PPARα-mediated lipid metabolism reprogramming supports anti-EGFR therapy resistance in head and neck squamous cell carcinoma

Abstract

Anti-epidermal growth factor receptor (EGFR) therapy (cetuximab) shows a limited clinical benefit for patients with locally advanced or recurrent/metastatic head and neck squamous cell carcinoma (HNSCC), due to the frequent occurrence of secondary resistance mechanisms. Here we report that cetuximab-resistant HNSCC cells display a peroxisome proliferator-activated receptor alpha (PPARα)-mediated lipid metabolism reprogramming, with increased fatty acid uptake and oxidation capacities, while glycolysis is not modified. This metabolic shift makes cetuximab-resistant HNSCC cells particularly sensitive to a pharmacological inhibition of either carnitine palmitoyltransferase 1A (CPT1A) or PPARα in 3D spheroids and tumor xenografts in mice. Importantly, the PPARα-related gene signature, in human clinical datasets, correlates with lower response to anti-EGFR therapy and poor survival in HNSCC patients, thereby validating its clinical relevance. This study points out lipid metabolism rewiring as a non-genetic resistance-causing mechanism in HNSCC that may be therapeutically targeted to overcome acquired resistance to anti-EGFR therapy.

Fig. 1. Acquired resistance to cetuximab is associated with expression changes for metabolism-related genes and proteins.

Growth of cetuximab-sensitive (-S) and -resistant (-R) SCC22b (a) and SC263 cells (b) after treatment with 1 µg/mL cetuximab (CTX) for 96 h (N = 3, n = 6). c General workflow of the integration of SCC22b-S/-R and SC263-S/-R proteomic analysis with a microarray-DNA data-based inference of protein activity by VIPER algorithm. Created in BioRender. https://BioRender.com/d17s016. d Heatmap showing the –log10 transformed enrichment p–values of different gene sets, indicated along the Y-axis, in each of the six identified network clusters (X-axis). A white color reflects a p-value of 0.05, hence all red colors indicate significant interactions. e Dot plot showing –log10 transformed enrichment p-values (X-axis) of gene sets (Y-axis) in clusters 1 (top panel) and 6 (bottom panel) of the protein-protein interaction network. Each dot is color-coded according to the enrichment odds ratio (OR). Small black dots indicate the enrichment p-value of that gene set in the entire network, in contrast to the cluster specific enrichment. A p-value of 0.05 is indicated using a vertical dashed line. f Minimal spanning tree representation of the protein-protein interaction network of cluster 1. Proteins in red and blue are respectively up- and downregulated in cetuximab-resistant HNSCC cells. The shape of the nodes reflects the protein class (i.e. circles=metabolic seeds, triangles=non-metabolic seeds). Grey nodes are intermediate proteins, that were identified using the shortest-paths algorithm and that are necessary to connect all proteins in the network. g Bar plot showing the node degree distribution of all proteins in the network of cluster 1. The node degree, being the number of connections each protein has in the network, is shown in the X-axis, the different proteins are listed along the Y-axis. Grey bars reflect intermediate proteins. Data are plotted as the means ± SEM (a, b). N indicates the number of independent biological experiments and n indicates the number of technical replicates. Significance was determined by two-way ANOVA with Sidák’s multiple comparison test (a, b). ***P < 0.001; ns not significant. Source data are provided as a Source Data file.

Fig. 2. Glycolytic activity is not changed in HNSCC cells upon acquired resistance to anti-EGFR therapy.

Glucose and lactate concentrations, at different timings, in the extracellular media from cetuximab-sensitive (-S) and -resistant (-R) FaDu (a) and SC263 (b) cells initially incubated at t0 in a medium containing 10 mM glucose (N = 3, n = 3). c Glucose-dependent extracellular acidification rate (ECAR) in cetuximab-sensitive (-S) and -resistant (-R) FaDu and SC263 cells (N = 3, n = 6). Growth of cetuximab-sensitive (-S) and -resistant (-R) FaDu (d) and SC263 cells (e) after treatment with 10 mM 2-deoxyglucose (2-DG) for 72 h (N = 3, n = 3). f ¹³C-lactate signal, detected by nuclear magnetic resonance, in 2D (N = 4), 3D N = 4) and ex vivo cultures (N = 6) of cetuximab-sensitive (-S) and -resistant (-R) FaDu cells after incubation with U-¹³C-glucose for 24 h. Data are plotted as the means ± SEM (a–f). N indicates the number of independent biological experiments and n indicates the number of technical replicates (when >1). Significance was determined by two-way ANOVA with Sidák’s multiple comparison test (c–f). ***P < 0.001; ns not significant. Source data are provided as a Source Data file.

