Antitumor Effects of Blocking Protein Neddylation in T315I-BCR-ABL Leukemia Cells and Leukemia Stem Cells

Chronic myeloid leukemia (CML) is a myeloproliferative disease, which is characterized by Philadelphia chromosome (Ph⁺), the genetic translocation t (9; 22) (q34; q11.2), involving the fusion of the Abelson oncogene (ABL) with the breakpoint cluster region (BCR) gene encoding the BCR-ABL fusion oncoprotein with constitutive tyrosine kinase activity (1). BCR-ABL is the single driving force in pathogenesis of CML. Tyrosine kinase inhibitors (TKI) including imatinib, dasatinib, and nilotinib are current first-line therapy to treat CML (2). Approximately 80% of patients with chronic phase (CP)-CML achieve a complete cytogenetic remission within 1 year of therapy with imatinib (3). Even though imatinib can effectively control CML development, drug resistance may occur. Approximately 50% of imatinib-resistant patients are due to emerging CML clones harboring BCR-ABL kinase mutants (4). T315I, the most vicious mutant to imatinib, accounts for at least 15% of relapse cases of CML (5). The application of the third-generation TKI ponatinib against T315I-BCR-ABL may be limited because of its severe toxic side effects (e.g., arterial thrombosis, myocardial infarction, and cerebrovascular events; ref. 6). Therefore, resistance to imatinib remains a therapeutic challenge in CML.

Reasons of the rest 50% of imatinib-resistant patients are various. CML, defined as a stem cell disease, has characteristic features of leukemia stem cells (LSC; ref. 7). The maintenance of LSCs is independent of BCR-ABL kinase activity (8), which is considered as roots of TKI resistance and CML relapse. Thus, it is urgent to seek novel agents to eliminate LSCs for CML cure.

Protein neddylation is a posttranslational modification that adds an ubiquitin-like NEDD8 (neural precursor cell expressed, developmentally downregulated 8) to target proteins. Neddylation cascade is catalyzed by three NEDD8-specific enzymes, an E1 NEDD8-activating enzyme (NAE1), an E2-conjugating enzyme (UBC12), and one of the several E3 ligases (9). It is implicated in the regulation of stability and subcellular localization of proteins, gene transcription, and DNA damage response (DDR), which eventually controls diverse physiologic and pathologic cellular behaviors such as survival, differentiation, senescence, and apoptosis (10). Overactivation of neddylation pathway is involved in tumorigenesis and tumor progression (11). Furthermore, aberrant neddylation confers drug resistance in ovarian cancer and multiple myeloma (12, 13), which makes NAE1 inhibition a novel strategy to overcome drug resistance in these cancers.

MLN4924 (Pevonedistat), the first-in-class NAE1 inhibitor, blocks the entire neddylation modification of proteins including cullins, which results in accumulation of many cullin-RING ligase (CRL) substrates including phospho-IκBα, p27^kip1, and CDT1. MLN4924 treatment induces apoptosis, DNA damage, and cell-cycle arrest in multiple types of solid and hematopoietic malignant cells (14). Pevonedistat is currently under phase I clinical trials (15, 16). Whether neddylation inhibition by MLN4924 induces apoptosis in T315I-BCR-ABL⁺ cells is elusive.

After obtaining the observation that NAE1 was overexpressed in the primary CD34⁺CD38⁻ cell subpopulation from CML patients when compared with the counterparts from normal bone marrow (NBM), we hypothesized that the function of certain neddylation-dependent protein substrates might be targeted to therapeutic ends in imatinib-resistant CML cells and LSCs. In support of this plausibility, neddylation is reported to regulate mouse embryonic stem cells and self-renewal property of cancer stem-like cells (CSC) in nasopharyngeal carcinoma (17, 18). The purpose of this study was to investigate whether blocking neddylation pathway was active against BCR-ABL mutational imatinib-resistant cells, self-renewal capacity of LSCs in CML and the underlying mechanisms. Our results showed that MLN4924 possessed an inhibitory activity against wild-type (WT) - and T315I-BCR-ABL⁺ cells with WT-p53. MLN4924 treatment significantly eliminated the LSCs from CML patients and CML mice transformed by retrovirus BCR-ABL transduction. Overall, our findings provide a rationale for targeting T315I-BCR-ABL⁺ bulk tumor cells and LSCs to overcome resistance to imatinib by neddylation inhibition with MLN4924.

