p53 remodels the lipidome of pancreatic cancer cells
To explore the role of p53 in remodelling the lipid profile of pancreatic cancer, we carried out a global untargeted lipidomic profiling in a cell line (KPshp53) derived from mouse PDAC, with the pancreas-specific expression of KRAS (LSL-KRASG12D) and a doxycycline-regulated short hairpin (sh)RNA targeting wild-type p53 expression. Lipidomic results indicate that loss of p53 extensively remodels the lipidome of pancreatic cancer cells (shp53, p < 0.05) (Fig. 1a, b). By clustering lipid species in biological classes, we observed that sphinganine, phosphatidylglycerol (PG), lysophosphatidylserine (LPS), lysophosphatidylcholine (LPC), lysophosphatidylethanolammine (LPE), lysophosphatidylinositol (LPI) and lysophosphatidylglycerol (LPG) were the most significantly altered classes, displaying a dramatic reduction of their abundancies upon p53 depletion (Fig. 1c, d and Additional file 1: Fig. S1). Conversely, other lipid classes including glucosylceramides, sphingosines, diacylglycerols, triglycerides, ceramides, hexosyl-ceramides, palmitate, and phospholipids were generally not comprehensively affected by p53 loss, despite displaying decreases of specific species (Additional file 1: Fig. S2a). Lysophospholipids are a class of lipids exerting signalling roles in a cell-autonomous and non-cell-autonomous manner [27], and can function also as signalling molecules in the microenvironment. Hence, we conducted a parallel global untargeted lipidomic profiling of conditioned media from KPshp53 cells. Consistently, we observed a massive reduction of lysophospholipid species (LPS, LPC, LPE) in the extracellular environment of doxycycline treated KPshp53 cells (p53 silenced) (Fig. 1e and Additional file 1: Fig. S2b). Hence, overall, these data clearly indicate that p53 has an essential role in controlling the lipidome of pancreatic cancer cells and, in particular, it can exert an important regulation of intracellular and extracellular signalling lysophospholipids.
p53 regulates production and secretion of lysophospholipidome
To better define the changes mediated by p53 on the lysophospholipidome, we next conducted a detailed analysis comparing the species displaying a differential abundance following p53 deficiency in the intracellular compartment and in the conditioned media. The most significantly altered LPC, showing consistent reductions, were the 14:0, 15:0, 17:0, 17:1 and 18:0e species (Fig. 2a,b). Our analysis however revealed a larger general cohort of intracellular and extracellular LPC species (Additional file 1: Figs. S1 and S3a), indicating overall the LPC among the mostly affected lysophospholipids. LPC is in general the most abundant class of lysophospholipids in plasma and body fluids and despite a clear dissection of the role of these molecules has not been conducted they appear to participate in cytotoxicity, haemostatis and inflammation [27]. Nanomolar concentrations of LPC can exert chemotactic roles for monocytes and macrophages, while saturated and monosaturated LPC can facilitate production of inflammatory redox oxygen species [28, 29]. Thus, p53 loss dependent reduction of LPC might underlie an immune evasion effect in cancer.
Similar pattern was observed in LPE and LPS classes, where saturated and monosaturated LPE 16:1e, 18:0, 18:1e and LPS18:0 were strongly reduced in the intracellular compartment and in the conditioned media of p53 depleted cells (Fig. 2c–f and Additional file 1: Fig. S4a, b). LPE is the second most abundant lysophospholipids in plasma. LPE can induce an increase of intracellular Ca2+ concentration producing proliferative and motilities activities in breast and ovarian cancer cell lines [30]. Remarkably, LPE was also shown to stimulate chemotactic migration and cellular invasion in ovarian cancer cells, indicating non-cell-autonomous properties [31]. LPS is within the less abundant lysophospholipids in the plasma, but it can display potent immunomodulatory activities. G-coupled receptors belonging to the P2Y purineceptor clusters have been identified as LPS receptors and their activation was associated to suppression of T cell and mast cell degranulation [32, 33]. Finally, specific reduction in the intracellular content of LPI and LPG was also detected, indicating a general alterations of lysophospholipids production (Additional file 1: Figure S4c, d).
