A chromatin-mediated reversible drug-tolerant state in cancer cell subpopulations

Abstract

Accumulating evidence implicates heterogeneity within cancer cell populations in the response to stressful exposures, including drug treatments. While modeling the acute response to various anticancer agents in drug-sensitive human tumor cell lines, we consistently detected a small subpopulation of reversibly "drug-tolerant" cells. These cells demonstrate >100-fold reduced drug sensitivity and maintain viability via engagement of IGF-1 receptor signaling and an altered chromatin state that requires the histone demethylase RBP2/KDM5A/Jarid1A. This drug-tolerant phenotype is transiently acquired and relinquished at low frequency by individual cells within the population, implicating the dynamic regulation of phenotypic heterogeneity in drug tolerance. The drug-tolerant subpopulation can be selectively ablated by treatment with IGF-1 receptor inhibitors or chromatin-modifying agents, potentially yielding a therapeutic opportunity. Together, these findings suggest that cancer cell populations employ a dynamic survival strategy in which individual cells transiently assume a reversibly drug-tolerant state to protect the population from eradication by potentially lethal exposures.

Copyright 2010 Elsevier Inc. All rights reserved.

Figures

Figure 1. Detection of a drug-tolerant subpopulation of cancer cells

(A) Survival curve describing the viability of PC9 NSCLC cells treated with the indicated concentrations of the EGFR TKI gefitinib for 72 hours (similar results were observed with erlotinib; Fig S1). Each data point represents the average value from 4 samples, and is expressed as a percentage of surviving cells relative to untreated controls. The dashed line corresponds to 50% cell killing. The grey arrow indicates the concentration of EGFR TKI (2μM) used to generate DTPs. (B) PC9 cells were plated and left either untreated (left), or treated with 2μM erlotinib (ERL) for 9 days (middle) or re-treated with drug every 3 days for 33 days (right). At the appropriate times, cells were fixed and stained with Giemsa (upper panel), or representative microscopic images were photographed (middle panel). In the lower panel, cells from the three different conditions were replated at equal densities, and 20 hours post-plating, cells were analyzed by FACS to determine the percentage of cells in the indicated cell cycle phases. Data are based on two independent experiments performed in duplicate. (C) NSCLC-derived cell lines PC9 and HCC827 (EGFR mutation ΔE746-A750) were treated with erlotinib (2μM). The melanoma-derived cell line M14 and the colorectal cancer-derived line Colo-205 (BRAF mutation V600E) were treated with the RAF kinase inhibitor AZ628 (2μM). The breast cancer cell lines MDA-MB175v2, SKBR3 and HCC1419 (activated Her2) were treated with Lapatinib (2μM). The gastric cancer-derived cell line KATO II (MET amplification) was treated with the MET inhibitor PF-2341066 (1μM). Single cell-derived PC9 clones are designated PC9 cl.A,B,C. Following 9 days of treatment (fresh drug was added every 3 days), surviving DTPs were quantified. Each experiment was performed in triplicate. (D) Survival curves describing PC9 cells and several PC9-derived DTEP clones generated by selection in 2μM gefitinib, and then treated with the indicated gefitinib concentrations for 72 hours. Data represent average values determined from four identically treated samples. Data are expressed as a percentage of surviving cells relative to untreated controls. The dashed line corresponds to 50% cell killing. (E) (Upper panels) Lysates from PC9 cells treated with increasing concentrations of erlotinib (0.01, 0.1 and 1μM) (left panel) or parental PC9 cells and PC9-derived DTPs in 2μM erlotinib (right panel) were analyzed by immunoblotting to detect phosphorylated EGFR (pEGFR) and total EGFR. (Lower panel) Lysates from PC9 cells and gefitinib-tolerant DTEPs after 3 passages in drug-free medium, and subsequently treated for 2h with the indicated gefitinib concentrations, were analyzed by immunoblotting to detect phospho-EGFR and total EGFR. (F) PC9 cells and PC9-derived DTEPs were either untreated (0) or treated with the indicated concentrations of cisplatin for 72h, after which cell numbers were determined. Each experiment was performed in triplicate, and the percentage of DTPs (relative to untreated controls) is presented. Error bars represent standard deviations from the mean.

