PTEN Is Fundamental for Elimination of Leukemia Stem Cells Mediated by GSK126 Targeting EZH2 in Chronic Myelogenous Leukemia

Chronic myelogenous leukemia (CML) is a malignant hematopoietic disorder originated from the pluripotent hematopoietic stem cells (HSCs) transformed by BCR–ABL fusion oncogene (1). There are three clinical phases in patients with CML: chronic phase (CP), accelerated phase, and terminal phase blast crisis (1). Tyrosine kinase inhibitor (TKI) imatinib mesylate (IM), the first-line drug for the treatment of CML (2), is highly effective and has greatly improved the event-free survival of patients with CP-CML (3). However, acquired resistance to IM can emerge, mainly owing to BCR–ABL mutations (e.g., T315I, G250E, Q252H, Y253H, and E255K/V), leading to disease relapse and progression (1, 2, 4). The second generation (e.g., nilotinib and dasatinib) and the third generation (ponatinib) of TKIs have been developed to override those point mutations in BCR–ABL, which have achieved a substantial curative effect in BCR–ABL mutational resistant patients (5, 6). Leukemia stem cells (LSCs), a rare population of CML cells with the traits of quiescence, self-renewal, and the differentiation disorder, are insensitive to TKIs (7). LSCs are believed at root to be another source of either primary or acquired resistance to IM (8). Accumulating evidence has demonstrated that BCR–ABL kinase-independent LSCs still persist even in patients with complete molecular response; such a small subset of LSCs may eventually lead to CML relapse (9, 10). Therefore, eradication of LSCs may be an effective strategy to cure CML.

The maintenance of LSCs is independent of BCR–ABL kinase activity (11), but regulated by multiple signaling pathways including the developmental signaling pathways (e.g., Wnt/β-catenin, Notch, Hedgehog), epigenetic regulators (e.g., SIRT1, PRMT5, EZH2), transcriptional factors (e.g., p53, NF-κB, FOXM1), cell metabolism regulators (e.g., Slc15A2, Alox5), and other cell regulators (e.g., PPARγ, Gas6/AXL, IL1RAP, PTEN; refs. 12–20). However, the signaling pathways that are responsible for LSC maintenance remain to be fully elucidated.

Enhancers of zeste homologue 2 (EZH2), a core catalytic subunit of polycomb repressive complex (PRC2), can trimethylate histone H3 at lysine 27 (H3K27) to silence gene transcription (21). EZH2 is aberrantly overexpressed in multiple types of cancer and correlates with poor prognosis (22–25). Moreover, recent reports have documented that depletion of EZH2 by genetic knockout eliminates CML LSCs and sensitizes them to IM treatment (26, 27). However, the mechanism by which EZH2 regulates stemness of LSCs remains unclear.

GSK126, an S-adenosylmethionine competitor, which can inhibit both wild-type and mutant EZH2, is a 150-fold selective inhibitor against EZH2 over EZH1 (25). The preliminary outcome of ongoing phase I clinical trial of GSK126 for relapsed/refractory lymphoma and solid tumors is encouraging (25). In this study, our purpose was to evaluate the effects of EZH2 inhibition by GSK126 or shRNA on CML LSCs and explore its underlying mechanism. Our results showed that pharmacologic inhibition of EZH2 by GSK126 not only significantly induced cell growth retardation and apoptosis in CML bulk leukemia cells, but also reduced CML LSCs in vitro and in vivo. Meanwhile, silencing EZH2 by specific shRNA elicited elimination of CML LSCs. Mechanistically, PTEN functioned at least in part as a mediator for the inhibitory effects of EZH2 repression on CML LSCs. The findings may facilitate the understanding of the regulatory role of EZH2 in CML LSC maintenance.

GSK126 (purity > 95%, HPLC) was purchased from Hope Chem Co., Ltd. IM was from Selleck. Annexin V–FITC and propidium iodide (PI) were from Sigma-Aldrich. GSK126 was dissolved in dimethylsulfoxide (DMSO) at 20 mmol/L, and IM was dissolved in DMSO at 10 mmol/L and stored at −20°C.

