A SORT1/EGFR molecular interplay as a prognostic biomarker in glioblastoma vasculogenic mimicry: implications in the transcriptional regulation of cancer stemness and drug resistance

Abstract

Background

The epidermal growth factor receptor (EGFR) plays a significant role in vasculogenic mimicry (VM), a process by which aggressive cancer cells within hypoxic solid tumors form blood vessel-like structures independent of endothelial cells. Mostly attributed to cancer stem cells (CSC), VM is strongly associated with chemoresistance and poor prognosis in glioblastomas (GBMs). The trafficking of EGFR from the plasma membrane is in part regulated by Sortilin (SORT1), a type I membrane glycoprotein with receptor sorting functions. Notably, EGFR and SORT1 have been reported to also localize within exosomes. This study questions whether the EGFR/SORT1 interplay impacts in vitro VM in human GBM-derived cell models.

Methods

In silico analysis was used to compare GBM to healthy brain tissue transcripts levels. cDNA arrays and RT-qPCR were performed to assess transcript levels in U87, U118, U138, and U251 GBM cells. Immunoblotting was used to assess protein expression levels. In vitro 3D VM was performed on Cultrex, while real-time chemotaxis was measured using the xCELLigence system. Transient gene silencing was achieved using sequence-specific siRNA.

Results

Increased levels of EGFR, along with tetraspanins exosomal markers CD9, CD63, and CD81 were observed in GBM. Interaction predictions further identified EGFR as a central hub linking SORT1 to exosomal biomarkers. Elevated SORT1 protein and gene expression were found in human stage IV GBM-derived U87, U118 and U251 cell lines. This was consistent with cDNA array analyses from clinically annotated GBM demonstrating higher SORT1 levels in stage III-IV compared to stage I-II tumors. Pharmacological inhibition of SORT1 function by AF38469 or siRNA-mediated transient silencing of SORT1 abrogated chemotactic cell migration and reduced in vitro VM in U87 and U251 GBM cells. SORT1 silencing prevented EGFR gene and protein expression upon in vitro VM, as well as reduced chemoresistance (ABCB1, ABCB5) and CSC (PROM1/CD133, SOX2) molecular signature.

Conclusions

These findings underscore the complex interplay of the EGFR/SORT1 axis in regulating VM and in driving to the transcriptional regulation of chemoresistance and CSC molecular signatures of GBM associated with VM. These insights could inspire novel therapeutic strategies targeting the EGFR/SORT1 complex in GBM.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12885-025-14789-3.

Introduction

Glioblastomas (GBMs) are the most aggressive and lethal primary tumor in adults, with a median survival of only 12–15 months despite maximal standard-of-care therapies. Resistance to conventional treatments, a highly heterogeneous tumor microenvironment, and often-inevitable tumor recurrence remain the most significant clinical challenges [1, 2]. GBMs are highly hypoxic and angiogenic tumors which can sustain nutrient supply through alternative vascularization mechanisms such as vasculogenic mimicry (VM), a phenomenon where highly plastic tumor cells form perfusable vessel-like structures without endothelial involvement [3, 4]. This process complements tumor angiogenesis and is strongly associated with tumor aggressiveness, hypoxia adaptation, metabolic reprogramming, and poor clinical outcome [5].

Among the key molecular regulators of VM, the epidermal growth factor receptor (EGFR) has emerged as a central effector [6, 7]. EGFR is frequently amplified, mutated or overexpressed in GBM and drives tumor progression through activation of different signaling pathways that promote proliferation, survival and invasion [8]. Interestingly, EGFR exhibits nuclear localization and is trafficked via endocytic and vesicular routes, suggesting a broader regulatory network involving intracellular transport and sorting mechanisms. EGFR has also been linked to the regulation of VM, but the underlying mechanisms remain incompletely documented [9]. These observations suggest that the intracellular trafficking and localization of EGFR, particularly its nuclear translocation, may influence transcriptional programs that drive VM and lead to therapy resistance.

Recent evidence have identified the involvement of Sortilin (SORT1), a sorting receptor involved in protein sorting and trafficking, as a potential regulator of EGFR dynamics [10]. SORT1 has been shown to mediate the internalization and intracellular routing of membrane-bound receptors and is overexpressed in several aggressive cancers [11–14]. The consequence of SORT1 overexpression has been observed to localize within exosomes, further suggesting a potential role in vesicle-mediated intercellular communication during GBM progression [15]. Based on these observations, we hypothesized that the EGFR/SORT1 interplay may therefore function as a master sensor-effector axis regulating VM in GBM.

Using in silico transcriptomic analysis, we demonstrate that EGFR and exosomal markers are significantly upregulated in GBM tissues while SORT1 expression is selectively increased in high-grade tumors. In vitro functional assays, pharmacological and genetic approaches reveal that functional inhibition or silencing of SORT1 disrupts VM formation, impairs GBM cell migration, and reduces EGFR expression along with molecular signatures associated with chemoresistance and cancer stem cells (CSC) phenotypes. Silencing EGFR recapitulates these effects, confirming the functional interdependence of SORt1 and EGFR in VM regulation. Altogether, our findings uncover a novel regulatory mechanism where SORT1 modulates EGFR expression, and consequently the VM capacity of GBM cells. This offers new insights into the molecular hub of chemoresistance and therapeutic escape in GBM. Targeting this axis may represent a promising therapeutic avenue to disrupt VM-associated resistance in GBM patients.