Figure 2. Heterogeneity and reversibility in the drug-tolerant cell populations

(A) PC9 cells and PC9-derived DTPs were processed for immunofluorescence using anti-CD133 and counterstained with Hoechst to visualize nuclei. Magnification, 20X. (B) Cell lysates from PC9 cells and PC9-derived DTPs were analyzed by immunoblotting with anti-CD133, and anti-GAPDH as loading control. (C) PC9 cells and PC9-derived DTPs were labeled with CD24 antibody conjugated to PE and a CD44 antibody conjugated to APC and analyzed by FACS. Note the enrichment of cells with surface expression of CD24 in DTPs (quantitation is shown in Fig. S4). (D) Uncloned PC9 cells or cells from two different single cell-derived PC9 clones (A and B) were treated with 2μM erlotinib for 33 days and then Giemsa stained. (E) Survival curves of PC9 cells and PC9-derived DTPs following recovery and re-expansion in drug-free medium and subsequent exposure to the indicated erlotinib concentrations for 72 hours, demonstrating the reversibility of drug tolerance. Each data point represents the average value determined from four samples. Data are expressed as percent surviving cells relative to untreated controls. Error bars represent standard deviations from the mean. The dashed line corresponds to 50% cell killing. (F) Survival curves of PC9 cells and a PC9-derived DTEP line, GR7, after gefitinib withdrawal for the indicated number of passages (P) and subsequent exposure to the indicated gefitinib concentrations for 72 hours. Each data point represents the average value determined from four samples. Data are expressed as percent surviving cells relative to untreated controls. Error bars represent standard deviations from the mean. The dashed line corresponds to 50% cell killing. (G) PC9 cells or PC9-derived DTEPs after gefitinib withdrawal for the indicated number of passages were either untreated or treated with 2 μM gefitinib (GEF) for 48 h, after which cells were subjected to FACS analysis. Presented in the graph is the percentage of dying cells in the population (sub-G1 DNA content). Error bars represent standard deviations from the mean from duplicate plates.

Figure 3. Chromatin alterations and a requirement for the KDM5A histone demethylase in the drug-tolerant state

(A) Non random expression differences between PC9 cells and PC9-derived DTEPs, indicative of global chromatin alterations. (Left panel) Non-random chromosomal arrangement of differentially up-regulated and down-regulated genes from an array analysis of PC9 cells versus PC9-derived DTEPs was revealed by counting the number of neighboring differentially expressed genes that are either both up-regulated or both down-regulated (104 total pairs examined). The distribution on the graph indicates the expected frequency of similarly regulated neighboring genes, assuming no relationship between expression status and chromosomal configuration, and the arrow indicates the observed number of neighboring genes that are similarly regulated. The p-value reflects the statistical likelihood that the observed number is different than the expected number. (Rght panel) Vertical ticks show the locations along different chromosomes of genes differentially expressed between PC9 cells and PC9-derived DTEPs (tolerant to either gefitinib or erlotinib). Genes whose expression is elevated in both the gefitinib- and erlotinib-tolerant PC9 cells (relative to untreated PC9) are shown above the horizontal lines corresponding to the chromosomes and genes whose expression is attenuated in these cells are shown below the horizontal lines. Individual genes are represented by vertical ticks of equal width. Consequently, ticks that appear wide represent a closely spaced cluster of genes. (B) The left panel shows increased expression of KDM5A in PC9-derived DTPs, relative to ERK1/2 as a loading control. The right panel shows the decrease in histone H3K4 methylation (H3K4 me3/2) in PC9-derived DTPs compared to parental PC9 cells either untreated (−) or treated for 20 hours with 2μM gefitinib (GEF), with total histone H3 and ERK1/2 serving as loading controls. (C) (Upper panel) Immunoblot demonstrating reduced expression of KDM5A in PC9 cells lentivirally-infected with shRNA targeting KDM5A (KDM5A k/d) compared to PC9 cells infected with a control shRNA (C). (Lower panel) Cell cycle profiles of these cells as assessed by FACS. Bar graphs represent 2 experiments performed in duplicate. (D) PC9 cells infected with control or KDM5A (KDM5A k/d) shRNAs were plated at equal density and treated with 2μM gefitinib (GEF) for 30 days with media/drug changes every 3 days. Surviving cells were counted (lower panel) or Giemsa stained and photographed (upper panel). Graphs represent two independent experiments performed in duplicate. (E) PC9 cells transduced with control or KDM5A (KDM5A k/d) shRNAs were plated at equal density and then treated with 2.5μM cisplatin with media changes ever 2 days. Plates were Giemsa stained 12 days post-treatment.

