Inhibiting the system xC-/glutathione axis selectively targets cancers with mutant-p53 accumulation

Abstract

TP53, a critical tumour suppressor gene, is mutated in over half of all cancers resulting in mutant-p53 protein accumulation and poor patient survival. Therapeutic strategies to target mutant-p53 cancers are urgently needed. We show that accumulated mutant-p53 protein suppresses the expression of SLC7A11, a component of the cystine/glutamate antiporter, system x_C^-, through binding to the master antioxidant transcription factor NRF2. This diminishes glutathione synthesis, rendering mutant-p53 tumours susceptible to oxidative damage. System x_C^- inhibitors specifically exploit this vulnerability to preferentially kill cancer cells with stabilized mutant-p53 protein. Moreover, we demonstrate that SLC7A11 expression is a novel and robust predictive biomarker for APR-246, a first-in-class mutant-p53 reactivator that also binds and depletes glutathione in tumours, triggering lipid peroxidative cell death. Importantly, system x_C^- antagonism strongly synergizes with APR-246 to induce apoptosis in mutant-p53 tumours. We propose a new paradigm for targeting cancers that accumulate mutant-p53 protein by inhibiting the SLC7A11-glutathione axis.

Figure 1. Glutathione depletion and ROS induction is central to the anti-tumour activity of APR-246.

(a) ROS detected using CellROX 6 h post 50 μM APR-246 treatment across a panel of oesophageal cell lines. Mean fluorescence unit (MFU)±s.e.m. From here on FLO-1 and JH-EsoAd1 cells were treated with their GI90 doses: 25 and 40 μM of APR-246, respectively. (b) Glutathione (GSH) concentrations post APR-246 treatment in FLO-1 and JH-EsoAd1 cells. (c,d) CellROX (c) and viability (d) analysis post treatment with APR-246 and/or 5 mM glutathione-monoethyl ester (GSH-MEE) and/or N-acetyl-cysteine (NAC) in FLO-1 and JH-EsoAd1 cells assayed at 10 and 96 h, respectively. (e) Structural formula of GS-MQ (left) following reaction of MQ with GSH. Typical mass spectrometry (MS) pattern of 100 μM GS-MQ (m/z=445.17 [m+H]) in 0.1% formic acid (right). (f) MS analysis of JH-EsoAd1 cell lysates collected 10 h post vehicle and APR-246 treatment. (g) Absorbance of 10 μM GSH or GS-MQ as measured with the Cayman GSH kit containing GSH reductase. Paired t-test (a), unpaired t-test (b,g), one-way ANOVA with Dunnett's multiple comparison post-test (c,d). Error bars=s.e.m., n=3 for all, *P<0.05, **P<0.01, ***P<0.001. See also Supplementary Fig. 1.

Figure 2. APR-246 triggers lipid peroxidative cell death through depleting glutathione.

(a,b) Detection of mitochondrial ROS using MitoSOX (a) and lipid peroxidation using C11-BODIPY (b) post APR-246 treatment in FLO-1 and JH-EsoAd1 cells. (c) Transmission electron microscopy of FLO-1 cells treated with APR-246 for 15 h. Red arrows: mitochondrial membrane rupture. A minimum of 10 cells were examined. Scale bar for × 10,000=2 μm, for × 80,000=200 nm. (d) Cytochrome c released from FLO-1 and JH-EsoAd1 cells measured using flow cytometry 20 h post APR-246 treatment. (e) Viability of FLO-1 and JH-EsoAd1 cells at 96 h post treatment with APR-246 and trolox (1 mM), ferrostatin-1 (Fer-1, 20 μM) or 2-merceptoethanol (2-ME, 100 μM). One-way ANOVA with Dunnett's multiple comparison post-test (e). Error bars=s.e.m., n=3 for all, *P<0.05, **P<0.01, ***P<0.001. See also Supplementary Fig. 2.

Figure 3. SLC7A11 expression predicts and modulates tumour sensitivity to APR-246.

