Functional Genomics In Vivo Reveal Metabolic Dependencies of Pancreatic Cancer Cells

Abstract

Pancreatic ductal adenocarcinoma (PDAC) cells require substantial metabolic rewiring to overcome nutrient limitations and immune surveillance. However, the metabolic pathways necessary for pancreatic tumor growth in vivo are poorly understood. To address this, we performed metabolism-focused CRISPR screens in PDAC cells grown in culture or engrafted in immunocompetent mice. While most metabolic gene essentialities are unexpectedly similar under these conditions, a small fraction of metabolic genes are differentially required for tumor progression. Among these, loss of heme synthesis reduces tumor growth due to a limiting role of heme in vivo, an effect independent of tissue origin or immune system. Our screens also identify autophagy as a metabolic requirement for pancreatic tumor immune evasion. Mechanistically, autophagy protects cancer cells from CD8+ T cell killing through TNFα-induced cell death in vitro. Altogether, this resource provides metabolic dependencies arising from microenvironmental limitations and the immune system, nominating potential anti-cancer targets.

Keywords: cancer metabolism; in vivo CRISPR screen; pancreatic cancer; tumor immune evasion.

Figure 1.. Metabolism-focused CRISPR screens in vivo reveal metabolic dependencies of pancreatic tumors

A. Schematic of genetics screens to identify metabolic dependencies of KP pancreatic cancer specifically in vivo. B. Cumulative frequency curve of represented guides in genetic screens. C. Gene scores of in vivo versus in vitro genetic screens of KP pancreatic cancer growth. D. Volcano plot of differential gene scores comparing in vivo against in vitro conditions (left). Top 20 genes scoring as differentially required in vivo. Gene involved in specific metabolic pathways are indicated (right). E. Gene sets enriched in differentially required genes in vivo versus in vitro for pancreatic cancer growth. The heatmap generated by iPAGE represents the extent to which each gene set is enriched among the genes that are essential for tumor growth in vivo. See also Figure S1.

Figure 2.. The tissue of origin partly dictates metabolic essentialities in Kras-driven cancers

A. Schematic of focused genetic screens to identify common and different essential metabolic genes for KP pancreatic and KP lung tumor growth in vivo. B. Top 40 genes scoring as differentially required in vivo in pancreatic tumors aligned to their differential gene scores in KP lung tumors. Genes involved in purine or heme synthesis are indicated. Bars are median differential gene scores with interquartile range. Dots are individual differential guide scores. C. Gene scores of in vivo KP pancreas tumor growth versus KP lung tumor growth in C57BL/6J mice. Genes involved in purine or heme synthesis are indicated. D. Guide scores of the indicated genes in the focused in vivo screens from KP pancreas and KP lung tumors.

Figure 3.. Heme synthesis is a metabolic dependency of Kras-driven tumors in vivo

A. Immunoblot of HMBS in the indicated KP pancreas and KP lung cancer cell lines. GAPDH was used as loading control. B. Fold change in cell number (log₂) of the indicated KP pancreas and KP lung cancer cell lines after culturing in vitro for the indicated durations (mean ± SD, n=3). ***p < 0.001 versus sgControl. C. Tumor weights of the indicated KP pancreas and KP lung tumors engrafted subcutaneously in C57BL/6J mice (box and whisker, n=8). *p < 0.05, ***p < 0.001 versus sgControl (left). Images of the indicated KP pancreas and KP lung tumors (right). D. Immunoblot of HMOX1 in KP pancreas and KP lung cancer cells grown in vitro under normoxia, hypoxia (0.5% oxygen) for 48 hrs and in subcutaneous tumors. GAPDH was used as loading control. E. Immunoblots of HMOX1 and HMBS in the indicated KP pancreas cell lines. GAPDH was used as loading control. F. Relative tumor weights of the indicated KP pancreas Hmbs_KO tumors engrafted subcutaneously in C57BL/6J mice (box and whisker, n=23). *p < 0.05 versus Control (top). Representative image of the indicated KP pancreas Hmbs_KO tumors (bottom). G. Schematic of competition assay using PDAC patient derived xenograft cells infected with the indicated sgRNAs. Cells were then engrafted subcutaneously in NSG mice (left). Relative fold change in sgRNA abundance (log2) from the PDX (mean ± SD, n=5). ***p < 0.001 versus sgControl (right). H. Disease-free survival rates of TCGA PDAC patients with high or low heme synthesis gene expressions. Weighted average expressions of CPOX, HMBS, PPOX and UROS was used (low heme n=83, high heme n=28). See also Figure S2.