Fig. 3. Cetuximab resistance is associated with increased fatty acid uptake and oxidation in HNSCC cells.

Oxygen consumption rate (OCR) of cetuximab-sensitive (-S) and -resistant (-R) FaDu (a) and SC263 cells (b) in media containing either glucose (Glc), glutamine (Gln), or palmitate (PA) (N = 3, n = 6). c Quantification for BODIPY FL C16 uptake in cetuximab-sensitive (-S) and -resistant (-R) FaDu and SC263 cells (N = 3). d Intracellular abundance of FAcyl-CoA (i.e. palmitoyl-CoA, palmitoleyl-CoA, oleyl-CoA, stearoyl-CoA) in cetuximab-sensitive (-S) and -resistant (-R) FaDu and SC263 cells (N = 3). mRNA expression (e) and representative immunoblotting (f) for CD36 in cetuximab-sensitive (-S) and -resistant (-R) FaDu and SC263 cells (N = 3). g Quantification for BODIPY FL C16 uptake in cetuximab-sensitive (-S) and -resistant (-R) FaDu cells upon transfection of siRNA against CD36 gene for 72 h (N = 2). mRNA expression (h) and representative immunoblotting (i) for CPT1A in cetuximab-sensitive (-S) and -resistant (-R) FaDu and SC263 cells (N = 3, n = 3). j Palmitate-dependent OCR in cetuximab-sensitive (-s) and -resistant (-R) FaDu cells upon treatment with 100 µM etomoxir for 30 min (N = 3, n = 6). k Intracellular abundance of acetyl-CoA in cetuximab-sensitive (-S) and -resistant (-R) FaDu cells upon treatment with 100 µM etomoxir for 3 h (N = 3). l Heatmap showing the upstream regulators (X-axis) of ACACA, CPT1A and FASN (Y-axis) in the protein-protein interaction network identified in Fig. 1. Upstream regulators directly interact with any of the proteins of interest (i.e. network distance of 1) and are differentially expressed or activated between cetuximab-resistant and -sensitive HNSCC cells. The direction and level of differential activation and/or repression is color-coded as shown in the legend underneath the heatmap. Data are plotted as the means ± SEM. N indicates the number of independent biological experiments and n indicates the number of technical replicates (when >1). Significance was determined by two-way ANOVA with Sidák’s multiple comparison test (a–e, h and j, k). P-values as indicated or ***P < 0.001; ns not significant. Source data are provided as a Source Data file.

Fig. 4. Increased stearoyl-CoA desaturase activity prevents lipotoxicity in cetuximab-resistant HNSCC cells.

Intracellular abundance of total FA (a) and relative proportion of SFA, MUFA and PUFA (b) in cetuximab-sensitive (-S) and -resistant (-R) FaDu and SC263 cells (N = 3). Intracellular FA abundance in neutral lipids (NL), free FA (FFA) and phospholipids (PL) fractions in cetuximab-sensitive (-S) and -resistant (-R) FaDu (c) and SC263 cells (d) upon treatment with 50 µM BSA-conjugated palmitate (PA) for 6 h (N = 3). e Δ9 FA desaturation index in cetuximab-sensitive (-S) and -resistant (-R) FaDu and SC263 cells (N = 3). Representative immunoblotting (f) and mRNA expression (g) for SCD1 and SCD5 in cetuximab-sensitive (-S) and -resistant (-R) FaDu and SC263 cells (N = 3, n = 3). Relative abundance of saturated FA (SFA), monounsaturated FA (MUFA) and polyunsaturated FA (PUFA) in FaDu and SC263 cells, either at total levels (h, i) or in the NL fraction ( j, k) after treatment with 40 µM A939572 for 24 h (N = 3). l Growth of cetuximab-sensitive (-S) and -resistant (-R) FaDu cells after treatment with 40 µM A939572 (SCD1 inhibitor) with or without the addition of 100 µM PA for 72 h (N = 3, n = 6). Data are plotted as the means ± SEM. N indicates the number of independent biological experiments and n indicates the number of technical replicates (when >1). Significance was determined by two-way ANOVA with Sidák’s multiple comparison test (a, b, e, and g–l). P-values as indicated or ***P < 0.001; ns not significant. Source data are provided as a Source Data file.

