Research ArticleCancer therapy

ELP-dependent expression of MCL1 promotes resistance to EGFR inhibition in triple-negative breast cancer cells

See allHide authors and affiliations

Science Signaling  17 Nov 2020:
Vol. 13, Issue 658, eabb9820
DOI: 10.1126/scisignal.abb9820
  • Fig. 1 TNBC cells are insensitive to EGFR inhibition despite high levels of EGFR-dependent signaling.

    (A and B) Erlotinib sensitivity in TNBC cells. (A) Normalized growth rate inhibition value (GR) after 72-hour exposure to erlotinib. Data are the means ± SD from three biological replicate wells. Statistical analyses for pairwise comparisons of area over the curve for TNBC compared to PC9 performed using ANOVA with Dunnett’s post hoc analysis test. (B) Erlotinib-induced lethal fraction kinetics. Data are means ± SD from three biological replicate wells. Statistical analyses of death rate for pairwise comparisons between vehicle and erlotinib-treated cells performed using ANOVA with Dunnett’s post hoc analysis test. (C and D) ERK phosphorylation (p-ERK) after 10 μM erlotinib exposure. Blots in (C) are representative of three biological replicates. (D) p-ERK quantified from Western blots in (C). Data are means ± SD from three biological replicate blots. Data in (D) are colored as in (A). Statistical analyses for pairwise comparisons of area over the curve for TNBC compared to PC9 performed using Dunnett’s test. (E and F) Gene set enrichment analysis (GSEA) of erlotinib-dependent mRNA expression changes in PC9 or BT20 cells. Sequencing data generated from three biological replicate samples. Rank-ordered log2 fold change (L2FC) for erlotinib- versus DMSO vehicle–treated samples used to evaluate genes that are induced (E) or depleted (F) by EGFR inhibition in bona fide EGFR-driven NSCLC cells. Gene signatures used in comparison are from msigDB [“Kobayashi EGFR Signaling 24-hour UP” in (E) and “Kobayashi EGFR Signaling 24-hour DN” in (F)]. FDR-adjusted P values shown based on 1000 permutations of gene sets. ***P < 0.001 (A, B, and D). n.s., not significant. See also figs. S1 and S2.

  • Fig. 2 Genome-wide screen using CRISPR-Cas9–mediated genome editing reveals that the ELP complex contributes to erlotinib insensitivity in TNBC.

    (A) Schematic overview of CRISPR screen. (B and C) Fold change in recovery of BT20 cells harboring gene knockouts when comparing erlotinib-treated to untreated cells. (B) Three hundred twenty-four genes were differentially recovered. Six ELP complex genes are highlighted (red), and nontargeting genes highlighted in black. (C) sgRNA-level data for the six ELP genes. FDR-corrected P values shown based on bootstrapping sgRNA level data. Gray curve is the distribution of all sgRNAs; red lines are the six individual sgRNAs for a given ELP gene (non–z-cored). (D and E) GSEA of CRISPR screening data. (D) Ten most enriched signatures within msigDB. FDR cutoff for significance shown. (E) Example signature shown for the GO term: TRNA Wobble Uridine. FDR P value shown. For (D) and (E), FDR was determined on the basis of 1000 permutations of gene sets. For (A) to (E), data are based on three biological replicate samples. (F and G) ELP validation using quantitative microscopy. (F) Population growth over time for BT20 cells with or without an ELP targeting siRNA. Gray, scrambled RNA untreated; blue, scrambled RNA + 10 μM erlotinib; red, ELP targeting siRNA pool; purple, ELP targeting siRNA pool + 10 μM erlotinib. See fig. S6 for knockdown validation. (G) Growth rates for curves in (F). Data are population growth per hour. Data in (F) and (G) are means ± SD for 16 biological replicate wells for scrambled RNA (SCX) condition and three biological replicate wells in ELP siRNA conditions. Growth rates for erlotinib-treated cells compared using Dunnett’s test. See also figs. S3 to S6. ***P < 0.001.

  • Fig. 3 ELP complex promotes survival of TNBC cells exposed to erlotinib by promoting expression of Mcl-1.

    (A) Lethal fraction kinetics after application of 10 μM erlotinib. Cells tested were BT20 + scrambled RNA (SCX) or BT20 + siRNA targeting ELP3, ELP4, ELP5, or ELP6. Data are means ± SD from three biological replicate wells. Death rates for each ELP KD compared to SCX (Dunnett’s test). (B and C) CellTrace proliferation dye dilution to determine growth rate of live cells. (B) BT20 cells with scrambled RNA or ELP4-targeted siRNA, untreated (Unt), or given 10 μM erlotinib (ERL). Samples collected before or 72 hours after drug application. Representative dye fluorescence distributions from flow cytometry shown based on three biological replicate samples with at least 10,000 cells measured per sample. (C) Quantification of cell growth rate from data in (B). Data are means ± SD of three biological replicate samples. (D) Mcl-1 protein expression in BT20 cells + SCX or + ELP4 siRNA, with or without 10 μM ERL. Blots are representative of three biological replicates that all showed similar results. (E) Proteome profiler apoptotic array. Data shown are representative of three biological replicate blots from ELP4 siRNA untreated or ELP4 siRNA + 10 μM ERL. (F) Quantification of (D) and (E). Data are means ± SD of three biological replicate blots. Labeled signals are significantly different in ELP4 KD cells compared to scrambled RNA cells (t test with FDR correction). See also figs. S6 to S9. *P < 0.05; **P < 0.01; ***P < 0.001 (A, C, and F); n.s., not significant.

