PLA2 driven lipid signaling drives ARMS tumorigenic cell properties

Abstract

Background

Advancements in chemotherapy have improved outcomes for rhabdomyosarcoma (RMS), the most common soft tissue sarcoma in children. However, relapse-free survival rates remain below 20% in metastatic alveolar rhabdomyosarcoma (ARMS). Tumor-initiating cells (TICs) drive recurrence. Targeting this pool of cells has been shown to reduce relapse rates and metastasis in various cancers. Emerging evidence implicates lipid metabolism in supporting stem-like properties in cancer. However, TICs remain poorly characterized in ARMS. This study investigates the transcriptomic and metabolic profile of TICs in ARMS.

Methods

Transcriptomic and lipidomic profiling were performed on tumorsphere-derived and adherent cells from ARMS cells. Differential expression of metabolic genes and lipid species was assessed via RNA-Sequencing and mass spectrometry. Functional dependence on Phospholipase A2 (PLA2) signaling was tested on TICs in vitro and in vivo. Seahorse assays were used to measure glycolytic and mitochondrial flux. Statistical significance was determined using Student’s t-test or one-way ANOVA.

Results

Transcriptomic analysis of ARMS tumorspheres showed upregulation of lipid metabolism especially different PLA2 enzyme subtypes, and suppression of glycolytic gene expression. Lipidomic analyses revealed enrichment of linoleic acid-derived triglycerides and lysophospholipids, paralleled by increased PLA2 activity. Tumorspheres also showed elevated expression of lipid storage regulators PPARG and CD36 and correspondingly, significant lipid droplet accumulation. Inhibition of PLA2 with Darapladib reduced tumorsphere formation and cellular motility in vitro and tumor volume in vivo. These defects were rescued by exogenous linoleic acid (LA) supplementation, implicating PLA2-driven lipid signalling in TIC maintenance. Interestingly, TICs display high metabolic flexibility in energy consumption, as neither glycolysis nor fatty acid oxidation (FAO) inhibition significantly impaired tumorsphere formation.

Conclusions

PLA2-mediated lipid remodelling supports TIC by promoting linoleic acid–linked lipid signalling. PLA2 inhibition impairs self-renewal, motility, and tumorigenic potential, highlighting it as a promising therapeutic target in ARMS. These findings uncover lipid remodelling as a key metabolic adaptation in ARMS TICs, independent of canonical energy pathways, with implications for targeting the stem-like compartment in pediatric sarcomas.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12964-025-02409-3.

Introduction

ARMS is a highly malignant paediatric soft tissue sarcoma that arises from skeletal muscle progenitor cells [1]. It accounts for a significant subset of RMS cases and is distinguished by a characteristic chromosomal translocation, most commonly PAX3–FOXO1 or PAX7–FOXO1, which gives rise to oncogenic fusion transcription factors [2, 3]. Fusion-positive tumors are associated with a more aggressive clinical course, increased metastatic potential, and resistance to standard therapy compared to their fusion-negative counterparts [4, 5]. Despite the use of intensive multimodal treatment regimens-including surgical resection, multi-agent chemotherapy (e.g., VAC: vincristine, actinomycin D, cyclophosphamide), and radiation therapy-the overall prognosis for patients with metastatic or relapsed ARMS remains poor, with a five-year survival rate falling below 30% [6–8].

The aggressive nature of advanced ARMS and its high relapse rate may, in part, be driven by a distinct subpopulation of TICs, also referred to as cancer stem-like cells. These cells are thought to underlie therapeutic resistance, disease progression, and metastatic dissemination, owing to their enhanced self-renewal capacity, tumorigenic potential, and ability to evade cytotoxic treatments [9–13]. Recent single-cell transcriptomic studies have begun to reveal the cellular heterogeneity within ARMS tumors, identifying subpopulations with stem-like signatures that may contribute to treatment failure and disease recurrence [14–18]. However, the functional and metabolic properties of these TICs in ARMS remain poorly characterized. In particular, the metabolic mechanisms that support TIC survival and plasticity are still not well understood. Given the central role of metabolism in regulating stemness, stress adaptation, and cell fate decisions, uncovering the metabolic dependencies of TICs in ARMS may provide novel therapeutic avenues for targeting this therapy-resistant population [19–24].

Emerging evidence suggests that lipid metabolism plays a pivotal role in regulating cancer cell plasticity, survival, and therapeutic resistance [25–27]. Lipid metabolic pathways contribute not only to membrane biosynthesis and energy production but also to the generation of bioactive lipid species that modulate intracellular signalling pathways central to oncogenesis and tumor progression [28, 29]. Among these, phospholipase A2 (PLA2) enzymes have gained increasing attention for their ability to hydrolyze membrane phospholipids, thereby releasing lipid mediators such as arachidonic acid (AA), linoleic acid (LA), and lysophospholipids [30–32]. These lipid products serve as precursors for various pro-tumorigenic signaling cascades, including those linked to inflammation, cell migration, and survival. However, the specific role of PLA2-driven lipid remodeling in the maintenance and function of TICs remains poorly defined, particularly in the context of ARMS.

In this study, using a combination of transcriptomic, lipidomic, and metabolic profiling, we examined the properties of ARMS TICs. We demonstrate that ARMS tumorspheres derived from cell lines and a patient derived xenograft (PDX) model have distinct signatures compared to adherent cells. Transcriptomic analysis demonstrates an upregulation of genes involved in lipid metabolism, specifically an enrichment in PLA2 enzyme expression in tumorspheres. Lipidomic analysis revealed a distinct triglyceride (TG) composition, with significant enrichment in lipid droplet-associated TGs, suggesting a reliance on lipid remodeling for metabolic adaptation. Consistent with these findings, inhibition of PLA2 activity with Darapladib impairs TIC function, leading to reduced tumorsphere formation, suppressed migration, and diminished tumorigenicity in vivo. Furthermore, supplementation with exogenous LA rescues TIC phenotypes in Darapladib-treated cells, highlighting the role of LA-derived lipid mediators in TIC maintenance. Interestingly, while ARMS TICs have a heightened dependence on long-chain fatty acid (LCFA) oxidation for energy production, they display metabolic flexibility as inhibition of LCFA oxidation does not abolish tumorsphere formation, suggesting that alternative metabolic pathways may compensate for lipid catabolism. Taken together, this study provides new insights into the metabolic vulnerabilities of ARMS TICs and underscores the therapeutic potential of targeting PLA2-driven lipid remodeling. By elucidating the interplay between lipid metabolism and TIC function, our work highlights new avenues for metabolic intervention in aggressive paediatric sarcomas.

