Background: The extracellular signal-regulated kinase (ERK) pathway regulates cell growth, and is hyper-activated and associated with drug resistance in hepatocellular carcinoma (HCC). Metabolic pathways are profoundly dysregulated in HCC. Whether an altered metabolic state is linked to activated ERK pathway and drug response in HCC is unaddressed. Methods: We deprived HCC cells of glutamine to induce metabolic alterations and performed various assays, including metabolomics (with 13 C-glucose isotope tracing), microarray analysis, and cell proliferation assays. Glutamine-deprived cells were also treated with kinase inhibitors (e.g. Sorafenib, Erlotinib, U0126 amongst other MEK inhibitors). We performed bioinformatics analysis and stratification of HCC tumour microarrays to determine upregulated ERK gene signatures in patients. Findings: In a subset of HCC cells, the withdrawal of glutamine triggers a severe metabolic alteration and ERK phosphorylation (pERK). This is accompanied by resistance to the anti-proliferative effect of kinase inhibitors, despite pERK inhibition. High intracellular serine is a consistent feature of an altered metabolic state and contributes to pERK induction and the kinase inhibitor resistance. Blocking the ERK pathway facilitates cell proliferation by reprogramming metabolism, notably enhancing aerobic glycolysis. We have identified 24 highly expressed ERK gene signatures that their combined expression strongly indicates a dysregulated metabolic gene network in human HCC tissues. Interpretation: A severely compromised metabolism lead to ERK pathway induction, and primes some HCC cells to pro-survival phenotypes upon ERK pathway blockade. Our findings offer novel insights for understanding, predicting and overcoming drug resistance in liver cancer patients.

The pancreatic ductal adenocarcinoma microenvironment is composed of a variety of cell types and marked by extensive fibrosis and inflammation. Tumor-associated macrophages (TAMs) are abundant, and they are important mediators of disease progression and invasion. TAMs are polarized in situ to a tumor promoting and immunosuppressive phenotype via cytokine signaling and metabolic crosstalk from malignant epithelial cells and other components of the tumor microenvironment. However, the specific distinguishing features and functions of TAMs remain poorly defined. Here, we generated tumor-educated macrophages (TEMs) in vitro and performed detailed, multiomic characterization (i.e., transcriptomics, proteomics, metabolomics). Our results reveal unique genetic and metabolic signatures of TEMs, the veracity of which were queried against our in-house single-cell RNA sequencing dataset of human pancreatic tumors. This analysis identified expression of novel, metabolic TEM markers in human pancreatic TAMs, including ARG1, ACLY, and TXNIP. We then utilized our TEM model system to study the role of mutant Kras signaling in cancer cells on TEM polarization. This revealed an important role for granulocyte–macrophage colony-stimulating factor (GM-CSF) and lactate on TEM polarization, molecules released from cancer cells in a mutant Kras-dependent manner. Lastly, we demonstrate that GM-CSF dysregulates TEM gene expression and metabolism through PI3K–AKT pathway signaling. Collectively, our results define new markers and programs to classify pancreatic TAMs, how these are engaged by cancer cells, and the precise signaling pathways mediating polarization.