Materials and methods

Materials

Sodium dodecyl sulfate (SDS) and bovine serum albumin (BSA) were obtained from Sigma-Aldrich (St. Louis, MO, USA). Eagle’s Minimum Essential Medium (EMEM) for cell culture was sourced from Wisent (catalog #320-005 CL). Electrophoresis reagents were purchased from Bio-Rad Laboratories (Hercules, CA, USA). HyGLO™ Chemiluminescent HRP Antibody Detection Reagents were supplied by Denville Scientific Inc. (Metuchen, NJ, USA). Protein quantification was performed using the micro BCA™ Protein Assay Kit from Pierce (Thermo Fisher Scientific, Waltham, MA, USA). The monoclonal anti-GAPDH antibody (D4C6R, 97166) and the polyclonal anti-EGFR antibody (D38B1) were acquired from Cell Signaling Technology (Danvers, MA, USA). The anti-SORT1-ECD antibody was obtained from BD Transduction Laboratories™ (#612100), while the anti-SORT1-ICD antibody was sourced from Abcam (#16640). HRP-conjugated donkey anti-rabbit and anti-mouse IgG secondary antibodies were purchased from Jackson ImmunoResearch Laboratories (West Grove, PA, USA).

Cell culture

Human glioblastoma (GBM) cell lines U87 (HTB-14), U118 (HTB-15), U138 (HTB-16), and U251 were obtained from the American Type Culture Collection (ATCC, Manassas, VA, USA). Cells were cultured in Eagle’s Minimum Essential Medium (EMEM; Wisent, 320-006CL) supplemented with 10% (v/v) calf serum (HyClone Laboratories, SH30541.03), 2 mM L-glutamine, 1 mM sodium pyruvate (Sigma-Aldrich Canada, P2256), 100 U/mL penicillin, and 100 µg/mL streptomycin (Wisent, 250-202-EL). Cultures were maintained at 37 °C in a humidified atmosphere containing 5% CO₂.

In vitro vasculogenic mimicry assay

Vasculogenic mimicry (VM) was assessed in vitro using Cultrex^® Basement Membrane Extract (R&D Systems, 3432-010-01; Toronto, ON, Canada) to evaluate the formation of 3D capillary-like structures. Briefly, 96-well plates were pre-coated with 50 µL of Cultrex and incubated to allow polymerization. Cells (2 × 10⁴ per well) were suspended in culture medium and seeded onto the polymerized matrix in a final volume of 100 µL. 3D capillary-like structure formation was monitored over time using a phase-contrast inverted microscope equipped with a digital camera.

Capillary-like structures image analysis

Quantification of in vitro VM structures was performed using either Wimasis Image Analysis software (Cordoba, Spain) or ImageJ (NIH, Bethesda, MD, USA) as previously described [16, 17].

RNA isolation, cDNA synthesis, and quantitative real-time PCR

Total RNA was isolated from cell monolayers using 1 mL of TRIzol™ reagent (Life Technologies, Gaithersburg, MD, USA) per 3 × 10⁶ cells, following the manufacturer’s protocol. For cDNA synthesis, 1–2 µg of total RNA was reverse-transcribed using either the High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems, Foster City, CA, USA) or the R2 First Strand Kit (QIAGEN, Valencia, CA, USA) for gene array applications. Synthesized cDNA was stored at −80 °C until further use. Quantitative real-time PCR (qPCR) was performed using iQ™ SYBR^® Green Supermix (Bio-Rad, Hercules, CA, USA) on an iCycler iQ5 Real-Time PCR Detection System (Bio-Rad). Amplification was monitored by SYBR Green I fluorescence binding to double-stranded DNA. The following primer sets were obtained from QIAGEN: GAPDH (Hs_GAPDH_1_SG, QT00079247), PPIA (Hs_PPIA_4_SG, QT01866137), SORT1 (Hs_SORT1_1_SG, QT0007331), SORCS1 (Hs_SORCS1_1_SG, QT00024290), SORCS2 (Hs_SORCS2_1_SG, QT00083104), SORCS3 (Hs_SORCS3_1_SG, QT00048965), SORL1 (Hs_SORL1_1_SG, QT00046830), CD133 (Hs_PROM1_1_SG, QT00075586), EGFR (Hs_EGFR_1_SG, QT00085701), ABCB1 (Hs_ABCB1_1_SG, QT00081928), ABCB5 (Hs_ABCB5_1_SG, QT00049448), and SOX2 (Hs_SOX2_1_SG, QT00237601). Gene expression levels were normalized to the housekeeping genes GAPDH and PPIA. Relative quantification was performed using the ΔCT method. The ΔCT value was calculated as the difference between the mean CT of the target gene and the mean CT of the reference genes from triplicate samples. Relative expression levels were expressed as 2^−ΔCT using CFX Manager Software version 2.1 (Bio-Rad).