Methods

Adherent cell culture and tumorsphere formation

Rh30 and Rh41 alveolar rhabdomyosarcoma (ARMS) cell lines, provided by Peter Houghton and Rosella Rota, were cultured in RPMI-1640 medium (Hyclone, Cytiva, #SH30027.01) supplemented with 10% fetal bovine serum (FBS) (HyClone, Cytiva, #SH30071.03) and 1% penicillin-streptomycin (Gibco, #15140-122) under standard conditions (37 °C, 5% CO₂). Normal human skeletal muscle myoblasts (HSMMs; ZenBio, #SKB-F) were cultured in Skeletal Muscle Cell Growth Medium (ZenBio, #SKM-M), containing DMEM, FBS, BSA, fetuin, hEGF, dexamethasone, insulin, penicillin, streptomycin, and amphotericin B. Mycoplasma contamination was routinely monitored. For tumorsphere assays, ARMS cells were dissociated and seeded at 10,000 cells/mL in ultra-low attachment plates with serum-free Neurobasal medium (Gibco, #21103049) supplemented with B27 (Gibco, #17504044), EGF (40 ng/mL, Gibco, #PHG0311), FGF (80 ng/mL, Gibco, #13256-029), and insulin (50 µg/mL, Sigma-Aldrich, #I9278). Tumorspheres (> 50 μm) were quantified after 72 h. To assess metabolic dependencies, single-cell suspensions from tumorspheres were seeded in 96-well ultra-low attachment plates in tumorsphere media containing Darapladib (1.2 µM), Etomoxir (50 µM), or 2-deoxy-D-glucose (2-DG, 2 mM), cultured for 7 days, and quantified. Cells were passaged enzymatically using Accutase (Innovative Cell Technologies, #AT104) every 5–7 days in the same media.

Patient-derived xenograft (PDX) model

PDX tumor samples were obtained from the VIVA-KKH Pediatric Solid Tumor Research Laboratory. Fresh tumor tissue was collected from patients with rhabdomyosarcoma who were prospectively recruited with informed consent (SingHealth Duke-NUS protocol 2014/2079). Leftover surgical biopsy and resection tissue was used to generate orthotopic PDX models. The PDX line RHB280217 was established from a pre-treatment biopsy specimen of a patient with alveolar rhabdomyosarcoma (ARMS). Tumors were excised, mechanically dissociated, and resuspended in Matrigel (Corning, #354234) (2:1) prior to implantation into NOD SCID mice. Mice were euthanized when tumor volume exceeded 15 mm in diameter. Engrafted tumors were evaluated histologically at each passage, and short tandem repeat (STR) microsatellite analysis was performed to confirm the preservation of the patient’s genomic profile. For in vitro studies, RHB280217 PDX-derived tumor-initiating cells (TICs) were cultured as tumorspheres in Neurobasal medium supplemented with B27, EGF (20 ng/mL), FGF (40 ng/mL), and Glutamax (Gibco, #35050061).

Cell viability

Cell viability following Darapladib treatment was assessed using the MTT assay. Rh30, Rh41, RHB280217, and HSMM cells were seeded in 96-well plates at a seeding density of 5000 cells per well and treated with increasing concentrations of Darapladib (ranging from 0.001 to 100 µM) for 72 h.

Tumorsphere-derived cells from Rh30 and RHB280217 were cultured as previously described, enzymatically dissociated using Accutase, and also seeded at a density of 5000 cells per well in 96-well plates.

MTT reagent (Sigma-Aldrich, #M5655) was added to each well and incubated for 3 h at 37 °C. Formazan crystals were dissolved in DMSO and absorbance was measured at 570 nm using a microplate reader. IC₅₀ values were calculated using non-linear regression in GraphPad Prism. Experiments were performed in triplicate for each cell line.

Transwell migration assay

The migratory ability of adherent and tumorsphere-derived cells was assessed using 8-µm pore size transwell inserts (Corning, #3422). Cells (50,000 per insert) were serum-starved for 6 h, seeded in serum-free RPMI-1640 in the upper chamber, while the lower chamber contained 10% FBS as a chemoattractant. After 24 h, migrated cells were fixed, stained with crystal violet, and counted. To test the effect of drugs, tumorsphere-derived cells were seeded in 6-well plates and treated with the respective drugs. Viable cells were collected and seeded in the upper chambers of transwell inserts in serum-free medium. Migration toward complete medium in the lower chamber was allowed for 16–24 h. Migrated cells were fixed, stained with crystal violet, and quantified.

Cell line xenograft models

Rh30 tumorsphere-derived cells (1 × 10⁶) in PBS were suspended with Matrigel^® Matrix in a 1:1 ratio (Corning, # 354234) and injected subcutaneously into NU/JInv (Jackson Laboratory; stock no. 002019) immunocompromised mice. Mice were randomly assigned to control vs. treatment group. To assess effects on tumor initiation, treatment with Darapladib (25 mg/kg/day) or vehicle control via oral gavage was initiated 7 days post-injection, and continued for 21 days. 8 mice were used in each group. As two mice died during the experiment, the data shown is from 6 mice. Tumor growth was monitored daily, and volume was calculated as V = (L × W²)/2. At endpoint when tumor reached 15 mm, they were excised, weighed, and analyzed histologically. All procedures were approved by Institutional Animal Care and Use Committee under protocol number R23-0697.