CML cells were cultured as previously reported (8). K562 cells purchased from the American Type Culture Collection (ATCC) were grown in RPMI1640 medium (Invitrogen) supplemented with 10% FBS (Biological Industries); KBM5 and KBM5-T315I kindly provided by Dr. Sai-Ching J. Yeung (University of Texas M.D. Anderson Cancer Center, Houston, TX) were grown in IMDM medium (Invitrogen) with 10% FBS; 293T and Plat-E cells were from the ATCC and cultured in DMEM (Invitrogen) with 10% FBS at 37°C humidified incubator containing 5% CO₂ as described previously (14). All the cell lines were tested and authenticated by using short tandem repeat matching analysis every 6 months. No mycoplasma contamination was detected.

Peripheral blood (PB) or bone marrow (BM) samples were obtained from patients with CML (Supplementary Table S1) and from healthy adult donors in the First Affiliated Hospital of Jinan University, The First Affiliated Hospital of Sun Yat-sen University, Sun Yat-sen Memorial Hospital, and Guangdong Provincial People's Hospital after informed consent according to the institutional guidelines and the Declaration of Helsinki principles. The primary human CD34⁺ cells were sorted by using a MACS bead kit (Miltenyi Biotec) as described previously (8, 14, 17).

Cell viability was detected by MTS assay (CellTiter 96 Aqueous One Solution reagent; Promega) as described previously (28). Briefly, 2 × 10⁴ cells plated in 50-μL culture medium in 96-well plates were exposed to varying concentrations of GSK126 for 72 hours. MTS (20 μL/well) was added to the culture and incubated for 4 hours at 37°C humidified incubator before the termination of treatment, and optical density was read with a wavelength of 490 nm. IC₅₀ value was determined by curve fitting of the sigmoidal dose-response curve.

The colony-forming capacity of CML cells was measured by use of a mono-layer soft-agar system as described (28).

Whole-cell lysates were prepared in RIPA buffer as described previously (8). To detect the level of cytochrome c in the cytosol, the cytosolic fraction was prepared with digitonin extraction buffer (10 mmol/L PIPES, 0.015% digitonin, 300 mmol/L sucrose, 100 mmol/L NaCl, 3 mmol/L MgCl₂, 5 mmol/L EDTA, and 1 mmol/L phenylmethylsulfonyl fluoride; ref. 29). Detailed information for antibodies and their sources is described in Supplementary Materials and Methods.

The CML cells were pretreated with GSK126 (25 μmol/L) at the indicated times, cells were harvested, incubated with MitoTracker probes (CMXRos and MTGreen; Invitrogen) at 37°C for 1 hour, and the inner mitochondrial transmembrane potential (⁠Ψm) was detected by flow cytometry as described previously (14, 30).

Scramble (pLKO.1-puro-nontarget shRNA) or human EZH2-specific target shRNA (pLKO.1-puro-hEZH2-target shRNA) were from Sigma-Aldrich (Supplementary Table S2). Lentivirus supernatants were produced by 293T cells as described previously (28). Virus-containing supernatants were collected and purified with 0.45-μm filter. CML cells were transduced by spinoculation (1,500 g, 90 minutes, 32°C) with virus-containing supernatants for two rounds and then incubated with puromycin (1 μg/mL) for 5 days to establish stable clones to silence EZH2.

After being treated with increasing concentrations of GSK126 for 24 hours, purified human CML CD34⁺ cells (5, 000/well) were seeded in 96-well plates supplemented with drug-free methylcellulose medium (MethoCult H4434; StemCell Technologies; ref. 8). On day 14, colonies consisting of ≥50 cells were counted. The cells derived from these colonies were replated for the secondary and tertiary rounds; colonies were counted on day 14 after each replating.

Retrovirally BCR–ABL-driven mouse model was employed as described in our previous report (8). Detailed information is described in Supplementary Materials and Methods.