Human cancer stem cell/drug resistance profiler PCR arrays

Premade RT² Profiler PCR arrays for Human Cancer Stem Cells (PAHS-176Z) and for Drug Resistance (PAHS-004Z) were purchased from QIAGEN and were used following the manufacturer’s instructions. Briefly, the genomic DNA was eliminated before reverse-transcribing 0.5 µg of total RNA using the RT² First Strand Kit (QIAGEN, 330404). Each plate was used to assess one cDNA sample prepared with the RT² SYBR Green qPCR Mastermix (QIAGEN, 330502). The real-time PCR system used was CFX (Bio-Rad). The relative expression analysis of 84 genes as well as controls was done through the GeneGlobe analysis center, a website provided by QIAGEN (https://geneglobe.qiagen.com/us/analyze), using the standard fold change 2^−ΔΔCq method. Ct values were normalized using the geometric mean of internal housekeeping genes present on the array. Analysis was performed using the ΔΔCt method, with fold-change threshold set at ± 2.0 and p-value < 0.05 considered significant. 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.

In silico analysis of SORT1 transcript levels in clinical glioblastoma and low-grade glioma tissues

Transcriptomic data for SORT1 expression in glioblastoma (GBM) and low-grade glioma (LGG) were analyzed using the Gene Expression Profiling Interactive Analysis (GEPIA) web server (http://gepia.cancer-pku.cn/detail.php; accessed November 18, 2024). GEPIA integrates RNA sequencing data from The Cancer Genome Atlas (TCGA) and the Genotype-Tissue Expression (GTEx) project, enabling comparative analysis between tumor and normal tissues. For this study, SORT1 expression was compared between GBM samples (n = 163) and normal brain tissues (n = 207), as well as between LGG samples (n = 518) and normal tissues. GEPIA’s differential expression analysis tool was used to generate box plots, with one-way ANOVA applied to assess statistical significance across disease states (GBM, LGG, and normal tissue). GEPIA also offers additional functionalities including survival analysis, correlation analysis, and dimensionality reduction, which were available for further exploration.

Transfection and RNA interference

Transient transfection for gene silencing was performed in U87 and U251 glioblastoma (GBM) cell lines using Lipofectamine™ 2000 transfection reagent (Thermo Fisher Scientific, Waltham, MA, USA), according to the manufacturer’s instructions. Cells were transfected with 1–20 nM small interfering RNA (siRNA) targeting either SORT1 (Hs_SORT_5 FlexiTube siRNA, SI03115168) or EGFR (Hs_EGFR_5 FlexiTube siRNA, SI00300104). A non-targeting scrambled siRNA (AllStars Negative Control siRNA, 1027281) was used as a control. All siRNA duplexes, including mismatch controls, were synthesized and annealed by QIAGEN (Valencia, CA, USA). The efficiency of gene silencing was evaluated by quantitative reverse transcription PCR (RT-qPCR) as described above.

Real-time cell migration assay

Cell migration was assessed using the xCELLigence Real-Time Cell Analyzer (RTCA) Dual-Plate (DP) Instrument (Roche Diagnostics, Laval, QC, Canada), following the manufacturer’s protocol. U87 or U251 cells were treated with vehicle control (DMSO) or varying concentrations of the SORT1 inhibitor AF38469. Prior to seeding, the underside of each well in the upper chamber of a CIM-Plate 16 was coated with 0.15% gelatin in PBS and incubated for 1 h at 37 °C. A total of 2.5 × 10⁴ cells per well were seeded into the upper chamber and incubated at 37 °C in a humidified atmosphere containing 5% CO₂. The lower chamber was filled with either serum-free or serum-enriched medium to establish a chemotactic gradient. After a 30-minute adhesion period, cell migration was monitored in real time for 2 h, with impedance measurements recorded every 5 min. Impedance values, expressed as the Cell Index, reflect the number of cells that migrated through the membrane. All experiments were performed in quadruplicate wells to ensure reproducibility.

Western blot analysis

Total protein lysates were prepared using a lysis buffer supplemented with 1 mM sodium fluoride (NaF) and 1 mM sodium orthovanadate (Na₃VO₄) as phosphatase inhibitors. Protein concentrations were determined, and 10–20 µg of total protein per sample was resolved by SDS-polyacrylamide gel electrophoresis (SDS-PAGE). Proteins were then transferred onto polyvinylidene difluoride (PVDF) membranes. Membranes were blocked for 1 h at room temperature in Tris-buffered saline containing 0.3% Tween-20 (TBST; 150 mM NaCl, 20 mM Tris-HCl, pH 7.5; Bioshop, TWN510-500) and 5% nonfat dry milk. After blocking, membranes were washed with TBST and incubated overnight at 4 °C with primary antibodies (1:1,000 dilution) in TBST supplemented with 3% bovine serum albumin (BSA) and 0.1% sodium azide (Sigma-Aldrich), with gentle agitation. Following three washes in TBST, membranes were incubated for 1 h at room temperature with horseradish peroxidase (HRP)-conjugated secondary antibodies (anti-rabbit or anti-mouse IgG; 1:2,500 dilution) in TBST containing 5% nonfat dry milk. Immunoreactive bands were visualized using enhanced chemiluminescence (ECL) detection.

Statistical analysis

All data are presented as the mean ± standard error of the mean (SEM) from at least three independent experiments, unless otherwise specified. Statistical comparisons were performed using non-parametric tests. For comparisons involving more than two groups, the Kruskal-Wallis test was applied, followed by Dunn’s multiple comparisons post hoc test. For two-group comparisons, the Mann–Whitney U test was used. A p-value of less than 0.05 (*) was considered statistically significant and is indicated in the figures. All statistical analyses were conducted using GraphPad Prism version 7 (GraphPad Software, San Diego, CA, USA).