Linoleic acid supplementation assay

Rh30 and Rh41 tumorsphere-derived cells were treated with Darapladib (1 µM) in the presence or absence of exogenous LA (Sigma-Aldrich, #L9530) added to tumorsphere media at a final concentration of 2.5 mg/ml. Cells were cultured under tumorsphere-forming conditions for 7 days in 6 wells and 96 well low attachment plates. For dose-response experiments, LA was supplemented at increasing concentrations (ranging from 1 mg/mL to 4 mg/mL) in Darapladib-treated cells and tumorsphere number was quantified. To evaluate cell motility, tumorsphere-derived Rh30 and Rh41 cells were dissociated into single cells and seeded in transwell with Darapladib ± LA and subjected to migration assays as previously described. Migrated cells were fixed, stained with crystal violet, and counted using light microscopy.

RNA sequencing and bioinformatic analysis

Total RNA was extracted from adherent and tumorsphere-derived Rh30, Rh41, and PDX samples using the RNeasy Mini Kit (Qiagen), followed by quality assessment with an Agilent 2100 Bioanalyzer. RNA libraries were prepared using the NEBNext Ultra II RNA Library Prep Kit and sequenced on an Illumina NovaSeq 6000 platform, generating 150-bp paired-end reads. Raw sequencing data were assessed for quality using FastQC, and reads were aligned to the human reference genome (GRCh38) using STAR aligner. Differential gene expression analysis was performed using DESeq2, with significance thresholds set at log2 fold change > 1 and adjusted p-value < 0.05. Gene Set Enrichment Analysis (GSEA) was conducted using the Molecular Signatures Database (MSigDB) to identify enriched biological pathways. Data visualization, including volcano plots, principal component analysis (PCA), and heatmaps, was performed using R packages ggplot2 and pheatmap. RNA-Seq data has been deposited in GEO database under accession number GSE295852. Private token for the reviewers is sdclqykqfbehbif.

Quantitative PCR (qPCR)

Total RNA was extracted using TRIzol, followed by cDNA synthesis with the SuperScript IV First-Strand Synthesis System. qPCR was performed using SYBR Green I Master Mix on a LightCycler 480 System (Roche, #05015278001). Gene expression was normalized to GAPDH using the 2 − ΔCt method.

Lipidomics analysis

One million Rh30 and Rh41 cells were spun and pelleted down at 5000 RPM. Lipid extraction was performed using the Bligh and Dyer method [33] and samples were analyzed via high-resolution liquid chromatography-mass spectrometry (LC-MS/MS) at the Singapore Lipidomics Incubator (SLING). Lipid species were identified using LipidSearch software and analyzed using LipidMaps and LipidSig for metabolic pathway assessment.

Seahorse metabolic analysis

Real-time metabolic profiling of tumorsphere-derived cells was performed using the Seahorse XFe96 Analyzer (Agilent Technologies), tumorsphere cells from Rh30, Rh41 and RHB280217 were collected, dissociated enzymatically with accutase, counted and seeded in the Agilent 24 well seahorse plate. Oxygen consumption rate (OCR) was measured using the Seahorse XF Cell Mito Stress Test Kit (#103015-100) to evaluate mitochondrial respiration and fatty acid oxidation capacity. 50,000 cells were seeded in Seahorse XF cell culture microplates and analyzed under basal and drug-treated conditions following sequential injections of oligomycin, FCCP, and rotenone/antimycin A. Where indicated, Etomoxir (4 µM) was used to assess dependence on CPT1A-mediated long-chain fatty acid oxidation.

Extracellular acidification rate (ECAR) was assessed using the glycolysis stress test to evaluate glycolytic flux. Cells were sequentially treated with glucose, oligomycin, and 2-deoxy-D-glucose (2-DG), and ECAR was recorded to determine glycolytic capacity and reserve. Data were normalized to cell number and analyzed using Wave software (Agilent Technologies).

Lipid droplet staining

Lipid droplet accumulation was assessed using Oil Red O (Sigma-Aldrich, #O0625) and BODIPY 493/503 (Thermo Fisher, #D3922) staining. Cells derived from adherent or tumorsphere cultures were seeded on coverslips and allowed to attach for 6–8 h prior to staining. For Oil Red O staining, cells were fixed and stained according to standard protocols and imaged using bright-field microscopy. For BODIPY staining, cells were fixed and stained with BODIPY 493/503, followed by imaging under fluorescence microscopy. Lipid droplet quantification was performed using ImageJ (version 1.54).

PLA2 activity assay

PLA2 activity was measured using a fluorescence-based Phospholipase A2 Assay Kit (Abcam, #ab133089), following the manufacturer’s instructions. Briefly, 1 × 10⁶ cells were harvested, washed with PBS, and lysed in the provided assay buffer supplemented with protease inhibitors. The lysates were centrifuged at 13,000 × g for 10 min at 4 °C to remove debris. Supernatants were collected, and protein concentrations were determined using the Bradford assay. Equal amounts of protein (typically 20–50 µg) were loaded into 96-well black plates in triplicate and incubated with the fluorescent PLA2 substrate for 1 h at 25 °C in the dark. Fluorescence was measured at Ex/Em = 485/535 nm using a microplate reader. PLA2 activity was normalized to total protein content.

Microscopy and image analysis

Tumorspheres were imaged using an inverted bright-field microscope at 10x magnification. Sphere size and number were quantified using ImageJ software. For stained cells (Oil Red O, BODIPY), images were acquired using bright-field or fluorescence microscopy respectively at 100x, under matched exposure settings across conditions.

Statistical analysis

All experiments were performed with a minimum of three biological replicates. Data are presented as mean ± SEM. Statistical analyses were conducted using GraphPad Prism software (version 9.0). Comparisons between two groups were performed using unpaired two-tailed Student’s t-tests. For cell viability assays (MTT), experiments were performed in technical triplicates and repeated independently at least three times. Dose–response curves and IC₅₀ values were generated using nonlinear regression with a variable slope model (log[inhibitor] vs. normalized response). For survival analysis the log-rank (Mantel–Cox) test was used. A p-value of less than 0.05 was considered statistically significant.