The in vivo frequency of LSCs was measured as described previously (14, 31). Briefly, BM and spleen cells from CML mice were transfected with scramble or shEZH2 lentivirus and transplanted into C57BL/6 recipient mice (irradiated at 550 cGy), and the CML mice were administrated with IM (100 mg/kg/day, gavage) or vehicle (double distilled water). Two weeks later, BM cells were harvested. The percentage of GFP⁺ nucleated cells in BM from the mice of control versus treated group was detected prior to transplantation, and a ration was calculated. Serial numbers of BM cells (2 × 10⁶, 1 × 10⁶, or 5 × 10⁵ cells/mouse) from treated mice were mixed with 2 × 10⁵ normal BM cells/mouse and were then transplanted into secondary recipient C57BL/6 mice irradiated at 550 cGy. Sixteen weeks later, GFP⁺ cells in the PB was monitored and normalized with the aforementioned ratio. The mice in which this adjusted percentage of GFP⁺ cells reached more than 0.5% was defined positive transplantation. The frequency of LSCs was calculated using Poisson statistics online from the Bioinformatics facility of The Walter & Eliza Hall Institute of Medical Research (32).

The qRT-PCR experiments were carried out as reported previously (8). Detailed information is described in Supplementary Materials and Methods.

Chromatin immunoprecipitation (ChIP) was performed using the EZ-ChIP Kit (EMD Millipore) according to our previous procedure (8). Briefly, K562 cells (1 × 10⁷) were cross-linked by formaldehyde for 10 minutes; glycine was added to quench unreacted formaldehyde. The cells were then washed, lysed, and sonicated to shear the cross-linked DNA to 200 to 1,000 bp. The cross-linked protein–DNA was immunoprecipitated with anti-EZH2 and anti-H3K27me3 antibodies (1:200) or normal rabbit IgG (negative control) at 4°C overnight with rotation. After brief centrifugation, pellets were harvested, washed, and protein–DNA complexes were eluted. The crosslink of protein–DNA complexes was reversed by 5-hour incubation at 65°C with proteinase K. DNA was purified by using spin columns, and the purified DNA was subjected to qRT-PCR assay. The primers for amplification of purified DNA are as follows. PTEN: (F) AGCAAGCCCCAGGCAGCTACACT, (R) GGTAGGAGGGGGCAGAGCGGTA (33).

All experiments were repeated three times, and data were expressed as mean ± SD. Comparisons between two groups were analyzed by two-tailed Student t test; differences among multiple groups were analyzed by one-way ANOVA with post hoc comparison by the Tukey test. GraphPad Prism 5.0 software (GraphPad) was used for statistical analysis. P < 0.05 was considered statistically significant.

With an approach of lentiviral shRNA, we first found that K562, KBM5, and KBM5-T315I cell clones with EZH2 stably silenced displayed a significant reduction in cell growth (Supplementary Fig. S1A and S1B) as well as clonogenicity (Supplementary Fig. S1C), which is consistent with previous report (27). Given that GSK126 is a potent selective small-molecule EZH2 inhibitor (Supplementary Fig. S1D; ref. 25) that is under clinical trial, we determined the effect of GSK126 on IM-sensitive (K562 and KBM5) and -resistant (KBM5-T315I) CML cells. Western blot analysis showed that GSK126 prominently diminished methylated histone 3 (H3K27me3) levels in these CML cells (Supplementary Fig. S1E and data not shown). GSK126 counteracted the cell viability in a dose-dependent manner, with IC₅₀ values of 11.3, 8.53, 8.65 μmol/L in K562, KBM5, and KBM5-T315I cells, respectively (Supplementary Fig. S1F). The inhibitory effect of GSK126 on growth of CML cells was further confirmed by clonogenicity assay (Supplementary Fig. S1G).