Results

Increased gene expression of EGFR and tetraspanins markers CD9, CD63 and CD81 in glioblastoma tumor tissues

Transcript levels of EGFR, SORT1, NTRK2, and exosomal tetraspanins biomarkers CD9, CD63 and CD81 were retrieved from clinically annotated glioblastoma (GBM) datasets (n = 163) and compared to healthy brain tissues (n = 207). In silico analyses revealed that EGFR, CD9, CD63 and CD81 were significantly upregulated in GBM tissues (Fig. 1A). Expression of SORT1 and NTRK2, both suggested to encode proteins that localize within exosomes, was however not different between both tissues possibly due to tumor tissue cell heterogeneity. STRING protein-protein interaction analysis interestingly confirmed EGFR as a central node connecting exosomal biomarkers to SORT1 and NTRK2 (Trk-B) suggesting its role in vesicle-mediated intercellular communication and GBM progression (Fig. 1B). While the primary aim was to link a transcriptional signature phenotype to exosome-associated biomarkers, further exosome characterization would provide additional validation and mechanistic insight. Nevertheless, the inherently correlative in silico analyses that we observe were validated in U87-derived exosomes (not shown). These analyses provide a valuable systems-level perspective and help prioritize candidate pathways for future mechanistic validation. SORT1 expression not statistically different between tumor and normal tissues at the bulk level could be attributed to tumor heterogeneity across GBM subtypes and cellular compartments, including its enrichment in specific tumor cell populations such as GBM stem-like cells (GSC), which may not be captured in bulk tissue comparisons. In contrast, EGFR is frequently amplified and overexpressed in a broader subset of GBM cases, leading to more consistent elevation at the population level. Therefore, the differential interpretive approach may reflect the distinct biological and genomic contexts of these genes.

Fig. 1.

Increased gene expression of EGFR and exosomal markers CD9, CD63 and CD81 in clinically annotated glioblastoma tumor tissues. A In silico analysis of transcript levels was performed using total RNA extracted from glioblastoma clinical samples (n = 163; red boxes) and compared to healthy tissues (n =207; grey boxes), (*p <0.05). B Indirect target proteins of EGFR were retrieved from STRING as described in the Methods section

SORT1 expression is upregulated in grade IV glioma cell lines and tumor tissues

The expression profile of SORT1, and its receptor family members SORCS1, SORCS2, SORCS3, and SORL1, was assessed across four human grade IV-derived glioma cell lines (U87, U118, U138, and U251). RT-qPCR analysis demonstrated elevated SORT1 expression in the four cell lines tested (Fig. 2A), with lower or variable expression levels of the family-related receptors tested (Supplemental Figure S1). Immunoblotting confirmed SORT1 and EGFR protein expression in these cell lines (Fig. 2B). Clinical cDNA arrays covering 43 pathologist-verified brain tumor samples showed that SORT1 gene expression was significantly higher in stage III-IV tumors (red box) compared to stage I-II tumors (green box) (Fig. 2C, p < 0.05). While transcriptional analyses and tumor grade-specific associations were intended as exploratory observations rather than definitive clinical correlations, we nevertheless attempted to validate these findings using publicly available TCGA patient-derived datasets. The expression patterns of SORT1/EGFR genes were found to increase across WHO glioma grades II, III, and IV as shown from the violin plots (Fig. 2D), and statistical significance found to support the trends observed in our experimental data (Fig. 2E). This strengthens the translational relevance of our findings.

Fig. 2.

Expression profile of SORT1 and EGFR in grade IV glioma cell lines and tissues. A Total RNA was extracted from the indicated four different human grade IV glioma cell lines. The gene expression levels of SORT1 and EGFR were assessed using RT-qPCR. B Total protein lysates were isolated from the four glioblastoma cell lines, and representative immunoblots of SORT1, EGFR, and GAPDH are presented from three independent experiments. C A cancer tissue cDNA array covering 43 clinical samples of pathologist-verified brain cancer tissues was analyzed for SORT1 gene expression by qPCR for stages I and II (green box) and for stages III and IV (red box). (*p <0.05).D Violin plots displaying the distribution of EGFR and SORT1 expression levels across WHO glioma grades (II, III, IV) based on TCGA-like data reveal a clear and E statistically significant increase in expression with advancing tumor grade. Inner lines represent the quartiles of each distribution

In vitro vasculogenic mimicry of glioblastoma cells

The ability of GBM cells to form in vitro VM was tested using a commercial Cultrex basement membrane extract. Among the tested cell lines, U87 and U251 were able to form capillary-like structures, while U118 and U138 cells were less efficient under the same conditions (not shown). Using U87 cells, three different cell densities (5,000, 10,000 and 20,000 cells per well) were tested to optimize and monitor structure formation for up to 10 h (Fig. 3A). VM parameters, including total tube length, branching points and total tubes increased progressively over 10 h (Fig. 3B), as quantified by Wimasis analysis. (Fig. 3C).

Fig. 3.