(a,b) Correlation between endogenous glutathione (GSH) levels (a) and mRNA expression of GSH pathway genes (b) with APR-246 sensitivity (GI50) in oesophageal cell lines. See also Supplementary Table 1. (c,d) Endogenous protein (c) and mRNA (d) levels of SLC7A11 in oesophageal cell lines. (e,f) Correlation between mRNA (e) and protein (f) levels of SLC7A11 with APR-246 GI50 in oesophageal cell lines. (g) SLC7A11 knockdown in p53^KO JH-EsoAd1 cells. APR-246 was applied 2 days post transfection of SLC7A11 and non-targeting control (NTC) siRNA with viability measured at 96 h post APR-246 (right). Knockdown was confirmed by western blot (left) 3 days post transfection. (h) SLC7A11 and red fluorescent protein (RFP) were overexpressed in mut-p53 JH-EsoAd1 cells. This was confirmed by western blot (left). Cell viability was measured 96 h post APR-246 (right). (i) Correlation between SLC7A11 protein level (left, n=4) and in vivo response to APR-246 (100 mg kg⁻¹, intraperitoneal injection, daily for 3 weeks. PDX4 was treated daily for 2 weeks) in cell line and patient-derived xenografts (PDX, right). % Tumour growth inhibition was quantified at endpoint. Pearson's correlation (a,b,e), unpaired t-test (f). Error bars=s.e.m., n=3 for all in vitro studies. See also Supplementary Fig. 3.

Figure 4. Accumulated mut-p53 entraps NRF2 and represses SLC7A11 expression.

(a,b) Correlation between p53 with SLC7A11 protein (a) and mRNA levels (b) in oesophageal cell lines. (c–e) SLC7A11 mRNA (bottom) and protein (top) levels post mut-p53 knockdown (c, non-targeting control (NTC) and p53 siRNA (sip53)) in FLO-1 cells, knockout (d) in JH-EsoAd1 cells and overexpression (e) in H1299 cells. (f) Immunoprecipitation (IP) and immunoblot (IB) of NRF2 with mut-p53 in FLO-1, Eso26, OACM5.1 and H1299 cells under basal growth conditions. Reverse IP and IB in Supplementary Fig. 4. (g) Chromatin IP of the SLC7A11 promoter using NRF2 and mut-p53 antibodies at rest and post H₂O₂ stress (4 h, 50 μM) in p53^Null (Par) and p53^R273H H1299 cells. (h) SLC7A11 mRNA expression following vehicle or H₂O₂ (50 μM) treatment in p53^Null (Par), p53^R273H and p53^R175H H1299 cells. Pearson's correlation (b), unpaired t-test (c,g), one-way ANOVA with Dunnett's multiple comparison post-test (h). Error bars=s.e.m., n=3 for all except (f) n=2, *P<0.05, ***P<0.001. See also Supplementary Fig. 4.

Figure 5. Mut-p53 accumulation sensitizes cancer cells to oxidative stress.

(a,b) Correlation of mut-p53 protein (normalized to β-actin) with ROS (a) and glutathione (GSH) (b) levels under basal growth conditions for all mut-p53 cancer cell lines used in this study. For further analysis cell lines were dichotomized into mut-p53 high and low expressers based on western blot in Fig. 4a. (c) System x_c⁻ activity as assayed by glutamate release from p53^Null, p53^R273H and p53^R175H H1299 cells. Cells were stimulated with 600 μM L-cystine for 3 h prior to glutamate assay. (d) GSH levels at rest in p53^Null, p53^R273H and p53^R175H H1299 cells. (e) Cell viability at 96 h post H₂O₂ treatment in p53^Null, p53^R273H and p53^R175H H1299 cells. (f) ROS levels at rest in p53^G266E and mut-p53^KO JH-EsoAd1 cells. (g) Cell viability at 96 h post H₂O₂ treatment in p53^G266E and mut-p53^KO JH-EsoAd1 cells. (h) ROS levels 6 h post 50 μM APR-246 treatment in all mut-p53 cancer cell lines grouped by endogenous mut-p53 protein levels as determined in a,b. (i) ROS levels 6 h post 50 μM APR-246 treatment in p53^G266E and mut-p53^KO JH-EsoAd1 cells. Pearson's correlation (a,b), one-way ANOVA with Dunnett's multiple comparison post-test (c,d), unpaired t-test (f–i). Error bars=s.e.m., n=3 for all except (d) n=4 and (f,i) n=5, *P<0.05, **P<0.01, ***P<0.001. See also Supplementary Fig. 5.