We next evaluated the apoptosis-inducing ability of GSK126 in CML cells. The results showed that GSK126 robustly induced apoptosis in a dose- and time-dependent manner in three CML cell lines as measured by Annexin V/PI dual staining assay (Supplementary Fig. S2A and S2B). Immunoblotting analysis revealed that GSK126 induced a dose- and time-dependent–specific PARP cleavage and caspase-3 activation in CML cells (Supplementary Fig. S2C and S2D), as well as a time-dependent release of cytochrome c into cytosol extracted by digitonin buffer (Supplementary Fig. S3A). The apoptosis-inducing ability of GSK126 in CML cells was further validated by depolarization of transmembrane potential (ΔΨm), as determined by CMXRos and MTGreen dual staining assay (Supplementary Fig. S3B). Western blot results showed that GSK126 displayed negligible effects on apoptosis-related family members (XIAP, Bcl-2, Bcl-X_L, and Mcl-1) but Survivin (Supplementary Fig. S3C). These data together reveal that GSK126 induces apoptotic cell death in IM-resistant and -sensitive CML cells.

After confirming that the mRNA level of EZH2 as examined by qRT-PCR analysis was much higher in the human primary CML CD34⁺ cells than that in NBM (normal BM from healthy donors) CD34⁺ cells (Fig. 1A), which is consistent with previous reports (26, 27), we next asked whether EZH2 inhibition by GSK126 would induce apoptosis in CML primitive/progenitor cells. Toward this end, purified CML CD34⁺ cells and NBM CD34⁺ cells were treated with GSK126 alone or in combination with IM for 24 hours. Flow cytometry analysis after dual staining of CD38-PE and Annexin V–FITC showed that GSK126 induced a remarkable increase in Annexin V–positive cells within the CD34⁺CD38⁻ population in CML patient samples (Fig. 1B), but not in NBM CD34⁺ cells (Fig. 1B). In an alternative set of experiments to evaluate the viable quiescent LSCs, purified primary human CML CD34⁺ cells were labeled with carboxyfluorescein diacetate succinimidyl ester (CFSE) followed by incubated with GSK126 alone or in combination with IM for 96 hours. The subsequent flow cytometry analysis revealed that GSK126 significantly led to a massive increase in Annexin V–positive cells in the CFSE-bright quiescent CML CD34⁺ population (Fig. 1C).

The colony-formation cells (CFC) and re-plating capacity were also drastically obviated in primary human CML CD34⁺ cells but not in NBM CD34⁺ cells after treatment with GSK126 for 24 hours (Fig. 1D). Collectively, these results suggest that pharmacologic inhibition of EZH2 by GSK126 impedes the survival and self-renewal capacities of human CML LSCs while sparing counterparts from NBM.

To assess the in vivo effect of GSK126 on CML LSCs, a retrovirally BCR–ABL-driven CML mouse model was employed (Fig. 2A). Splenomegaly and nodules in spleens were thwarted in the CML mice those received treatment with GSK126 or in combination with IM in comparison with those mice received from treatment with vehicle for 2 weeks (Supplementary Fig. S4A). Meanwhile, administration of GSK126 alone or in combination with IM for 2 weeks remarkably prolonged survival of CML mice (Fig. 2B). Of note, after a span of 39 days of drug discontinuation, three of 10 mice in the GSK126-treated group and five of 10 mice in the GSK126+IM combinationally treated group were still alive at the end point of observation, respectively (data not shown). This implicates that GSK126 or GSK126+IM may induce long-term sustained remission.

Flow cytometry analysis revealed that GSK126 alone or in combination with IM drastically reduced the population of leukemic BCR-ABL-GFP⁺ cells (Fig. 2C and Supplementary Fig. S4B) and GFP⁺ myeloid cells (Mac1⁺Gr-1⁺) in BM (Fig. 2D) and spleens (Supplementary Fig. S4C). Furthermore, the proportions of leukemia GFP⁺LSK cells (Lin⁻Sca-1⁺c-Kit⁺), GFP⁺LT-HSCs (LSK Flt3⁻CD150⁺CD48⁻), and GFP⁺ST-HSCs (LSK Flt3⁻CD150⁻CD48⁻) in BM (Fig. 2E–H) and spleen cells (Supplementary Fig. S4D–S4G) were significantly reduced in the mice administrated with GSK126 alone or with IM in CML mice when compared with those in the vehicle-treated mice.

