1 Introduction
Retroperitoneal liposarcoma (RLPS) is a type of malignant sarcoma of adipocytic differentiation that originates from mesenchymal progenitor cells (MPCs) [
1]. RLPS accounts for more than 40% of all retroperitoneal sarcomas and 15%–20% of all sarcomas [
2]. RLPS is characterized by a high rate of local recurrence after surgery but relatively few distant metastases. RLPS is poorly responsive to chemotherapy and radiotherapy. Surgery is still the main approach for both primary and recurrent RLPS treatment [
3], and therefore there is an urgent demand for effective therapeutic strategies to treat RLPS. Determining the pathogenesis of RLPS, and identifying therapeutic targets are necessary to improve the long-term survival rate and quality of life of patients with RLPS.
Histologically, retroperitoneal well-differentiated liposarcoma (RWDLPS) closely resembles mature adipose tissue with fibrous septa and variable nuclear atypia, while retroperitoneal dedifferentiated liposarcoma (RDDLPS) is more akin to an undifferentiated or spindle cell sarcoma. RDDLPS and RWDLPS are characterized by giant marker chromosomes that originate from the 12q13–15 region, which contains highly amplified genes [
4].
MDM2 and
CDK4 are two of the most well-studied genes. Recent progress in the last decade using targeted sequencing, RNA sequencing (RNA-seq), whole-exome sequencing (WES) or whole-genome sequencing (WGS) identified
MDM2,
CDK4,
HMGA2,
TSPAN31,
CPM,
YEATS4,
FRS2 and
PTPRB to be amplified in RDDLPS and RWDLPS [
5]. With the completion of the WES and WGS projects of liposarcoma, we have made tremendous progress in our understanding of the genetic basis driving the development of RWDLPS and RDDLPS. However, the proteomics, metabolomics and lipidomics of RWDLPS/RDDLPS have remained unclear. Increasing evidence suggests that metabolites and lipids are important for tumorigenesis and tumor development [
6,
7]. Therefore, the metabolic changes between RLPS and adjacent adipose tissue and a combined multi-omics joint analysis may help provide an in-depth understanding of liposarcoma. Therefore, the exploration of proteomics, metabolomics, and lipidomics can help us understand the molecules affecting RWDLPS and RDDLPS, such as key proteins, metabolites, and lipids that mediate progression, drug resistance, differentiation, and metastasis of sarcoma.
In this work, we present proteomic, metabolomic, and lipidomic analyses of RDDLPS and RWDLPS and of adjacent adipose tissue. Based on the results, our study not only reveals the metabolic profile of RLPS, but also provides an important resource for data mining and guidance for clinical validation and basic research. These results will help find more promising metabolic therapeutic potentials for patients with RLPS.
2 Materials and methods
2.1 Study population
This study aimed to reveal the metabolic profile of RLPS through multi-omics joint analysis. We collected matched pairs of frozen and fixed normal and tumor samples from 51 patients with DDLPS or WDLPS. All the samples were collected from Xiang’an Hospital of Xiamen University and Peking University International Hospital. Fixed samples were transferred to Shanghai Outdo Biotech Co., Ltd., for tissue chip assays. The frozen samples were prepared for proteomic, metabolic, and lipidomic analysis. The present protocols were reviewed and approved by the Ethics Committees of all participating institutions, including the Xiang’an Hospital of Xiamen University (No. XAHLL2021024) and Peking University International Hospital (No. WA2020RW29). All participants were enrolled and anonymized after approval by the institutional review board. We obtained written informed consent from all participants, except for those we could not contact due to lack of follow-up. In these cases, the institutional review boards at each participating institution granted permission for existing tissue samples to be used for research purposes.