Results

ARMS tumorsphere-derived cells show enrichment in stemness factors with a distinct transcriptome

To investigate the transcriptomic differences between tumorsphere-derived TICs and adherent cells, we cultured Rh30, Rh41 cell lines, and an ARMS PDX model RHB280217 as adherent and tumorsphere cultures according to established protocols [34] (Fig. 1A). As expected, stemness-associated transcription factors OCT4 (POU5F1), SOX2, NANOG, and NOTCH were upregulated in tumorsphere-derived cells (Fig. 1B), confirming the stemness potential of tumorsphere-derived cells.

Fig. 1.

Transcriptomic characterization of ARMS tumorsphere-derived cells. A Representative bright-field images showing the morphology of adherent and tumorspheres from Rh30, Rh41, and RHB280217 cells. B Relative mRNA fold change of stemness-associated transcription factors (Oct4, Sox2, Nanog, and Notch3) tumorsphere-derived cells from Rh30, Rh41, and RHB280217 relative to adherent counterparts. Gene expression was normalized to GAPDH and analyzed by qPCR. Values represent mean ± SEM from n = 3 biological replicates, each with 3 technical replicates. C PCA showing distinct clustering of tumorsphere-derived cells versus adherent cells for Rh30 and Rh41 cells. D Volcano plots showing DEGs between tumorsphere and adherent cells for Rh30 and Rh41 cells. Red: significantly upregulated genes (adjusted p < 0.05, log2 fold change > 1); blue: significantly downregulated genes (adjusted p < 0.05, log2 fold change < −1) E Hierarchical clustering of DEGs in Rh30 and Rh41 tumorsphere versus adherent cells. F Venn diagram of upregulated genes showing overlap between Rh30 and Rh41 tumorsphere-derived cells revealing a conserved TIC-specific gene expression signature. Statistical comparisons between adherent and tumorsphere groups were performed using unpaired two-tailed Student’s t-tests in GraphPad Prism. (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001). Scale bar represents 100 μm

We then performed RNA sequencing (RNA-seq) on tumorsphere-derived and adherent Rh30, Rh41, and RHB280217 cells. Principal Component Analysis (PCA) demonstrated a clear separation between tumorsphere and adherent cell transcriptomes, confirming distinct molecular identities (Fig. 1C and Supplemental Fig. 1 A). Differential gene expression analysis revealed tumorsphere-specific transcriptional programs as visualized in volcano plots for Rh30, Rh41, and RHB280217 (Fig. 1D and Supplemental Fig. 1B). Hierarchical clustering of differentially expressed genes (DEGs) confirmed the distinct transcriptional landscape of tumorsphere-derived cells (Fig. 1E and Supplemental Fig. 1 C). A significant set of upregulated genes was shared between Rh30 and Rh41 tumorspheres indicating a core set of shared transcriptional alterations (Fig. 1F).

Lipid metabolic pathways are upregulated, while Glycolysis is suppressed in tumorspheres

Gene set enrichment and pathway-level analyses of the differentially expressed genes in tumorsphere-derived cells revealed several upregulated programs commonly associated with stem-like behaviour, including cytokine-cytokine receptor interaction and cell adhesion (Fig. 2A). Representative genes within these enriched pathways include CXCL8, TNFRSF1B, and IL11RA (cytokine interaction). These genes are associated with tumor–niche communication, inflammation, and survival signalling, consistent with a broader TIC regulatory state. These findings support the notion that ARMS TICs adopt a survival and plasticity program. Among these, lipid metabolism emerged as a particularly prominent axis, with consistent upregulation of genes involved in α-linolenic acid (ALA), linoleic acid (LA), and arachidonic acid metabolism (AA) (Fig. 2A-B). Notably, LA and AA metabolism were also upregulated in RHB280217 tumorspheres (Supplemental Fig. 1D-E). To identify key regulators of this altered metabolic state, we analyzed DEGs associated with lipid metabolism. PLA2 family members were consistently upregulated in tumorspheres, with PLA2G5, PLA2G6, PLA2G2D, and PLA2G7 showing significant overexpression compared to adherent cells across Rh30, Rh41, and PDX models (Fig. 2C-D). Consistently, PLA2 enzymatic activity was elevated in tumorsphere-derived cells relative to adherent counterparts (Fig. 2E). Given their role in hydrolyzing membrane phospholipids to release free fatty acids, including ALA, LA, and AA, we postulated that these enzymes likely act as major drivers of the observed metabolic remodeling. In addition to lipid metabolism, a significant downregulation of glycolysis-associated pathways in tumorspheres was seen. KEGG pathway over-representation analysis (ORA) revealed suppression of glycolysis in Rh30 and Rh41 tumorspheres (Fig. 2F). GSEA further validated this trend, showing a significant negative enrichment score for glycolysis gene sets in tumorspheres (Fig. 2G). We validated the transcriptional downregulation of key glycolytic genes, including ENO1, PFKM, and LDHA, further supporting the suppression of glycolysis in TICs (Fig. 2H). Together, these findings suggest that ARMS tumorsphere-derived cells undergo metabolic reprogramming, shifting from glycolysis toward lipid metabolism and suggest a potential dependency of TICs on lipid metabolism for survival and tumorigenic potential. In support of PLA2 activity as a broader TIC-associated feature, we also assessed PLA2 isoform expression in tumorsphere-derived cells from embryonal rhabdomyosarcoma (ERMS) cell lines JR1 and RD18. Notably, these ERMS tumorspheres exhibited consistent upregulation of stemness markers (OCT4, SOX2, NANOG, NOTCH3) as well as multiple PLA2 family members (Supplementary Fig. S2).

Fig. 2.