Similarly, the proportions of leukemic granulocyte-macrophage progenitors (GFP⁺GMPs, GFP⁺Lin⁻Sca-1⁻c-Kit⁺CD34⁺FcγRII/III^hig) and common myeloid progenitors (GFP⁺CMPs, GFP⁺Lin⁻Sca-1⁻c-Kit⁺CD34⁺FcγRII/III^low) in BM cells (Fig. 2I–K) and splenic cells (Supplementary Fig. S4H–S4J) were greatly reduced in the mice those received administration of GSK126 alone or in combination with IM.

Taken together, these data indicate that pharmacologic inhibition of EZH2 by GSK126 markedly prolongs the survival of CML mice and reduces the pool size of LSCs and myeloid progenitors in vivo.

In order to characterize the in vivo regulatory role of EZH2 in CML LSCs, we assessed whether knockdown of EZH2 in the primary CML cells affected the in vivo frequency of LSCs and overall survival of the secondary recipient mice (Fig. 3A). qRT-PCR analysis validated that EZH2 mRNA level was conspicuously reduced after the second round of transduction with lentiviral mouse EZH2 shRNA prior to transplantation (Supplementary Fig. S5A). Western blot monitoring further confirmed an appreciably decreased EZH2 in the splenic cells from the secondary recipient mice received from transplantation of EZH2-silenced CML cells (Fig. 3B). The splenomegaly and its surface nodules were diminished by treatment with EZH2 knockdown alone or in combination with IM when compared with scramble control (Fig. 3C). Meanwhile, knockdown of EZH2 alone or in combination with IM conspicuously prolonged survival of CML mice (Fig. 3D). Furthermore, EZH2 knockdown alone or in combination with IM prominently reduced the populations of leukemic BCR-ABL-GFP⁺ cells (Fig. 3E and Supplementary Fig. S5B) and GFP⁺ myeloid cells (Mac1⁺Gr-1⁺; Fig. 3F and Supplementary Fig. S5C) in BM and spleen cells. These data suggest that EZH2 knockdown alone or in combination with IM greatly restrains the leukemogenesis and prolongs the survival in CML mice.

In such CML mice, the proportions of leukemic GFP⁺LSK cells, GFP⁺LT-HSCs, and GFP⁺ST-HSCs in BM and spleen cells were markedly decreased after treated with EZH2 knockdown alone or in combination with IM (Fig. 3G–J and Supplementary Fig. S5D–S5G).

In addition, the populations of GFP⁺GMPs and GFP⁺CMPs in BM cells (Fig. 3K–M) and splenic cells (Supplementary Fig. S5H–S5J) were significantly reduced after knockdown of EZH2 alone or with IM in CML mice.

Collectively, these data indicate that silencing EZH2 by shRNA prolongs the survival of CML mice and reduces the pool size of LSCs and myeloid progenitors in vivo.

When analyzing HSCs and progenitors out of the gated GFP⁻ population, the percentages of GFP⁻LSK cells, GFP⁻LT-HSCs, and GFP⁻ST-HSCs, as well as GFP⁻GMPs and GFP⁻CMPs in BM and spleen were not significantly decreased in the mice that received administration of GSK126 versus of vehicle control (Supplementary Fig. S6A and S6B) as well as in the secondary recipient mice that received transplantation of EZH2-silenced murine CML cells (Supplementary Fig. S6C). These results suggest that blocking EZH2 methyltransferase activity by GSK126 or targeting EZH2 by shRNA has minimal cytotoxic effect on normal hematopoiesis in the same CML mice.

We further quantify the frequency of CML LSCs with in vivo limiting dilution assay in CML mice (Fig. 4A). The results showed that knockdown of EZH2 alone or combined with IM significantly reduced CML LSCs frequency in vivo (Fig. 4B and Supplementary Table S3). Meanwhile, knockdown of EZH2 alone or combined with IM pronouncedly blocked the engraftment of GFP⁺ leukemia cells in the secondary recipient mice at 16 weeks (Fig. 4C), implicating the impaired secondary leukemogenesis.