Overall, these data indicate the p53-dependent lipidome significantly impinge on production and secretion of lysophospholipids. This might represent an unexpected, novel level of regulation exerted by p53 on the tumour microenvironment and immunity.
p53 transcriptionally regulates phospholipases
In the last decade cancer genomic sequencing studies have experienced a massive growth, resulting into the formation of datasets containing huge amount of open-access data [34]. By performing a deep analysis of publicly available datasets of pancreatic cancer, we aimed to identify putative phospholipases (PLs) that might account for the observed p53-dependent lysophospholipidome. Through this approach, we selected three putative phospholipases on the basis of the potential regulation by p53 and their clinical relevance for pancreatic cancer. These were the phospholipase C delta 4 (PLCD4), phospholipase C beta 4 (PLCB4) and phospholipase D 3 (PLD3). Interestingly, the expression of the three phospholipases was strongly affected in p53 deficient cells (Fig. 3a), suggesting the existence of molecular axis downstream of p53 function. We next asked whether PLCD4, PLCB4 and PLD3 might be directly regulated by p53 via a transcriptional control. To address our hypothesis, we looked for binding enrichment of p53 on PLCD4, PLCB4 and PLD3 genomic loci, querying available ChIP-seq datasets. Notably, we identified several peaks for p53 in the genomic regions of the three phospholipases, which strongly suggests a direct involvement of p53 in their transcription regulation. Furthermore, we observed that p53 peaks broadly overlapped with regions of enhanced chromatin accessibility (ATAC), which were also enriched for permissive histone modifications such as trimethylation of lysine 4 of histone 3 (H3K4me3), acetylation of histone 4 (H4ac) and acetylation of lysine 9 of histone 3 (H3K9ac) (Fig. 3b). Thus, these data indicate a direct molecular axis p53/PLs, that might underlie a transcriptional reprogramming mediated by p53 for the regulation of enzymes involved in lipid metabolism.
PLCD4, PLCB4 and PLD3 correlates with p53 status and prognosis of pancreatic cancer patients
By analysing PanCancer genomic data, we then asked whether PLCD4, PLCB4 and PLD3 levels correlate to the pathogenesis of human pancreatic ductal carcinoma. To address this, we performed a bioinformatic analysis of available datasets of human pancreatic adenocarcinoma (PAAC). Interestingly, with the help of Gene Expression Profiling Interactive Analysis (GEPIA), we observed that the expression levels of the three phospholipases underwent a decrease through the different stages of PAAC (Fig. 4a). Next, we sought to determine the relationship of the selected phospholipases with p53 mutational status, which is highly mutated in pancreatic cancer. We therefore analysed the expression levels of PLCD4, PLCB4 and PLD3 in a cohort of 184 patients belonging to the TCGA PanCancer Atlas dataset. Interestingly, PLCD4, PLCB4 and PLD3 mRNA levels correlate with p53 status in PDAC patients (Fig. 4b). These data strongly indicate a biologically relevant p53/PLs axis in the pathogenesis of PDAC. To further extend our study, we next focused on the prognostic significance of this molecular markers. By stratifying the patients’ cohort according to the mRNA expression of PLCD4, PLCB4 and PLD3 (Low, High), we computed a Kaplan–Meier survival analysis. The results indicate that higher expression of the PLs represented a good prognostic factor (Fig. 4c). Thus, the p53/PLs axis displays clinical significance for PDAC pathogenesis and integrated with our lipidomic analysis suggests that p53 mediates a transcriptional programme that influence synthesis and secretion of signalling lysophospholipids. These data can therefore indicate the lysophospholipids as novel mediators of cell-autonomous and non-cell-autonomous tumour suppressive function of p53.