Figure 4. Drug-tolerant cancer cells are sensitive to HDAC inhibition

(A) (Upper panel) Immunoblot demonstrating reduced H3K14 acetylation in PC9-derived DTPs treated with 2μM gefitinib. GAPDH serves as a loading control. (Lower panel) Immunoblots to detect histone H3K14 acetylation in PC9 cells and PC9-derived DTEPs either untreated (−) or treated for 24h with 1μM of gefitinib (+GEF). Total histone H3 (H3) and ERK1/2 serve as loading controls. (B) (Upper panel) Sensitivity of PC9 cells and PC9-derived DTPs to Trichostatin A (TSA). Cells were plated at the same density and either untreated or treated with the indicated TSA concentrations for 20 hours, after which cells were subjected to FACS analysis. Graphs indicate the percentage of dying cells in the population (sub-G1 DNA content), based on 2 independent experiments. (Lower panel) TSA sensitivity of PC9 cells, PC9-derived DTEPs and re-sensitized DTEPs (cells that re-acquired sensitivity to EGFR TKIs following 30+ passages in drug-free medium). Cells were either untreated or treated with the indicated TSA concentrations for 24 h, after which they were subjected to FACS analysis. Presented in the graph is the percentage of dying cells with sub-G1 DNA content, based on 3 independent experiments. Error bars represent standard deviations from the mean. (C) (Upper panel) TSA sensitivity of M14-derived DTPs. Parental M14 melanoma cells and M14-derived DTPs were plated at equal densities and were either untreated (−), treated with 2μM AZ628 or 100nM TSA for 36 hours, after which they were subjected to FACS analysis. Indicated in the graph is the percentage of cells that display a sub-G1 DNA content indicative of apoptotic cells. The values presented are the average of 2 separate experiments. Error bars represent standard deviations from the mean. (Lower panel) TSA sensitivity of M14 and M14-derived DTEPs to AZ628 and TSA. M14 cells or M14-derived DTEPs were plated at equal density and were either untreated (−), treated with 2μM AZ628 or 100nM TSA, and then analyzed as described above. The values presented are the average of 2 separate experiments. Error bars represent standard deviations from the mean. (D) COLO 205 colorectal cancer cells or cisplatin-tolerant COLO 205 cells were plated at equal density and were either untreated (−) or treated with 50 nM TSA. 36 hours post-plating cells were subjected to FACS analysis. Indicated in the graph is the percentage of cell with a sub-G1 DNA content indicative of apoptosis. The values presented are the average of 2 separate experiments. Error bars represent standard deviations from the mean. (E) Model describing the emergence and reversibility of drug-tolerant cancer cells from a largely drug-sensitive population. (F) Lysates from PC9-derived DTPs either untreated (−) or treated with 50nM TSA for 20 hours were analyzed by immunoblotting using antibodies directed against γH2AX and ERK1/2 as a loading control. (G) PC9 cells and PC9-derived DTEPs were either untreated (−), treated for 24h with 1μM gefitinib (+GEF) or 50 nM TSA (+TSA), in the presence or absence of the caspase inhibitor, Z-VAD, and lysates were analyzed by immunoblotting using antibodies directed against γ-H2AX. p44/42 ERK was immunoblotted as a loading control. (H) PC9 cells and PC9-derived DTEPs were either untreated, treated with 2mM caffeine, 75nM TSA alone or, 75nM TSA + caffeine for 36h, after which cells were subjected to FACS analysis. Indicated on the plot is the percentage of cells displaying sub-G1 DNA content (apoptotic cells; top histogram) and the percentage of cells in S phase (bottom histogram).

Figure 5. Preventing the establishment of drug-tolerant clones

(A) PC9 cells were either untreated or treated singly with the indicated pharmacological agents for 6 days (top rows) or with erlotinib (ERL) alone or the combination of the indicated pharmacological agent with erlotinib for 33 days (bottom rows). Fresh media with drugs was provided every three days. Following treatment, plates were fixed and Giemsa stained. All experiments were performed in triplicate and representative plates are shown. (B) Individual colonies were counted and the quantified results were graphed. In some cases the colonies were too numerous to count (indicated as >500 colonies). Results reflect the mean and standard deviation from triplicate samples. (C) PC9 cells were either treated with 2μM of erlotinib (ERL) for 25 days or pre-treated for 9 days with 20nM of TSA and then treated with 2μM of erlotinib for 25 days. In parallel, PC9 cells were also treated continuously with 20nM TSA for 25 days (~ 4 passages). After 25 days, plates were fixed and Giemsa stained. The experiment was performed in triplicate and representative plates are shown.