Enrichment of lipid metabolism pathways in tumorsphere-derived cells. A KEGG ORA of upregulated genes in Rh30 and Rh41 tumorsphere-derived cells showing enrichment of lipid metabolism pathways, including LA, ALA, and AA metabolism. B GSEA plots showing significant enrichment of LA, ALA, and AA metabolism in tumorsphere-derived cells versus adherent controls for Rh30 and Rh41 cells. C Venn diagram representing the overlap of DEGs in LA, ALA, and AA metabolic pathways, with common upregulated targets including multiple PLA2 family members. D Relative mRNA fold change of PLA2G5, PLA2G6, PLA2G2D, and PLA2G7 in Rh30, Rh41, and RHB280217 tumorsphere-derived cells compared to adherent controls, as measured by qPCR. E Quantification of PLA2 enzymatic activity in tumorsphere-derived versus adherent cells from Rh30, Rh41, and RHB280217. F KEGG ORA of downregulated genes in Rh30 and Rh41 tumorsphere-derived cells identifying significant suppression of glycolysis-related metabolic pathways. G GSEA plots of downregulated genes showing significant suppression of glycolysis/gluconeogenesis pathways in tumorsphere-derived cells from Rh30 and Rh41 cells. H qPCR validation of key downregulated glycolytic genes (ENO1, PKM, LDHA) in Rh30 and Rh41 tumorsphere-derived cells compared to adherent counterparts. For panels D, E, and H, values represent mean ± SEM from n = 3 biological replicates, each with 3 technical replicates. Statistical significance was determined using unpaired two-tailed Student’s t-tests in GraphPad Prism. (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001)

Lipidomic profiling reveals distinct triglyceride composition and high turnover of PLA2-regulated lipid species in tumorspheres

To understand the metabolic consequences of PLA2-driven lipid remodeling in TICs, we performed lipidomic profiling of tumorsphere-derived versus adherent cells. Differential lipid abundance analysis revealed a broad shift in the lipid profile of TICs, with a significant number of upregulated and downregulated lipid species (Fig. 3A). Notably, several lipid species were commonly enriched across both Rh30 and Rh41 tumorspheres, indicating a shared pattern of lipid alterations characteristic of TICs. (Fig. 3B).

Fig. 3.

A, B Lipidomic profiling identifies enrichment of LA-derived TG and lysophospholipids in tumorspheres. A Volcano plots showing differentially abundant lipid species in tumorsphere-derived versus adherent cells from Rh30 and Rh41 cells. Significantly upregulated lipids are shown in red, and significantly downregulated lipids are shown in blue (threshold: adjusted p < 0.05, log₂ fold change > 1 or < −1). Statistical significance was determined using unpaired two-tailed t-tests and corrected for multiple comparisons using the Benjamini-Hochberg method. B Venn diagram showing commonly upregulated lipid species between Rh30 and Rh41 tumorspheres, indicating conserved lipid alterations. C Bar graph depicting the relative abundance of LA-derived lipid species in Rh30 and Rh41 tumorspheres, highlighting selective accumulation of LA derivatives. D Comparative analysis of lysophospholipid enrichment in Rh30 and Rh41 tumorspheres, consistent with increased PLA2 activity. E Commonly upregulated lipid species in Rh30 and Rh41 tumorspheres categorized by lipid class, with TG emerging as the dominant species. F Heatmap displaying the distribution of differentially expressed TG species in tumorspheres versus adherent cells across Rh30 and Rh41 cells. G LipidMaps-based metabolic network visualization illustrating altered lipid flux and interconversion of TG, DG, and phospholipid species in tumorsphere-derived cells

A closer examination of the lipid composition in tumorspheres identified a preferential enrichment of linoleic acid (LA)-derived triglycerides (TGs) and lysophospholipids (Fig. 3C-D). The accumulation of LA-derived TGs and lysophospholipids suggests a PLA2-mediated shift in lipid metabolism characterized by increased lipid turnover and altered storage and signalling in TICs (Fig. 3C). Similarly, the increased presence of lysophospholipids, a direct byproduct of PLA2 activity, underscores the role of PLA2 in TIC lipid remodeling (Fig. 3D). Further analysis of the commonly upregulated lipid classes in TICs identified triglycerides (TGs) as the top enriched lipid species (Fig. 3E). A comparative analysis of overall TG enrichment revealed a substantial change in different species (Fig. 3F). Pathway analysis using LipidMaps demonstrated alterations in lipid metabolic flux pathways, indicating increased turnover of TG species through PLA2-mediated remodeling (Fig. 3G). Together, these findings suggest that PLA2-mediated lipid remodeling in ARMS TICs results in a distinct lipid profile, characterized by linoleic acid-derived triglyceride accumulation, lysophospholipid enrichment, and enhanced turnover of TGs.

ARMS TICs show increased lipid droplet accumulation

Lipidomic profiling also revealed a significant enrichment of lipid droplet-associated TGs in tumorsphere-derived cells compared to adherent cells. Analysis of lipid complexity distribution showed that while the overall TG complexity did not change significantly between conditions, the profile of specific TG species was altered (Fig. 4A). Given this observation, we examined the subset of TGs associated with lipid droplets and found them to be significantly enriched in tumorspheres, reinforcing the role of lipid storage in TIC biology (Fig. 4B). To investigate the molecular mechanisms underlying lipid accumulation, we examined whether expression of fatty acid uptake and lipid metabolism regulators were apparent in our RNA-Seq analysis. PPARG, a key transcriptional regulator of lipid metabolism, was significantly upregulated in tumorspheres, along with CD36, which mediates extracellular fatty acid uptake (Fig. 4C). These findings suggest that TICs adopt an enhanced lipid uptake and storage as part of their metabolic reprogramming. Lipid droplet accumulation was confirmed through Oil Red O and BODIPY staining, which demonstrated increased intracellular lipid droplets in tumorsphere-derived TICs compared to adherent cells (Fig. 4D, E).

Fig. 4.