In vitro human analysis of the U87 human glioblastoma cell line’s capacity to form capillary-like structures on Cultrex. Cells were trypsinized and seeded on top of Cultrex to generate three-dimensional capillary-like structures as described in the Methods section. A Representative phase-contrast images were captured and Wimasis analysis (represented in blue) on the structures was performed. Three different cell densities were used and structure formation monitored for up to 10 hours. Scale bar = 1000 μm. B VM parameters were extracted from the Wimasis analysis of (A) and representative quantification provided for total tube length, total branching and total tubes as a function of cell density. C VM parameters were extracted from the Wimasis analysis of (A), with representative quantification provided for total tube length, total branching and total tubes as a function of time during structure formation

Silencing of SORT1 disrupts VM in glioblastoma cells

To evaluate the functional role of SORT1 in VM, transient siRNA-mediated silencing of SORT1 was performed in U87 and U251 cells. SORT1 gene expression levels were validated by RT-qPCR, which confirmed a significant reduction in SORT1 transcripts (Fig. 4A). Immunoblotting further revealed decreased SORT1 (using antibodies recognizing either the extracellular or the intracellular domains) and EGFR protein expressions in siSORT1-transfected cells (Fig. 4B). EGFR decrease upon SORT1 silencing was also validated in U251 cells (Supplemental Figure S2). VM formation was impaired in siSORT1-transfected cells (Fig. 4C) with reduced total loops and loop areas in a dose-dependent manner, as quantified by Wimasis analysis (Fig. 4D). We did not include a scrambled siRNA control because prior VM optimization experiments have consistently shown that the scrambled sequence used in our system did not elicit any measurable effects [16, 17]. We also note that the observed effects of siSORT1 on VM were robust and consistent across multiple assays, further supporting the specificity of the response.

Fig. 4.

Silencing of SORT1 alters in vitro VM in glioblastoma cells. A Transient gene silencing was performed using 20 nM siRNA for SORT1 (siSORT1) silencing or a non-specific scrambled sequence (siScrambled) in U87 and U251 glioblastoma cells and confirmed by RT-qPCR. B Total protein lysates were isolated from U87 transfected cells, and representative immunoblots of SORT1 (anti-SORT1-ECD, Ab recognizing the extracellular domain of SORT1 and anti-SORT1-ICD, Ab recognizing the intracellular domain of SORT1), EGFR, and GAPDH are presented from three independent experiments. C U87 cells were transfected with the indicated siRNA concentrations, trypsinized, and seeded on top of Cultrex to monitor three-dimensional capillary-like structure formation as described in the Methods section. Representative phase-contrast images were captured at 24 hours (upper panels), and Wimasis analysis (lower panels) was performed on the structures. Scale bar = 1000 μm. D VM parameters were extracted from the Wimasis analysis of (C), with representative quantification provided for total loops and loop areas as a function of the indicated siSORT1 concentrations

Pharmacological inhibition of SORT1 reduces glioblastoma cell migration

The effects of the SORT1 functional inhibitor AF38469 on glioblastoma cell migration were assessed using the xCELLigence system. A screen was first attempted with U87 cells treated with increasing concentrations of AF38469 and which exhibited a dose-dependent inhibition of cell migration in response to serum stimulation (Fig. 5A). IC₅₀ values of AF38469 were further extracted from the inhibition curves for U87 and U251 cells and found to be in the low µM range (0.32–1.75 µM) confirming the pharmacological efficacy in reducing cell migration (Fig. 5B). Similarly, siRNA-mediated silencing of SORT1 significantly reduced cell migration under serum-induced chemotactic conditions in U87 and U251 cells (Fig. 5C). AF38469-mediated SORT1 functional inhibition was also found to alter dose-dependently VM formation albeit at low nM concentrations (1.06–1.45 nM) suggesting that SORT1 functions may be more sensitive to pharmacological targeting than the subsequent cell chemotaxis (Supplemental Figure S3).

Fig. 5.

Pharmacological targeting of Sortilin function inhibits U87 and U251 glioblastoma cell migration. A U87 glioblastoma cells were trypsinized and resuspended in the indicated concentrations of Sortilin functional inhibitor AF38469. Treated cells were then assessed for real-time migration using the xCELLigence system in response to serum (closed circles). The absence of serum was performed to assess basal cell migration under no chemotactic cues (open circles). B IC₅₀ values from the respective cell migration were extracted from (A) for U87 and U251 glioblastoma cells at the indicated AF38469 concentrations. C Transient siRNA-mediated gene silencing of SORT1 (siSORT1) or a non-specific scrambled sequence (siScrambled) was performed in U87 and U251 glioblastoma cells. Transfected cells were further assessed for real-time cell migration using the xCELLigence system in response to serum (closed circles) in U87 (left panel) and U251 (right panel) cells

SORT1 silencing alters EGFR expression

To investigate the regulatory relationship between SORT1 and EGFR, lysates from U87 cells transfected with siSORT1 were analyzed for EGFR protein expression. Immunoblotting revealed a reduction in EGFR protein levels in 10–20 nM siSORT1 transfected cells (Fig. 6A). Densitometric analysis further showed a significant decrease in both SORT1 and EGFR protein expression at these concentrations highlighting a dose-dependent regulatory relationship between the two proteins (Fig. 6B). It is important to note that the relationship between gene knockdown efficiency and phenotypic response is not always linear. While SORT1 protein levels may appear substantially reduced at lower doses, residual expression or functional activity may still be sufficient to partially sustain VM. The dose-dependent reduction in VM observed in Fig. 4D likely reflects a threshold effect, where incremental decreases in SORT1 expression progressively impair VM capacity until a critical level is reached. Additionally, differences in assay sensitivity and timing between Figs. 4D and 6A may contribute to this apparent discrepancy. Figure 6A reflects endpoint protein levels, while Fig. 4D captures a dynamic, functional phenotype that may be influenced by cumulative effects over time.