KBM5 and KBM5-T315I cells were kindly provided by Dr. Sai-Ching J. Yeung (University of Texas MD Anderson Cancer Center, Houston, TX) and cultured in Iscove's modified Dulbecco's medium (Thermo Fisher Scientific) with 10% FBS (Biological Industries; ref. 19). K562, KU812, M2-10B4, and HL-60 cells were purchased from the ATCC and cultured in RPMI1640 medium (Thermo Fisher Scientific) supplemented with 10% FBS. 293T, Plat-E, BT-549, MCF-7, and U2OS cells were obtained from ATCC and cultured in DMEM (Thermo Fisher Scientific) supplemented with 10% FBS. BV173 cells were obtained from DSMZ and cultured in RPMI1640 with 10% FBS. The parental BaF3 cells were cultured in RPMI1640 with IL3. The BaF3 cells stably expressing either 210-kDa WT-BCR-ABL (BaF3-BCR-ABL) or T315I-BCR-ABL (BaF3-T315I) and KCL-22 cells were generously provided by Dr. Jia Fei (Medical College of Jinan University, Guangzhou, China) and maintained in RPMI1640 with 10% FBS (20). Cells were incubated at 37°C in a humidified incubator containing 5% CO₂. All the cell lines were authenticated by using short tandem repeat (STR) matching analysis last month. No mycoplasma contamination was detected.

MSCV-BCR-ABL-IRES-EGFP construct was described previously (21). HA-tagged p53 in pcDNA3.1 was a gift from Dr. Tiebang Kang (Sun Yat-Sen University Cancer Center, Guangzhou, China; ref. 22). Human p53 shRNA and mouse p27^kip1 shRNA were from Sigma-Aldrich. siRNA duplexes against p27^kip1, NAE1, and control (Mock) were purchased from Transheep. Detailed information for transfection of plasmids and siRNA duplexes is described in Supplementary Methods.

Whole-cell lysates were prepared in RIPA buffer (23). Cytosolic fraction for detection of cytochrome c was prepared with digitonin extraction buffer. Cytoplasmic and nuclear fractions were obtained as described previously (24), with details shown in Supplementary Methods and Supplementary Table S1.

After CML cells were treated with MLN4924, apoptosis was measured by Annexin V-fluoresceinisothiocyanate (FITC)/propidium iodide (PI) apoptosis detection kit (Sigma-Aldrich), and analyzed with a FACS C6 flow cytometry.

Primary human CML CD34⁺ cells were treated with MLN4924, imatinib alone or in combination for 24 hours, then collected and washed. Cells were stained with Annexin V-FITC and CD38-PE for 30 minutes, apoptotic cells (CD34⁺CD38⁻Annexin V⁺) were detected by flow cytometry (BD LSRFortessa; ref. 21).

The real-time PCR was performed as described previously (24), with details provided in Supplementary Methods.

Mitochondrial transmembrane potential (ΔΨm) was detected as described previously (25), with details provided in Supplementary Methods.

Peripheral blood or bone marrow samples were obtained from patients with CML (Supplementary Table S2) and healthy adult donors in Sun Yat-sen Memorial Hospital of Sun Yat-sen University, The First Affiliated Hospital of Sun Yat-sen University, Guangdong General Hospital/Guangdong Academy of Medical Sciences, and The First Affiliated Hospital of Jinan University after written informed consent according to the institutional guidelines and the Declaration of Helsinki principles. Primary human CD34⁺ cells were isolated by using CD34 Microbeads Kit (Miltenyi Biotec) according to the instructions of the manufacturer and incubated in Iscove's modified Dulbecco's medium supplemented with 10% FBS (21, 24). The studies were approved by Institutional Review Board, Sun Yat-sen University (Guangzhou, China).

Primary human CML CD34⁺ cells were pretreated with MLN4924 for 24 hours, then washed with PBS, and 5,000 cells were seeded in the methylcellulose medium (MethoCult H4434, STEMCELL Technologies). After incubation for 14 days, colonies were counted. The cells were harvested and replated (5,000 cells per well) for the secondary and tertiary rounds, respectively. Colonies were counted on day 14 (21, 24).