Tumorspheres exhibit increased accumulation of lipid droplet-associated triglycerides and lipid storage. A Distribution of TG species by adjusted complexity score across Rh30 and Rh41 tumorsphere and adherent cells, showing no major change in triglyceride complexity despite altered lipid profiles. B Bar graph illustrating the selective enrichment of lipid droplet-associated TG in tumorsphere-derived cells from Rh30 and Rh41 cells compared to adherent counterparts. C Relative mRNA fold change of PPARG and CD36 expression in tumorsphere versus adherent cells, showing upregulation of lipid uptake and storage genes. D Representative images of Oil-Red-O-stained cells showing increased intracellular lipid droplets in tumorspheres compared to adherent counterparts across Rh30, Rh41, and RHB280217. Quantification of lipid droplet number per cell is shown on the right. E Representative BODIPY 493/503 fluorescence images confirming elevated lipid droplet accumulation in tumorsphere-derived cells. Corresponding quantification of lipid droplet count per cell is shown for all three models. For panels B–E, values represent mean ± SEM from n = 3 biological replicates, each with 3 technical replicates. For panels C-E statistical significance was determined using unpaired two-tailed Student’s t-tests in GraphPad Prism. (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001). Scale bar represents 20 μm

PLA2-mediated lipid remodeling supports tumorsphere formation and cellular motility

To functionally validate the role of PLA2-driven lipid metabolism in ARMS TICs, we investigated the effect of Darapladib, a PLA2 inhibitor. MTT assays showed that ARMS cells displayed greater sensitivity to Darapladib compared to normal human skeletal myoblasts (HSMM) (Fig. 5A). Importantly, tumorsphere-derived cells exhibited lower IC₅₀ values compared to their respective adherent counterparts, further supporting the selective vulnerability of TICs to PLA2 inhibition (Supplemental Fig. 3 A). Treatment with Darapladib significantly reduced tumorsphere number and size across Rh30, Rh41, and RHB280217 compared to vehicle-treated controls (Fig. 5B), indicating that PLA2 inhibition impairs self-renewal capacity. Moreover, PLA2 inhibition significantly reduced the cell motility of Rh30 and Rh41 cells (Fig. 5C).

Fig. 5.

PLA2 inhibition by Darapladib impairs tumorsphere formation, cellular motility, and tumor progression. A MTT assay showing dose–response curves and IC₅₀ values for Darapladib in Rh30, Rh41, RHB280217 and HSMM cells, indicating cytotoxicity in ARMS tumor cells. B Tumorsphere formation assay showing a marked reduction in sphere number and size in Rh30, Rh41, and RHB280217 cells upon Darapladib (1 µM) treatment. C Transwell migration assay revealing impaired migratory capacity of Rh30 and Rh41 TICs following PLA2 inhibition. Quantification of migration is shown in the right panels. D Body weight of mice upon treatment with Darapladib (25 mg/kg/day), showing no major differences between treatment and control groups. E Kaplan–Meier survival analysis indicating prolonged tumor latency in Darapladib-treated mice compared to vehicle-treated controls. F Representative image of excised tumors from DMSO and Darapladib-treated groups. Quantification of tumor weight and volume shows significant reduction following PLA2 inhibition. G qPCR analysis of stemness-associated genes (Oct4, Sox2, Nanog) and lipid uptake marker (CD36) in tumors from DMSO versus Darapladib-treated mice. Cell viability (MTT) assay in Fig. 5A were performed in technical triplicates and repeated in three independent experiments. IC₅₀ values were calculated using nonlinear regression curve fitting (log[inhibitor] vs. normalized response – variable slope) in GraphPad Prism. For in vitro assays (B, C, G), values represent mean ± SEM from n = 3 biological replicates, each with 3 technical replicates. Comparisons were made using unpaired two-tailed Student’s t-tests in GraphPad Prism. For in vivo tumor weight and volume comparisons (F), data from n = 6 mice per group were analyzed using unpaired two-tailed Student’s t-tests. Survival (E) was compared using the log-rank (Mantel–Cox) test. (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001). Scale bar represents 100 μm

We next evaluated PLA2-driven lipid metabolism in vivo using xenograft models. Rh30 tumorsphere-derived cells were injected into immunocompromised mice, and tumor-bearing mice were treated with Darapladib or DMSO. No significant loss in body weight was apparent between Darapladib-treated and control groups (Fig. 5D). Kaplan-Meier survival analysis showed that Darapladib treatment significantly prolonged the time required for tumors to reach the euthanasia threshold size, demonstrating that PLA2 inhibition extends survival in vivo (Fig. 5E). Additionally, tumors in Darapladib-treated mice were significantly smaller in volume and weight compared to controls (Fig. 5F), demonstrating that PLA2 inhibition reduces the tumor progression capacity of ARMS TICs. Importantly, quantitative PCR analysis of tumors from treated mice revealed a sharp downregulation of key stemness-associated transcription factors-including Oct4, Sox2, and Nanog-as well as the lipid transporter CD36 in Darapladib-treated tumors compared to controls (Fig. 5G). These findings establish PLA2 activity as a crucial regulator of self-renewal, motility, with potential to prolong survival by delaying tumor progression.

Linoleic acid signalling and metabolic flexibility support PLA2-driven tumorigenicity

Since tumorspheres showed an enrichment of LA species indicative of lipid signalling, along with an increased lipid droplets which might be utilised as an energy source through FAO, we investigated the relevance of this axis in tumorspheres.

To determine whether LA signalling contributes to PLA2-driven tumorigenicity, we supplemented Darapladib-treated cells with exogenous LA. LA supplementation significantly, although not completely, rescued tumorsphere formation in both Rh30 and Rh41 TICs demonstrating that LA metabolism contributes significantly to TIC self-renewal capacity (Fig. 6A). A dose-dependent increase in tumorsphere number was observed with LA supplementation in Darapladib-treated TICs which peaked between 2 and 3 mg/ml LA (Fig. 6B). We next assessed whether LA supplementation rescues the impaired migration caused by PLA2 inhibition. LA supplementation reversed Darapladib-mediated impairment of migration to levels observed in untreated controls (Fig. 6C). To further investigate whether LA supplementation affects PLA2 expression in Darapladib-treated cells, we examined mRNA levels of PLA2G5, PLA2G6, and PLA2G7 in Rh30 and Rh41 TICs. LA supplementation partially restored PLA2G5 and PLA2G6 expression compared to Darapladib treatment alone, while PLA2G7 expression was almost completely rescued (Supplemental Fig. 3B, C). Together, these findings indicate that LA metabolism contributes to TIC-driven tumor progression by modulating both self-renewal and motility.

