Metabolic reprogramming from glycolysis to fatty acid uptake and beta-oxidation in platinum-resistant cancer cells

Abstract

Increased glycolysis is considered as a hallmark of cancer. Yet, cancer cell metabolic reprograming during therapeutic resistance development is under-studied. Here, through high-throughput stimulated Raman scattering imaging and single cell analysis, we find that cisplatin-resistant cells exhibit increased fatty acids (FA) uptake, accompanied by decreased glucose uptake and lipogenesis, indicating reprogramming from glucose to FA dependent anabolic and energy metabolism. A metabolic index incorporating glucose derived anabolism and FA uptake correlates linearly to the level of cisplatin resistance in ovarian cancer (OC) cell lines and primary cells. The increased FA uptake facilitates cancer cell survival under cisplatin-induced oxidative stress by enhancing beta-oxidation. Consequently, blocking beta-oxidation by a small molecule inhibitor combined with cisplatin or carboplatin synergistically suppresses OC proliferation in vitro and growth of patient-derived xenografts in vivo. Collectively, these findings support a rapid detection method of cisplatin-resistance at single cell level and a strategy for treating cisplatin-resistant tumors.

Fig. 1. High-throughput imaging of lipid metabolism in isogenic pairs of cisplatin-sensitive and -resistant OC cells.

a Representative large-area SRS images of parental PEO1 and cisplatin-resistant PEO4 cells. b Histograms of integrated cellular lipid intensity in PEO1 and PEO4 cells generated through high-throughput single-cell analysis. c Representative large-area SRS images of parental SKOV3 and cisplatin-resistant SKOV3-cisR cells. d Histograms of integrated cellular lipid intensity in SKOV3 and SKOV3-cisR cells. e Histograms of integrated cellular lipid intensity in SKOV3 cells treated with or without cisplatin. f Histograms of integrated cellular lipid intensity in SKOV3-cisR cells treated with or without cisplatin. g Representative hyperspectral SRS image (sum of all channels) and Phasor mapped lipid image of sliced OVCAR5 xenograft tumor tissue from mouse treated with vehicle (sensitive) or carboplatin (resistant). h Quantitative analysis of SRS signal from lipid in carboplatin sensitive and resistant ovarian tumor tissue by mean intensity. Data are presented as means + SD; n = 3 animals; two-sided Student’s t test; P = 0.043; *P < 0.05. Scale bar: 20 µm. Source data are provided in the Source Data file.

Fig. 2. Increased FA uptake, not de novo lipogenesis, is the major contributor to lipid accumulation in cisplatin-resistant OC cells.

a Representative bright field and SRS images of PEO1 and PEO4 cells fed with glucose-d₇ for 3 days. b Quantitative analysis of SRS signal of C-D bonds in glucose-d₇ fed PEO1 and PEO4 cells by mean intensity and area fraction. n = 5. P = 0.0076 and 0.0083. c Representative bright field and SRS images of PEO1 and PEO4 cells fed with PA-d₃₁ for 6 h. d Quantitative analysis of SRS signal of C-D bonds in PA-d₃₁ fed PEO1 and PEO4 cells by mean intensity and area fraction. n = 6. P = 0.0051 and 3 × 10⁻⁵. e Representative bright field and SRS images of PEO1 and PEO4 cells fed with OA-d₃₄ for 6 h. f Quantitative analysis of SRS signal of C-D bonds in OA-d₃₄ fed PEO1 (n = 6) and PEO4 (n = 7) cells by mean intensity and area fraction. P = 0.030 and 0.0048. g Representative SRS images of SKOV3 and SKOV3-cisR cells fed with glucose-d₇ for 3 days and quantitative analysis of SRS signal of C-D bonds by mean intensity. n = 7. P = 0.00075. h Representative SRS images of SKOV3 and SKOV3-cisR cells fed with PA-d₃₁ for 6 h and quantitative analysis of SRS signal of C-D bonds by mean intensity. n = 5. P = 0.0010. i Representative SRS images of SKOV3 and SKOV3-cisR cells fed with OA-d₃₄ for 6 h and quantitative analysis of SRS signal of C-D bonds by mean intensity. n = 8. P = 2.2 × 10⁻⁵. Data in all the bar charts (b, d, f and g–i) are shown as means + SD. All n represents technical replicates. Statistical significance was analyzed using one-sided Student’s t test. *P < 0.05, **P < 0.01, and ***P < 0.001. Scale bar: 20 µm. Source data are provided in the Source Data file.

