CRISPR-Cas9 screen identifies oxidative phosphorylation as essential for cancer cell survival at low extracellular pH

Abstract

Unlike most cell types, many cancer cells survive at low extracellular pH (pHe), a chemical signature of tumors. Genes that facilitate survival under acid stress are therefore potential targets for cancer therapies. We performed a genome-wide CRISPR-Cas9 cell viability screen at physiological and acidic conditions to systematically identify gene knockouts associated with pH-related fitness defects in colorectal cancer cells. Knockouts of genes involved in oxidative phosphorylation (NDUFS1) and iron-sulfur cluster biogenesis (IBA57, NFU1) grew well at physiological pHe, but underwent profound cell death under acidic conditions. We identified several small-molecule inhibitors of mitochondrial metabolism that can kill cancer cells at low pHe only. Xenografts established from NDUFS1^-/- cells grew considerably slower than their wild-type controls, but growth could be stimulated with systemic bicarbonate therapy that lessens the tumoral acid stress. These findings raise the possibility of therapeutically targeting mitochondrial metabolism in combination with acid stress as a cancer treatment option.

Keywords: CRISPR-Cas9 screen; acidosis; oxidative phosphorylation; tumor acidity.

Graphical abstract

Figure 1

Whole-genome CRISPR-Cas9 cell viability screen identifies genes underpinning acid sensitivity and acid resistance (A–C) Cell growth reported over a range extracellular pH (pHe) in SW1222, SW480, and COLO320DM cells, measured using SRB absorbance (mean ± SEM of five independent repeats, with three technical replicates each). (D) Scheme of experimental design for genome-wide screen. Three medium pHe conditions were tested at day 11 of treatment (two independent repeats). (E) Cumulative proliferation of cells at pHe 6.6, 6.9, 7.4 (mean of n = 2). (F) All 18,049 genes are ranked by their selective essentiality at pHe 6.6 versus pHe 7.4. Red and blue color indicates indicate significantly depleted and enriched genes at 10% false discovery rate (FDR). Dashed line represents threshold level of 10% FDR. (G) Genes of the Gene Ontology “intracellular pH regulation” pathway (extended with additional pHi-regulating genes) highlighted as purple circles along the overall ranking of all genes, as in (F). Dashed line represents threshold level of 10% FDR.

Figure 2

Genes involved in mitochondrial metabolism are essential for survival under acidic conditions (A) Venn diagram showing the number of genes that are essential under mildly acidic (pHe 6.9) and highly acidic (pHe 6.6) conditions, indicating the degree of overlap. (B) KEGG OXPHOS genes highlighted in red along the overall ranking of genes, as in Figure 1F. Dashed line represents threshold level of 10% FDR. (C and D) Experimental validation of screen hits in SW480 and SW1222 KO cell lines. Growth was assayed at day 6 as a function of pHe (mean ± SEM of three to four independent repeats, with three technical replicates each). pH₅₀ value represents the pHe at which growth is halved relative to that at the optimum pHe. pH₅₀ values for wild-type cells represent average values obtained from different batches of experiments. Dotted line indicates pH₅₀ value of non-transduced wild-type cells. (E) sgRNA abundance at different pHe levels of the screen for the NDUFS1 gene coding for a complex I subunit. Mean relative abundance (± SEM) shown across four guides per gene across two screen replicates. (F and G) Normalized growth rates (measured by SRB absorbance) at 6 days as a function of pHe of wild-type, NDUFS1 sg1-infected, and NDUFS1 sg2-infected SW480 and SW1222 cell pools. Data are plotted as relative cell growth normalized to optimum pHe (mean ± SEM of three to four independent repeats, with three technical replicates each). Significance determined with two-way ANOVA using Šídák's multiple comparisons test (^∗p < 0.05, ^∗∗p < 0.01, ^∗∗∗p < 0.001; ns, non-significant, p > 0.05).

Figure 3

Glycolysis is essential at physiological pHe but dispensable at low pHe (A) Venn diagram showing number of gene knockouts enriched in mildly acidic (pHe 6.9) and highly acidic (pHe 6.6) conditions, indicating the degree of overlap. (B) KEGG glycolysis genes highlighted in red along the overall ranking of genes, as in Figure 1F. Dashed line represents threshold level of 10% FDR. (C and D) Experimental validation in SW480 and COLO320DM KO cell lines. Growth was assayed at day 6 as a function of pHe (mean ± SEM of three independent repeats, with three technical replicates each). pH₅₀ value represents the pHe at which growth is halved relative to that at the optimum pHe. pH₅₀ values for wild-type cells represent average values obtained from different batches of experiments. Dotted line indicates pH₅₀ value of non-transduced wild-type cells. gRNAs for the same gene are shown with a unique color. (E) sgRNA abundance at different pHe levels of the screen for ALDOA. Mean relative abundance (± SEM) shown across four guides per gene across two screen replicates. (F and G) Normalized growth rates at 6 days as a function of pHe (measured by SRB absorbance) of wild-type, ALDOA sg1-infected, and ALDOA sg2-infected SW480 or COLO320DM cell pools. Data are plotted as relative cell growth normalized to optimum pHe (mean ± SEM of three independent repeats, with three technical replicates each). Significance determined with two-way ANOVA using Šídák's multiple comparisons test. (H and I) Absolute cell growth (measured by SRB absorbance) in wild-type and ALDOA sg1-infected SW480 cells cultured for 6 days at pHe = 7.4 or pHe = 6.6 (mean ± SEM of three independent repeats, with three technical replicates each). Significance determined with two-tailed unpaired t test (^∗p < 0.05, ^∗∗p < 0.01, ^∗∗∗p < 0.001; ns, non-significant, p > 0.05).