Primary human CML-nucleated cells (2 × 10⁶) were cocultured with preestablished and irradiated (80 Gy) M2-10B4 cells in LTC-IC medium (H5100, STEMCELL Technologies) in the presence of MLN4924 (500 nmol/L) in combination with or without imatinib (2,500 nmol/L) for 1 week. Cultures were maintained in LTC-IC medium for 5 weeks with weekly half-medium replacement. The cells were harvested, counted, and plated into methylcellulose medium (MethoCult H4435, STEMCELL Technologies) for colony-forming assay. After incubation for 14 days, LTC-IC–derived colonies were counted (21).

High-titer helper-free retroviruses were produced by transient transfection of Plat-E cells with the retroviral construct MSCV-BCR-ABL-IRES-EGFP as described previously (26). Donor male C57BL/6 mice (Guangdong Medical Laboratory Animal Center) were pretreated with 5-fluorouracil (5-FU, 200 mg/kg). Five days later, bone marrow cells were harvested and transduced two rounds with MSCV-BCR-ABL-IRES-EGFP retrovirus in the presence of cytokines (SCF, IL3, and IL6). The bone marrow cells were then transplanted into sublethally irradiated (550 cGy) recipient female C57BL/6 mice (21). Following transplantation, the mice were treated with vehicle, MLN4924 (60 mg/kg/day, i.p.), imatinib (100 mg/kg/day, gavage) or their combination for 14 days.

To examine the in vivo effect of p27^kip1 knockdown, splenic cells from CML mice transduced with scramble or murine p27^kip1 shRNA lentivirus were transplanted into sublethally irradiated recipient C57BL/6 mice followed by administration of MLN4924 for 14 days (24). The detailed information for the detection of LSK, LT-HSC, and ST-HSC cells is documented in Supplementary Methods.

Bone marrow cells from CML mice treated with or without MLN4924 were harvested and transplanted via tail vein into sublethally irradiated recipient C57BL/6 mice at a serial concentrations of cells (2 × 10⁶, 1 × 10⁶, 5 × 10⁵ cells/mouse) in conjunction with normal bone marrow cells (2 × 10⁵ cells/mouse). GFP⁺ cells in peripheral blood were monitored for 16 weeks by flow cytometry once a week, and GFP⁺ cells > 0.5% were considered as a CML mouse. At week 16, LSC frequency was determined using Poisson statistics online at the Bioinformatics facility of The Walter & Eliza Hall Institute of Medical Research (21, 27). All animal studies were conducted with the approval of the Sun Yat-sen University Institutional Animal Care and Use Committee.

Statistical analyses were performed using GraphPad Prism 5.0 Software (GraphPad). All experiments were carried out at least three times, and data are presented as mean ± SEM unless otherwise specified. Paired analyses were calculated using Student t test, and comparison of multiple groups by one-way ANOVA, post hoc intergroup comparisons, Tukey test. P < 0.05 was considered statistically significant. Kaplan–Meier survival curves were analyzed by log-rank test.

We first compared NAE1 levels in CML cells and other cancerous cells with normal human peripheral blood mononuclear cells (PBMC). Immunoblotting analysis showed that NAE1 levels were higher in the tested 6 lines of CML cells and some Ph⁻ tumor cells (HL-60, BT-549, and MCF-7 but not U2OS) than normal PBMCs (Fig. 1A). Consistently, NAE1 was increased in BaF3-BCR-ABL and BaF3-T315I cells relative to their parent BaF3 cells (Fig. 1A). We next evaluated the specificity of MLN4924 inhibitory effect on neddylation pathway in CML cells harboring either WT-BCR-ABL or T315I-BCR-ABL. The results showed that MLN4924, but not MG132 and bortezomib (two proteasome inhibitors), inhibited global protein neddylation (Fig. 1B). MLN4924 dramatically suppressed cullin1 neddylation and provoked stabilization of key CRL substrates (e.g., phospho-IκBα and p27^kip1) in a dose-dependent manner (Fig. 1C).