2.2 TMT (tandem mass tag) based proteomic analysis
Tissue samples were ground using liquid nitrogen into a cell powder and then transferred to a 5 mL centrifuge tube. After that, four volumes of lysis buffer (8 mol/L urea, 1% protease inhibitor cocktail) was added to the cell powder, followed by sonication three times on ice using a high intensity ultrasonic processor (Scientz). Of note, for the PTM experiments, inhibitors were also added to the lysis buffer: 3 μmol/L TSA and 50 mmol/L NAM for acetylation. The remaining debris was removed by centrifugation at 12 000 g at 4 °C for 10 min. Finally, the supernatant was collected and the protein concentration was determined with a BCA kit according to the manufacturer’s instructions. For digestion, the protein solution was reduced with 5 mmol/L dithiothreitol for 30 min at 56 °C and alkylated with 11 mmol/L iodoacetamide for 15 min at room temperature in the dark. The protein sample was then diluted by adding 100 mmol/L TEAB to urea to a concentration of less than 2 mol/L. Finally, trypsin was added at a 1:50 trypsin-to-protein mass ratio for the first digestion overnight, and at a 1:100 trypsin-to-protein mass ratio was used for a second 4 h-digestion. After trypsin digestion, the peptides were desalted using a Strata X C18 SPE column (Phenomenex) and vacuum dried. The peptides were reconstituted in 0.5 mol/L TEAB and processed according to the manufacturer’s protocol for the TMT and iTRAQ kit. Briefly, one unit of TMT and iTRAQ reagents was thawed and reconstituted in acetonitrile. The peptide mixtures were then incubated for 2 h at room temperature and pooled, desalted, and dried by vacuum centrifugation. The tryptic peptides were fractionated by high pH reverse-phase HPLC using a Thermo Betasil C18 column (5 μm particles, 10 mm ID, 250 mm length). Briefly, peptides were first separated with a gradient of 8%−32% acetonitrile (pH 9.0) over 60 min into 60 fractions. Then, the peptides were combined into six fractions and dried by vacuum centrifugation. The tryptic peptides were dissolved in 0.1% formic acid (solvent A), and directly loaded onto a homemade reversed-phase analytical column (15-cm length, 75 μm i.d.). The gradient comprised increasing concentrations of solvent B (0.1% formic acid in 98% acetonitrile) from 6% to 23% over 26 min, 23% to 35% over 8 min, increased to 80% over 3 min, and then held at 80% for the last 3 min, all at a constant flow rate of 400 nL/min on an EASY-nLC 1000 UPLC system. The peptides were subjected to a NSI source followed by tandem mass spectrometry (MS/MS) in Q ExactiveTM Plus (Thermo Fisher) coupled with a UPLC system. The electrospray voltage applied was 2.0 kV. The m/z scan range was 350–1800 for full scan, and intact peptides were detected in the Orbitrap at a resolution of 70 000. Peptides were then selected for MS/MS using a NCE setting of 28, and the fragments were detected in the Orbitrap at a resolution of 17 500. A data-dependent procedure that alternated between one MS scan followed by 20 MS/MS scans with a 15.0 s dynamic exclusion was used. Automatic gain control (AGC) was set at 5E4. The fixed first mass was set as 100 m/z. The resulting MS/MS data were processed using the Maxquant search engine (v.1.5.2.8). Tandem mass spectra were searched against the human Uniprot database concatenated with reverse decoy database. Trypsin/P was specified as the cleavage enzyme, allowing up to four missing cleavages. The mass tolerance for precursor ions was set as 20 ppm in the first search and 5 ppm in the main search, and the mass tolerance for fragment ions was set as 0.02 Da. Carbamidomethyl on Cys was a specified as fixed modification, and acetylation and oxidation of Met were specified as variable modifications. The FDR was adjusted to < 1%, and the minimum score for modified peptides was > 40. The Kyoto Encyclopedia of Genes and Genomes (KEGG) database was used to analyze and annotate protein pathways based on the MS results.
2.3 LC-MS/MS for targeted metabolomics
The LC-MS system was performed using an Exion LC system (AB SCIEX) with a ZIC-pHILIC column (100 mm × 2.1 mm, Millipore) connected to a QTRAP-5500 mass spectrometer (AB SCIEX). For LC conditions, 2 µL samples were injected and analyzed. The flow rate was 0.2 mL/min. The column and tray temperatures were set at 40 °C and 4 °C, respectively. Mobile phase A contained 15 mmol/L ammonium acetate, and 3 mL/L ammonium hydrate (> 28%) in water, and mobile phase B was a 90% acetonitrile aqueous solution. The gradient elution was set as follows: 95% B maintained for 1 min, decreased to 45% in 14 min and held for 2 min, and increased to 95% in 0.5 min and held for 4.5 min. The ESI voltage was set to 4500 V in negative ion MRM mode and 5500 V in positive MRM mode, respectively. The ion temperature and curtain gas were set at 500 °C and 35 µL/min, respectively. The LC-MS/MS conditions were controlled by Analyst 1.7.1 software, and the final data were processed by Mutiquant 3.0.3 software.
