pellet enriched with exosomes was resuspended in 500 µL of PBS for differential light scattering (DLS) particle size analysis (see below) or in TRIzol for total RNA isolation and subsequent gene array analysis.

Flow cytometry characterization

The cell culture media supernatant was first filtered using a 1.2 μm syringe filter to remove residual cellular debris. Prior to antibody labeling, 1 µL of TruStain FcX™ (BioLegend) was added to each sample to block non-specific binding. For each sample, 5 µL of the filtered supernatant was then combined with 1 µL of anti-CD105-BV421 (BioLegend) and 93 µL of 0.2 μm filtered PBS. The mixture was incubated for 15 min at room temperature in the dark to allow antibody binding. Samples were analyzed using the Attune NxT flow cytometer (ThermoFisher Scientific) for quantification of EVs [27]. Gating strategies were established using a non-relevant anti-CD45-BV421 antibody (BioLegend) to define background signal for the U87-derived EVs.

Dynamic light scattering

EVs size/diameter was assessed by DLS as previously described [28]. The mean hydrodynamic diameter of EVs was calculated by fitting a Gaussian function to the measured size distribution. Prior to DLS measurements, each sample was centrifuged at 300 g for 10 s to pellet large aggregates; 50 µL of the sample was added to a ZEN00400 cuvette, and DLS measurements were conducted at 25 °C using a Nano ZSP Zetasizer (Malvern Instruments Ltd., UK) operating at 633 nm and recording the back-scattered light at an angle of 175°. The sample was allowed to equilibrate for 2 min before each measurement. DLS was recorded for 200 s with three replicate measurements. Signal intensity was transformed to volume distribution assuming a spherical shape of EVs and using the Malvern Instruments Ltd. software.

Statistical data analysis

All statistical analyses were conducted using the GraphPad Prism 7 software (https://www.graphpad.com; San Diego, CA). Data and error bars are presented as the mean ± standard error of the mean (SEM) from three or more independent experiments, unless otherwise specified. Analysis was performed using the one-way ANOVA (Figs. 2A, 4D and 6) or the two-way ANOVA (Figs. 2B and 7E). Probability values of less than 0.05 (*) were considered significant.

Fig. 2.

VT107 inhibits TNFα-induced expression of the downstream Hippo signaling effector CYR61 in human U87 glioblastoma cells. (A) Serum-starved human U87 glioblastoma cells were treated with vehicle (white bars) or with 30 ng/ml TNFα (black bars) for 24 h, and total RNA isolated as described in the Methods section. Expression levels of the indicated genes were assessed by RT-qPCR. (B) Serum-starved human U87 glioblastoma cells were treated with vehicle (white bars) or with 30 ng/ml TNFα (black bars) for 24 h in the presence or not of 1 µM of the indicated Hippo signaling pathway inhibitors (GNE7883, IAG933 and VT107), and total RNA isolated as described in the Methods section. Expression levels of CYR61 and of GAPDH were assessed by RT-qPCR. (*p < 0.05)

Fig. 4.

VT107 alters the COX2 and IL6 gene expression in extracellular vesicles cargo isolated from TNFα-treated U87 glioblastoma cells. Serum-starved U87 glioblastoma cells were treated or not with 30 ng/ml TNFα, or with a combination of 30 ng/ml TNFα and 1 µM VT107. Conditioned media was next used to isolated extracellular vesicles (EVs) as described in the Methods section. (A) A representative staining is shown of U87-derived EVs with anti-CD105 BV421 antibody to assess the presence of mesenchymal markers (left panel). A clear population of CD105⁺ events is observed in the designated gate. As a specificity control, an irrelevant antibody (anti-CD45 BV421) was used on an identical sample (right panel), showing minimal background staining. Events were acquired on a logarithmic scale using FSC-H versus VL1-H, and gating was applied to identify positive EVs based on fluorescence intensity. (B) Representative differential light scattering profiles of EVs size distribution. (C) Protein levels of the CD9 and CD63 exosomal biomarkers was assessed in EVs by Western blotting. (D) Total RNA was extracted from untreated, TNFα-, or TNFα/VT107-treated U87 glioblastoma cells-derived EVs and RT-qPCR used to assess the expression of COX2 and IL6 transcript levels

Fig. 6.

VT107 prevents the transcriptional regulation of TNFa-regulated genes in U87 glioblastoma cells. Total RNA was extracted from EVs secreted upon untreated, TNFα-treated, or TNFα/VT107-treated U87 glioblastoma cells. Gene expression levels of the indicated (A) CCL family members, (B) cytokines or growth factors, or (C) CXCL5 is shown

Fig. 7.