Figure 6. System x_C⁻ inhibition selectively targets cancer cells with mut-p53 accumulation.

(a) Plot of 389 anti-ROS stress genes identified in an unbiased multi-omics analysis of the Broad Institute's Project Achilles v2.4 dataset (See Methods and Supplementary Fig. 9 for analysis workflow), correlating statistical significance with cancer cell killing following shRNA-mediated gene knockdown. Vertical axis quantifies the extent of preferential killing of missense mut-p53 cells (which typically accumulate mut-p53 protein) versus non-missense p53 cells (including wt-p53, splice, frameshift (Fs) and deletion (Del) variants). Vertical dotted line: Bonferroni corrected P-value=1.1 × 10⁻⁴ comparing the effect of gene knockdown on missense mut-p53 versus non-missense p53 cancer cells. See also Supplementary Data 1. (b) Cell viability following shRNA-mediated knockdown of SLC7A11 in cell lines from Project Achilles v2.4. Each box/whiskers plot=median/5–95 percentiles. (c) SLC7A11 mRNA levels 48 h post transfection with non-targeting control (NTC) siRNA or siSLC7A11 in oesophageal cell lines. (d,e) Viability at 96 h post SLC7A11 knockdown (d) and grouped by each cell line's endogenous p53 (e, left) or SLC7A11 (e, right) protein expression. (f) Erastin sensitivity (GI50) in oesophageal cell lines. (g) SLC7A11 knockdown in parental (Par) p53^Null and mut-p53 overexpressing H1299 cells. SLC7A11 mRNA (left) and viability (right) was measured at 48 and 96 h post siRNA transfection, respectively. (h) Viability at 96 h post erastin treatment in p53^Null and mut-p53 overexpressing H1299 cells. For all studies n=3 except (h) n=5, and (a,b). Unpaired t-test (b,e), one-way ANOVA with Dunnett's multiple comparison post-test (g). Error bars=s.e.m., *P<0.05, **P<0.01, ***P<0.01. See also Supplementary Fig. 6.

Figure 7. System x_C⁻ antagonists synergize with APR-246 to inhibit mut-p53 cancer cells.

(a) Viability at 96 h post APR-246 treatment in non-targeting control (NTC) or siSLC7A11 transfected mut-p53 high (left) and low (right) expressing cell lines. Cells were transfected with siRNA 48 h prior to APR-246. (b,c) CellROX (b) and Annexin-V (c) analysis in cells treated with 10 μM APR-246 48 h post NTC or siSLC7A11 transfection. CellROX and Annexin-V were assayed at 24 and 48 h post APR-246 respectively. Mean fluorescence unit (MFU). (d) Combination index (CI) plot of cell lines treated with APR-246 and sulfasalazine (SAS) for 96 h. For CI analysis, cells were treated with a range of doses for each agent alone and in combination. Synergistic interaction was quantified using CalcuSyn v2, where a CI<0.9: synergism, CI>1.1: antagonism and 0.9≤CI≤1.1 (Grey area): additive effect. CI plot shows the extent of drug interaction to achieve 50% cell death. (e) Extent of synergistic interaction between SAS and APR-246 to achieve 50% (ED50), 75% (ED75) and 90% (ED90) cell death in oesophageal cell lines with different levels of p53 protein. Each point=mean CI per cell line (n=3). Bars=mean of each group. (f) Viability at 96 h post treatment with 10 μM APR-246 and/or 400 μM SAS in parental (Par) p53^Null and mut-p53 overexpressing H1299 cells. (g) Viability at 96 h post treatment with 7.5 μM APR-246 and/or 200 μM SAS in FLO-1 cells with either SLC7A11 or red fluorescence protein (RFP) overexpression. (h–k) GSH (h), MitoSOX (i), C11-BODIPY (j) and Annexin-V/PI (k) analysis of FLO-1 cells treated with 7.5 μM APR-246 and/or 0.5 μM erastin or 200 μM SAS. GSH, MitoSOX, C11-BODIPY and Annexin-V/PI were assayed at 15, 24, 24 and 48 h post treatment, respectively. For b,c,f–k the dose of APR-246, erastin or SAS was deliberately chosen to have low cytotoxicity on its own to highlight the combinatory effect. One-way ANOVA with Dunnett's multiple comparison post-test (b,c,e,f,h,i,k), unpaired t-test (g). Error bars=s.e.m., n=3 for all, *P<0.05, **P<0.01, ***P<0.01. See also Supplementary Fig. 7.

