CRISPR screens with trastuzumab emtansine in HER2-positive breast cancer cell lines reveal new insights into drug resistance

Abstract

Background: Trastuzumab emtansine (T-DM1) is an antibody-drug conjugate that is an effective therapy for HER2-positive breast cancer; however, its efficacy is limited by drug resistance. While multiple mechanisms of resistance have been proposed, these are not yet well understood. Greater understanding of T-DM1 sensitivity and resistance could provide new combination strategies to overcome resistance or predictive biomarkers to guide therapy.

Methods: We have conducted CRISPR/Cas9 functional genomics modifier screens in HER2-positive breast cancer cell lines to allow for unbiased discovery of T-DM1 sensitivity and resistance genes. Whole-genome knockout screens were carried out in MDA-MB-361 and MDA-MB-453 cells treated with T-DM1 and its payload cytotoxin DM1. Hits were validated in secondary T-DM1 screens using a focused single-guide RNA (sgRNA) library and subsequently by individual gene knockout.

Results: The whole-genome CRISPR screens with T-DM1 and DM1 identified 599 genes as potential modifiers of T-DM1 sensitivity and resistance. Of these, 17 genes were significantly enriched and 3 genes depleted at P < 0.001 in either or both MDA-MB-361 and MDA-MB-453 libraries in the secondary screens. Among the top hits, were known T-DM1 sensitivity genes ERBB2 and SLC46A3, in addition to negative regulators of mTOR complex 1: TSC1 and TSC2. MDA-MB-453 clones with knockout of TSC1 or partial knockout of TSC2 were more resistant to T-DM1 than wild type cells in competition growth assays and to T-DM1 and other HER2 targeting therapies (T-DXd, lapatinib and neratinib) in growth inhibition assays, and had increased internalisation of T-DM1 at 6 h. T-DM1 and the mTOR inhibitor everolimus demonstrated synergistic activity at inhibiting cell proliferation at multiple T-DM1 concentrations across four HER2-positive breast cancer cell lines.

Conclusions: Our CRISPR screening approach with T-DM1 in HER2-positive breast cancer cell lines identified genes not previously implicated in T-DM1 sensitivity or resistance, including TSC1 and TSC2. These genes may inform new strategies to enhance T-DM1 therapy in the clinic.

Keywords: Antibody-drug conjugates; CRISPR/Cas9; Drug resistance; Functional genomics; HER2-positive breast cancer; T-DM1; TSC1; TSC2.

Fig. 1

Whole-genome CRISPR/Cas9 knockout screens with T-DM1 and DM1 in MDA-MB-453 and MDA-MB-361 cells. A Schematic of whole-genome screens. B Concentration and schedule of T-DM1 and DM1 treatment during the screens. Arrows denote the time at which the screens were terminated. C Cell growth for the control (Ctrl) and drug-treated cultures during the screens (mean ± SEM, n = 3). D Emergence of resistance to T-DM1 and/or DM1 at the end of the screens in cultures that were treated with T-DM1 or DM1 during the screens relative to untreated (naïve) cultures in MDA-MB-453 and E MDA-MB-361 cells (n = 3). Cultures at the end of the screens were treated with DM1 or T-DM1 continuously for 6 days. F HER2 protein expression in control and drug-treated cultures at the end of the screen. β-actin was used as a protein loading control

Fig. 2

Identification of enriched or depleted genes in T-DM1 and DM1 whole-genome CRISPR/Cas9 knockout screens. A Volcano plots in MDA-MB-453 and B MDA-MB-361 cells indicating genes that were positively (sgRNA knockouts depleted) or negatively (sgRNA knockouts enriched) selected following T-DM1 or DM1 treatment in the knockout screens based on median log₂-fold change in the representation of sgRNA against each gene in drug-treated cultures relative to untreated cultures. The statistical significance of each gene was determined using the MAGeCK statistical algorithm. Select high-ranking findings are highlighted. C Gene ontology (GO) analysis of the gene pathways selected negatively (knockouts thereof enriched) at P ≤ 0.05 (MaGeCK) in the course of both screens with T-DM1. No gene pathways were positively selected across both screens. False discovery rate (FDR) is based on nominal P-value from the hypergeometric test. Numerical values corresponding to a pathway report fold enrichment that is the number of genes that were selected divided by all genes in the pathway

Fig. 3

Secondary CRISPR/Cas9 knockout screens using a focused library to target 599 candidate T-DM1 sensitivity/resistance genes. A Schematic of the secondary screens. B Cell growth for the control and T-DM1-treated cultures during the screens. Each condition was performed with duplicate cultures derived from a single (MDA-MB-361) or distinct (MDA-MB-453) founder libraries and the mean growth curves are plotted. C Emergence of resistance to T-DM1 at the end of the screens (day 32) in cultures that were treated with T-DM1 during the screens relative to untreated (naïve) cultures in MDA-MB-361 cells, as assessed by clonogenic survival of cultures exposed continuously to 1 or 1.4 nM T-DM1 over 21 days. D Volcano plot of the statistical significance of gene-level enrichment or depletion as a function of the median log₂ fold-change in normalised read counts (T-DM1/control) of all sgRNA against each target (4 sgRNA per gene) in all replicate screens performed in MDA-MB-361 and MDA-MB-453 cells

Fig. 4

Validation of TSC1 and TSC2 as T-DM1 sensitivity genes. A Protein expression of TSC1, TSC2 and HER2 by Western blotting in TSC1-knockout and partial TSC2-knockout MDA-MB-453 clonal cell lines. Each blot represents different cell lysates. β-actin was used as a protein loading control. B Competition growth assay of GFP-negative TSC1-knockout MDA-MB-453 clonal cell lines #1, #3 and #4 cultured 1:1 with GFP-positive MDA-MB-453 wild type cells and treated with T-DM1 at the indicated concentrations for 10 days. Cells either received the same concentration of T-DM1 for 10 days (d0–10), or an initial concentration for days 0–5 that was subsequently doubled for days 5–10. Lines represent the mean ± SEM for three separate experiments, with each separate experiment represented as a different symbol. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001. C Growth inhibition curves of T-DM1 in TSC1-knockout and partial TSC2-knockout MDA-MB-453 clonal cell lines. D Growth inhibition curves of DM1, T-DXd, lapatinib and neratinib in TSC1-knockout and partial TSC2-knockout MDA-MB-453 clonal cell lines. Plots in C) and D) are representative images of growth inhibition plots of three separate experiments. Symbols represent mean ± SEM of two technical replicates

Fig. 5

Internalisation of T-DM1 increases with TSC1 knockout and partial TSC2 knockout. MDA-MB-453 wild type, TSC1 knockout (clone #4) and TSC2 partial knockout (clone #8) cells were incubated with T-DM1 conjugated to Alexa Fluor 488 (T-DM1-488) and imaged as live cells over time by confocal microscopy. A Confocal microscopy images at 20 × objective. Arrows indicate cells with internalisation of T-DM1-488 into the cytoplasm at 6 h. Scale bar = 20 µm. B T-DM1-488 uptake into cells over time. Bars represent mean ± SEM of 24 tracked cells. ns, nonsignificant; **, P < 0.01, ****, P < 0.0001 at 6 h vs 0 h by one-way ANOVA

Fig. 6

Everolimus has synergistic antiproliferative activity with T-DM1 Combination indices of antiproliferative activity of T-DM1 and everolimus. Everolimus (EVL) was tested at different concentrations to cause 20–80% inhibition of cell growth. Average indicates the average combination index across all five everolimus concentrations. Symbols represent the mean ± SEM of n = 2–4. The green box indicates synergy with combination index < 0.9