VT107 alters the capacity of extracellular vesicles isolated from TNFα-treated U87 glioblastoma cells to trigger in vitro mesenchymal stem cells vasculogenic mimicry. (A) Representative phase contrast images were taken to monitor the formation of mesenchymal stem/stromal cells capillary-like structures for up to 6 h (upper panels) and were analyzed via Wimasis (lower panels). (B) Quantification from Wimasis analysis of the total loops over time. (C) Cultrex was depleted from soluble growth factors, then used to monitor the capacity to re-induce capillary-like structure formation upon 6 h (upper panels). The structures were analyzed via Wimasis (lower panels). Total loops were quantified to assess in vitro VM. (D) The impact of serum-free media (closed triangles), EVs (open circles) and of serum-enriched media (closed circles), or (E) the impact of EVs isolated from control, 30 ng/ml TNFα-, 1 µM VT107-, or TNFα/VT107-treated cells, was assessed on in vitro VM

Results

TNFα triggers the phosphorylation of ERK, IκB, and NFκB p105 in human U87 glioblastoma cells

U87 cell responsiveness to increasing concentrations of TNFα was first assessed (Fig. 1A). ERK (extracellular signal-regulated kinase) phosphorylation, a key event in the mitogen-activated protein kinase (MAPK) pathway which regulates processes like cell division, differentiation, and survival, was found to be phosphorylated by TNFα, along with activation of the canonical nuclear factor-kappa B (NFκB) pathway, leading to the phosphorylation of IκB and NFκB p105 (precursor of p50) (Fig. 1B, closed circles). Interestingly, signal transducer and activator of transcription 3 (STAT3) as well as Src phosphorylation was not triggered by TNFα in U87 cells (Fig. 1B, open circles).

Fig. 1.

TNFα triggers the phosphorylation of ERK, IκB, and NFκB p105 in human U87 glioblastoma cells. (A) Serum-starved human U87 glioblastoma cells were treated with the indicated increasing concentrations of TNFα for 30 min, then cell lysates (20 µg) assessed for the indicated phosphoprotein and total protein expression by SDS-PAGE. (B) Representative densitometric analysis of the ratios of phosphoprotein/protein expression were plotted

Pharmacological effects of Hippo signaling inhibitors GNE7883, IAG933 and VT107 in TNFα-primed human U87 glioblastoma cells

The Hippo signaling pathway controls the upstream activity of YAP and TAZ [32, 33]. Moreover, some key downstream effectors of the Hippo pathway further include CYR61 and CTGF, as these are direct transcriptional targets of YAP/TAZ and play roles in angiogenesis, fibrosis, and tumor progression [34]; AXL, a receptor tyrosine kinase that promotes tumor metastasis and therapy resistance when upregulated by YAP/TAZ [35, 36]; and NF2 (Merlin). YAP has also been shown to play a role in sustaining resistance to targeted therapies as well [37]. When human U87 cells were treated with TNFα, an upregulation of the relative YAP1, CTGF, AXL, and CYR61 transcript levels was observed with the best induction being that of CYR61 by TNFα (Fig. 2A). Three pharmacological Hippo signaling inhibitors were next tested for their potential to alter TNFα response. These included VT107, a pan-TEAD auto-palmitoylation inhibitor which interferes with TEAD-mediated gene transcription and Hippo pathway signaling currently being studied in clinical trials against mesothelioma [38], GNE7883, a pan-TEAD inhibitor that blocks the association of YAP/TAZ with TEAD, inhibiting cell proliferation in multiple cancer cell models [39], and IAG933, a TEAD inhibitor developed by Novartis and currently being evaluated in a Phase I clinical trial for patients with advanced mesothelioma and other solid tumors [40, 41]. We found that all three were capable to reduce TNFα-induced CYR61 transcript levels, with VT107 being the best inhibitor among them all (Fig. 2B).

The Hippo pathway inhibitor VT107 prevents the TNFα-mediated phosphorylation of ERK, IκB, and NFκB-P105 in human U87 glioblastoma cells

To assess whether any crosstalk with the Hippo signaling pathway regulates TNFα signaling, U87 cells were treated or not with increasing concentrations of VT107 and primed with TNFα (Fig. 3A). When cells were treated with TNFα, phosphorylation of ERK, IκB and NFκB p105 was reduced dose-dependently by VT107 with maximal inhibition obtained between 1 and 3 µM (Fig. 3B). Altogether, this confirms that a signaling crosstalk links Hippo signaling to TNFα-mediated pro-inflammatory signaling.

Fig. 3.

