Volume 19, Issue 11, Pages 2289-2303 (June 2017) PHGDH Defines a Metabolic Subtype in Lung Adenocarcinomas with Poor Prognosis  Boxi Zhang, Adi Zheng, Per Hydbring, Gorbatchev Ambroise, Amanda Tomie Ouchida, Michel Goiny, Helin Vakifahmetoglu-Norberg, Erik Norberg  Cell Reports  Volume 19, Issue 11, Pages 2289-2303 (June 2017) DOI: 10.1016/j.celrep.2017.05.067 Copyright © 2017 The Author(s) Terms and Conditions

Cell Reports 2017 19, 2289-2303DOI: (10.1016/j.celrep.2017.05.067) Copyright © 2017 The Author(s) Terms and Conditions

Figure 1 PHGDH Is Differentially Expressed in Primary Lung Adenocarcinomas, and High Levels Are Associated with Poor Prognosis (A) Kaplan-Meier analysis of PHGDH expression and overall survival in 720 lung adenocarcinoma patients. (B) Tissue-microarray analysis (TMA) of the PHGDH protein expression level in 75 lung adenocarcinomas and 75 normal adjacent tissues. Three representative cores are shown in the figure, and tumor expression level was diverging into low- and high-expression subgroups. (C) Percentage distribution of the distinct PHGDH subgroups from the TMA analysis. (D) Differential expression of the PHGDH protein level was quantified in tumor and normal cases. Error bars ± SEM (∗∗∗p < 0.0001, ANOVA). Cell Reports 2017 19, 2289-2303DOI: (10.1016/j.celrep.2017.05.067) Copyright © 2017 The Author(s) Terms and Conditions

Figure 2 PHGDH Is Selectively Required for the Viability of Lung Adenocarcinoma Cells with High Endogenous Levels (A) Expression level of the serine biosynthetic pathway in a cell line panel consisting of adenocarcinomas and normal (non-transformed) lung cells. (B) The effect of genetic depletion, by two independent siRNAs targeting PHGDH, on cell viability in the cell line panel. (C and D) The effect of overexpression of PHGDH on (C) cell proliferation and (D) migration in the cell line panel. (E) Six lung adenocarcinoma cell lines (as indicated in the figure) were xenografted and tumor volume was assessed. (F) Expression of PHGDH in six normal lung (N1–N6) and 11 PHGDH-high tumors (T1–T11) after 42 days of growth. Error bars ± SEM (∗p < 0.01, ∗∗p < 0.001, and ∗∗∗p < 0.0001, two-tailed Student’s t test). Cell Reports 2017 19, 2289-2303DOI: (10.1016/j.celrep.2017.05.067) Copyright © 2017 The Author(s) Terms and Conditions

Figure 3 Utilization of Glucose-Derived Carbons in Lung Adenocarcinomas 13C isotopomer flux analysis of uniformly labeled glucose. (A) Schematic overview of the de novo serine biosynthesis pathway. (B) U-13C-glucose in media. (C) 13C enrichment of glucose-derived 13C-lactate. (D) The glycolytic function was analyzed using an Extracellular Flux Analyzer (five to seven independent ECAR measurements for each cell line). Error bars ± SEM (∗∗∗p < 0.001, two-tailed Student’s t test). (E) 13C enrichment of glucose-derived carbon in serine. For each metabolite, cumulative data were obtained from two cell lines with high endogenous PHGDH (NCI-H1437 and NCI-H1792) and two low (NCI-H1838 and NCI-H1563) lung adenocarcinoma cell lines, labeled for 8 and 16 hr. Error bars ± SEM (∗p < 0.01, ∗∗p < 0.001, and ∗∗∗p < 0.0001; ns, non-significant; two-tailed Student’s t test). Cell Reports 2017 19, 2289-2303DOI: (10.1016/j.celrep.2017.05.067) Copyright © 2017 The Author(s) Terms and Conditions

