Construction of a lipid metabolism-based prognostic gene signature in cervical squamous cell carcinoma and validation of LIPG’s oncogenic role

Figures

• 

  Fig. 1. Flowchart of the study

• 

  Fig. 2. Differentially expressed lipid metabolism-related genes (LMGs) in CSCC. A. Principal components analysis (PCA) of CSCC tissues (n = 253) and normal tissues (n = 13) based on LMGs. B. Volcano plot of the differentially expressed LMGs (DLMGs). C. Venn Diagram of differentially expressed genes (DEGs) and LMGs (dataset 1: batch removed with “limma” package, dataset 2: batch removed with “SVA” package, dataset 3: without batch removed). D. Gene ontology (GO) enrichment analysis of the DLMGs (“BP” refers to “biological process”, “CC” refers to “cellular component”, “MF” refers to “molecular function”). E. Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis of the DLMGs

• 

  Fig. 3. Construction of the Lipid Metabolism Prognostic Signature (LMPS). A. Forest plot of 9 prognostic differentially expressed LMGs (DLMGs). B. C-index and average C-index of prognostic models constructed by 24 machine learning methods of training set (TCGA cohort) and validation set (GEO cohort). C. The least absolute shrinkage and selection operator (Lasso) coefficient profiles of 9 prognostic DLMGs. D. Tenfold cross-validation for Lasso model parameter adjustment using minimum criteria. E, F. Distribution of survival status and duration among patients categorized into the high- and low- LMPS groups

• 

  Fig. 4. Prognostic prediction value of the LMPS in CSCC patient survival. A. The Kaplan–Meier (KM) curves based on the high- and low- LMPS groups in the TCGA cohort. B. Receiver operating characteristic (ROC) curves for overall survival (OS) at 1-year, 3-year, and 5-year intervals in the TCGA cohort. C. The KM curves based on the high- and low- LMPS groups in the HTMCP cohort. D. Time-independent ROC curves of the LMPS and 15 metabolism-associated prognostic signatures reviewed from the literature in the TCGA cohort. E-L. The KM curves of the high- and low- LMPS groups stratified by age (old vs. young; E-H) and FIGO stage (I-II vs. III-IV; I-L) in the TCGA and HTMCP cohorts

• 

  Fig. 5. Association of the LMPS with immune microenvironment and immunotherapy. A. Heatmap showing the normalized infiltration levels of immune and stromal cells estimated by the xCell algorithm. B. The correlation between the LMPS, the 6 prognostic differentially expressed LMGs (DLMGs), and the cancer-immunity cycle, with significant correlations (P < 0.05) displayed. The values shown in the figure represent the correlation coefficients. C. The different expression levels of human leukocyte antigen (HLA) genes between the high- and low-LMPS groups. D. Top 15 mutated genes in the high- and low- LMPS groups. E-G. Correlations between the LMPS and immune score, stromal score, and ESTIMATE score calculated using the ESTIMATE algorithm. H. Comparison of immune scores between the high- and low-LMPS groups across three GEO cohorts. I. Different expression levels of immune checkpoint genes between the high- and low-LMPS groups. J-M. Different levels of CTL proportion (J), TIDE score (K), exclusion score (L), dysfunction score (M) between the high- and low-LMPS groups

• 

  Fig. 6. Association of the LMPS with drug sensitivity. A-G. Comparison of IC50 values of drugs recommended in cervical cancer treatment guidelines, including Paclitaxel (A), Cisplatin (B), Oxaliplatin (C), 5-Fluorouracil (D), Gemcitabine (E), Topotecan (F), and Irinotecan (G), between the high- and low-LMPS groups. H. Comparison of the IC50 value of BI-2536 between the high- and low-LMPS groups. I. Correlation between the LMPS and the IC50 value of BI-2536. J-L. CCK8 viability assay of SW756 (J), Siha (K), Siha-NC vs. Siha-OE (L) cells treated with BI-2536