The Hippo pathway inhibitor VT107 inhibits the TNFα-mediated phosphorylation of ERK, IκB, and NFκB p105 in human U87 glioblastoma cells. (A) Serum-starved human U87 glioblastoma cells were treated with the indicated increasing concentrations of VT107 for 24 h, then stimulated with 30 ng/ml TNFα for 30 min, and cell lysates (20 µg) assessed for the indicated phosphoprotein and total protein expression by SDS-PAGE. (B) Representative densitometric analysis of the ratios of phosphoprotein/protein expression were plotted as a percent of maximal TNFα effect

VT107 alters the pro-inflammatory COX2 and IL6 cargo expression in extracellular vesicles isolated from TNFα-primed U87 glioblastoma cells

Studying extracellular vesicles (EVs) provides original insights into how inflammation fuels cancer growth and resistance to therapy. Isolating EVs further becomes crucial for studying cancer-related inflammation as they further play a major role in intercellular communication within the TME [42]. EVs were isolated from TNFα-primed U87 glioblastoma cells as described in the Methods section, and confirmed the presence of CD105⁺ EVs to validate our isolation method, with significantly higher detection compared to labeling with the irrelevant anti-CD45 control (Fig. 4A) supporting the specificity of CD105 expression on U87-derived EVs. Importantly, as shown from the differential light scattering (DLS) profiles, neither TNFα nor combined TNFα/VT107 treatments altered the size of the EVs as their diameters ranged from 30 to 1000 nm, with a majority around 300 nm (Fig. 4B). Final characterization was performed at the protein level as the CD9 and CD63 exosomal biomarkers expression also remained unchanged (Fig. 4C). When total RNA was extracted from EVs of the respective conditions, RT-qPCR showed that transcript levels of COX2 and IL6 pro-inflammatory biomarkers were induced upon TNFα treatment, while VT107 significantly prevented such increase (Fig. 4D). This prompted us to further explore the potential VT107-mediated alterations in the EVs pro-inflammatory cargo.

VT107 alters the transcriptomic pro-inflammatory signature in the extracellular vesicles cargo isolated from TNFα-treated U87 glioblastoma cells

To address the potential anti-inflammatory molecular signature that VT107 exerts against TNFα-mediated transcriptional reprogramming in EVs, a qPCR screen was performed using total RNA isolated from TNFα- or TNFα/VT107-treated U87 cells. Among the 84 genes investigated, 23 were significantly upregulated, including members of the CXCL (CXCL1), CCL (CCL2, CCL3, CCL5, and CCL20), and some interleukins/interleukin receptor (IL1A, IL1B, IL3, IL13, IL15, IL-21, IL10RA, and IL1R1) (Fig. 5A, green). In addition, 18 genes were concomitantly downregulated upon TNFα treatment, including several CCLs and CXCLs, as well as some of their receptors, and some interleukins (IL7, IL17C, IL27, and IL33) (Fig. 5A, red). When EVs were isolated from cells treated with the Hippo signaling inhibitor VT107 and cargo assessed, 20 of the TNFα-upregulated genes were found inhibited (Fig. 5B), and the top 10 inhibited genes described in (Supplementary Table 1). Similarly, 13 of the TNFα-downregulated genes were found inhibited by VT107 (Fig. 5D), and the top 10 inhibited genes also described in (Supplementary Table 1). The protein-to-protein interaction network predictions confirmed the multiple interactions among the 10 upregulated genes that were under the control of Hippo signaling (Fig. 5C), as well as those existing among the top 10 most downregulated genes (Fig. 5E). Validation of the screen was next performed by qPCR for CCL2, CCL3, CCL5, and CCL20 genes (Fig. 6A), and CSF3, IL1B, and VEGF (Fig. 6B), where TNFα-induced gene expression was effectively prevented by VT107. CXCL5 gene downregulation was also validated in response to TNFα, with VT107 preventing that reduction (Fig. 6C). Collectively, this suggests that targeting the Hippo signaling pathway can effectively alter the inflammatory molecular signature of EVs and likely to control intercellular events.

Fig. 5.