Figure 4 Biosynthesis of Glutathione and Pyrimidine Precursors Derived from Serine (A) Schematic overview of biosynthetic pathways downstream of serine metabolism. White circles represent non-labeled carbons and filled black is 13C labeled. (B) Uptake of the 13C-serine isotopomer. (C) 13C enrichment of serine-derived glycine. (D) 13C enrichment of serine-derived glutathione. (E) Glutathione levels in lung adenocarcinoma cell lines and normal cells. (F) 13C enrichment of serine-derived orotate. For each metabolite, cumulative data were obtained from two cell lines with high endogenous PHGDH (NCI-H1437 and NCI-H1792) and two low (NCI-H1838 and NCI-H1563) lung adenocarcinoma cell lines, labeled for 8 and 16 hr. Error bars ± SEM (∗p < 0.01, ∗∗p < 0.001, and ∗∗∗p < 0.0001; ns, non-significant; two-tailed Student’s t test). Cell Reports 2017 19, 2289-2303DOI: (10.1016/j.celrep.2017.05.067) Copyright © 2017 The Author(s) Terms and Conditions

Figure 5 The DNA Damage Response Is Regulated by Serine Metabolism in Lung Adenocarcinomas Cells with High Endogenous PHGDH Levels Determination of the relative levels of purines, pyrimidines in two cell lines with high endogenous PHGDH (NCI-H1437 and NCI-H1792) and two low (NCI-H1838 and NCI-H1563) lung adenocarcinoma cell lines using LC-MS. (A) Uracil, orotate, adenine, and xanthine levels presented as fold change (PHGDH high over PHGDH low). (B) The effect of genetic depletion (20 hr) of PHGDH on DNA damage as analyzed for ƴ-H2AX staining using immunoblotting. (C) High-performance chromatography analysis of purines, pyrimidines, and serine upon siRNA targeting PHGDH in four cell lines, as indicated in the figure. (D) The effect of genetic depletion (20 hr) of PHGDH on DNA damage as analyzed for ƴ-H2AX staining using immunostaining. Rescue experiments with a pool 10 μg/mL of the identified purines (Pur) and pyrimidines (Pyr) (in C) were supplemented prior to siRNA-mediated knockdown of PHGDH. Error bars ± SEM (∗p < 0.01 and ∗∗∗p < 0.0001). Cell Reports 2017 19, 2289-2303DOI: (10.1016/j.celrep.2017.05.067) Copyright © 2017 The Author(s) Terms and Conditions

Figure 6 Expression of the Serine Pathway Correlates with Glutathione, Purine, and Pyrimidine Biosynthesis Pathways in Primary Lung Adenocarcinomas (A) Heatmap representation of the relative mRNA levels of genes corresponding to the serine pathway and the rate-limiting enzymes of purine, pyrimidine, and glutathione biosynthesis pathways. (B) Transcript abundance (probe intensity) of the indicated genes in primary lung adenocarcinomas from the GEO: GSE31210 dataset. Differential expression was determined using the Mann-Whitney U test combined with the false discovery rate (FDR) correction. (C) Spearman’s correlation of the indicated genes. (D) Reactome analysis for the identified upregulated genes from patients displaying high PHGDH expression. The p values were corrected for multiple hypothesis testing by the FDR. Error bars ± SEM (∗p < 0.01, ∗∗p < 0.001, and ∗∗∗p < 0.0001; FDR < 0.05). Cell Reports 2017 19, 2289-2303DOI: (10.1016/j.celrep.2017.05.067) Copyright © 2017 The Author(s) Terms and Conditions

Figure 7 Enrichment of PHGDH with Gene Sets Required for Proliferation Gene set enrichment analysis of primary lung adenocarcinoma dataset GEO: GSE31210. (A and B) The (A) enriched pathways and (B) GSEA plots of enriched gene sets. Gene set analysis was tested for multiple hypothesis testing with p < 0.05; FDR < 0.2 are reported. Cell Reports 2017 19, 2289-2303DOI: (10.1016/j.celrep.2017.05.067) Copyright © 2017 The Author(s) Terms and Conditions

Figure 8 Differential PHGDH Levels Can Be Defined by Proteasomal Degradation and the Deubiquitinase JOSD2 (A) The effect of proteasome inhibition by 2 μM MLN9708 on PHGDH protein levels. (B) Immunoprecipitation of PHGDH and analysis of its ubiquitination level. (C) Genome-wide siRNA screen targeting DUBs in A549 cells. (D and E) The effect of JOSD2 depletion on JOSD2 (D) mRNA levels and PHGDH protein levels using two independent siRNAs targeting JOSD2 and (E) cell death. Error bars ± SEM (∗∗∗p < 0.0001; ns, non-significant; two-tailed Student’s t test). Cell Reports 2017 19, 2289-2303DOI: (10.1016/j.celrep.2017.05.067) Copyright © 2017 The Author(s) Terms and Conditions