2.4 LC-MS/MS for lipidomics
The supernatant was injected into a Thermo AccucoreTMC30 (2.6 μm, 2.1 mm × 100 mm) column. The column temperature, flow rate, and injection volume were set to 45 °C, 0.35 mL/min, and 2 μL, respectively. The mobile phase consisted of acetonitrile: water (60:40, v/v) containing 0.1% formic acid and 10 mmol/L ammonium formate (A) and acetonitrile: isopropanol (10:90, v/v) containing 0.1% formic acid and 10 mmol/L ammonium formate (B). The gradient program initiated from A:B (80:20, v/v) at 0 min, A:B (70:30, v/v) at 2.0 min, A:B (40:60, v/v) at 4 min, A:B (15:85, v/v) at 9 min, A:B (10:90, v/v) at 14 min, A:B (5:95, v/v) at 15.5 min, held for 1.8 min, and returned to A:B (80:20, v/v) at 20 min. Mass spectra were acquired on a SCIEX triple quadrupole-linear ion trap mass spectrometer (QTRAP) 6500 + LC-MS/MS system, equipped with an ESI turbo ion spray interface, operating in positive and negative ion modes and controlled by Analyst 1.6.3 software (SCIEX). The ESI source temperature and ion spray voltage were set at 500 °C and 5500 V (positive) and −4500 V (negative), respectively. The ion source gas I, gas II, and curtain gas were set at 45, 55, and 35 psi, respectively. Metabolomic and lipidomic data were imported into the online software MetaboAnalyst for multivariate statistical analysis. The number of biological replicates in each experiment is listed in the Figure legends. Statistical analyses of LC-MS data are described in the figure legends. Except where indicated, statistically significant differences were assessed by a two-tailed paired t-test.
2.5 Immunohistochemical staining
Tissue sections were dewaxed in xylene and then hydrated in graded alcohol solutions and distilled water. Immunohistochemical staining was performed using antibodies against a proto-oncogene commonly expressed in RLPS (MDM2), the key protein in fat digestion and absorption (CD36), the key protein in pentose phosphate pathway (PPP, G6PD), and the key protein in lactic acid metabolism (LDHA), according to the manufacturers’ recommendations. Sepia staining was considered positive staining. A digital tissue biopsy scanner (3DHISTECH, Hungary) was used to scan the tissue microarray, and Aipathwell (Service bio, China), a digital pathological image analysis software was used to automate the analysis of the above collected images. The evaluation items included: positive cell ratio = number of positive cells/total number of cells; H-SCORE = ∑ (pi × i) = (percentage of weak intensity × 1) + (percentage of moderate intensity × 2) + (percentage of strong intensity × 3); and IRS = SI (positive intensity) × PP (positive cell ratio). SI was divided into three grades: grade 0, no positive staining; grade 1, light yellow weak positive; grade 2, brown yellow medium positive; grade 3, tan strong positive. PP was divided into four levels: 0, 0%–5%; 1, 6%–25%; 2, 26%–50%; 3, 51%–75%; 4, > 75%.
2.6 Cell lines and regents
SW872 (HTB-92) cell line was purchased from ATCC. LPS510 (IM-H539) cell line was purchased from Xiamen Immocell Biotechnology Co. Ltd. Cells were cultured in DMEM (IMC-201, Xiamen Immocell Biotechnology Co. Ltd.) containing penicillin and streptomycin (IMC-601, Xiamen Immocell Biotechnology Co. Ltd.), and 10% fetal bovine serum (FBS, P30-3302, PAN). Cisplatin (HY-17394), doxorubicin (HY-15142A), gemcitabine (HY-17026), RG7112 (HY-10959), abemaciclib (HY-16297A), 2-deoxy-D-glucose (HY-13966), and RRX-001 (HY-16438) were purchased from MedChemExpress LLC(MCE). Antibodies were purchased from the following sources: CD36 (10752-RP02, SinoBiological, China), LDHA (GB11342, Servicebio, China), G6PD (GB111797, Servicebio, China), MDM2 (ab16895, Abcam, UK).
2.7 Cell counting kit-8 assay (CCK-8 assay) and cell viability assay
The proliferation of cells was detected using a cell counting kit-8 (RM02823, Abclonal, Wuhan, China) assay according to the manufacturer’s instructions. The cells (1 × 103 cells/well) were seeded in a 96-well plate with 100 µL of DMEM supplemented with 10% FBS. After 48 h of incubation, the CCK-8 reagent (10 µL) was added to each well and continuously cultured for 1 h at 5% CO2. The absorbance rate at 450 nm was measured using a microplate reader (TECAN, F50). All experiments were repeated three times. Cell viability was determined by the CCK8 assay, liposarcoma cells (5000 cells per well) were seeded in a 96-well plate and cultured with or without drug treated medium for a set time. The CCK8 kit was used according to the manufacturer’s instructions.
2.8 Statistical analysis
All data were collected from more than three independent experiments and expressed as the mean ± standard error of the mean. All statistical significance was determined by t-tests. Graphing was performed in GraphPad Prism 8 software (GraphPad Software, San Diego, CA, USA). The data are presented as means ± SEM. A P value < 0.05 was considered statistically significant and is labeled with one asterisk (*), a P value < 0.01 is summarized with two asterisks (**), and a P value < 0.001 is summarized with three asterisks (***).
