Exploiting Temporal Collateral Sensitivity in Tumor Clonal Evolution - PubMed (original) (raw)

Exploiting Temporal Collateral Sensitivity in Tumor Clonal Evolution

Boyang Zhao et al. Cell. 2016.

Abstract

The prevailing approach to addressing secondary drug resistance in cancer focuses on treating the resistance mechanisms at relapse. However, the dynamic nature of clonal evolution, along with potential fitness costs and cost compensations, may present exploitable vulnerabilities-a notion that we term "temporal collateral sensitivity." Using a combined pharmacological screen and drug resistance selection approach in a murine model of Ph(+) acute lymphoblastic leukemia, we indeed find that temporal and/or persistent collateral sensitivity to non-classical BCR-ABL1 drugs arises in emergent tumor subpopulations during the evolution of resistance toward initial treatment with BCR-ABL1-targeted inhibitors. We determined the sensitization mechanism via genotypic, phenotypic, signaling, and binding measurements in combination with computational models and demonstrated significant overall survival extension in mice. Additional stochastic mathematical models and small-molecule screens extended our insights, indicating the value of focusing on evolutionary trajectories and pharmacological profiles to identify new strategies to treat dynamic tumor vulnerabilities.

Copyright © 2016 Elsevier Inc. All rights reserved.

PubMed Disclaimer

Figures

Figure 1

Figure 1. Conceptual fitness landscapes with clonal intermediates

Predefined fitness landscapes can be visualized with a z-axis corresponding to a fitness of the population under a given environmental condition, and x- and y- corresponding to a two-dimensional coordinate of the genotype of each subpopulation. The actual genotype can be in a high-dimensional space, but is explicitly represented here in two-dimensions. The fitness landscape for drug A is composed of two Gaussian peaks for intermediate and terminal stage. In contrast, at the location of the intermediate peak, the corresponding fitness landscape for drug B contains a valley. Initial population is a homogeneous population starting at a low fitness, as indicated by the white asterisk. See also Figure S1.

Figure 2

Figure 2. A pharmacological screen of each distinct evolutionary stage to identify persistent and temporal collateral resistance and sensitivity

(A). Schematic of experimental setup for drug resistance selection experiment (see Methods for details). Briefly, a murine derived Ph+ acute lymphoblastic leukemia cell line is treated at IC90 1x drug concentration. Upon recovery and outgrowth, the population is dose escalated to 2x the previous drug concentration. The derived cell line, mimicking a specific stage of the clonal evolution, was allowed to recover and profiled based on viability assays across a panel of targeted and chemotherapeutics. Each selection experiment terminates upon either no outgrowth at the given drug concentration or until IC90 16x. (B) Preliminary drug selection experiment with DMSO control and three independent dasatinib selection at IC90 1x concentration, illustrating collateral sensitivity and resistance. (C) A complete overview of the drug selection experiments for vehicle, dasatinib, and bosutinib, showing diverse collateral resistance and sensitivity patterns. The black triangles illustrate each independent series of dose escalating concentrations, and hence also an indicator of time. The heatmap shows the log2 transform of the ratio of the EC50s for each drug of given cell line relative to parental cell line. Cell lines with DMSO control grown in parallel had similar EC50s as the parental. The kinase domain of ABL1 was also Sanger sequenced, the subpanel to the right of heatmap illustrates complete concurrence between the sensitization to crizotinib, foretinib, cabozantinib, and vandetanib and single mutational V299L in ABL1. (D) A representative Sanger sequencing of ABL1 V299L. (E) Dose responses of BCR-ABL1 inhibitors and collaterally sensitive inhibitors crizotinib, foretinib, cabozantinib, and vandetanib in Ba/F3 isogenic parental, BCR-ABL1 WT, and BCR-ABL1 V299L cell lines. The sensitization was consistently observed and suggests that V299L is a causative determinant for the sensitization phenotype. See also Figure S2 and S3.