Fig. 6.

LA supplementation rescues PLA2 inhibition, and TICs exhibit metabolic flexibility. A Tumorsphere formation assay in Rh30 and Rh41 cells treated with Darapladib, showing a reduction in sphere number, which is rescued by co-treatment with LA. Quantification of sphere number is shown on the right. B Dose-dependent rescue of tumorsphere formation by increasing concentrations of LA in Darapladib-treated TICs. C Transwell migration assay showing that LA supplementation restores the migratory capacity of Darapladib-treated Rh30 and Rh41 TICs. Quantification of migrated cells is shown on the right. D ECAR measurements reveal reduced glycolytic flux in tumorsphere-derived cells compared to adherent controls. E OCR measurements show only partial suppression of mitochondrial respiration upon etomoxir treatment in Rh30 (left) and Rh41 (right) tumorspheres, suggesting limited reliance on long-chain fatty acid oxidation. F Tumorsphere formation assay with 2-deoxy-D-glucose (2-DG) treatment shows no significant effect on tumorsphere supporting glycolytic independence of TICs. G Treatment with Etomoxir does not significantly impair tumorsphere formation. For panels A, C, F, and G, values represent mean ± SEM from n = 3 biological replicates. Statistical comparisons were performed using unpaired two-tailed Student’s t-tests in GraphPad Prism unless otherwise noted. For Seahorse ECAR and OCR experiments (panels D and E), data represent mean ± SEM from technical triplicates in at least 3 independent experiments. (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001; ns = not significant). Scale bar represents 100 μm

To characterize the substrate dependence of TICs to meet energy needs, we assessed glycolytic and mitochondrial metabolism. Extracellular acidification rate (ECAR) measurements revealed a global downregulation of glycolytic flux in TICs compared to adherent cells, consistent with previous transcriptomic findings of downregulated glycolytic genes (Fig. 6D). Given the enrichment of lipid-associated TGs and the metabolic reliance on fatty acids, we next examined the contribution of long-chain fatty acid (LCFA) oxidation to mitochondrial respiration. Oxygen consumption rate (OCR) analysis revealed that treatment with etomoxir, an inhibitor of CPT1A-dependent fatty acid oxidation, resulted in a partial reduction in OCR, indicating that tumorspheres do not exclusively rely on LCFA metabolism for mitochondrial respiration (Fig. 6E and Supplemental Fig. 3D). This suggests that TICs maintain a degree of metabolic flexibility, enabling them to adapt to different energy sources. To directly assess whether energy production through glycolysis or FAO is required for tumorsphere formation, we treated TICs with 2-deoxy-D-glucose (2-DG), an inhibitor of glycolysis, and etomoxir. 2-DG treatment did not significantly impair tumorsphere formation in Rh30 and Rh41 cells, suggesting that glycolysis is not a dominant metabolic pathway supporting TIC propagation under these conditions (Fig. 6F). Interestingly, etomoxir treatment also did not significantly reduce tumorsphere formation in Rh30 or Rh41 cells, indicating that FAO is not the sole metabolic driver of TIC maintenance under these conditions (Fig. 6G).

Taken together, our data demonstrate a critical role for LA-driven lipid signalling downstream of PLA2 activity in supporting TIC function, but TICs do not depend on any single metabolic pathway for energy production.

Discussion

Fusion-positive ARMS remains a highly aggressive paediatric malignancy with limited therapeutic options in relapsed and metastatic settings [3]. Recent single-cell transcriptomic and lineage-tracing studies have uncovered dynamic cellular states in FP-ARMS, including proliferative, stress-resistant, and immune-evasive populations. However, the metabolic features underpinning these states remain poorly defined [15, 18]. Tumorsphere culture systems have been widely employed to enrich for tumor-initiating or stem-like cells in various cancers, including sarcomas [35–37]. In ARMS, tumorsphere-derived cells have been shown to exhibit enhanced self-renewal, resistance to therapy, and tumorigenic potential in vivo, supporting their use as a TIC-enriched model system [34, 38]. Moreover, despite growing evidence that lipid metabolism supports therapy resistance and metastatic seeding in several cancers [39], its role in ARMS has remained largely unexplored. Our study addresses these gaps by identifying PLA2-mediated lipid remodeling as a key metabolic adaptation that enriches bioactive lipid species involved in signalling pathways critical for TIC maintenance and tumor progression in ARMS. By linking PLA2 activity to bioactive lipid turnover and plasticity, we highlight a non-canonical metabolic vulnerability of ARMS TICs.

We show that TIC-enriched tumorspheres exhibit transcriptional upregulation of multiple PLA2 isoforms and suppression of glycolytic genes, consistent with a metabolic shift toward lipid metabolism. This is functionally supported by lipidomic data showing accumulation of lysophospholipids and linoleic acid (LA)-derived triglycerides-direct or indirect PLA2 products. These alterations coincide with upregulation of lipid storage regulators such as CD36 and PPARG and increased lipid droplet accumulation, suggesting that ARMS TICs store and mobilize lipid reserves to support survival and tumorigenicity. PLA2 inhibition via Darapladib significantly impaired tumorsphere formation, migration, and tumor progression. Strikingly, supplementation with exogenous LA restored these phenotypes, suggesting that lipid signalling-not energy production per se-is central to TIC maintenance. Unlike solid tumors that rely heavily on aerobic glycolysis known as the Warburg effect [40–42], ARMS TICs exhibited suppressed glycolytic flux and were minimally affected by FAO inhibition.