3 Results
3.1 TMT-based quantitative proteomics of clinical RWDLPS and RDDLPS tissue and adjacent adipose tissue
RLPS is associated with a high rate of recurrence and mortality. To date, there is no effective treatment strategy beyond surgery. To delineate the proteomic, metabolic, and lipidomic signatures of RLPS, we enrolled RLPS patients in this study to harvest postoperative tumor tissue and adjacent adipose tissue to perform multi-omics joint analysis (Fig.1). There was a total of 35 RDDLPS and 16 RWDLPS patients in the cohort. The clinical features of the sample sets are displayed in Tab.1.
To comprehensively explore the proteomic profiles of RWDLPS and RDDLPS tissue and adjacent adipose tissue, five paired samples from RWDLPS patients and five paired samples from RDDLPS patients were used to perform a TMT-based quantitative proteome analysis. In total, we identified 4996 proteins from 57 092 peptides, 4533 of which were quantitatively analyzed. The mapping of partial least squares-discriminant analysis (PLS-DA) showed a clear separation between liposarcoma tissue and adipose tissue (Fig.2). The relative quantitative value of each sample was taken as log
2 and then the
P value was calculated by the two-sample two-tailed
t-test method. When a
P value < 0.05 and a change in differential expression greater than 1.5 was considered the threshold for significant upregulation, while less than 1/1.5 was considered the threshold for significant downregulation. Among them, 475 proteins were upregulated and 354 proteins were downregulated (Fig.2, Table S1). As shown in Fig. S1, the top 15 upregulated proteins in both RDDLPS and RWDLPS were HMGA2, SEC14L4, FCER1G, PIP4K2C, SASH1, APOC1, MARCKS, APOA2, FKBP10, POSTN, IGLV4-69, ARHGEF18, IGLV2-23, CNTRL, and IGKV3D-20. The top 15 downregulated proteins in both RDDLPS and RWDLPS were TRARG1, GPR4, S100B, CAV1, WISP2, PLIN4, GPD1, LTF, CALB2, MYO3A, CD36, CAVIN1, NQO1, CRYAB, and S100A1. It is worth noting that HMGA2 (S/
N = 5.486, ***
P < 0.001), CDK4 (S/
N = 1.744, ***
P < 0.001), and TSPAN31 (S/
N = 1.992, ***
P < 0.001) (Table S1) that were identified in our study were consistent with the results summarized by Lu
et al. in a genomic study [
5].
Next, the upregulated and downregulated proteins separately as the inputs and searched against databases, including gene ontology (GO) biological processes, KEGG pathways, and molecular function. As shown in the bubble diagram of enrichment and the distribution of differentially expressed proteins in the KEGG pathway, the upregulated proteins associated with tumorigenesis were significantly enriched in DNA replication, protein export, primary immunodeficiency, transcriptional deregulation in cancer, intestinal immune network for IgA production, ribosome biogenesis in eukaryotes, cell cycle, and the complement and coagulation cascades (Fig.2). Downregulated proteins were enriched in bioquinone and other terpenoid-quinone biosynthesis, ECM-receptor interaction, proximal tubule bicarbonate reclamation, bile secretion, PPAR signaling pathway, focal adhesion and multiple metabolic pathways (bioquinone and other terpenoid-quinone biosynthesis, nitrogen metabolism, glyoxylate and dicarboxylate metabolism, phenylalanine metabolism, regulation of lipolysis in adipocytes, glycine, serine, and threonine metabolism, fatty acid biosynthesis, pyruvate metabolism, tryptophan metabolism, fatty acid degradation, glycerolipid metabolism, glycolysis and gluconeogenesis) (Fig.2). To further understand the strength of the change of the differentially expressed proteins, we divided them into four groups according to their differential expression multiples (S/N), designated Q1 to Q4 (Q1: < 0.500; Q2: 0.500–0.667; Q3: 1.5–2.0; Q4: > 2.0). Based on Fisher’s exact test, using the P value obtained from enrichment analysis, the relevant functions in different groups were clustered together by a hierarchical clustering method and drawn as a heatmap. The horizontal axis of the heat map represents enrichment test results of different groups, and the vertical axis represents the description of differentially expressed enrichment-related functions (GO, KEGG pathway). The color blocks correspond to the differentially expressed proteins in different groups and functional descriptions indicate the intensity of enrichment. Red represents strong enrichment, while blue represents weak enrichment (Fig.2–2G). Consistent with the above results, we found downregulation of identified metabolism-related events in almost all molecular functions, biological processes, and KEGG enrichments except for DNA metabolic process in biological process. In addition, in terms of biological processes, the most significantly upregulated processes were chromosome separation and DNA metabolic process (Fig.2). In molecular function enrichment, the most significantly upregulated functions were DNA-dependent ATPase activity, DNA topoisomerase II activity, and DNA topoisomerase activity (Fig.2). All these results suggested that DNA synthesis was active in liposarcoma.