Figure 4.. Autophagy is an immune-dependent metabolic liability and enables immune evasion in PDAC

A. Schematic of focused genetic screens to identify immune-dependent metabolic liabilities of KP pancreatic tumor growth in vivo. B. Gene scores of in vivo KP pancreas tumor growth in immunodeficient NSG mice versus immunocompetent C57BL/6J mice. C. Top 5 genes scoring as differentially required for KP pancreas tumor growth in immunocompetent mice compared to immunodeficient mice. Genes involved in TAP complex are indicated in blue. The autophagy gene Atg7 is indicated in red. D. Immunoblots of ATG7 and LC3B in the indicated KP pancreas cell lines. GAPDH was used as loading control. E. Tumor weights of the indicated KP pancreas Atg7_KO tumors engrafted subcutaneously in the indicated mice (box and whisker, n=8). **p < 0.01, ***p < 0.001 versus Atg7 addback (left). Image of the indicated KP pancreas Atg7_KO tumors from the indicated mice (right). F. Immunohistochemical staining of cleaved-CASPASE-3 in the indicated KP pancreas Atg7_KO tumors engrafted subcutaneously in C57BL/6J mice. Representative images are shown. Scale bar, 230 μm. G. Percentage of IFNγ+ activated CD8+ T cells (left) or and NK cells (right) extracted from the indicated KP pancreas Atg7_KO tumors engrafted subcutaneously in C57BL/6J mice (mean ± SD, n=5). *p < 0.05, ***p < 0.001 versus Atg7 addback. H. Tumor weights of the indicated KP pancreas Atg7_KO tumors engrafted orthotopically in the pancreas of C57BL/6J mice (box and whisker, n=12). **p < 0.01 versus Atg7 addback (left). Representative image of the orthotopic KP pancreas Atg7_KO tumors from the indicated mice (right). I. Relative cell count of the indicated OVA-expressing KP pancreas Atg7_KO cell lines after co-culturing with activated OT-I CD8+ T cells for 48 hrs at the indicated E:T ratios (mean ± SD, n=3). Counts were normalized to the average of the monocultured cells of the same line. ***p < 0.001 versus Atg7 addback. See also Figure S3.

Figure 5.. Autophagy enables tumor immune evasion by increasing TNFα resistance

A. RNAseq analysis of KP pancreas Atg7_KO tumors. Gene sets enriched in transcriptome of KP pancreas Atg7_KO tumors compared to Atg7 addback tumors engrafted subcutaneously in C57BL/6J mice (n=3). Immune-related gene sets are boxed in red. B. Relative cell count of indicated KP pancreas Atg7_KO cell lines treated for 48 hrs with 100 ng/mL TNFα or 100 ng/mL IFNγ (mean ± SD, n=3). Counts were normalized to the average of the untreated cells of the same line. ***p < 0.001 versus Atg7 addback. C. Relative cell count of indicated human pancreatic cancer MIA PaCa-2, PANC-1 and PATU-8988T cell lines treated for 48 hrs with 100 ng/mL, 200 ng/mL and 600 ng/mL TNFα respectively (mean ± SD, n=3). Counts were normalized to the average of the untreated cells of the same line. **p < 0.01, ***p < 0.001 versus wild type. D. Immunoblot of CASPASE-8, cleaved CASPASE-8 and CASPASE-3 in the indicated KP pancreas Atg7_KO cell lines treated for 24 hrs with 100 ng/mL TNFα. GAPDH was used as loading control. E. Immunoblot of TNFRSF1A in the indicated KP pancreas Atg7_KO cell lines. GAPDH was used as loading control. F. Relative cell count of indicated KP pancreas Atg7_KO cell lines treated for 48 hrs with 100 ng/mL TNFα (mean ± SD, n=3). Counts were normalized to the average of the untreated cells of the same line. ***p < 0.001. G. Relative cell count of KP pancreas cells treated for 48 hrs with 100 ng/mL TNFα or 50nM Bafilomycin A1 (BafA) (mean ± SD, n=3). Counts were normalized to the average of the untreated cells or those treated with BafA. ***p < 0.001. H. Relative cell count of the indicated OVA-expressing KP pancreas Atg7_KO cell lines after co-culturing with activated OT-I CD8+ T cells for 48 hrs at the indicated E:T ratios with or without 40 μg/mL anti-TNFα (mean ± SD, n=3). Counts were normalized to the average of the untreated monocultured cells of the same line. ***p < 0.001 versus untreated cells. See also Figures S4 & S5.