Figure 6. Inhibition of HDAC activity disrupts tolerance to a variety of cancer drugs

(A) The EGFR-mutant NSCLC cell line NCI-HCC827 was either untreated, treated with TSA (50nM TSA), erlotinib (2μM ERL), or erlotinib plus TSA (2μM ERL + 50nM TSA). After 6 days, the untreated and TSA-treated plates had reached confluence and were Giemsa stained. The erlotinib and erlotinib + TSA-treated plates were similarly fixed and stained after 19 days of treatment. BRAF-mutant M14 melanoma cells were either untreated, treated with TSA (100nM TSA), treated with AZ628 (2μM AZ628), or treated with AZ628 + TSA (2μM AZ628 + 100nM TSA). After 6 days, the untreated and TSA-treated plates had reached confluence and were Giemsa stained. The AZ628 and AZ628 + TSA treated plates were similarly fixed and stained after 39 days of treatment. PC9 cells were either left untreated, treated with TSA (20nM TSA), cisplatin (2μM CIS), or cisplatin plus TSA (2μM CIS + 20nM TSA). After 6 days, the untreated and TSA treated plates had reached confluence and were Giemsa stained. The CIS + TSA treated plates were similarly fixed and stained after 45 days of treatment. Drug treatments were repeated every three days. The experiment was performed in triplicate and representative stained plates are shown. (B) Quantification of the lower panels in (A) expressed as the number of drug-tolerant colonies observed, with erlotinib-, AZ628- and cisplatin-only treated cells shown in black and the combination of these drugs with TSA shown in grey. Error bars represent the standard deviation from the mean (n=3).

Figure 7. IGF-1R activity is required for the chromatin state in drug tolerant cells

(A) Immunoblot demonstrating phosphorylation of IGF-1R and expression of IGFBP3 in DTPs versus parental PC9 cells. Total IGF-1R is shown as a loading control. (B) Immunoblot of extracts from an erlotinib-treated (or untreated) mouse lung tumor from transgenic mice expressing an activating EGFR mutant. After 2 days of treatment, EGFR phosphorylation is reduced, and phospho-IGFR is detected in residual tumor material. ERK1/2 is shown as a loading control. (C) Immunoblot demonstrating reduced phosphorylation of IGF-1R following treatment with 1μm of the IGF-1R inhibitor AEW541 for 2 hours. ERK1/2 is used as a loading control. (D) Photomicrographs of PC9-derived DTPs following 4 rounds of treatment (12 days) with 1μM gefitinib (GEF), 1μM gefitinib plus 1μM AEW541, or 1μM gefitinib plus 50ng/ml human IGF-1. Representative fields are shown. (E) PC9 cells were treated with 2.5μM cisplatin or 2.5μM cisplatin in combination with 1μM AEW541. After 30 days the plates were Giemsa stained. A representative example is shown. (F) M14 melanoma cells were treated with 2μM AZ628 or 2 μM AZ628 in combination with 1μM AEW541 in triplicate. After 30 days cells were counted and the bar graphs reflect the number of remaining cells per dish. Error bars represent standard deviations from the mean. (G) PC9 cells and PC9 cells in which KDM5A was stably depleted by shRNA (k/d) were either untreated or treated with gefitinib (GEF) or gefitinib plus IGF-1 for 48 hours and then analyzed by FACS to determine the percentage of cells with a sub-G1 content (apoptotic). Graphs reflect the average of two independent experiments performed in duplicate. Error bars represent standard deviations from the mean. (H) Immunoblot of extracts from parental PC9 cells or PC9-derived DTPs plated at equal density and treated with 1μM of AEW541 for 16 hours. Note that H3K4me2 (dimethylated) levels increase upon addition of AEW541 to PC9-derived DTPs. (I) Immunoblot of lysates from PC9-derived DTPs either untreated or treated with 1μM of AEW541 for 24 hours. ERK1/2 is shown as a loading control. (J) PC9 cells, PC9-derived DTEPs (gefitinib-treated), or PC9-derived DTEPs (gefitinib- and AEW541-treated) were either untreated or treated with gefitinib (GEF) or TSA for 48 hours and then BrdU labeled and analyzed by FACS to determine the percentage of cell with a sub-G1 content (apoptotic). Graphs reflect the average of two independent experiments performed in duplicate. Error bars represent standard deviations from the mean.