Figure 3

Figure 3. Mathematical models of tumor kinetics predicts pre-existing subpopulations and treatment window

(A–B) Representative stochastic birth/death model simulation results with Monte Carlo sampling of parameters for the first round of selection with dasatinib at IC90 1x concentration. The total population size is shown in (A) and the corresponding subpopulation fractions shown in (B). Simulation results were constrained to those fitting experimental observations (in terms of total population size and tumor composition) at day 0 and 9. Stochastic model predictions at ten evenly distributed time points (connected by line) are shown in plots. (C–D) Distribution of pre-existing subpopulation percentages of BCR-ABL1 V299L (C) and V299L compound (D) for those simulation results that fit our experimental observations. The histogram includes all parameter combinations with at least one simulation run (out of the 50 per parameter combination) that fit the data. The only way to explain our observed kinetics was with the pre-existence of BCR-ABL1 V299L. (E) Sensitivity analyses based on Monte Carlo sampling and stochastic birth/death model showing the effects of each parameter on final tumor population size and tumor composition, as measured by Kendall correlation. Blue and red indicate positive and negative correlation, respectively. The major determinant of final subpopulation sizes was the pre-existing subpopulation sizes. Birth/death rates and background mutation rates had minimal effects. (F–G) Representative ODE simulation kinetics of long-term drug resistance selection with dose escalating concentrations of dasatinib. The dose schedule is shown in (F) and corresponding subpopulation fractions in (G). (H-I) Given the dose escalation simulations, we also predicted the EC50s for the overall population at each time point over the course of dasatinib selection. This provides an approximate treatment window for which we can observe temporal collateral sensitivity to drugs such as foretinib. The predicted EC50 for the overall population was based on a weighted sum of the known EC50s for individual subpopulations. See also Figure S4.

Figure 4

Figure 4. Sensitivity of BCR-ABL1 V299L acts through on-target inhibition of BCR-ABL1

(A). Representative cell cycle profiles taken at 12 hours post treatment in vitro for Ph+ ALL cell lines derived from drug selection experiments, either with BCR-ABL1 WT or V299L. Treatments with ABL1 inhibitors led to a G1 arrest, albeit at higher concentrations for V299L cell lines due to resistance. While no G1 arrest was observe upon treatment with crizotinib, foretinib, cabozantinib, and vandetanib in the BCR-ABL1 WT cell line, G1 arrest was observed in the presence of BCR-ABL1 V299L. (B) Representative flow cytometry analysis of phospho-Stat5 (a measure of ABL1 activity) and cleaved-PARP (a measure of apoptosis). Similar to the cell cycle profile phenotypes, we observed an inhibition of pStat5 in the presence of V299L upon treatment with the collaterally sensitives, supporting an on-target ABL1 inhibition as the mechanism of action. (C) In vitro kinase assay at 10 μM ATP with recombinant active ABL1 WT or V299L against vandetanib, showing a preferential inhibition of kinase assay against ABL1 V299L relative to WT. Results for other small molecules are shown in Supp Fig S6A. Data are shown as mean ± s.d. from three independent experiments. (D-E) Models of vandetanib docked to ABL1 WT and V299L. Vandetanib and bosutinib are shown in blue and orange, respectively. V299L causes steric hindrance to nitrile group of bosutinib, whereas it provides additional van der Waals contact to quinazoline group of vandetanib. See also Figure S5, S6, and S7.