Figure 4

Low pHe suppresses glycolysis and stimulates the production of reactive oxygen species (A and B) Relationship between lactate production and glucose consumption (measured by biochemical assay) as a function of pHe in SW480 and SW1222 cells (mean n = 3 independent repeats ± SEM). (C and D) Time courses of medium pH and O₂ as a function of pHe for SW1222 wild-type cells. pHe and O₂ were measured using HPTS and RuBP fluorescence, respectively, in media buffered with 10 mM HEPES and 10 mM MES. (mean n = 4 independent repeats, carried out in technical triplicate). (E and F) Reactive oxygen species (ROS) levels in SW480 and SW1222 cells cultured for 6 days at varying pHe. ROS levels expressed as H₂DCFDA fluorescence normalized to Hoechst 33342 fluorescence (mean ± SEM of five independent repeats, with three technical replicates each). (G and H) Normalized growth (measured by SRB absorbance) of SW480 and SW1222 cells cultured for 6 days at 21% O₂ versus 2% O₂. Data are plotted as relative cell growth normalized to optimum pHe (mean ± SEM of three to four independent repeats, with three technical replicates each). Significance determined with two-way ANOVA using Šídák's multiple comparisons test (^∗p < 0.05, ^∗∗p < 0.01, ^∗∗∗p < 0.001; ns, non-significant, p > 0.05).

Figure 5

Inhibition of OXPHOS selectively kills cancer cells cultured under acidic conditions (A) Western blot of lysates from SW1222 wild-type and NDUFS1^−/− clones. (B) Normalized growth rates (measured by SRB absorbance) of SW1222 wild-type and NDUFS1^−/− clonal cell lines cultured for 6 days. Data are plotted as relative cell growth normalized to optimum pHe (mean ± SEM of three independent repeats, with three technical replicates each). Significance determined with two-way ANOVA using Šídák's multiple comparisons test. (C and D) Time courses of medium pH and O₂ as a function of pHe for SW1222 wild-type and NDUFS1^−/− cells. pHe and O₂ were measured using HPTS and RuBP fluorescence, respectively, in media buffered with 2 mM HEPES and 2 mM MES. cells (mean ± SEM of eight independent repeats, with six technical replicates each). (E and F) Absolute cell growth (measured by SRB absorbance) in WT and NDUFS1^−/− cells cultured for 6 days at 21% O₂ versus 2% O₂ at pHe 7.7 (mean ± SEM of six independent repeats, with three technical replicates each). Significance determined with two-tailed unpaired t test. (G and J) Normalized growth rates (measured by SRB absorbance) of SW480 and SW1222 and cells cultured for 6 days with 10 nM rotenone, 10 μM atovaquone (ATQ), or vehicle. Data are plotted as relative cell growth normalized to optimum pHe (mean n = 3–4 independent repeats ± SEM; carried out in technical triplicates). Significance determined with two-way ANOVA using Šídák's multiple comparisons test (^∗p < 0.05, ^∗∗p < 0.01, ^∗∗∗p < 0.001; ns, non-significant, p > 0.05).

Figure 6

Iron-sulfur cluster biogenesis is essential for survival under acidic conditions (A) sgRNA abundance at different time pHe levels of the screen for NFU1. Mean relative abundance (± SEM) shown across four guides per gene across all screen replicates. (B and C) Normalized growth rates (measured by SRB absorbance) of wild-type, NFU1 sg1-infected, and NFU1 sg2-infected cell pools. Data are plotted as relative cell growth normalized to optimum pHe (mean ± SEM of three to four independent repeats, with three technical replicates each). (D and E) Normalized growth rates (measured by SRB absorbance) of SW480 and SW1222 and cells cultured for 6 days with 50 μM pioglitazone or vehicle. Data are plotted as relative cell growth normalized to optimum pHe (mean ± SEM of three independent repeats, with three technical replicates each). Significance determined with two-way ANOVA using Šídák's multiple comparisons test (^∗p < 0.05, ^∗∗p < 0.01, ^∗∗∗p < 0.001; ns, non-significant, p > 0.05).

Figure 7

Xenografts of NDUFS1^−/− cells show reduced tumor growth compared with wild-type cells (A) SW1222 wild-type and NDUFS1^−/− tumor volume in mice receiving regular drinking water. Data represents paired measurements from injected with wild-type cells in their left flank and NDUFS1^−/− in their right flank (mean ± SEM of six animals). (B) SW1222 wild-type and NDUFS1^−/− tumor volume in mice receiving oral sodium bicarbonate treatment. Data represents paired measurements from injected with wild-type cells in their left flank and NDUFS1^−/− in their right flank (mean ± SEM of six animals). (C) Comparison of tumor volume between NDUFS1^−/− in control and sodium-bicarbonate-treated animals (mean ± SEM of six animals per group). Significance determined with two-way ANOVA using Šídák's multiple comparisons test (^∗p < 0.05, ^∗∗p < 0.01, ^∗∗∗p < 0.001; ns, non-significant, p > 0.05). (D) Representative images of Cy5.5-conjugated pH-(low)-insertion peptide (pHLIP, red) and Hoechst-33342 (blue) in the fresh frozen tumor sections of wild-type and NDUFS1^−/− SW1222 xenografts in animals receiving oral bicarbonate or water (control). Scale bar, 500 μm. Note that only tiles containing the tumor were scanned.