GCN2 adapts protein synthesis to scavenging-dependent growth

Abstract

Pancreatic cancer cells with limited access to free amino acids can grow by scavenging extracellular protein. In a murine model of pancreatic cancer, we performed a genome-wide CRISPR screen for genes required for scavenging-dependent growth. The screen identified key mediators of macropinocytosis, peripheral lysosome positioning, endosome-lysosome fusion, lysosomal protein catabolism, and translational control. The top hit was GCN2, a kinase that suppresses translation initiation upon amino acid depletion. Using isotope tracers, we show that GCN2 is not required for protein scavenging. Instead, GCN2 prevents ribosome stalling but without slowing protein synthesis; cells still use all of the limiting amino acids as they emerge from lysosomes. GCN2 also adapts gene expression to the nutrient-poor environment, reorienting protein synthesis away from ribosomes and toward lysosomal hydrolases, such as cathepsin L. GCN2, cathepsin L, and the other genes identified in the screen are potential therapeutic targets in pancreatic cancer.

Keywords: Cathepsin L; GCN2; PDAC; lysosomes; macropinocytosis; protein scavenging; protein synthesis; translation.

Figure 1.. Genome-wide loss-of-function screen illuminates cellular machinery required for scavenging-driven growth

(A) Screen design. KRPC-A cells were transduced with a genome-scale lentiviral sgRNA library to generate a population of pooled knockouts. Each gene was targeted by at least 10 unique sgRNAs. Infected cells were switched to standard amino-acid-replete medium (DMEM), amino-acid-replete medium supplemented with 50 g/L cell-culture-grade bovine serum albumin, or leucine-free medium supplemented with 50 g/L albumin. After 12 population doublings, knockout frequencies in each condition were determined using high-throughput sequencing. (B) For each gene, a specific essentiality score was calculated. Genes with negative specific essentiality scores were more essential in the leucine-free medium than in the amino-acid-replete medium. To identify genes with statistically significant specific essentiality scores, we compared the distribution of scores for all genes with the distribution of scores for non-detected genes (as determined by RNA sequencing). 413 genes were specifically essential in the leucine-free medium at FDR < 0.1. (C–F) The top hits of the screen (the 100 genes with the highest specific essentiality scores) fell into two categories—protein scavenging machinery (C and D) and regulators of protein synthesis (E and F). Selected top hits are highlighted in scatter plots of specific essentiality scores from two independent replicates of the genome-wide screen (C and E) and depicted in schematics (D and F). (G–I) Gcn2 knockout cells (G), Vasp knockout cells (H), and Vps39 knockout cells (I) were transduced with retroviral vectors expressing either enhanced green fluorescent protein (EGFP) or the corresponding sgRNA-resistant human cDNA. These paired cell lines were cultured in either amino-acid-replete medium for 24 h or leucine-free medium for 48 h, and growth was measured by comparing the total cell volume at initial and final time points. All media were supplemented with 50 g/L albumin (BSA). Error bars represent 95% confidence intervals (n = 3). *p < 0.05, **p < 0.01, ***p < 0.001.

Figure 2.. GCN2 prevents ribosome stalling in cells growing on extracellular protein

(A) Gcn2 WT and Gcn2 KO cells were switched to either amino-acid-replete medium or leucine-free medium, each supplemented with 50 g/L bovine serum albumin (BSA). Total cell volumes were measured using packed cell volume tubes. Error bars show 95% confidence intervals. *p < 0.05, **p < 0.01, ***p < 0.001. (B) Gcn2 WT and Gcn2 KO cells were switched to either the amino-acid-replete medium2 or the leucine-free medium, each supplemented with 50 g/L BSA. After 2 h, cellular protein was extracted, and GCN2 abundance and signaling was analyzed by western blotting. Results are representative of 3 independent experiments. (C) Polysome profiles of Gcn2 WT and Gcn2 KO cells switched to either amino-acid-replete medium or leucine-free medium, each supplemented with 50 g/L BSA, for 1 h. Cell lysates were loaded onto 10%–50% sucrose gradients and spun at 35,000 g for 2.5 h at 4°C before measurement. Profiles were normalized such that the area under each curve was equal. (D) Ribosome profiling was performed in each of the cell lines and conditions in (C), and codon occupancies were calculated such that for each sample, thepercent occupancies for all 64 codons sum to 100%. Differences in occupancy between leucine-deprived cells and cells cultured in the amino-acid-rich medium are plotted. CUC and CUU are two of the six leucine codons.