VT107 alters the transcriptomic pro-inflammatory signature in extracellular vesicles cargo isolated from TNFα-treated U87 glioblastoma cells. (A) Total RNA was extracted from EVs secreted from untreated or TNFα-treated U87 glioblastoma cells. Gene expression levels of upregulated genes (green arrow) and downregulated genes (red arrow) are represented as the fold regulation from inflammation and cytokines gene arrays. (B) Histograms representing the number of genes that were induced upon TNFα, and the extent of their inhibition when cells were co-treated with 1 µM VT107. (C) Protein-to-protein interaction network of the 10 most TNFα-induced genes whose expression was inhibited by VT107, as retrieved with STRING. (D) Histograms representing the number of genes that were reduced upon TNFα, and the extent of their inhibition when cells were co-treated with 1 µM VT107. (E) Protein-to-protein interaction network of the 10 most TNFα-downregulated genes whose expression was also inhibited by VT107, as determined with STRING

VT107 alters the capacity of extracellular vesicles isolated from TNFα-primed U87 glioblastoma cells to trigger mesenchymal stem/stromal cells in vitro vasculogenic mimicry

Paracrine regulation of EVs-mediated intercellular communication has recently been inferred in different biological processes [43, 44]. Given evidence that vascular progenitors derived from bone marrow stem/stromal cells (MSC) are avidly recruited by vascularizing tumors like GBM [45, 46], we hypothesized whether EVs’ cargo could influence MSC VM. In vitro VM was performed with MSC seeded on top of a Cultrex matrix as described in the Methods section (Fig. 7A), and total loops formation with time found to increase as analyzed using WIMASIS (Fig. 7B). In order to isolate and assess the sole impact of EVs’ cargo, we depleted Cultrex from its soluble growth factors to prevent in vitro VM (Fig. 7C, - serum). When MSC were next seeded on growth factors-depleted Cultrex, but in the presence of either serum or EVs, capillary-like structures were found re-induced (Fig. 7D). Interestingly, EVs isolated from TNFα-primed U87 cells were found to trigger more loops, while those isolated from VT107-treated cells prevented that increase (Fig. 7E). This suggests that EVs’ cargo can rapidly exert paracrine regulation of VM and potentially contribute to a more vascularized solid tumor within a pro-inflammatory TME.

Discussion

Glioblastoma (GBM) remains one of the most challenging malignancies to treat due to its aggressive nature and ability to evade therapeutic interventions. One of the key mechanisms underlying GBM progression involves the secretion of EVs, which facilitate intercellular communication within the TME. These EVs transport a variety of biomolecules - including proteins, lipids, and nucleic acids - that contribute to tumor immune evasion, therapy resistance, and enhanced proliferation. While the Hippo signaling pathway has been extensively studied in cancer biology for its role in regulating cell proliferation and apoptosis, its precise involvement in shaping EVs cargo and tumor paracrine signaling has remained unexplored. The findings of this study highlight a significant impact of Hippo inhibition on EVs-mediated VM and suggest VT107 as a promising candidate in modulating GBM progression within a pro-inflammatory TME.

In fact, we demonstrate that TNFα stimulation induces a pro-inflammatory response in U87 GBM cells, as evidenced by the increased phosphorylation status of ERK, IκB, and NFκB. This validated that GBM cells respond robustly to inflammatory cues, triggering pathways that can enhance their survival, and that could alter the composition of secreted EVs. Importantly, treatment with the Hippo pathway inhibitor VT107 effectively suppressed TNFα-induced signaling events, particularly reducing the downstream expression of CYR61, a key mediator of Hippo-associated oncogenic effects [11, 47, 48]. This inhibition also led to a marked reduction in pro-inflammatory factors within EVs’ cargo, including genes which encode COX2, IL6, IL1B, and multiple members of the CCL family. Such findings indicate that Hippo inhibition could not only disrupt direct intracellular signaling in GBM cells but also modify the extracellular communication landscape by altering the EVs cargo.

One of the most notable observations from our study is the impact of EVs isolated from VT107-treated cells on MSC and their ability to undergo VM. The ability of solid tumors to recruit MSC and to form vascular-like structures independently of an endothelial cell phenotype, can facilitate nutrient delivery and tumor expansion. The reduced paracrine regulation of MSCs in VM assays following VT107 treatment suggests that Hippo inhibition has the potential to impair tumor-associated neovascularization processes mediated upon EVs signaling. Given that GBM cells rely on both traditional angiogenesis and VM to sustain their rapid growth, targeting pathways that regulate these processes is critical for improving future treatment strategies.

The findings of this study further offer multiple therapeutic implications. Firstly, targeting Hippo signaling through VT107 may not only suppress GBM cell survival but also reprogram the TME by altering EVs-mediated signaling. This dual mechanism of action highlights the potential of Hippo inhibition as a strategy for disrupting communication networks that sustain GBM progression. Secondly, by limiting the inflammatory cargo within EVs, VT107 may enhance the efficacy of existing anti-inflammatory or immunotherapeutic approaches, potentially rendering GBM cells more vulnerable to immune system attack. Lastly, the suppression of MSC-driven VM suggests that Hippo inhibition could serve as an adjunct to anti-angiogenic therapies, further restricting GBM vascular adaptability and reducing treatment resistance.