Fig. 3. Metabolic index by integrating glucose-derived lipogenesis and FA uptake directly correlates with cisplatin resistance.

a A linear regression of glucose-d₇ intensity to IC₅₀s of cisplatin in various OC cell lines. b A linear regression of PA-d₃₁ intensity to IC₅₀s of cisplatin in various OC cell lines. c A linear regression of the ratio of PA-d₃₁/(PA-d₃₁ + Glucose-d₇) to IC₅₀s of cisplatin in various OC cell lines. n = 6 technical replicates for (a–c). d Representative bright field images, raw SRS images, and processed SRS images of ODYA and glucose-d₇ in OVCAR5 and -cisR cells. e A linear regression of the metabolic index, as defined by the ratio of C ≡ C/(C ≡ C + C-D) to IC₅₀s of cisplatin in various OC cell line pairs (COV362 (n = 8), PEO (n = 6 and 7) and OVCAR5 (n = 4)). n represents technical replicates. R² = 0.9235. Data is shown as mean ± SEM. f Representative bright field images, raw SRS images, and processed SRS images of ODYA and glucose-d₇ in primary OC cells from cisplatin treatment-resistant or sensitive patient. g Quantitative analysis of metabolic index (the ratio of C ≡ C/(C ≡ C + C-D) for primary OC cells from cisplatin treatment-resistant or sensitive patient. Each data point represents the average metabolic index of individual cancer cells from a patient and its error bar indicates the SEM; n = 30, 31, 19, 25, 27, 33, 12, 11, 24, 20 and 30 cells. The box plot indicates the analysis for each group (sensitive (n = 7 biological replicates) vs. resistant (n = 4 biological replicates)). The bound of outer box, inner box, lines, whiskers, circles represent SEM, mean, medium, 25% to 75% of data, maxima and minima, respectively. Statistical significance used two-sided Student’s t test; P = 0.011. *P < 0.05. Scale bar: 20 µm. Source data are provided in the Source Data file.

Fig. 4. FA uptake directly contributes to cisplatin resistance.

a SRS image of OVCAR5-cisR cell cultured with control serum (FBS), delipid serum or control serum supplemented with 1% lipid mixture for 24 h. b Quantitative C-H signal from lipid droplet for a. The outer box, inner box, lines, whiskers and circles indicates 25% to 75% of data, mean, medium, SD, maxima and minima respectively. n = 15 technical replicates. P = 0.0043 and 0.0072. c Dose-response to cisplatin under culture environment with control, reduced (medium containing delipid serum) and increased (control serum supplemented with 1% lipid mixture) lipid content for OVCAR5-cisR cells. d, e Relative mRNA expression level of FABP5 (d) and FABP(PM) (e) in OVCAR5 and -cisR cells. n = 4. P = 0.0037 and 0.0018. f Relative mRNA expression level of FABP5 and FABP(PM) in OVCAR5 cells treated with cisplatin for 0, 6, 12 or 24 h. n = 3. P = 0.00016, 2.4 × 10⁻⁶, 0.00037, 0.0069, 0.00037 and 0.052. For the mRNA expression levels measurement (d–f), the results are shown as means + SD; n represents biological replicates. g Representative bright field and SRS images of SKOV3-cisR cell after BMS treatment at 10 μM for 24 h during concomitant incubation with 100 μM PA-d₃₁ for 6 h. h Quantification of C-D SRS signal intensity for (g). n = 9 and 8 technical replicates. P = 0.0026. i–k Dose-response to cisplatin with or without supplemental BMS treatment for PEO4 (i), SKOV3-cisR (j) and OVCAR5-cisR (k) cell. The results in all the does-response curves (c and i–k) are shown as means ± SD; n = 3 biological replicates. Data in all the bar charts (d–f and h) are shown as means + SD. All statistical significance was analyzed using two-sided Student’s t test. **P < 0.01, and ***P < 0.001. All scale bar: 20 µm. Source data are provided in Source Data file.

Fig. 5. FA uptake contributes to cisplatin resistance by increasing FAO.