Next, we examined the effect of MLN4924 on the proliferation of CML cells. CML cells and normal PBMCs were treated with MLN4924 at incremental concentrations for 72 hours. MTS assay showed that MLN4924 markedly inhibited the cell viability of KU812, BV173, and KBM5 cells all harboring WT-BCR-ABL with IC₅₀ values of 204 nmol/L, 233 nmol/L, and 354 nmol/L, respectively (Fig. 1D). In contrast, MLN4924 did not reduce the cell viability at even up to 10, 000 nmol/L in PBMCs, which expressed extremely low NAE1 (Fig. 1D and A). MLN4924 also suppressed the cell viability of KBM5-T315I to the similar extent as KBM5 (Fig. 1D). Similarly, MLN4924 potently decreased the cell viability of BaF3-T315I as well as BaF3-BCR-ABL but not their parental BaF3 cells (Fig. 1D).

To our surprise, MLN4924 did not counteract the cell viability of K562 and KCL-22 cells that are p53-null CML cells (IC₅₀ values of 20,000 nmol/L and 16,000 nmol/L, respectively, Supplementary Fig. S1A; ref. 28). Given that p53 stabilization induces apoptosis in CML cells (29), we speculated that MLN4924 might exert its antitumor activity in CML cells in a p53-dependent manner. To test this hypothesis, K562 and KCL-22 cells transfected with full-length WT-p53 cDNA construct were exposed to MLN4924, and subjected to MTS assay. The results showed that ectopically restoring p53-sensitized K562 and KCL-22 cells to MLN4924 as demonstrated by approximately 3-fold decrease of IC₅₀ values relative to the corresponding CML cells transfected with empty vector (Supplementary Fig. S1B and S1C). In an alternative set of experiments, KBM5-T315I cells transfected with shRNA against p53 were treated with MLN4924. As illustrated in Supplementary Fig. S1D and S1E, silencing p53 obviously attenuated the apoptosis-inducing ability of MLN4924 in KBM5-T315I cells, suggesting a requirement of p53 for the antitumor effect of MLN4924 in CML cells.

To determine the antitumor activity specificity of MLN4924 in the cancerous cells harboring BCR-ABL relative to Ph⁻ cancerous cells, we included a panel of Ph⁻ cancerous cells (e.g., HL-60, BT-549, MCF-7, and U2OS). The MTS results showed that MLN4924 remarkably inhibited the cell viability of those cells (e.g., HL-60, BT-549, and MCF-7) overexpressing NAE1 but not U2OS weak-expressing NAE1 (Fig. 1D). The sensitivity of cancer cells to MLN4924 appeared to be associated with their NAE1 levels, which was consistent with previous reports (15–17).

In accord with the results in PBMCs, the cell viability of parental BaF3 cells was not diminished when MLN4924 was tested at the concentrations even up to 20,000 nmol/L (Fig. 1D). These results imply existence of a therapeutic window of MLN4924.

Colony-forming assay revealed that MLN4924 selectively inhibited the clonogenicity in the BCR-ABL⁺ cells (KBM5, KBM5-T315I, BV173, and KU812) as well as BaF3-BCR-ABL and BaF3-T315I cells with NAE1 overexpressed and WT-p53, but not in parental BaF3 cells (Fig. 1E). Collectively, our results suggest that single treatment of MLN4924 exhibits equal efficiency against imatinib-resistant as well as -sensitive BCR-ABL⁺ cells with NAE1 overexpressed and WT-p53 while sparing normal PBMCs.

Given that MLN4924 leads to abnormal accumulation of the CRL substrates, which subsequently induces DDR and cell-cycle arrest to inhibit cancer cell growth (12, 30), we examined the effect of MLN4924 on cell-cycle distribution. Flow cytometry analysis revealed that MLN4924 elicited G₂–M-phase arrest in KBM5 and KBM5-T315I cells (Fig. 1F and G). MLN4924 also induced DDR in KBM5 and KBM5-T315I cells, as reflected by the increase in CDT1, γH2AX, and phospho-Chk2 (Fig. 1H). Besides, MLN4924 induced accumulation of endogenous inhibitor WEE1 of G₂–M-phase transition and downregulation of M-phase indicator phospho-H3 in CML cells (Fig. 1H).