3.2 Characterization of the metabolome and lipidome of clinical RWDLPS and RDDLPS tissue and adjacent adipose tissue
Based on the above proteomic results, we found that there were significant differences in multiple metabolic processes and signal pathways, including DNA metabolic processes between liposarcoma tissue and paired adjacent adipose tissue. To further verify and expand the proteomic studies, we then performed a targeted metabolomic and lipidomic analysis of RLPS tissue and paired adjacent adipose tissue. Using the LC-MS workflow, we identified 127 metabolites from all 51 paired samples (Table S2). PLS-DA of all the samples showed a significant difference between the liposarcoma tissue and paired adipose tissues (Fig.3). The differential metabolites included amino acids, organic acids, fatty acids, carbohydrates, and other metabolites. As shown in the volcano plots and heatmaps, the liposarcoma tissue had altered expression of 67 metabolites by 1.4 folds (45 increased and 26 decreased metabolites) compared with the adipose tissue (Fig.3, Fig. S2). We imported the altered metabolites between the RWDLPS and RDDLPS tissue and the adipose tissue into the KEGG database for metabolic pathway enrichment analysis, which showed 10 significantly upregulated pathways, including glycerophospholipid metabolism; pyrimidine metabolism; amino sugar and nucleotide sugar metabolism; beta-alanine metabolism; purine metabolism; glycolysis and gluconeogenesis; alanine, aspartate, and glutamate metabolism; arginine biosynthesis; histidine metabolism; and neomycin, kanamycin, and gentamicin biosynthesis. The 10 significantly downregulated pathways included aminoacyl-tRNA biosynthesis; glycine, serine, and threonine metabolism; arginine biosynthesis; purine metabolism; glutathione metabolism; glyoxylate and dicarboxylate metabolism; phenylalanine, tyrosine, and tryptophan biosynthesis; arginine and proline metabolism; valine, leucine, and isoleucine biosynthesis; and ubiquinone and other terpenoid-quinone biosynthesis (Fig.3 and 3D). The downregulated pathways, as marked by red text in Fig.3, were consistent with the proteomic results. Interestingly, among the top 15 upregulated metabolites, butyryl carnnitine, isobutyryl-carnitine, hexanoyl-carnitine, and lauroylcarnitine suggested disturbed fatty acid oxidation whereas among the top 15 downregulated metabolites, such as serine, methionine, creatine, and creatinine indicated that hypomethylation was a characteristic of RLPS (Fig. S3).
Moreover, we identified 593 lipids from all 50 paired samples using the LC/MS workflow (Fig. S4, Table S3). PLS-DA of all samples illustrated a significant difference between the RWDLPS and RDDLPS tissue and the paired adipose tissue (Fig.4). A total of 593 lipid molecules were identified with a 1.4-fold change. Among them, 366 lipid molecules were upregulated and 277 lipid molecules were downregulated (Fig.4). The upregulated lipids mainly include diacylglycerophosphocholines, diacylglycerophosphoethanolamines, lysophosphatidylcholines, fatty acyl carnitines, monacylglycerophosphoethanolamines, glycerophosphoinositols, 1-(1z-alkenyl)-2-acylglycerophosphoethanolamines, ceramide phosphocholines, fatty acid esters, lysophosphatidyl ethanolamine. The downregulated lipids mainly included triacylglycerols, diacylglycerols, ubiquinones, quinone and hydroquinone lipids, and diradylglycerols (Fig.4 and 4D). We imported the altered lipids data between the RWDLPS and RDDLPS tissue and adipose tissue into the KEGG database for lipid pathway enrichment analysis. The result showed significant biosynthesis of unsaturated fatty acids, primary bile acid biosynthesis, sphingolipid metabolism, arachidonic acid metabolism, taurine and hypotaurine metabolism, and fatty acid degradation and significantly downregulated pathways including phosphatidylinositol phosphate metabolism, mitochondrial electron, transport chain, citric acid cycle and the Warburg effect (Fig.4 and 4F). Taken together, metabolomics and lipidomics data suggest that the upregulated lipogenesis and nucleotide biosynthesis whereas the reduced energy metabolism linking to amino acid metabolism and mitochondrial activity were found in both RWDLPS and RDDLPS.
3.3 Multi-omics analysis of clinical RWDLPS and RDDLPS tissue and adjacent adipose tissue associated with metabolism
To explore the features of metabolism of RDDLPS and RWDLPS by multi-mics analysis, we identified clusters of orthologous (COG) groups of proteins to determine the functional classification of the differential proteins, as shown in Table S4. Next, the proteins associated with metabolism, the metabolites identified by targeted metabolomics, and the lipids identified by lipidomics were classified according to metabolic pathways (Tab.2). Among the top 15 upregulated proteins, we found PIP4K2C associated with inositol phosphate metabolism, APOC1 associated with cholesterol metabolism, and APOA2 associated with cholesterol metabolism. Among the top 15 downregulated proteins, we found PLIN1 associated with regulation of lipolysis in adipocytes, CD36 associated with fat digestion and absorption, NOQ1 associated with ubiquinone and other terpenoid-quinone biosynthesis, and GPD1 associated with glycerophospholipid metabolism.