To clarify the underlying action mechanism of EZH2 on CML LSCs, we detected several intrinsic regulators to modulate the destiny of CSCs in K562 cells (Fig. 5A and data not shown). Among them, only PTEN, p27, and c-Myc were observed to be considerably changed in K562 cells (Fig. 5A and data not shown). Given that PTEN can restrict survival and self-renewal of CML LSCs (19), and that EZH2 can bind to PTEN gene promoter to repress the transcription of PTEN in solid cancer cells (34, 35), we chose PTEN as a candidate EZH2 mediator in CML LSCs in the subsequent experiments. We noted that EZH2 knockdown significantly elevated the protein level of PTEN in all three lines of human CML cells (Fig. 5A) as well as in the murine LSK (Lin⁻Sca-1⁺c-Kit⁺) cells from CML mice (Fig. 5B). Because EZH2 has the capacity to trimethylate histone H3 at lysine 27 (H3K27) to silence genes transcription (36), we envisaged that EZH2 trimethylated H3K27 at the promoter of PTEN and thereby repressed its transcription. To test this hypothesis, we first detected the mRNA levels of PTEN in K562 cell clones with EZH2 stably silenced. The EZH2-silenced K562 cells exhibited a significantly increased PTEN transcription as evaluated by qRT-PCR when compared with scramble control (Fig. 5C). Similarly, pharmacologic inhibition of EZH2 by GSK126 elicited an elevation in mRNA level of PTEN gene in the purified primary human CML CD34⁺ cells in comparison with control (Fig. 5D). To further define the regulatory role of EZH2 in PTEN gene transcription in CML cells, we searched for a putative EZH2-binding site in the PTEN gene promoter (33). ChIP assay showed that knockdown of EZH2 disturbed the EZH2 recruitment to the promoter of PTEN gene in CML cells (Fig. 5E). We also observed a decrease in the H3K27me3 enrichment at the PTEN gene promoter in the EZH2-depleted CML cells (Fig. 5F). Taken together, it is plausible that EZH2 binds to the promoter of PTEN and repressing its gene transcription.

Because the constitutive activation of AKT/mTOR, major downstreams of PTEN, is critically involved in the survival of CML cells (37), we next determined the effect of EZH2 knockdown on the activity of AKT/mTOR signaling in CML cells. The results showed that the phosphorylation of the key components in the AKT/mTOR signaling [e.g., phospho-AKT (S473), phospho-AKT (T308), phospho-mTOR (S2448), phospho-p70S6K (T389), and phospho-S6 (S235/236)] was greatly restrained by knockdown of EZH2 in CML cells (Fig. 5G). Furthermore, the EZH2 knockdown–enabled AKT/mTOR signaling blockage was further obstructed by combinational treatment with IM in K562 cells (Fig. 5H).

Because inhibition of BCR–ABL could increase the expression of PTEN (19), we asked whether BCR–ABL was involved in the EZH2 knockdown–mediated PTEN transcriptional elevation. Immunoblotting analysis revealed that knockdown of EZH2 hardly affected the levels of total BCR–ABL and phospho-BCR–ABL (Y245) in all three lines of CML cells (Fig. 5G), which was in concord with the effect of EZH2 inhibition by JQEZ5 and UNC1999 on BCR–ABL in CML cells (27). Alternatively, knockdown of BCR–ABL by shRNA did not potentiate the EZH2 silencing–mediated elevation in PTEN transcription (Supplementary Fig. S7A). In aggregate, these data suggest that EZH2 represses PTEN gene transcription independently of BCR–ABL.