Because of their sensitivity to MLN4924, we chose the CML cells (e.g., KBM5, KBM5-T315I, KU812, and BV173) with WT-p53 in the subsequent experiments. We explored apoptosis-inducing ability of MLN4924 in CML cells. Analysis of flow cytometry using Annexin V/PI double staining showed that MLN4924 induced apoptosis in a dose- and time-dependent manner in these 4 lines of CML cells bearing WT-BCR-ABL or T315I-BCR-ABL (Fig. 2A; Supplementary Fig. S2A). Furthermore, MLN4924 induced specific cleavage of PARP and caspase-3, and decrease in the proform of caspase-9 (Fig. 2B; Supplementary Fig. S2B). MLN4924, whereas, had minimal effect on BCR-ABL and phospho-BCR-ABL in the CML cells (Fig. 2B).

We also detected the apoptosis-related proteins by immunoblotting analysis. The results showed that downregulation of Mcl-1, XIAP, Bid, and upregulation of Bim, and tBid, but no alteration in Bcl-2, Bcl-X_L, and Survivin in the MLN4924-treated CML cells (Fig. 2C). The mRNA levels of Noxa were also increased in the MLN4924-treated CML cells (Supplementary Fig. S2C). We conclude that disturbance of balance between the pro- and antiapoptotic proteins caused by MLN4924 treatment may favor onset of apoptosis in CML cells.

Because the activation of caspase-9, -3 (Fig. 2B), and cytoplasmic translocation of cytochrome c (Fig. 2D) were observed in MLN4924-treated cells, the effect of MLN4924 on mitochondrial transmembrane potential (ΔΨm) was further investigated in KBM5 and KBM5-T315I cells. The results showed that MLN4924 significantly increased the proportions of CML cells with ΔΨm loss (Fig. 2E and F). These data suggest that MLN4924 triggers intrinsic apoptosis pathway in CML cells.

Because the third-generation TKI ponatinib is active against T315I-BCR-ABL (6), we examined whether MLN4924 was synergistic with ponatinib in T315I-BCR-ABL⁺ CML cells. The results indicated that the combination between MLN4924 and ponatinib synergistically retarded the KBM5-T315I cell growth as measured by MTS assay (combinational index, CI<1, Supplementary Fig. S2D).

To determine the on-target effect of the MLN4924, NAE1 knockdown with siRNA duplexes in KU812 cells was performed. NAE1 silencing induced not only cleavage of caspase-3, but also DDR as reflected by increase in γH2AX and phospho-Chk2 (Fig. 2G). Similarly, NAE1 knockdown inhibited the colony-forming ability in CML cells (Fig. 2H). These data reveals that NAE1 knockdown mimicks the effect of the MLN4924 treatment, suggesting the on-target effect of the compound.

Because aberrant neddylation activation has been implicated in regulation of survival and self-renewal of CSCs (17, 18, 31), we investigated the potential impact of MLN4924 on stemness of CML LSCs. First, we examined the NAE1 expression in the CD34⁺ cells isolated from CML patients or NBM from healthy donors. qRT-PCR analysis showed that the mRNA levels of NAE1 were significantly higher in CML CD34⁺ cells than in NBM CD34⁺ cells (Fig. 3A). In support of this finding, analysis with a set of publicly available GSE47927 database also revealed that the mRNA levels of NAE1 were remarkably increased in CML CD34⁺CD38⁻ cells relative to normal counterparts (Fig. 3B).

We next determined the effect of MLN4924 on the survival of CML CD34⁺ cells. Flow cytometry analysis showed that single treatment with MLN4924 significantly induced apoptosis of CML CD34⁺ cells, while imatinib alone was ineffective (Fig. 3C). Remarkably, inhibitory effect of MLN4924 on CFC/replating ability of human CML CD34⁺ cells was observed (Fig. 3D). Similarly, treatment with MLN4924 alone or in combination with imatinib but not imatinib alone led to a reduced LTC-IC derived from primary human CML CD34⁺ cells (Fig. 3E). Taken together, our results suggest that pharmacologic inhibition of NAE1 by MLN4924 reduces survival and self-renewal of primary human CML CD34⁺ cells.