Based on the malignant proliferation and high recurrence rate of RLPS, we next focused on energy and cell cycle related metabolic process. As can be seen from the network diagram (Fig.5), glycolysis, purine, and pyrimidine metabolic pathways were enhanced in RLPS tissue compared with adjacent adipose tissue, however, the TCA cycle, arginine biosynthesis, and glycine, serine, and threonine metabolism decreased. To be specific, in terms of metabolites, glucose-6p and fructose-6p in glycolysis were upregulated; malate, citrate and fumaric acid in the TCA cycle were downregulated. Of great interest, GMP, xanthine, AMP, guanosine, uracil, uridine, CMP, and cytidine in purine and pyrimidine metabolism were upregulated. In terms of proteins, ADP-dependent glucokinase (ADPGK) in glycolysis was upregulated, pyruvate carboxylase (PC), malate dehydrogenase (MDH1), citrate synthase (CS), isocitrate dehydrogenase (IDH1), cytoplasmic aconitate hydratase (ACO1), 2-oxoglutarate dehydrogenase (OGDH), and succinate dehydrogenase (SDHA) in the TCA cycle were downregulated and PPAT in purine metabolism was upregulated. It was suggested that liposarcoma tissue requires glucose for glycolysis and the metabolites in glycolysis tend to flow to pyrimidine, purine and lactate metabolic pathways rather than the TCA cycle. In addition, aspartate and N-acetylaspartate in aspartate metabolism were upregulated; citrulline, arginine, urea and ornithine in arginine biosynthesis were downregulated; and glycine, serine, and threonine in the glycine, serine and threonine metabolic pathway were downregulated. Simultaneously, the proteins ASS1 and ARG1 were also downregulated. It was suggested that aspartate tends to flow to pyrimidine metabolic pathways rather than the TCA cycle, arginine biosynthesis, or glycine, serine, or threonine metabolism. TAGs and MAGs were downregulated, with PUFAs, MUFAs, SFAs and phospholipids upregulated, but DAGs were partially downregulated and partially upregulated (Fig.5). Furthermore, proteins related to fatty acid digestion and absorption, fatty acid degradation and fatty acid elongation were downregulated, such as FABP, CD36, ACSL1, FASN, DGAT1, ATGL, LIPE and MGLL (Tab.2). It was suggested that PUFAs, MUFAs and SFAs were not used for TAGs and lipid droplet synthesis, but instead for phospholipid synthesis.
In general, our results revealed that glycolysis, purine metabolism, pyrimidine metabolism and phospholipid formation were upregulated in both RDDLPS and RWDLPS tissue compared with the adjacent adipose tissue, whereas the TCA cycle, lipid absorption and synthesis, fatty acid degradation and biosynthesis. These metabolic reprogramming may be required to meet nucleotide, biological membrane, and energy during the malignant proliferation of RLPS.
3.4 Multi-omics analysis of WDLPS and DDLPS tissue
RDDLPS has a higher degree of malignancy and poorer prognosis than RWDLPS. Clinically, there are different degrees of differentiation of the primary and recurrence, and even conversion to each other. Therefore, it is also very important to analyze the differences in proteins and metabolites between RWDLPS and RDDLPS. Based on the gene set associated with 85 metabolic pathways, we used the GSVA algorithm to score the relevant pathways in the samples. The results showed that arginine and proline metabolism, histidine metabolism, oxidative phosphorylation, n-glycan biosynthesis, the PPP, and lysine degradation were significantly upregulated in RDDLPS tissue compared with RWDLPS tissues (Fig.6). In terms of proteins, DOK2, MYH9, NAA10, THBS2, DNPH4, and FCGR1A had significantly higher expression, and CD36, PTN, IGLV2-23, GC, CAMKK2, A1BG, HMGA2. APOC1, APOA2, F12, HRG, IGF2, C1RL, GPC6, TF, SERPINA1, ISLR, and KNG1 had significantly lower expression in RDDLPS compared with RWDLPS (Fig.6). Metabolite and lipidomic analysis showed glycerophosphocholines, glycerophosphoethanolamines, fatty esters, glycerophosphoserines, sphingomyelins, and sterol esters were upregulated in RDDLPS compared with RWDLPS (Fig.6), while glycerolipids and triacylglycerols were downregulated (Fig.6). The top five most significantly downregulated lipids were all triacylglycerols. It is worth noting that among the significantly upregulated lipids, ursocholic acid, taurolithocholic acid-3-sulfate, taurochenodeoxycholic acid, cholesterol, and taurocholic acid are all sterols. Coenzyme Q8, Q9, and Q10 were significantly downregulated (Fig.6).