To further sharpen the role of PTEN in the EZH2 knockdown–mediated elimination of CML LSCs, we knocked down EZH2 alone or in combination with PTEN by the corresponding lentiviral shRNA in splenic cells harvested from the first generation of CML mice, and then transplanted these cells into the sublethally irradiated (550 cGy) C57BL/6 recipient mice (Fig. 6A). qRT-PCR analysis confirmed that EZH2 or PTEN mRNA level was drastically decreased after the second round of transduction with mouse EZH2, PTEN, or their combination shRNA lentivirus prior to transplantation (Supplementary Fig. S8A). Two weeks later, Western blot analysis also confirmed that the lentiviral shRNA against EZH2 or PTEN was remarkably effective (Fig. 6B). In contrast, the protein level of PTEN in the splenic cells was remarkably increased in the secondary recipient mice those received transplantation of EZH2 knockdown cells (Fig. 6B). The EZH2 knockdown–mediated blockage in splenomegaly and the nodule number appeared to be considerably attenuated by PTEN silencing (Supplementary Fig. S8B). Impressively, the knockdown of EZH2-mediated prolonged-survival effect in CML mice was impaired by concurrent shRNA transduction against PTEN (Fig. 6C). Furthermore, EZH2 knockdown–mediated reduction of the populations of leukemic BCR-ABL-GFP⁺ cells (Fig. 6D and Supplementary Fig. S8C) and GFP⁺ myeloid cells (Fig. 6E and Supplementary Fig. S8D) in BM and spleen cells was also crippled by concurrent shRNA transduction against PTEN.

Further analysis indicated that the EZH2 knockdown–mediated decrease in the subpopulations of leukemic GFP⁺LSK cells and GFP⁺LT-HSCs in BM and spleen cells was at least partially reversed by PTEN knockdown (Fig. 6F–I and Supplementary Fig. S8E–S8H).

In addition, the EZH2 knockdown–mediated obviation of the populations of leukemic GFP⁺GMPs and GFP⁺CMPs in BM cells (Fig. 6J and K) and splenic cells (Supplementary Fig. S8I–S8K) was at least partially counteracted by simultaneous PTEN knockdown. Taken together, these data reveal that EZH2 mediates prolonged survival and LSC's reduction at least in part through the elevation of PTEN.

In the present study, we discovered that inhibition of EZH2 by its methyltransferase activity–specific inhibitor GSK126 was efficient in repressing the growth, inducing apoptosis and impairing LSC's survival and self-renewal of primary CD34⁺ CML cells but sparing those from NBM. Silencing EZH2 by shRNA lentivirus greatly reduced growth and frequency of CML LSCs in CML mice. EZH2 inhibition by GSK126 or specific shRNA remarkably alleviates splenomegaly and prolongs the survival of CML mice. Mechanistically, EZH2 knockdown significantly elevated the expression of PTEN in CML cells and in LSKs of CML mice. Knockdown of PTEN at least partially reversed the inhibitory effects of EZH2 knockdown on LSCs in CML mice.

We found that EZH2 mRNA level was much higher in the CML CD34⁺ cells than NBM CD34⁺ cells, which was consistent with the previous reports (26, 27). Accumulating evidence indicates that EZH2 supports aberrant survival and self-renewal of CSCs in a wide spectrum of cancer such as breast, MLL-AF9⁺ AML, and glioblastoma (22–24). In consensus, the body of evidence from us and others during preparation of this article (26, 27) demonstrated that the catalytic activity of EZH2 is required for survival and self-renewal of CML LSCs. Inhibition of EZH2 by GSK126, a highly selective inhibitor of the catalytic activity of EZH2, significantly cleared the quiescent LSCs and hampered their self-renewal capacity in human CML LSCs and in CML mice.