We employed a human BCR-ABL gene–driven CML mouse model to evaluate the in vivo effect of MLN4924 on CML (21). CML mice were randomized into 4 groups to be treated with either vehicle, MLN4924, imatinib or their combination for 14 days (Fig. 4A). MLN4924 alone or in combination with imatinib inhibited the splenomegaly of CML mice (Supplementary Fig. S3A). Correspondingly, MLN4924 or imatinib sufficiently prolonged the survival of CML mice, while the combinational treatment of MLN4924 and imatinib elicited an enhanced extension of survival in CML mice (Fig. 4B). Consistently, the populations of GFP⁺ leukemia cells and myeloid leukemic cells in bone marrow and spleen were significantly reduced in the CML mice treated with MLN4924, and further reduced in the group treated with MLN4924 + imatinib (Fig. 4C; Supplementary Fig. S3B).

We next measured the effect of MLN4924 on the proportions of stem/progenitor cells in CML mice. MLN4924 alone or in combination with imatinib dramatically reduced the proportions of LSK, LT-HSC, and ST-HSC cells, as well as GMP and CMP in bone marrow and spleen (Fig. 4D–G; Supplementary Fig. S3C–S3J). Consistent with previous studies (21, 32), imatinib alone did not eliminate LSCs (Fig. 4E). We also examined the effect of MLN4924 on the proportions of quiescent LSK cells. A significant decrease of the subset of quiescent LSK cells as defined by Ki67^lowHoechst3342^low G₀-phase population was observed in the mice treated with MLN4924 (Supplementary Fig. S3K). Meanwhile, MLN4924 treatment inhibited the global neddylation in splenic nucleated cells from CML mice (Supplementary Fig. S3L).

We further investigated the exact in vivo impact of MLN4924 on the frequency of CML LSCs in the secondary transplantation by limiting dilution assay (Fig. 4H). The secondary recipient mice received transplantation of the bone marrow cells isolated from the primary CML mice that were treated with MLN4924 alone or in combination with imatinib showed significantly lower engraftment of GFP⁺ cells and the frequency of CML LSCs at week 16 posttransplantation (Fig. 4I and J; Supplementary Table S3).

Given that cell-cycle inhibitor p27^kip1 is a substrate of CRL, and that p27^kip1 deficiency increases LSK cell population and accelerates leukemogenesis in CML mice (33), we examined the effect of MLN4924 on the expression of p27^kip1 and its E3 ligase Skp2. The results revealed that MLN4924 treatment caused a decrease in Skp2, while an accumulation in p27^kip1 protein levels, in CML cells (Fig. 5A). Further immunoblotting analysis with nuclear and cytoplasm extractions revealed that increased accumulation of p27^kip1 protein caused by MLN4924 treatment predominantly occurred in nucleus but not in cytoplasm (Fig. 5B). To examine whether downregulation of Skp2 protein by MLN4924 is due to proteasome-dependent degradation, the in vivo ubiquitination assay was performed. MLN4924 treatment in KBM5-T315I cells significantly increased Skp2 ubiquitination (Fig. 5C). Taken together, our results suggest that MLN4924 induces Skp2 degradation in a proteasome-dependent manner with concurrent p27^kip1 accumulation in nucleus.

Flow cytometry analysis further confirmed that MLN4924 treatment significantly increased protein levels of intracellular p27^kip1 in the gated LSK cells from CML mice (Fig. 5D). Increased protein levels of intracellular p27^kip1 detected by flow cytometry analysis were also identified in the primary human CML CD34⁺ from 2 of 3 patients (Fig. 5E). Furthermore, the nucleus relocation of p27^kip1 was also exceptional in the MLN4924-treated KBM5 and KBM5-T315I cells (Fig. 5F) and primary human CML CD34⁺ cells as observed with immunofluorescence staining (Fig. 5G).

To define the role of p27^kip1 in MLN4924-induced apoptosis, we knocked down p27^kip1 by siRNA duplexes in KU812 cells, and then treated these cells with MLN4924. The results showed that silencing p27^kip1 at least partially attenuated the MLN4924-induced apoptosis in CML cells (Fig. 5H).