• 

  Fig. 7. Selection of the key gene LIPG and its expression in CSCC. A. Ranking of the importance of six genes in the LMPS using the Random Forest methods. B. Expression levels of LIPG and PLA2G3 in cervical cancer (n = 253) and normal cervix tissues (n = 13). C. Kaplan–Meier (KM) curves based on expression levels of PLA2G3 and LIPG. D. Expression levels of LIPG in cervical cancer (n = 13) and normal cervix tissues (n = 5), and comparison between patients with FIGO (2018) stage I-II (n = 9) and stage III (n = 4). E. Endogenous LIPG protein levels in Siha, SW756 and C33a cell lines. F. mRNA and protein levels of LIPG in Siha cells overexpressing LIPG. G. mRNA and protein levels of LIPG in SW756 cells with LIPG knockdown

• 

  Fig. 8. LIPG affects lipid metabolism of CSCC cells. A-D. The total cholesterol (TCHO) and triglycerides (TG) levels in cervical cancer cells with LIPG overexpression or knockdown. E. Score scatter plot of OPLS-DA model comparing Siha-NC and Siha-OE cells. F. Heatmap of hierarchical clustering analysis of the top 10 differential metabolites between Siha-NC and Siha-OE groups. G. Enrichment analysis of differential metabolites

• 

  Fig. 9. The pro-tumor effects of LIPG in both in vitro and in vivo experiments. A. CCK8 viability assay of Siha-NC and Siha-OE cells. B. CCK8 viability assay of SW756-NC, SW756-Sh1 and SW756-Sh2 cells. C. Colony formation assay of Siha-NC and Siha-OE cells. D. Colony formation assay of SW756-NC, SW756-Sh1 and SW756-Sh2 cells. E. Transwell cell migration ability assays of Siha-NC and Siha-OE cells. F. Transwell cell migration ability assays of SW756-NC, SW756-Sh1 and SW756-Sh2 cells. G. Wound healing ability assays of Siha-NC and Siha-OE cells. H. Wound healing ability assays of SW756-NC, SW756-Sh1 and SW756-Sh2 cells. I. Apoptosis assays of Siha-NC and Siha-OE cells. J. Apoptosis assays of SW756-NC, SW756-Sh1 and SW756-Sh2 cells. K. Photographs of nude mice injected subcutaneously with Siha-NC and Siha-OE cells. L. Photographs of excised subcutaneous tumors. M. Tumor growth curves of subcutaneous tumor in nude mice

• 

  Fig. 10. LIPG promotes CSCC progression through the MAPK-p38 pathway. A. Gene Ontology (GO) enrichment analysis of the LIPG gene, displaying the top 5 enriched pathways. GSEA enrichment plot of the pathway “Positive regulation of MAPK pathway”. B-D. Western blot analysis of p-JNK/JNK (B), p-Erk/Erk (C), and p-p38/p38 (D) levels in LIPG-overexpression and LIPG-knockdown cells. E. Western blot analysis showing the effect of p38 inhibition on p-p38/p38 levels in LIPG-overexpression cells. F. Cell proliferation of Siha-NC, Siha-OE and Siha-OE cells treated with p38 inhibitor measured with CCK8 assay. G. Transwell cell migration ability assays of Siha-NC, Siha-OE and Siha-OE cells treated with p38 inhibitor. H. Wound healing ability assays of Siha-NC, Siha-OE and Siha-OE cells treated with p38 inhibitor. I. Colony formation assay of Siha-NC, Siha-OE and Siha-OE cells treated with p38 inhibitor. J-K. Lysophosphatidylcholine (LPC) levels in the supernatant of SW756-NC, SW756-Sh1/Sh2 cells (J) and Siha-NC, Siha-OE cells (K) at 24, 48 and 72 hours after cell attachment. L-M. Western blot analysis of p-p38/p38 levels in Siha cells (L) and SW756 cells (M) treated with LPC