Fig. 6.

SORT1 expression levels impact EGFR expression in U87 glioblastoma cells. A Transient siRNA-mediated gene silencing was performed with increasing concentrations of siSORT1, or a non-specific scrambled sequence (siScrambled), in U87 glioblastoma cells. Total protein lysates were isolated from the transfected cells, and representative immunoblots of SORT1 (anti-SORT1-ECD and anti-SORT1-ICD), EGFR, and GAPDH are presented. B A representative densitometry quantification of protein expression of (A) was performed using the ImageJ software for SORT1 (anti-SORT1-ECD and anti-SORT1-ICD) dimer and monomer forms as well as for EGFR, and is presented as a function of siSORT1 concentrations

Impact of SORT1 and EGFR gene silencing upon the transcriptomic drug resistance and cancer stem cells signature of VM in U87 glioblastoma cells

In an attempt to monitor the impact of SORT1 or EGFR silencing upon VM, transient siRNA-mediated gene silencing of SORT1 or EGFR was performed in U87 glioblastoma cells and total RNA extracted upon 3D VM and RT-qPCR performed using a RT² gene array. Expression was compared to 2D U87 cell monolayers. The gene expression profiles of 84 drug resistance genes are represented (Fig. 7A, upper/lower panel in black). The impact of SORT1 silencing (Fig. 7A, upper panel, red), and of EGFR silencing (Fig. 7A, lower panel, green) is shown. The extent of gene inhibition of the respective silencing of SORT1 or EGFR is further presented (Fig. 7B) and shows a tendency for genes to share common inhibitory profiles in response to either silencing. A protein-protein interaction profiling is presented with the top 15 genes inhibited upon both EGFR and SORT1 silencing and confirms their inter-relationship in the drug resistance phenotype (Fig. 7C). Similarly, an approach to screen for the acquisition of a CSC molecular signature was performed where the respective silencing of either SORT1 (Fig. 8A, upper panel) or EGFR (Fig. 8A, lower panel) reduced the up- or down-regulation profiles of genes upon VM. Inhibitory correlation confirms that similar genes are regulated by both SORT1 and EGFR (Fig. 8B), and that they are part of a coherent CSC phenotypic inter-relationship (Fig. 8C). In the future, RNA-seq would offer a comprehensive and unbiased approach to the transcriptomic profiling we performed herein and could reveal additional pathways and targets regulated by SORT1. Finally, a validation of drug resistance and CSC gene expression upon VM in siSORT1-transfected U87 glioblastoma cells confirms several of the genes identified from the screen performed above (SORT1, ABCB1, ABCB5, PROM1, and SOX2). Transient siRNA-mediated gene silencing of SORT1 was performed at two concentrations (1 and 20 nM) and RT-qPCR performed to assess the expression of the indicated genes upon VM (3D, black bars) or monolayers (2D, white bars) (Fig. 9A and B). Additional functional validation should help confirm the biological relevance of these gene expression changes. Given the scope and focus of the current study, we prioritized transcriptomic profiling to identify key regulatory patterns and generate hypotheses for future mechanistic exploration.

Fig. 7.

Impact of SORT1 and EGFR gene silencing upon the transcriptomic drug resistance signature of VM in U87 glioblastoma cells. Transient siRNA-mediated gene silencing of (A) SORT1 (red plot, upper panel) or EGFR (green plot, lower panel) was performed in U87 glioblastoma cells, total RNA extracted upon VM, and RT-qPCR performed using a RT² gene array. Expression was compared to their respective U87 cell 2D monolayers. The gene expression profiles of 84 drug resistance genes are represented. B The extent of gene inhibition of the respective silencing of SORT1 or EGFR is presented. C A protein-protein interaction profiling is represented with the top 15 genes inhibited both upon EGFR and upon SORT1 silencing

Fig. 8.

Impact of SORT1 and EGFR gene silencing upon the transcriptomic cancer stem cells signature of VM in U87 glioblastoma cells. Transient siRNA-mediated gene silencing of (A) SORT1 (red plot, upper panel) or EGFR (green plot, lower panel) was performed in U87 glioblastoma cells, total RNA extracted upon VM, and RT-qPCR performed using a RT² gene array. Expression was compared to their respective U87 cell 2D monolayers. The gene expression profiles of 84 cancer stem cells genes are represented. B The extent of gene inhibition of the respective silencing of SORT1 or EGFR is presented. C A protein-protein interaction profiling is represented with the top 15 genes inhibited both upon EGFR and upon SORT1 silencing

Fig. 9.