  • Fig. 4 Mcl-1 inhibition synergistically enhances sensitivity to erlotinib in TNBC.

    (A) Full dose titration of erlotinib (ERL) and S63845 (Mcl-1i) in BT20 cells. Data are relative viability (RV) of drug-treated cells compared to untreated cells at 72 hours. Heatmap is scaled according to mean values from three biological replicate wells. (B) Isobologram analysis for data in (A). Data are arrayed in linear scale with linear interpolation. White dashed line represents ERL + Mcl-1i combinations that result in 50% response (50% isobol). (C) ERL sensitivity at 72 hours with or without 1 μM Mcl-1i. Data are means ± SD from three biological replicate wells. Responses evaluated based on area over curves, compared using a t test. (D) Lethal fraction kinetic responses in BT20 cells treated with 1 μM ERL, 1 μM Mcl-1i, or both. Data are means ± SD of three biological replicate wells. Statistical analyses performed using Dunnett’s test for pairwise comparisons of death rates. (E and F) ERL, Mcl-1i, or combination responses at varied doses in a panel of TNBCs. Drugs were tested as single agents or in 1:1 fixed ratio combinations across seven doses. (E) Area over the dose-response curve (AOC). Data are means ± SD for three biological replicate wells. Individual replicates shown (white dots). (F) Combination index (CI) and deviation from Bliss independence (DBI) computed for combinations of ERL and Mcl-1i in TNBC cells. For DBI, statistical analysis performed for observed data compared to a Bliss reference model for response independence. For CI, statistical analysis performed relative to the predicted IC50 (median inhibitory concentration) dose, given dose additivity. For (E) and (F), full dose-response profiles and statistical analysis are in fig. S10. *P < 0.05; ***P < 0.001 (C, D, and F); n.s., not significant.

Supplementary Materials

  • stke.sciencemag.org/cgi/content/full/13/658/eabb9820/DC1

    Fig. S1. Erlotinib drug GRADE in TNBC compared to PC9 cells.

    Fig. S2. ERK activation upon erlotinib treatment in TNBC cells.

    Fig. S3. RTK expression changes upon erlotinib exposure.

    Fig. S4. Overview of genome-wide screen and analysis.

    Fig. S5. Varied levels of ELP complex essentiality in TNBC cells.

    Fig. S6. Validation of ELP knockdown.

    Fig. S7. Correlations between erlotinib sensitivity and EGFR, MCL1, and ELP expression.

    Fig. S8. ELP complex facilitates expression of MCL1 upon erlotinib exposure.

    Fig. S9. Expression of apoptotic regulatory proteins upon erlotinib exposure.

    Fig. S10. Mcl-1 inhibition synergistically enhances sensitivity to erlotinib in a panel of TNBC cells.

    Table S1. Sequences for siRNAs targeting ELP3 to ELP6.

    Data file S1. Raw counts from genome-wide screen.

    Data file S2. Gene-level fold change from genome-wide screen.

  • The PDF file includes:

    • Fig. S1. Erlotinib drug GRADE in TNBC compared to PC9 cells.
    • Fig. S2. ERK activation upon erlotinib treatment in TNBC cells.
    • Fig. S3. RTK expression changes upon erlotinib exposure.
    • Fig. S4. Overview of genome-wide screen and analysis.
    • Fig. S5. Varied levels of ELP complex essentiality in TNBC cells.
    • Fig. S6. Validation of ELP knockdown.
    • Fig. S7. Correlations between erlotinib sensitivity and EGFR, MCL1, and ELP expression.
    • Fig. S8. ELP complex facilitates expression of MCL1 upon erlotinib exposure.
    • Fig. S9. Expression of apoptotic regulatory proteins upon erlotinib exposure.
    • Fig. S10. Mcl-1 inhibition synergistically enhances sensitivity to erlotinib in a panel of TNBC cells.
    • Table S1. Sequences for siRNAs targeting ELP3 to ELP6.
    • Legends for data files S1 and S2

    [Download PDF]

    Other Supplementary Material for this manuscript includes the following:

    • Data file S1 (Microsoft Excel format). Raw counts from genome-wide screen.
    • Data file S2 (Microsoft Excel format). Gene-level fold change from genome-wide screen.

Stay Connected to Science Signaling

Navigate This Article