Figure 3.. GCN2 is not required for protein scavenging

(A) Stable isotope tracer strategy to measure protein scavenging rate. Cells are grown for 5 population doublings with labeled essential amino acids to label cellular proteins. Unlabeled serum albumin is then added to the medium, and uptake and degradation of this extracellular protein yields unlabeled amino acid monomers. Because cytosolic amino acids are in rapid exchange with a much bigger pool of extracellular amino acids, most of these unlabeled amino acids end up in the medium, where they can be readily sampled and measured. Thus, the rate of secretion of unlabeled amino acids serves as a proxy for protein scavenging rate. (B) Unlabeled valine secretion over 24 h in Gcn2, Vasp, and Vps39 knockout cells expressing either EGFP or the corresponding sgRNA-resistant human cDNA. For each cell line, rates were normalized to the amino-acid-replete condition with re-expression. All media were supplemented with 50 g/L bovine serum albumin (BSA). See Figure S3 for phenylalanine and lysine secretion. Error bars represent 95% confidence intervals (n = 3). *p < 0.05, **p < 0.01, ***p < 0.001.

Figure 4.. In cells reliant on protein scavenging, protein synthesis rate is limited by scavenging, not GCN2 signaling

(A) Stable isotope tracer strategy to simultaneously measure the protein synthesis rate and protein scavenging rate. Cells are grown for 5 population doublings inmedium with ¹³C- and ¹⁵N-labeled essential amino acids (isotope flavor 1). After five population doublings, free amino acids and cellular proteins are almost completely labeled. Cells are then switched to medium containing a second, distinct set of amino acid tracers (isotope flavor 2). This medium is also supplemented with unlabeled serum albumin. As cells grow, new protein is marked by a combination of Flavor 2 and unlabeled amino acids (from protein scavenging), whereas old protein is marked by Flavor 1, enabling calculation of protein synthesis rate. Protein scavenging rate can be calculated by summing the unlabeled amino acids incorporated into cellular proteins and secreted into the medium. (B) Example unprocessed valine data from Gcn2 WT and Gcn2 KO cells cultured in the leucine-free medium supplemented with 50 g/L bovine serum albumin (BSA). Ion counts are converted to nmol by comparing with known standards. Complete unprocessed valine data underlying the rates in (C–E) are shown in the Figure S4. (C and D) (C) Protein synthesis rates and (D) protein scavenging rates of Gcn2 WT and Gcn2 KO cells cultured in amino-acid-replete and leucine-free media supplemented with the indicated amounts of serum albumin. (E) Direct comparison of protein synthesis rates and scavenging rates of Gcn2 WT and KO cells cultured in leucine-free media. Error bars represent 95% confidence intervals (n = 3). Error was calculated by averaging three independent rate estimates derived from three different amino acids (lysine, phenylalanine, and valine).