RT-qPCR validation of drug resistance and cancer stem cells gene expression upon VM in siSORT1-transfected U87 glioblastoma cells. Transient siRNA-mediated gene silencing of SORT1 was performed at two concentrations (1 and 20 nM) in U87 glioblastoma cells, total RNA extracted upon VM, and RT-qPCR performed to assess the expression of the indicated genes upon VM (3D, black bars) or monolayers (2D, white bars) in (A) drug resistance phenotype, and (B) cancer stem cells phenotype

Discussion

SORT1 is involved in various cellular processes, including intracellular trafficking and signaling regulation [18], and it is suggested to contribute to the acquisition of chemoresistance in GBM [19]. Such resistance to therapeutic stress is a common feature of the tumor’s adaptive capacity, enabling the cancer cells to evade apoptosis in order to survive under treatment conditions. SORT1 may also influence chemoresistance by modulating key signaling pathways and trafficking of membrane receptors such as EGFR, which promote cell survival and proliferation, counteracting the effects of chemotherapy. Ultimately, the combined regulation of chemoresistance and CSC phenotypes could collectively affect the tumor microenvironment as SORT1’s role in cellular communication might further enable GBM cells to adapt to and thrive in hostile conditions triggered by treatment. Here, we addressed an intricate role of SORT1/EGFR interplay emphasizing their contribution to GBM therapy resistance, in part through VM, and highlighting their potential as therapeutic targets. Given no endothelial cells or co-culture systems was used herein, the vascular-like structures observed were formed exclusively by GBM cells, consistent with established in vitro models of VM. The experimental context of our study, lacking any endothelial cell input, supports the interpretation that these structures are tumor cell-derived.

The basis of the EGFR and SORT1 functional interaction in GBM lies in their converging roles in promoting tumor progression and resistance to therapy. As EGFR is frequently altered in GBM [20], driving oncogenic signaling that supports tumor growth, survival, and resistance to treatments [21], we demonstrated here for the first time that SORT1 could alter EGFR expression at both transcriptional and translational levels. Such molecular interplay was further proved to significantly impact the capacity to perform VM, a process closely linked to chemoresistance and especially to the CSC phenotype [22]. Molecular evidence has shown that EGFR plays a significant role in regulating CSC markers expression, thereby contributing to tumor initiation, progression, and resistance to therapy [23]. Notably, EGFR activation has been linked to upregulation of stemness genes such as SOX2, OCT4, and NANOG [24, 25]. These markers are crucial for maintaining the self-renewal and tumorigenic potential of CSC. Studies have also demonstrated that EGFR activation enhances the ability of cancer cells to form tumorspheres, a characteristic feature of CSC [26]. This process is accompanied by increased expression of CSC markers like CD44 and BMI-1 [27]. EGFR-driven CSC also exhibit resistance to conventional therapies, including chemotherapy and radiotherapy [28].

Interestingly, while the GBM cell lines used in our study do not harbor EGFR amplification, which is a common feature in a subset of primary GBM tumors, one can conclude that our findings regarding EGFR do not solely dependent on amplification status but rather focus on its expression dynamics and potential regulation by SORT1. EGFR signaling can be active and biologically relevant even in the absence of gene amplification, particularly through ligand-mediated activation or post-transcriptional regulation. Our data suggest that SORT1 may influence EGFR-related pathways through such mechanisms, which are still clinically relevant. Future studies using EGFR-amplified models (e.g., patient-derived xenografts or primary cultures) will be essential to further validate and extend these findings. To provide more direct mechanistic evidence that would further strengthen the proposed relationship between SORT1 and EGFR trafficking, co-immunoprecipitation, proximity ligation assay, or trafficking studies could be pursued in future work.

Our study is the first to demonstrate that the SORT1/EGFR interplay orchestrates the parallel transcriptional regulation of both CSC-related phenotype (CD133/PROM1, SOX2, CD44, ABCB5) and chemoresistance-associated phenotype (BCL2, ABCB1, ABCG2, HIF1A). Supporting our results, previous studies have reported that CD133 + GBM stem-like cells are capable of initiating VM in vivo [29]. Together, these findings highlight the critical role of EGFR regulation in modulating CSC features that drive VM, and suggest that targeting the SORT1/EGFR interplay could be a promising strategy to counteract CSC-driven tumor progression and treatment resistance in GBM. Validation of CSC properties will further require functional assays such as sphere formation, limiting dilution, or tumor initiation studies. Here, we used PROM1/CD133 and SOX2 as established molecular biomarkers to infer stem-like characteristics, based on their well-documented association with GBM stem-like cells in the literature. While we did not perform functional CSC assays in this study, our intention was to highlight transcriptional trends suggestive of a stem-like phenotype rather than to definitively classify cells as CSCs. Future studies will also be needed to functionally validate the CSC potential of these populations. Direct mechanistic experiments including EGFR promoter reporter assays, mRNA stability assays, and rescue assays should provide valuable future confirmation of the interaction between SORT1 and EGFR, and a better understanding of the SORT1/EGFR axis.