3.5 Inhibiting glycolysis and the PPP enhance antitumor effects of MDM2 and CDK4 inhibitors
To verify the above results, we used immunohistochemistry (IHC) to identify the key proteins of relevant pathways in tissue microarray, including the RDDLPS and RWDLPS marker protein (MDM2), the key protein in fat digestion and absorption (CD36), the key protein in lactate metabolism (LDHA), and the key protein in the PPP (G6PD). IHC assays showed that MDM2, LDHA, and G6PD were upregulated in RLPS tissue samples compared with normal adipose tissue, but CD36 was downregulated (Fig.7–7D)
Similar to previous studies, we found that glycolysis and the PPP were enhanced in liposarcoma tissue compare with adjacent adipose tissue. Furthermore, we explored the effects of glycolysis and PPP inhibitors on tumor cells. RRX-001 (inhibitor of the PPP), 2-deoxy-D-glucose (competitive inhibitor of glycolysis), cisplatin (commonly used chemotherapy drug), doxorubicin and gemcitabine (first-line drugs of liposarcoma), RG7112 (small molecule inhibitor targeting MDM2), and abemaciclib (small molecule inhibitor targeting CDK4) were selected to test for antitumor effects in DDLPS cell lines (SW872, LPS510). According to clinical trial data, we choose a lower concentration for the test. As the results show, both 2-deoxy-D-glucose and RRX-001 significantly promoted the antitumor effects of RG7112 and abemaciclib (Fig.7 and 7F).
4 Discussion
RLPS usually reach a large size with high proliferation and recurrence, appearing to form an anoxic environment, but still with a strong ability to proliferate. To meet the demands of the malignant proliferation of tumors, large amounts of energy, biomass and nucleotides are required. Glucose can generate pyruvate and NADH by glycolysis, moreover, pyruvate is able to generate lactic acid, which can be released outside the cell. Pyruvate is consumed by mitochondria via phosphonic acid oxidation. The metabolites in glycolysis can be used for nucleotide synthesis, NADPH synthesis and hexosamine biosynthesis through the PPP [
8–
10]. Warburg found an abnormal increase in glycolysis in tumor cells, known as the “Warburg effect.” While earlier reports found abnormally increasing glycolysis along with impaired oxidative phosphorylation, a growing number of studies have shown that genetic defects associated with oxidative phosphorylation increase glycolysis, but increased glycolysis does not mean that oxidative phosphorylation is disrupted [
11,
12].
According to our results, glycolysis, purine metabolism and pyrimidine metabolism were enhanced while oxidative phosphorylation was weakened in liposarcoma compared with adjacent adipose tissue. RLPS tend to be large size that are more likely to form anoxic environments. Under anaerobic conditions, glucose is not completely degraded, but is partially degraded to pyruvate and then converted to lactic acid. This process produces 2 mol of ATP per 1 mol of glucose, which is much lower than oxidative phosphorylation. Unlike normal cells, tumor cells prefer glycolysis to provide a rapid supply of ATP and intermediates because of their rapid proliferation [
13]. In addition, metabolic intermediates of glycolysis also play an important role in cell anabolism and resistance to reactive oxygen species (ROS). In the case of glucose starvation, the intermediates used for the PPP pathway and carbon metabolism reduced, resulting in insufficient production of NADPH and a nucleotide. Thus, the levels of ROS increase. The proliferation of tumor cells will decrease and the tumor cells become be more sensitive to cell death induced by ROS [
14]. The PPP, which is associated with NADPH production, purine metabolism, and pyrimidine metabolism, was enhanced in RLPS. As mentioned above, ROS in tumor cells is always at high levels. If glycolysis and PPP can be targeted and blocked, this may increase tumor cell death or sensitivity to chemotherapy drugs.
We also found significant upregulation of lactic acid (Fig.5) in liposarcoma tissue. Both tumor and stromal cells co-exist in the microenvironment. Tumor cells can adapt or remodel the microenvironment to promote malignant proliferation and metastasis [
15]. An increasing number of studies have shown that blocking communication between the microenvironment and the tumor inhibited tumor progression, providing new strategies for tumor therapy [
16]. Lactic acid used to be considered metabolic waste, but recent studies have found that tumors produce a large amount of lactic acid through glycolysis, which is very important for tumor malignancy. For example, lactic acid can be used as a direct nutrient for deep tumor tissue with high glucose consumption [
17,
18]. Lactic acid-mediated histone lactate modifications activated macrophage differentiation into M2, which promoted tumor progression. Lactic acid also caused Treg cells to act as an immunosuppressor [
19]. Tumors have an advantage in nutrient competition by reprogramming their metabolism. According to the different metabolic characteristics of tumor cells and normal cells, we can kill tumor cells by targeting metabolic pathways.