GFP⁻HSCs and progenitors out of the gated GFP⁻ population may reflect the normal hematopoiesis in the CML mice. We found that the subpopulations of GFP⁻LSK cells, GFP⁻LT-HSCs, and GFP⁻ST-HSCs, as well as GFP⁻GMPs and GFP⁻CMPs in BM were not significantly decreased by targeting EZH2 via either GSK126 or specific lentiviral shRNA in mice (Supplementary Fig. S6). The results suggest that blocking EZH2 methyltransferase activity by GSK126 or biological depletion has minimal cytotoxic effect on normal hematopoiesis in the same CML mice. Similarly, despite regulating normal HSC function in a developmental-stage–specific manner, EZH2 is, however, not essential for the maintenance of normal HSCs, which indicates that pharmacologic inhibition of EZH2 may incur low toxic side effect (38, 39). Furthermore, our data showed that primary NBM CD34⁺ cells were spared from the deleterious effects upon GSK126 treatment, which was consistent with the effect of GSK343 as previously reported (26). Taken together, these data hinted that targeting EZH2 selectively kills LSCs while sparing HSCs.

PTEN, a tumor-suppressor gene with frequent loss-of-functional mutation in variety cancers, is an endogenous inhibitor of PI3K/AKT signaling pathway (40, 41). PTEN deletion results in an increased maintenance of quiescent LSCs and promotion of leukemogenesis of AML (42). Conversely, inhibition of the PI3K pathway by rapamycin attenuates such an outcome (42). In CML, PTEN has been demonstrated to restrict survival and self-renewal of LSCs (19). In our study, knockdown of EZH2 increased the expression of PTEN in CML bulk leukemia cells as well as in murine LSKs. In addition, knockdown of EZH2 led to decreased recruitment of EZH2 and H3K27me3 to the promoter of PTEN in CML cells. Furthermore, knockdown of PTEN attenuated the potency of EZH2 knockdown in LSCs in CML mice. Therefore, PTEN likely functions as a bridge between EZH2 and LSCs stemness. Of interest, BCR–ABL, the original driver to transform HSCs, appears not involved in the regulatory network composed of EZH2-PTEN-LSCs stemness.

The results from us (Fig. 5G and Fig. 5H) and others support that PI3K/AKT/mTOR signaling is regulated at least two different layers (i.e., BCR–ABL and PTEN; refs. 19, 43). Simultaneous interference at such two layers is predicted to provoke an enhanced effect in blocking of PI3K/AKT/mTOR signaling pathway. Indeed, the improved survival was observed in the CML mice cotreated with GSK126 and IM.

Considering the efficacy of knockdown, the reversed effect of EZH2 knockdown by PTEN is of importance. However, other factors to confer the effect of EZH2 knockdown on LSCs cannot be excluded. For instance, increased p27 and reduced c-Myc as revealed in our results (data not shown) may also contribute to the influence of EZH2 knockdown in CML LSCs. In addition, the elevation of p53, p16, and p21 may also be involved (26, 27).

In conclusion, EZH2 is required for survival and self-renewal of CML LSCs. Pharmacologic inhibition of EZH2 by GSK126 drastically inhibited the survival and self-renewal of CML LSCs and remarkably prolonged the survival of CML mice. EZH2 silencing-mediated CML LSC's suppression at least in part through the elevation of PTEN (proposed model, Fig. 6L). These findings improve the understanding of the molecular mechanism of EZH2 in CML LSCs, and EZH2 may be a potential target to eliminate CML LSCs.

No potential conflicts of interest were disclosed.

Conception and design: J. Zhou, J. Pan

Development of methodology: J. Zhou, C. Liu, Y. Jin

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): J. Zhou, D. Nie, J. Li, X. Du, Y. Lu, Y. Li, C. Liu, W. Dai, Y. Wang, Y. Jin

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): J. Zhou, Y. Jin, J. Pan

Writing, review, and/or revision of the manuscript: J. Zhou, J. Pan

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): J. Zhou

Study supervision: J. Pan

This study was supported by grants from National Natural Science Funds (No. U1301226 and No. 81373434 to J. Pan; No. 81473247 and No. 81673451 to Y. Jin); The Natural Science Funds of Guangdong Province for Distinguished Young Scholars (grant No. 2016A030306036 to Y. Jin); the Research Foundation of Education Bureau of Guangdong Province, China (grant cxzd1103 to J. Pan); and Natural Science Foundation of Guangdong province (grant 2015A030312014 to J. Pan).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.