Fig. 5. Cetuximab-resistant HNSCC cells rely on the availability and utilization of exogenous FA for their growth in vitro and in vivo.

a Growth of cetuximab-sensitive (-S) and resistant (-R) FaDu cells upon transfection of siRNA against CD36 for 72 h (N = 3, n = 3). Growth of cetuximab-resistant (-R) FaDu (b) and SC263 cells (c) upon treatment with 1 µg/ml cetuximab for 72 h, in a medium supplemented with normal or delipidated FBS (N = 3, n = 3). Growth of FaDu-R (d) and SC263-R cells (e) upon treatment with 1 µg/ml cetuximab for 72 h, in a delipidated FBS-containing medium supplemented with 1% chemically defined lipid concentrate (LC) or a mix of BSA-conjugated palmitate and oleate (50 µM of each FA) (N = 3, n = 6). Growth of cetuximab-sensitive FaDu cells upon incubation up to 6 weeks with 1 µg/mL cetuximab in full or lipid-depleted culture medium (f) and comparison of cell viability at 1 week and 6 weeks of cetuximab treatment (g) (N = 2, n = 3). Growth of cetuximab-sensitive (-S) and -resistant (-R) FaDu cells in 2D (h) and 3D conditions (i) upon treatment with 100 µM etomoxir for 3 or 14 days respectively (N = 3, n = 3). j Volume at day 17 of in vivo tumor xenografts from FaDu-R cells in nude mice daily treated with 40 mg/kg etomoxir (or DMSO as vehicle) alone or in combination with 30 mg/kg cetuximab (or 0.9% NaCl as vehicle) (N = 6). Data are plotted as the means ± SEM. N indicates the number of independent biological experiments and n indicates the number of technical replicates (when >1). Significance was determined by two-way ANOVA with Sidák’s multiple comparison test (a–e, h and j). P-values as indicated or ***P < 0.001; ns not significant. Source data are provided as a Source Data file.

Fig. 6. A PPARα-dependent transcriptional program governs lipid metabolism rewiring in cetuximab-resistant HNSCC cells.

a Bar plot showing –log10 transformed enrichment p-values (X-axis) of PPAR-associated gene sets (Y-axis) between cetuximab-sensitive and resistant SCC22b and SC263 cells. A dashed vertical line indicates a p-value of 0.05. PPARα DNA-binding activity (b) and nuclear localization (c) in cetuximab-sensitive (-S) and -resistant (-R) FaDu and SC263 cells (N = 2, n = 2). mRNA expression for CD36 (d) and palmitate-dependent OCR (e) in cetuximab-sensitive (-S) FaDu and SC263 cells upon treatment with 20 µM pemafibrate for 48 h (N = 3, n = 3). Growth of cetuximab-sensitive (-S) FaDu (f) and SC263 cells (g) upon treatment with 20 µM pemafibrate for 48 h and then with 1 µg/mL cetuximab for 72 h (N = 3, n = 6). mRNA expression for CD36 and CPT1A (h) and palmitate-dependent OCR (i) in cetuximab-resistant (-R) FaDu and SC263 cells upon treatment with 10 µM GW6471 for 24 h (N = 3, n = 3). Growth of cetuximab-sensitive (-S) and -resistant (-R) FaDu (j) and SC263 cells (k) upon treatment with 10 µM GW6471 for 72 h (N = 3, n = 6). l Follow-up of spheroid growth from cetuximab-sensitive (-S) and -resistant (-R) FaDu cells upon treatment with 10 µM GW6471 for 14 days (N = 3, n = 6). Volume at days 20 or 25 of in vivo tumor xenografts from cetuximab-resistant (-R) FaDu (m) and SCC22b cells (n), respectively, in nude mice daily treated with 20 mg/kg GW6471 (or DMSO as vehicle) alone or in combination with 30 mg/kg cetuximab (or 0.9% NaCl as vehicle) (N = 6). Data are plotted as the means ± SEM. N indicates the number of independent biological experiments and n indicates the number of technical replicates (when >1). Significance was determined by two-way ANOVA with Sidák’s multiple comparison test (c–k and m, n). P-values as indicated or ***P < 0.001; ns not significant. Source data are provided as a Source Data file.