Figure 5.. GCN2 suppresses the synthesis of anabolic proteins in favor of catabolic proteins

(A) Stable isotope tracer strategy to measure the synthesis of individual proteins by proteomics. Cells are grown for 5 population doublings in medium with ¹³Clabeled lysine and arginine. After five population doublings, free amino acids and cellular proteins are almost completely labeled. Cells are then switched to medium containing unlabeled amino acids and supplemented with unlabeled serum albumin. As cells grow, new protein is marked by unlabeled lysine and arginine, whereas old protein is marked by labeled lysine and arginine, enabling independent measurement of each by quantitative proteomics. After 24 h, protein is extracted, digested, TMT-labeled, and analyzed by mass spectrometry (STAR Methods). (B) Unlabeled (newly synthesized) fractions of all measured proteins in Gcn2 WT and Gcn2 KO cells grown in amino-acid-replete or leucine-free medium supplemented with 50 g/L serum albumin for the indicated times. (C) Histograms show differences in unlabeled protein present after 24 h in leucine-free medium supplemented with 50 g/L bovine serum albumin (BSA) in Gcn2 WT versus Gcn2 KO cells (top) or untreated KRPC-A cells versus KRPC-A cells treated with 5 μM GCN2iB (bottom). The top row (Gcn2 WT versus Gcn2 KO) is an average of three independent biological replicates; the bottom is a single biological replicate. (D) Rank plot showing the differences in unlabeled protein present after 24 h in the leucine-free medium supplemented with 50 g/L BSA in Gcn2 WT versus Gcn2 KO cells for each top screen hit measured (74 of the top 100). The leucine transporter SLC7A5; the mannose-6-phosphate receptor IGF2R, which transports lysosomal hydrolases to the lysosome; and the lysosomal hydrolase cathepsin L are three of the top seven most upregulated proteins by GCN2. (E) Cathepsin L abundances, measured by quantitative proteomics, in Gcn2 WT and Gcn2 KO cells cultured in amino-acid-replete or leucine-free medium supplemented with 50 g/L BSA for 24 h. Measurements from three biological replicates are shown for the leucine-free condition, and a pair-wise t test was performed to show significance. (F) Western blot showing cathepsin L levels after switching from the amino-acid-replete medium to the leucine-free medium supplemented with 50 g/L BSA. The36 kDa band represents inactive pro-cathepsin L and the 30 kDa band represents mature cathepsin L.

Figure 6.. Small-molecule targeting of GCN2 and cathepsin L

KRPC-A cells were cultured in amino-acid-replete and leucine-free media in varying concentrations of GCN2iB (GCN2 inhibitor); GDC0941 (PI3-kinase class 1A inhibitor); cathepsin L inhibitor; and hydroxychloroquine. All media were supplemented with 50 g/L bovine serum albumin (BSA). Error bars represent 95% confidence intervals (n = 3).

Figure 7.. Cellular proteins critical for growth fueled by protein scavenging

Growth dependent on protein scavenging as an amino acid supply route relies on a few major cellular activities, which are illustrated with select hits of our screen depicted in red and proteins previously identified as critical depicted in blue. Extracellular protein is taken up by macropinocytosis, which is stimulated by oncogenic K-Ras signaling (Commisso et al., 2013). Two recent papers identified syndecan 1 and plasma membrane-localized V-ATPase as key mediators of this process. Both stimulate Rac1, which then mobilizes the actin cytoskeleton to achieve protein uptake (Ramirez et al., 2019; Yao et al., 2019). Our screen identified Vasp as a key cytoskeletal protein and Rabankyrin-5 as another key protein involved in macropinocytosis. Protein taken in by macropinocytosis must then be catabolized in lysosomes. Our screen identified multiple members of the BORC complex, which mediates peripheral positioning of lysosomes, and of the HOPS complex, which is required for efficient endosome-lysosome fusion. Other hits included the lysosomal hydrolases, cathepsin B and L, which convert lysosomal protein into free amino acids. Finally, amino acids generated by lysosomal catabolism must be used productively. To enable efficient elongation, aminoacid-limited cells must suppress translation inhibition, either by tuning mTORC1 activity (Palm et al., 2015) or by activating GCN2, which minimizes ribosome stalling without suppressing the overall protein synthesis. GCN2 also promotes the synthesis of catabolic proteins critical for growth in amino-acid-poor conditions. Together, these five cellular activities enable sustained growth using extracellular protein as an amino acid source.