SORT1-mediated mechanisms, particularly their role in the regulation of extracellular vesicle secretion, have been associated with therapy resistance and with the expansion of CSC populations [30]. In addition, SORT1’s role in secretion-induced progranulin pathways has been shown to promote CSC expansion in breast cancer [31]. As molecular evidence also supports a role for SORT1 in CSC biology through the increased expression of CSC markers such as CD133 [32, 33], we further demonstrate here a link between SORT1 expression and VM, suggesting that SORT1 contributes to GBM plasticity and CSC-driven vascularization. Interestingly, EGFR has also been found to locate in exosomes secreted by cancer cells. These EGFR-containing exosomes can play a crucial role in tumor progression, metastasis, and drug resistance [34]. For instance, studies have shown that exosomes derived from EGFR-mutated lung cancer cells carry EGFR, which enhances tumor growth and therapy resistance [35, 36]. Exosomal wild-type EGFR has also been shown to be transferred between cells, contributing to acquired resistance against targeted therapies like osimertinib in non-small cell lung cancer [37]. These findings underscore the potential of both exosomal EGFR as well as SORT1 as biomarkers for early cancer detection and prognosis.

The role of extracellular vesicles and biomimetic nanoplatforms in targeting GBM via the EGFR axis could become an important emerging area that aligns directly with our findings. In particular, a recent study by Cheng et al. demonstrates how engineered extracellular vesicles targeting EGFR-positive GBM provide a novel and safe delivery strategy for overcoming therapy resistance [38]. This work reinforces the therapeutic relevance of the EGFR axis and supports the current proposition that vesicle-mediated communication and receptor trafficking are promising therapeutic avenues. While our findings suggest that exosome-mediated regulation may be involved in these phenotypes, we did not directly isolate or characterize exosomes in this study. This decision was based on the scope of our investigation, which focused on transcriptional profiling and in vitro functional assays to uncover key regulatory mechanisms. Direct exosome isolation, characterization (e.g., nanoparticle tracking analysis, Western blot for exosomal markers), and functional assays (e.g., exosome transfer experiments) will become important future directions.

By identifying SORT1- and EGFR-associated transcriptional signatures and functional outcomes, our study provides foundation for future investigations into the role of exosomes in GBM progression. We believe these findings may help guide more targeted mechanistic studies, including those focused on extracellular vesicle biology. From a therapeutic standpoint, targeting SORT1 has shown promise. Studies have demonstrated that either silencing SORT1 or delivering peptide-drug conjugates directed against SORT1 can inhibit CSC-associated VM and circumvent drug resistance [39]. Collectively, these observations position SORT1 not only as a marker of CSC-driven progression and resistance, but also as a viable therapeutic target in GBM and potentially other aggressive SORT1-positive cancers.

Conclusion

Although the direct interactions between EGFR and SORT1 have been inferred but not fully characterized, emerging evidence suggests that their pathways may converge to drive tumor aggressiveness. SORT1 is involved in the turnover of EGFR at the plasma membrane, and its expression was reported to increase in EGFR-overexpressing tumors and in patients with EGFR amplification [40]. The impact of SORT1 silencing on VM inhibition therefore reconciles with the fact that downregulation of SORT1 affects EGFR internalization. EGFR activation can influence downstream pathways that SORT1 might also regulate, such as those involved in cell survival and proliferation. Accordingly, SORT1 expression in tumors potentially represents a useful predictive marker of patient outcome. Both EGFR and SORT1 are associated with mechanisms that enable GBM cells to evade the effects of chemotherapy and targeted therapies. Further research is required to clarify how the EGFR/SORT1 interplay impacts the cancer cell transcriptional reprogramming, and whether targeting the EGFR/SORT1 axis may represent a novel therapeutic avenue to overcome resistance and improve clinical outcomes in GBM patients. In terms of limitations, while we acknowledge the importance of in vivo validation to further substantiate the clinical relevance of our in vitro findings, such experiments would fall outside the scope of our current molecular study. Nonetheless, we are convinced that our findings provide a strong foundation for future in vivo investigations, and we hope they will serve as a valuable reference for subsequent translational research.

Abbreviations

BSA

Bovine serum albumin

CSC

Cancer stem cells

EGFR

Epidermal growth factor receptor

GAPDH

Glyceraldehyde-3-phosphate dehydrogenase

GBM

Glioblastoma

GCS

Glioblastoma stem-like cells

LGG

Low-grade glioma

NaF

Sodium fluoride

PPIA

Peptidylprolyl isomerase A

SDS

Sodium dodecyl sulfate

SEM

Standard error of the mean

SORT1

Sortilin

VM

Vasculogenic mimicry

Authors’ contributions

Conceptualization, Marie-Eve Roy, Borhane Annabi; Data curation, Marie-Eve Roy; Formal Analysis, Marie-Eve Roy, Nicoletta Eliopoulos, and Borhane Annabi; Funding acquisition, Borhane Annabi; Investigation, Marie-Eve Roy, Alain Zgheib; Methodology, Marie-Eve Roy, Alain Zgheib; Supervision, Borhane Annabi; Writing – original draft, Marie-Eve Roy, Nicoletta Eliopoulos, and Borhane Annabi; Writing – review and editing, Marie-Eve Roy, Alain Zgheib, Nicoletta Eliopoulos, and Borhane Annabi. All authors read and approved the final manuscript.

Funding

This work was funded by the Institutional Research Chair in Cancer Prevention and Treatment held by Dr Borhane Annabi, and by a grant from the Natural Sciences and Engineering Research Council of Canada (NSERC, RGPIN-2024-04541).

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.