Fig. 7. An active crosstalk between EGFR signaling and lipid metabolism reprogramming supports acquired resistance to cetuximab in HNSCC.

Growth of cetuximab-resistant (-R) FaDu (a) and SC263 cells (b) upon 5 weeks of drug withdrawal (w/d) from the routine culture medium and retreatment with 10 µM GW6471 or 100 µM etomoxir for 72 h. As a comparison, cetuximab-treated parental cell populations (-S) are also indicated (N = 2, n = 3). c Representative immunoblotting for CPT1A in cetuximab-resistant FaDu and SC263 cells upon 5 weeks of drug withdrawal (w/d). mRNA expression (d) and representative immunoblotting (e) for EGFR in cetuximab-sensitive (-S) and -resistant (-R) FaDu and SC263 cells (N = 3, n = 3). Expression levels of surface-localized EGFR in cetuximab-sensitive (-S) and -resistant (-R) FaDu (f) and SC263 cells (g) upon treatment with 10 µM GW6471 or incubation in lipid-depleted medium for 72 h (N = 2). h Growth of cetuximab-resistant FaDu cells, pre-exposed to 10 µM GW6471 or lipid-depleted medium for 72 h, before treatment with 1 µg/mL cetuximab for 72 h (N = 2, n = 3). mRNA expression for CPT1A in cetuximab-sensitive (-S) and -resistant (-R) FaDu (i) and SC263 cells (j) upon treatment with gefitinib, afatinib, selumetinib or buparlisib (2 µM each, for 24 h) (N = 3, n = 3). Data are plotted as the means ± SEM. N indicates the number of independent biological experiments and n indicates the number of technical replicates (when >1). The samples in c derive from the same experiment but different gels for CPT1A were processed in parallel for the 2 cell lines. The samples in e derive from the same experiment and one gel for EGFR was run. Significance was determined by two-way ANOVA with Sidák’s multiple comparison test (d and i, j). ***P < 0.001; ns not significant. Source data are provided as a Source Data file.

Fig. 8. PPAR-dependent FA metabolism dysregulation correlates with cetuximab resistance in patient-derived HNSCC models.

a Schematic representation for the establishment of cetuximab-sensitive and -resistant patient-derived xenograft (PDX) models from treatment-naive HNSCC clinical specimens. Created in BioRender. https://BioRender.com/b56r875. Tumor growth of cetuximab-sensitive (b) and -resistant (c) UCLHN4 PDX in nude mice weekly treated with 30 mg/kg cetuximab or vehicle. Xenografts were considered resistant when tumors reached 200% ±10% of their initial size. d, e Individual GSEA plots of FA metabolism-related pathways in RNA-seq data from cetuximab-sensitive and -resistant UCLHN4 PDX (N = 3 for each condition). f mRNA expression for CD36 in cetuximab-sensitive and -resistant UCLHN4 PDX (N = 3). Representative tumor slides (g) and quantification (h) for anti-CD36 immunohistochemistry on tissue sections from cetuximab-sensitive and -resistant UCLHN4 PDX. Scale bar: 100 µm (N = 4). Follow-up of tumor growth (i) and relative tumor size at day 15 (j) for cetuximab-resistant UCLHN4 PDX in nude mice treated with 30 mg/kg cetuximab in combination with either 40 mg/kg etomoxir or 20 mg/kg GW6471. k Schematic representation for the clinical workflow of the window-of-opportunity study with HNSCC patients treated for 14 days with cetuximab. ¹⁸FDG-PET scan was carried out before and after treatment and enabled to discriminate between partial metabolic response (PMR) and stable metabolic disease (SMD). Created in BioRender. https://BioRender.com/o74h220. l GSEA plot of the PPARα-related gene signature in RNA-seq data from cetuximab-treated HNSCC patients (SMD vs PMR) (N = 15). Data are plotted as the means ± SEM. N indicates the number of independent biological experiments and n indicates the number of technical replicates (when >1). Significance was determined by two-tailed unpaired Student’s t-test (h), one-way ANOVA with Dunnett’s multiple comparison test (j) or two-way ANOVA with Sidák’s multiple comparison test (f). P-values as indicated or ***P < 0.001. Source data are provided as a Source Data file.