Despite these promising findings, additional investigations are required to fully elucidate the molecular mechanisms linking Hippo inhibition to EVs cargo modulation. Future studies should explore the broader transcriptomic and proteomic alterations induced by VT107 in both GBM cells and their associated EVs. Understanding whether VT107 can synergize with existing therapeutic modalities, including chemotherapy and immune checkpoint inhibitors, will also be essential for determining its clinical viability. Additionally, in vivo studies assessing the impact of VT107 on tumor progression and microenvironmental remodeling will provide critical insights into its translational potential.

EVs cargo is highly influenced by the cellular origin, microenvironmental conditions, and external stimuli. EVs profiles observed in our cell model system therefore reflect such specific cellular context. Accordingly, our conclusions are not intended to be universally representative of all GBM-derived EVs. Rather than asserting that all GBM subtypes will exhibit identical EVs-mediated effects, our approach emphasizes potential mechanistic insights that may be conserved across contexts in modulating tumor-stromal interactions. Our current findings should therefore serve as a foundation for hypothesis generation, not as definitive generalizations.

Finally, several studies have demonstrated that U87 cells and patient-derived GBM models (PDGMs) share overlapping molecular and phenotypic features, despite their differences in origin and complexity. For instance, both models exhibit expression of canonical GBM markers, including GFAP, EGFR, and CD44, which are commonly used to define glial identity and tumor aggressiveness [49]. Activation of key oncogenic pathways, such as PI3K/AKT, MAPK/ERK, and NFκB, which are central to GBM proliferation, survival, and resistance to therapy have also been highlighted [50]. These shared features make U87 cells a useful, though simplified, model for studying conserved GBM biology, particularly in early-stage mechanistic studies. However, we acknowledge that PDGMs retain greater heterogeneity and better reflect the tumor microenvironment, making them essential for follow-up translational validation.

Conclusions

In conclusion, the goal of this early-stage investigation was to elucidate EVs-driven mechanisms in a controlled in vitro setting that can inform future validation in patient-derived models and in vivo systems that capture GBM’s complexity. While we agree that EVs and tumor heterogeneity limit broad extrapolation, our study provides mechanistic evidence that contributes to the evolving understanding of GBM biology. In addition, this study underscores the importance of targeting EVs-mediated signaling in GBM and identifies Hippo pathway inhibition as a promising approach for modulating tumor progression. By reducing pro-inflammatory EV cargo and impairing paracrine-driven VM, VT107 represents a potential candidate for overcoming therapy resistance and improving patient outcomes. Continued research into Hippo-targeted therapies could pave the way for novel multimodal strategies aimed at disrupting GBM tumor ecosystems and enhancing treatment efficacy.

Abbreviations

ATCC

American type culture collection

DLS

Differential light scattering

ERK

Extracellular signal-regulated kinases

EVs

Extracellular vesicles

FBS

Fetal bovine serum

GBM

Glioblastoma

MAPK

Mitogen-activated protein kinase

MSC

Mesenchymal stem/stromal cells

NFκB

Nuclear factor-kappa B

PDGMs

Patient-derived GBM models

STAT3

Signal transducer and activator of transcription 3

TAZ

Transcriptional coactivator with a PDZ-binding domain

TME

Tumor microenvironment

TNF

Tumor necrosis factor

VM

Vasculogenic mimicry

YAP

Yes-associated protein

Authors’ contributions

Conceptualization : Rosalie Zilinski, Borhane AnnabiData curation : Rosalie Zilinski, Alain Zgheib, Angélique Sabaoth KonanFormal analysis : Rosalie Zilinski, Nicoletta Eliopoulos, Luc H Boudreau, Borhane AnnabiFunding acquisition : Borhane AnnabiInvestigation : Rosalie Zilinski, Alain Zgheib, Angélique Sabaoth KonanMethodology : Alain Zgheib, Angélique Sabaoth Konan, Luc H BoudreauSupervision : Borhane AnnabiWriting - original draft : Rosalie Zilinski, Nicoletta Eliopoulos, Borhane AnnabiWriting - review & editing : Alain Zgheib, Rosalie Zilinski, Angélique Sabaoth Konan, Nicoletta Eliopoulos, Luc H Boudreau, Borhane AnnabiAll the authors read and approved the final manuscript.

Funding

This work was funded by a grant from the Natural Sciences and Engineering Research Council of Canada (NSERC, RGPIN-2024-04541) to BA.

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.