The effects of PLA2 inhibition in our study align with emerging evidence that PLA2 enzymes, particularly Lp-PLA2 encoded by PLA2G7, play multifaceted roles in cancer biology. For instance, Darapladib has been shown to sensitize cancer cells to ferroptosis by remodeling lipid metabolism, leading to increased phosphatidylethanolamine species and reduced lysophosphatidylethanolamine levels [43]. Additionally, Darapladib has demonstrated antitumor activity in glioma models by inducing apoptosis and mitochondrial dysfunction [44]. LA-derived oxylipins, such as 9- and 13-hydroxyoctadecadienoic acid (HODE), have been shown to promote tumor growth and metastasis by activating signalling pathways such as peroxisome proliferator-activated receptors (PPARs) and G protein-coupled receptors like GPR132 [45–47]. These metabolites can influence processes such as cell proliferation, apoptosis, and inflammation, which are critical in tumor development and progression.

Our findings are consistent with a broader shift in cancer biology that recognizes lipid metabolism as a hallmark of tumor progression and stemness [48–50]. CD36 has been shown to mediate fatty acid uptake and promote metastasis [51, 52], while PPARG is a key transcriptional regulator of lipid storage and stem-like plasticity [53]. Our study extends previous observations in RMS showing antitumor effects of general lipid metabolism inhibition, where pharmacologic suppression of FAO via malonyl-CoA decarboxylase inhibition (MCDi) led to malonyl-CoA accumulation, CPT1 inhibition, and reduced tumor growth in vivo [54]. That study demonstrated metabolic reprogramming characterized by increased glycolysis, suppression of the pentose phosphate pathway, and induction of autophagy and p21. However, the focus of that study remained on bulk tumor populations and did not address the role of lipid metabolism in TICs.

Recent single-cell studies on FP-ARMS developmental plasticity have characterized the tumor as transcriptionally locked in a progenitor-like or regenerative muscle state, with cycling between proliferative and stress-resistant subpopulations. Our findings suggest that PLA2-driven lipid remodeling may provide the molecular substrate enabling such transitions, facilitating membrane remodeling, stress adaptation, and signal transduction necessary for state switching. Moreover, lipid metabolism has been implicated in the creation of immunosuppressive niches in other cancers, raising the possibility that PLA2 activity may also modulate tumor-immune interactions in FP-ARMS.

Several PLA2 enzymes were found to be upregulated in tumorspheres including PLA2G5, PLA2G6, PLA2G2D and PLA2G7. PLA2 activity assays confirmed functional upregulation in tumorspheres. However, since Darapladib exhibits greater selectivity for PLA2G7, the role of the other isoforms remains to be investigated. Future studies employing broader PLA2 inhibitors and metabolic flux tracing using stable isotope-labeled fatty acids will be needed to define the specific contributions of each enzyme and lipid species. Additionally, whether LA functions through its incorporation into membranes, generation of signaling lipids, or restoration of structural lipid balance remains to be investigated. Testing a broader panel of fatty acids and combining PLA2 inhibitors with inhibitors of lipid uptake (e.g., CD36) or lipolysis could clarify this mechanism.

In conclusion, our study identifies PLA2-driven lipid metabolism as a central adaptation in ARMS TICs. This remodeling supports lipid storage, signaling, and tumor progression in metabolically flexible cells. This work provides a foundation for targeting TICs through PLA2-mediated lipid remodeling in high-risk ARMS.

Abbreviations

AA

Arachidonic acid

ALA

Alpha-linolenic acid

ANOVA

Analysis of variance

ARMS

Alveolar rhabdomyosarcoma

BODIPY

Boron-dipyrromethene (lipid staining dye)

CD36

Cluster of differentiation 36

CPT1A

Carnitine palmitoyltransferase 1A

DEGs

Differentially expressed genes

DMSO

Dimethyl sulfoxide

ECAR

Extracellular acidification rate

EGF

Epidermal growth factor

FAO

Fatty acid oxidation

FBS

Fetal bovine serum

FCCP

Carbonyl cyanide-p-trifluoromethoxyphenylhydrazone

FGF

Fibroblast growth factor

GEMM

Genetically engineered mouse model

GPR132

G protein-coupled receptor 132

GSEA

Gene set enrichment analysis

HODE

Hydroxyoctadecadienoic acid

HSMM

Human skeletal muscle myoblasts

IC₅₀

Half-maximal inhibitory concentration

LA

Linoleic acid

LCFA

Long-chain fatty acid

LC-MS/MS

Liquid chromatography tandem mass spectrometry

LDHA

Lactate dehydrogenase A

MCDi

Malonyl-CoA decarboxylase inhibitor

MTT

Methyl thiazolyl tetrazolium

NB

Neurobasal (medium)

OCT4

Octamer-binding transcription factor 4 (POU5F1)

OCR

Oxygen consumption rate

ORA

Over-representation analysis

PBS

Phosphate-buffered saline

PCA

Principal component analysis

PDX

Patient-derived xenograft

PLA2

Phospholipase A2

PLA2G5/6/7/2D

Specific phospholipase A2 isoforms

PPARG

Peroxisome proliferator-activated receptor gamma

qPCR

Quantitative polymerase chain reaction

RNA-seq

RNA sequencing

RMS

Rhabdomyosarcoma

SOX2

SRY-box transcription factor 2

TICs

Tumor-initiating cells

TGs

Triglycerides

2-DG

2-Deoxy-d-glucose

Author contributions

Amogh Gupta: conceptualization, methodology, validation, writing and editing. Bharathi Ramanathan: validation. Dipanwita Das: investigation, methodology. Amos Loh: resources, writing and editing. Aiswariya Vadivellu: resources, methodology. Reshma Taneja: conceptualization, project administration, supervision, writing.

Funding

This work was supported by a Ministry of Education MOE-T2EP30222-0014 grant to R. Taneja.

Data availability

RNA-Seq data has been deposited in GEO database under accession number GSE295852. Private token for the reviewers is sdclqykqfbehbif.

Declarations

Competing interests

The authors declare no competing interests.