Diverse drug-resistance mechanisms can emerge from drug-tolerant cancer persister cells

Abstract

Cancer therapy has traditionally focused on eliminating fast-growing populations of cells. Yet, an increasing body of evidence suggests that small subpopulations of cancer cells can evade strong selective drug pressure by entering a 'persister' state of negligible growth. This drug-tolerant state has been hypothesized to be part of an initial strategy towards eventual acquisition of bona fide drug-resistance mechanisms. However, the diversity of drug-resistance mechanisms that can expand from a persister bottleneck is unknown. Here we compare persister-derived, erlotinib-resistant colonies that arose from a single, EGFR-addicted lung cancer cell. We find, using a combination of large-scale drug screening and whole-exome sequencing, that our erlotinib-resistant colonies acquired diverse resistance mechanisms, including the most commonly observed clinical resistance mechanisms. Thus, the drug-tolerant persister state does not limit--and may even provide a latent reservoir of cells for--the emergence of heterogeneous drug-resistance mechanisms.

Figure 1. The emergence of persister-derived erlotinib-resistant colonies from PC9-1 cells.

(a) Schematic outline of the emergence of drug-resistant cancer cell populations (right), originating from a common clone (left), through the bottleneck of drug-tolerant, slow-growing persisters (middle grey lines). The vertical axis indicates population size; the horizontal axis is time. (b) Evolution of PERC sensitivity to erlotinib after removal from drug treatment. PERCs were grown in erlotinib-free media and periodically retested over ∼40 weeks for erlotinib sensitivity (2.5 μM erlotinib, 72 h CellTiter-Glo assay; Methods). Black: PC9-1; grey: PERCs. Viability values are calculated as the mean of technical replicates (n=3); average s.d. between replicates is 1.23. Dotted red line marks 50% viability as compared with drug-free growth; we note that when a PERC response curve crosses this line it has an IC50 of 2.5 μM erlotinib (which is 100 times the IC50 of PC9). (c) Short-term regrowth of PERCs in erlotinib after drug holiday. After ∼46 weeks of growth in erlotinib-free media, selected PERCs were grown in 2.5 μM erlotinib and imaged daily for 2 weeks. Growth is quantified in terms of percentage of field of view covered by cells (Methods). For each cell line, at each time point, 484 images were analysed and the mean fraction of cellular area calculated is reported here (error bars denote s.d. across the 484 images). A typical single cell covers ∼0.05% of the field of view as defined.

Figure 2. Identification of PERC drug-resistance mechanisms via drug screening for therapeutic vulnerabilities.

Response of PERCs versus PC9-1 to a diverse drug library. (a) Heatmap: drug-response scores of the PERCs (screen performed in erlotinib-containing media) relative to PC9-1 (screen performed in drug-free media). Rows: PERCs. Columns: 560 anticancer compounds. Scores (green/red colours): based on signed-area differences between drug-response curves of PERCs versus PC9-1 (Methods). Each response score reflects six doses performed in duplicate (n=2). (b–d) Shown are smoothed response curves with respect to selected drugs (corresponding to black triangles in a). Graphs: PERCs (drug+2.5 μM erlotinib; response curves coloured according to scores) versus PC9-1 (only drug no erlotinib; black). Green/red: drug-response scores of PERC compared with PC9-1. Smoothed curves were constructed by fitting the mean viability (n=2) at each dose to a sigmoidal function using an unweighted least-squares fit (Methods). (e) Heatmap: enrichment of PERCs (rows) for strong response to specific drug categories (columns). Drug-category-response scores are based on a hypergeometric test for varying drug-response scores (Methods; Supplementary Fig. 3). Green/red: colours as in a. (f) Annotation of drugs (columns) to specific drug categories (rows).

Figure 3. Assessment of PERC drug-resistance mechanisms via genetics and specific perturbations.

(a) Comparison of genetic alterations in PERCs (rows) for selected genes implicated in erlotinib resistance (columns). Black: presence of a non-synonymous single-nucleotide variant (SNV) versus PC9-1; red/blue: CNVs corresponding to amplification/deletions of genetic regions (Methods) versus PC9-1. (b–f) Corroboration of genetic information with predicted vulnerabilities. (PERCs were in erlotinib media; PC9-1 cells were in erlotinib-free media). (b) Response among PERCs to drugs targeting the EGFR T790M mutation: (left) heatmap: relative response of PERCs (rows) to EGFR T790M drugs (columns). Drug response (blue/yellow) is assessed by change in the AUC of a PERC from the mean AUC among all PERCs (Methods). (Right) Bar plot: row median response across drugs. *: PERCs with EGFR T790M mutation. (c) PERC dose–response curves to SGX-523, targeting cMET. Colours: PC9-1 (black); cMET-amplified PERC17 (green); other PERCs (grey). Curves are generated as in Fig. 2b. (d) Transient siRNA-mediated knockdown of MET in PC9-1 and a subset of PERCs. MET knockdown in PERC17 (only cell line with MET amplification) induces an increased cleavage of PARP, a marker for apoptosis. GAPDH: loading control. (e) PERC dose–response curves to Selumetinib, which targets MEK. Colours: PC9-1 (black); PERCs with mutations relevant to MEK sensitivity (legend); other PERCs (grey). Curves are generated as in Fig. 2b. (f) Test for role of MEK as an erlotinib-bypass mechanism. Bars indicate the strength of MEK bypass; height measures the differences in AUC between response curves for Selumetinib (n=4) and Selumetinib+erlotinib (n=3; six doses as in e). Colours: PERCs with mutations in NRAS (blue) and RAF1 (teal), both upstream of MEK; other PERCs (legend).