Cysteine is a crucial amino acid for glutathione biosynthesis, which is necessary for the maintenance of cellular antioxidant defense. Recent studies have shown that cysteine deprivation induces ferroptotic cell death in cancer cells, indicating an exploitable metabolic vulnerability for cancer therapy. However, the molecular variables that underpin cysteine dependency or can be selectively targeted to mimic cysteine deprivation are largely unknown. Here, we have studied cysteine dependency using over 20 pancreatic cancer cells (~35% of which are patient-derived), by measuring cell viability upon cystine withdrawal. In addition, we measured metabolite signatures by mass spectrometry, assessed genomic profile by bioinformatic analysis and qPCR, and also analyzed protein-level expression of cystine-glutamate antiporter SLC7A11 (xCT)—a known transporter for cystine uptake. We found that pancreatic cancer cells cluster at least into three groups, namely, cysteine-independent, moderately dependent, and highly dependent cells. The highly dependent group consistently showed low intracellular cysteine, methionine, and glutamine and a high level of intermediary metabolites of glycolytic pathway. This is accompanied by a downregulation of genes involved in apoptosis, lysosomal function, and the upregulation of cell cycle gene network as revealed by altered pathway enrichment. Further, while the cysteine-independent cells express high gene and protein levels of SLC7A11, which was further upregulated upon cysteine deprivation, the highly dependent cells expressed a very low level of SLC7A11 regardless of cysteine availability. Importantly, by analyzing the cysteine-dependent cell lines, we identified several metabolic genes that are upregulated in human pancreatic cancer and predict patient overall survival (e.g., HIF1A, LDHA, SLC16A1). In conclusion, we have uncovered differential cysteine dependency and the accompanying hallmarks in pancreatic cancer cells, thus providing impetus for targeting cysteine metabolism in a specific subset of pancreatic cancer. Citation Format: Zeribe C. Nwosu, Matthew H. Ward, Peter Sajjakulnukit, Li Zhang, Marina Pasca di Magliano, Costas A. Lyssiotis. A roadmap for targeting cysteine dependency in a subset of pancreatic cancer [abstract]. In: Proceedings of the AACR Special Conference on Pancreatic Cancer: Advances in Science and Clinical Care; 2019 Sept 6-9; Boston, MA. Philadelphia (PA): AACR; Cancer Res 2019;79(24 Suppl):Abstract nr C36.

Pancreatic ductal adenocarcinoma (PDA) is a lethal disease characterized by high invasiveness, therapeutic resistance, and metabolic aberrations. Although altered metabolism drives PDA growth and survival, the complete spectrum of metabolites used as nutrients by PDA remains largely unknown. Here, we aimed to determine novel nutrients utilized by PDA. We assessed how >175 metabolites impacted metabolic activity in 19 PDA cell lines under nutrient-restricted conditions. This analysis identified uridine as a novel metabolite driver of PDA survival in glucose-deprived conditions. Uridine utilization strongly correlated with expression of the enzyme uridine phosphorylase 1 (UPP1). Metabolomics profiling, notably 13C-stable isotope tracing, revealed that uridine-derived ribose is the relevant component supporting redox balance, survival, and proliferation in glucose-deprived PDA cells. We demonstrate that UPP1 catabolizes uridine, shunting its ribose component into central carbon metabolism to support glycolysis, the tricarboxylic acid (TCA) cycle and nucleotide biosynthesis. Compared to non-tumoral tissues, we show that PDA tumors express high UPP1, which correlated with poor overall survival in multiple patient cohorts. Further, uridine is enriched in the pancreatic tumor microenvironment, and we demonstrate that this may be provided in part by tumor associated macrophages. Finally, we found that inhibition of UPP1 restricted the ability of PDA cells to use uridine, and that UPP1 knockout impairs tumor growth in vivo. Our data identifies uridine catabolism as a critical aspect of compensatory metabolism in nutrient-deprived PDA cells, suggesting a novel metabolic axis for PDA therapy.

The pancreatic ductal adenocarcinoma (PDA) microenvironment is composed of a variety of cell types and marked by extensive fibrosis and inflammation. Tumor-associated macrophages (TAM) are abundant, and they are important mediators of disease progression and invasion. TAMs are polarized in situ to a tumor promoting and immunosuppressive phenotype via cytokine signaling and metabolic crosstalk from malignant epithelial cells and other components of the tumor microenvironment (TME). However, the specific distinguishing features and functions of TAMs remain poorly defined. Here, we generated tumor-educated macrophages (TEM) in vitro and performed detailed, multi-omic characterization (i.e. transcriptomics, proteomics, metabolomics). Our results reveal unique genetic and metabolic signatures of TEMs, the veracity of which were queried against our in-house single cell RNA sequencing (scRNA-seq) dataset of human pancreatic tumors. This analysis identified expression of novel, metabolic TEM markers in human pancreatic TAMs, including ARG1, ACLY, and TXNIP. We then utilized our TEM model system to study the role of mutant Kras signaling in cancer cells on TEM polarization. This revealed an important role for GM-CSF and lactate on TEM polarization, molecules released from cancer cells in a mutant Kras-dependent manner. Lastly, we demonstrate that GM-CSF dysregulates TEM gene expression and metabolism through PI3K-AKT pathway signaling. Collectively, our results define new markers and programs to classify pancreatic TAMs, how these are engaged by cancer cells, and the precise signaling pathways mediating polarization.