Figure 5

Figure 5. Non-canonical BCR-ABL1 inhibitors demonstrates in vivo efficacy

(A) Representative in vivo bioluminescence of mice at and during time of treatment. Derived cell lines with either BCR-ABL1 WT or V299L was tail-vein injected into immunocompetent recipient mice. Initial imaging was performed at day 10 post transplantation. Mice were subsequently treated once daily with vehicle, 10 mg/kg dasatinib, 50 mg/kg imatinib, 50 mg/kg vandetanib, or 50 mg/kg foretinib. (B) Fold change in total whole-mouse bioluminescence signal between post and pre- treatment. Mice bearing BCR-ABL1 V299L ALLs showed significant tumor burden reduction upon treatment with foretinib or vandetanib. Statistical significance determined by Mann-Whitney test. * P < 0.05 and ** P < 0.01. (C–D) Spleen from the same cohort of mice was also imaged and weighted (4 days and 8 days post initial treatment for BCR-ABL1 WT and V299L, respectively). Treatment of BCR-ABL1 V299L in vivo with foretinib and vandetanib showed strong reduction in spleen size. Scale bar indicates 1 cm. Statistical significance determined by Mann-Whitney test. ** P < 0.01. (E) Kaplan-Meier overall survival of immunocompetent recipient mice transplanted with BCR-ABL1 WT or V299L. Treatment of mice with foretinib or vandetanib led to significant extension in overall survival. Data presented was compiled from two independent injection experiments. Statistical significance determined with log-rank rest. *** P < 0.001.

Figure 6

Figure 6. Sequential drug switching changes clonal trajectories. (A) Experimental setup of sequential drug selection

Parental ALL cell lines were initially selected with drug A at IC90 1x the concentration. Upon outgrowth, drug B was used at dose escalating concentrations for continued resistance selection. (B) Pharmacological profile depicting collateral resistance and/or sensitivity for recovered cells upon initial selection with drug A. Heatmap shows the log2 transform of the ratio in EC50s between given representative cell line with unique ABL1 mutation and the DMSO control cell line (similar as the parental). (C) Pharmacological profile for representative cell lines with initial unique ABL1 mutations following drug A → drug B selection. The sequence of drugs used for selection can diversify resulting resistant ABL1 mutations.

Figure 7

Figure 7. Small molecule screen reveals compounds with diverse fitness landscapes

(A) High-throughput small molecule screen with 391 compounds against parental, BCR-ABL1 V299L, BCR-ABL1 V299L/E255K derived ALL cell lines, as a model of the initial, intermediate, and terminal stages of clonal evolution. Plot shows the log2 ratio in EC50 between V299L and parental (for x-axis) and between V299L and V299l/E255K (for y-axis). Data points colored in orange are BCR-ABL1 inhibitor positive controls, and in red are other positive controls. (B–G) Conceptual fitness landscapes with predefined positions for the three clonal stages. Height of the peak/valleys determined based on actual EC50 values for given drug and cell line.

Similar articles

Cited by

References

    1. Andersson DI, Hughes D. Antibiotic resistance and its cost: is it possible to reverse resistance? Nat Rev Microbiol. 2010;8:260–271. - PubMed
    1. Andreu EJ, Lledó E, Poch E, Ivorra C, Albero MP, Martínez-Climent JA, Montiel-Duarte C, Rifón J, Pérez-Calvo J, Arbona C, et al. BCR-ABL induces the expression of Skp2 through the PI3K pathway to promote p27Kip1 degradation and proliferation of chronic myelogenous leukemia cells. Cancer Res. 2005;65:3264–3272. - PubMed
    1. Bhang HC, Ruddy Da, Krishnamurthy Radhakrishna V, Caushi JX, Zhao R, Hims MM, Singh AP, Kao I, Rakiec D, Shaw P, et al. Studying clonal dynamics in response to cancer therapy using high-complexity barcoding. Nat Med 2015 - PubMed
    1. Boulos N, Mulder HL, Calabrese CR, Morrison JB, Rehg JE, Relling MV, Sherr CJ, Williams RT. Chemotherapeutic agents circumvent emergence of dasatinib-resistant BCR-ABL kinase mutations in a precise mouse model of Philadelphia chromosome-positive acute lymphoblastic leukemia. Blood. 2011;117:3585–3595. - PMC - PubMed
    1. Chen G, Mulla WA, Kucharavy A, Tsai HJ, Rubinstein B, Conkright J, McCroskey S, Bradford WD, Weems L, Haug JS, et al. Targeting the Adaptability of Heterogeneous Aneuploids. Cell. 2015;160:771–784. - PMC - PubMed

Publication types

MeSH terms

Substances

Grants and funding

LinkOut - more resources