Functional CRISPR knockout screens for modifiers of sensitivity to trastuzumab emtansine

Abstract

Background Therapies targeting HER2 amplification are a success of modern oncology, yet resistance to these agents remains a major challenge. For example, the mertansine- (DM1-) armed antibody-drug conjugate trastuzumab emtansine (T-DM1; Kadcyla) extends the survival of HER2-positive metastatic breast cancer patients after failure of prior trastuzumab-based therapy. However, T-DM1 provides objective responses in less than half of patients and its effectiveness is limited by both acquired and intrinsic resistance. Methods We employed whole-genome CRISPR knockout screens using the GeCKOv2 sgRNA library in two trastuzumab-refractory breast cancer cell lines (MDA-MB-361 and MDA-MB-453) to identify and characterise genomic modifiers of sensitivity to T-DM1 and free DM1 in vitro. Genome-scale knockout libraries developed resistance to T-DM1 and DM1 after eight and thirteen weeks of escalating drug treatment, respectively. Positive and negative selection screen hits were identified by sequencing and bioinformatics analysis using MaGECK and PinAPL-Py. In parallel, we undertook RNA sequencing analysis of the DM1 and T-DM1 resistant cells to identify transcriptional adaptations to therapy. Results Fourteen genes, highly enriched for positive or negative selection after T-DM1 or DM1 treatment, were selected to assess the impact of individual knockout on T-DM1 sensitivity. This confirmed the role of the putative lysosomal DM1 transporter SLC46A3 in T-DM1 resistance, and identified a set of genes potentially involved in T-DM1 activity. This study revealed consistent changes for both drugs with the genes involved in mitosis and cell cycle regulation highly overrepresented among differentially expressed transcripts. Adaptation to T-DM1 alone involved upregulation of genes related to proteolysis, acidification of vacuoles, intracellular trafficking of vesicles and maturation of lysosomes. Conclusions A panel of 550 candidate genes from both the genome-wide knockout screen and transcriptomic profiling informed the generation of a custom sgRNA library for secondary in vitro and in vivo validation screens. Finally, the predictive utility of these genes is being explored in clinical datasets.