Lipid metabolism is greatly altered in cancer, such as enhanced synthesis and uptake of lipids to meet the needs of rapid cancer growth and metastasis. Lipids not only constitute the structural basis of membranes, but also produce energy and participate in signal transduction. Lipidomic analysis showed that fatty acids, phospholipids, and sterols were significantly higher in liposarcoma tissue than in adjacent adipose tissue, but triacylglycerols, glycerolipids, and diacylglycerols were significantly lower. Moreover, fatty acids, phospholipids, and sterols were significantly higher in RDDLPS than in RWDLPS, but triacylglycerols, glycerolipids, and diacylglycerols were significantly lower. Fatty acids (FAs) oxidation produces acetyl-CoA, which enter into the TCA cycle to produce isocitrate and malate along with the production of NADPH. Tumor cells can reduce fatty acid synthesis to reduce NADPH consumption [
20]. PUFAs are essential and must be provided endogenously to cells. We found that PUFAs, MUFAs, and SFAs were significantly higher in RLPS tissue compared with adipose tissue, such as arachidonic acid (C20:4). They are not used in the synthesis of triradylglycerols, glycerolipids, and diradylglycerols, but more likely are used in the synthesis of phospholipids and sterols. Phospholipids are the most prominent type of membrane molecule and contain serine, ethanolamine, choline, glycerol, or inositol, and glycolipids are another type of membrane lipid that contains glucose or galactose [
21]. Sterols are another major type of membrane lipid [
22]. Multiple studies have shown that blocking cholesterol synthesis is harmful to cancers because HMG-CoA reductase is inhibited in the mevalonate pathway by statins [
23,
24].
Two main types of fatty acid transporters, CD36 and FABPs are significantly lower in liposarcoma than in adipose cells. IHC results showed that CD36 exhibited strong expression in liposarcoma tissue, although lower than in adipose tissue. Given the homology of liposarcoma and adipocytes, blocking the absorption of FAs through CD36 and inducing synthesis of triacylglycerols, glycerolipids, and diacylglycerols may induce DDLPS to transform into WDLPS or even adipocytes which is a therapeutic direction for liposarcoma. Meantime, High amounts of PUFA-containing membrane lipids sensitize cancer cells to ROS and ferroptosis [
25]. Both enzymes prefer PUFAs as a substrate and sensitize cancer cells to ferroptosis by increasing the proportion of PUFA-containing phospholipids [
26,
27]. The use of more PUFAs in our diet, combined with the induction of more ROS, may be beneficial for liposarcoma treatment.
Current available therapeutic drugs for liposarcoma include doxorubicin, trabectedin, eribulin, gemcitabine, and dacarbazine. There are also several emerging systemic therapies including tyrosine kinase inhibitors, CDK4 inhibitors, MDM2 inhibitors, and mTOR inhibitors. There remains an unmet need for effective treatments for advanced and metastatic liposarcoma [
28]. 2-deoxy-D-glucose is a glucose analog that acts as a competitive inhibitor of glucose metabolism, inhibiting glycolysis via its actions on hexokinase [
29]. A number of tumor-related clinical trials of 2-deoxy-D-glucose have been conducted, mainly in the application of [
18F]-fluoro-2-deoxy-D-glucose in Positron Emission Tomography Computed Tomography [
30]. RRx-001, a hypoxia-selective epigenetic agent and studied as a radio- and chemosensitizer, triggered apoptosis and overcome drug resistance in myeloma. RRx-001 is a potent inhibitor of G6PD and showed potent antimalarial activity [
31]. Therefore, we first wondered whether blocking glycolysis and PPP would promote the antitumor effect of the above drugs. As our results showed, 2-deoxy-D-glucose and RRX-001 significantly promoted the antitumor effects of MDM2 and CDK4 inhibitors. These findings suggest that more combinations based on glycolysis and the PPP are worth exploring to achieve optimal therapeutic outcomes.
With the deepening of research and the development of technology, the understanding of tumor metabolism has been extended from a simple single line to a complex metabolic network since the discovery of the “Warburg effect”. In the present work, we conducted a multi-omics analysis to compare the molecular alterations in RDDLPS and RWDLPS and adjacent adipose tissue. The data sets provide a rich resource of information and knowledge of liposarcomas and a basis for us to analyze drug therapy and drug resistance for RLPS.
5 Conclusions
Our study not only described the proteomic profiles of RDDLPS/RWDLPS, but also depicted the major metabolic features of RLPS. The data sets provide a rich resource of information and knowledge of RLPS, suggesting far-reaching implications or clinical treatment in the future.
PDF
(10507KB)