a Oxygen consumption curves of OVCAR5-cisR and OVCAR5 over 3 h. n = 4 biological replicates. b, c Oxygen consumption curves of OVCAR5-cisR (b) and OVCAR5 (c) with 40 μM etomoxir treatment over 3 h. n = 4 biological replicates. d Seahorse measured OCR profile of OVCAR5 and OVCAR5-cisR cells with or without etomoxir treatment, followed by injections of mitochondrial respiration inhibitors oligomycin, FCCP, rotenone and antimycin A indicated by arrows. n = 3 technical replicates. e Quantified etomoxir induced basal respiration, ATP production and maximal respiration reduction in OVCAR5 and OVCAR5-cisR cell. Data are presented as means + SD; n = 3 technical replicates; two-sided Student’s t test; P = 0.0021, 0.018 and 0.0046; *P < 0.05 and **P < 0.01. f–h Dose-response to etomoxir for cisplatin resistant cell lines and their parental cell lines including PEO1 and PEO4 (f), OVCAR5 and -cisR (g) and COV362 and -cisR (h). i–k Dose-response to cisplatin with or without supplemental etomoxir treatment at 40 μM for PEO4 (i), OVCAR5-cisR (j) and COV362 -cisR (k) cell. l Dose-response to cisplatin for OVCAR5-cisR shCtrl and shCPT1a cell. n = 3 biological replicates for does-response curves (f–l). The data in all curves chart (a–d and f–l) are shown as means ± SD. m Total tumor volume growth curve from day 14 to 37 after tumor cell inoculation for vehicle (n = 3), carboplatin (n = 3), etomoxir (n = 4) and combinational (n = 6) treatment groups. n Mice body weight record since tumor inoculation for vehicle (n = 3), carboplatin (n = 3), etomoxir (n = 4) and combinational (n = 6) treatment groups. The data for PDX in vivo experiment (m and n) are shown as means ± SEM; n represents the number of animals. Source data are provided in Source Data file.

Fig. 6. Increased FA uptake and oxidation supports cancer cell survival under cisplatin-induced oxidative stress.

a Representative bright field and fluorescent images of OVCAR5 (n = 55) and OVCAR5-cisR (n = 49) treated with DCFDA cellular ROS assay kit. b Quantification of fluorescent signal intensity for (a). P = 4.6 × 10⁻²⁹. c Quantification of DCF fluorescent signal intensity of OVCAR5 and -cisR with cisplatin treatment at 1.6 μM or 3.3 μM for 24 h. n = 2. P = 0.0064, 0.040 and 0.016. d, e Quantified NADPH/NADP ratio of PEO1 and PEO4 (d), and OVCAR5 and -cisR (e). n = 3. P = 2.4 × 10⁻⁵ and 0.048. f Representative bright field and fluorescent images of OVCAR5 and -cisR with 100 μM glucose analog 2-NBDG treatment for 2 h after incubation with 3.3 μM cisplatin for 24 h. g Quantified fluorescent signal intensity for (f). n = 13, 16, 16 and 17. P = 0.019 and 8.2 × 10⁻⁷. h, i Quantified ATP/ADP ratio of cisplatin-resistant cell lines and their parental cell lines involving OVCAR5 (n = 3) and -cisR (n = 2) (h), and PEO1 (n = 5) and PEO4 (n = 6) (i). P = 0.0071 and 0.039. j Quantified ATP/ADP ratio of OVCAR5 and -cisR treated with 3.3 μM cisplatin with or without supplement of 100 μM palmitic acid for 6 h. n = 3. P = 0.046 (k) Proposed mechanism about cisplatin effect on cellular metabolism and cell proliferation. All n in fluorescent measurement (a–c and f–g) represents technical replicates, n in the assay measurement (d, e, h–j) represents biological replicates. All Scale bar: 30 µm. For box plots (b and g), the bound of outer box indicates 25% to 75% of data; inner box indicates mean; lines represent medium; whiskers indicate SD; circles indicate maxima and minima of data. Data in all the bar charts (c–e and h–j) are shown as means + SD. Statistical significance was analyzed using one-sided Student’s t test. *P < 0.05, **P < 0.01, and ***P < 0.001. Source data are provided in Source Data file.

Fig. 7. Cisplatin induced FA uptake is a universal metabolic feature in multiple types of cancers.

a Representative bright field and SRS images of MIA PaCa2 cells treated with 6.6 μM cisplatin for 24 h followed by 100 μM PA-d₃₁ or OA-d₃₄ incubation for 6 h. b Quantitation of C-D signal in MIA PaCa-2 cells treated with or without cisplatin by mean intensity. n = 7 for PA-d₃₁ and n = 8 for OA-d₃₄. P = 0.00029 and 0.026. c Representative bright field and SRS images of A549 cells treated with 13.2 μM cisplatin for 48 h followed by 100 μM PA-d₃₁ or OA-d₃₄ incubation for 6 h. d Quantitation of C-D signal in A549 cells treated with or without cisplatin by mean intensity. n = 8. P = 0.0031 and 0.016. e Representative bright field and SRS images of MDA-MB-231 cells treated with 6.6 μM cisplatin for 24 h followed by 100 μM PA-d₃₁ or OA-d₃₄ incubation for 6 h. f Quantitation of C-D signal in MDA-MB-231 cells treated with or without cisplatin by mean intensity. n = 6. All n represents technical replicates. P = 0.0024. Data in all bar charts (b, d, f) are shown as means + SD; All statistical significance was analyzed using one-sided Student’s t test. *P < 0.05. **P < 0.01. ***P < 0.001. Scale bar: 20 µm. Source data are provided in Source Data file.