Figure 8. System x_C⁻ antagonists synergize with APR-246 in vivo.

(a) Example fluorescence imaging of mice bearing FLO-1 xenografts transduced with a doxycycline (dox)-inducible dual fluorescent vector with shSLC7A11 cloned in (top). Western blot of FLO-1 xenografts 72 h post dox (Food: 600 mg kg⁻¹, water: 2 mg ml⁻¹) induction to assess the extent of knockdown using two hairpins (sh) against SLC7A11 (Bottom). (b,c) Growth curves of FLO-1 tumours transduced with shSLC7A11_1 (b) and shSLC7A11_2 (c) hairpins in mice treated with APR-246 (100 mg kg⁻¹, daily) and/or dox for 28 days. n=5 per group. (d) Kaplan–Meier plot of time to reach 1,500 mm³ tumour volume in mice bearing FLO-1 tumours transduced with shSLC7A11_1 and treated according to conditions detailed in b. (e) Intratumoral GSH levels in FLO-1 xenografts transduced with shSLC7A11_2 at 28 days post treatment. (f) Growth curves of FLO-1 xenografts in mice treated with either vehicle (0.9% saline, daily, n=7), APR-246 (100 mg kg⁻¹, daily, n=7), sulfasalazine (SAS, 6 mg, twice daily, n=6) or APR-246 and SAS (n=6) for 28 days (top). Body weight at 28 days post treatment as a percentage change from baseline (bottom). Ethically acceptable weight loss is defined by the Peter MacCallum Cancer Centre Animal Experimentation Ethics Committee as <20% compared to pre-treatment body weight (within the dotted lines). (g) Growth curves of Patient-derived xenograft 4 (PDX4) in mice treated with the same drug regimen as (f) for 14 days. Vehicle: n=9, all other groups: n=8. (h) Representative western blot of PDX4 demonstrating the accumulation of mut-p53 protein. (i) Intratumoral GSH levels in PDX4 at 14 days post treatment with drug regimens detailed in g. (j) Representative H&E, Ki67 and Cleave caspase 3 staining of PDX4 tumours (left) and their respective quantification (right). Scale bars=100 μm. Grey shading: treatment period. One-way ANOVA with Dunnett's multiple comparison post-test (b–g,i,j). Error bars=s.e.m., *P<0.05, **P<0.01, ***P<0.01. See also Supplementary Fig. 8.

Figure 9. Model representation of mut-p53 entrapment of NRF2 and its implications for cellular redox balance and therapeutic intervention.

(a) In the absence of mut-p53 accumulation, NRF2 is able to transcriptionally regulate cellular redox balance by binding to antioxidant responsive elements (ARE) on antioxidative stress (AOS) genes. A crucial component of NRF2-mediated redox regulation is the transactivation of SLC7A11, a key component of the glutamate (Glu)/cystine (Cys) antiporter, system x_c⁻ (Sx_c⁻), resulting in maintenance of intracellular glutathione (GSH) reserves. (b) Cells with no mut-p53 accumulation have sufficient GSH reserves, and can mount a normal NRF2-mediated defence against oxidative stress, such as that induced by APR-246 (active compound: MQ). Therefore, these cells are relatively resistant to APR-246. (c) In cancer cells with mut-p53 accumulation, mut-p53 entraps NRF2 and impairs its canonical transcriptional activity, resulting in suppressed expression of SLC7A11 and other AOS genes. This reduces GSH reserves and increases resting levels of ROS. Although these cells are tolerant of this, they are susceptible to further oxidative stress and genomic instability. (d) Cancer cells with mut-p53 accumulation are therefore highly sensitive to system x_c⁻ inhibitors (for example sulfasalazine (SAS) and erastin) and APR-246. In combination, these agents synergistically deplete mut-p53 cancer cells of GSH, resulting in significant oxidative stress and massive cell death.