Collectively, these data suggest that MLN4924 leads to Skp2 degradation and p27^kip1 accumulation in nucleus in CML bulk leukemia cells as well as LSCs.

The fact that MLN4924 induces nuclear accumulation of p27^kip1 in CML LSCs prompted us to assess whether the repressive effect of MLN4924 on CML LSCs in vivo was mediated by p27^kip1. We knocked down p27^kip1 by lentiviral shRNA in splenic GFP⁺ cells collected from the first generation of CML mice, and then transplanted such cells into the secondary recipient C57BL/6 mice sublethally irradiated. The mice were then treated with MLN4924 or vehicle for 14 days (Fig. 6A and B). As anticipated, the secondary recipient mice received with MLN4924 did not display splenomegaly (Supplementary Fig. S4A). However, p27^kip1-knockdown alone or p27^kip1-knockdown combined with MLN4924 exhibited splenomegaly, suggesting that p27^kip1-knockdown attenuated the efficacy of MLN4924 in CML mice (Supplementary Fig. S4A). This was confirmed by the observation that MLN4924 significantly reduced the percentages of GFP⁺ and myeloid (Mac-1⁺Gr-1⁺) cells in bone marrow, while p27^kip1-knockdown efficiently rescued such effects of MLN4924 (Fig. 6C and D). Of note, flow cytometry analysis revealed that the percentages of LSKs, LT-HSCs, and ST-HSCs, as well as GMP and CMP cells in bone marrow were also decreased by MLN4924 treatment, which was obviously reversed by p27^kip1-knockdown (Fig. 6E–I). Similar results were obtained in the splenic cells of CML mice (Supplementary Fig. S4B–S4H). Of importance, the recipient mice received injection of p27^kip1-knockdown CML cells in combination with MLN4924 treatment showed a significantly decreased survival when compared with the CML mice treated with MLN4924 alone (Fig. 6J), further confirming that p27^kip1-knockdown reversed the in vivo suppressive effect of MLN4924 in CML. Collectively, these data suggest that MLN4924 inhibits the maintenance of CML LSK cells at least partially through p27^kip1 protein accumulation in nucleus.

To evaluate the impact of MLN4924 on the long-term in vivo repopulating potential, human CML CD34⁺ cells were transplanted into sublethally irradiated NOD-scid-IL2Rg^−/− (NSI) mice (Fig. 7A). MLN4924 treatment reduced the percentages of human CD45⁺ cells in bone marrow and spleen (Fig. 7B–D). In addition, the proportions of human myeloid cells (CD33⁺ and CD14⁺) and lymphocytes (B- and T-cells) were also reduced (Fig. 7E–G). These results indicated that MLN4924 suppressed the long-term engraftment capacity of human CML CD34⁺ cells in vivo.

Resistance to imatinib due to T315I mutation and LSCs remains a challenge in patients with CML. In this study, we discovered that targeting neddylation pathway by MLN4924 led to cell-cycle arrest at G₂–M-phase, and apoptosis of CML cells harboring either WT-BCR-ABL or T315I-BCR-ABL in a WT-p53–dependent manner. Furthermore, MLN4924 inhibited the survival and self-renewal of primary human CML CD34⁺ cells. MLN4924 significantly prolonged the survival of CML mice and reduced LSK cells from CML mice at least partially through p27^kip1 protein accumulation in nucleus. In addition, MLN4924 inhibited the engraftment capacity of human CML CD34⁺ cells in NSI mice.

In our study, we found that MLN4924 efficiently blocked cullin1 neddylation, which confers stabilization of various proteins including phospho-IκBα, p27^kip1, and activation of the DDR effector Chk2 in WT- and T315I-BCR-ABL⁺ cells. Our results showed that MLN4924 treatment in CML cells induced G₂–M-phase arrest, which is consistent with the observations in pancreatic cancer, gastric cancer cells, as well as acute leukemia cells (30, 34). Distinctly, MLN4924 elicits S-phase arrest and DNA rereplication in other type of tumor cells (e.g., colorectal carcinoma cells), which may be explained by accumulated CDT1 that is required for loading of the DNA replication helicase complex (MCM2-7) onto chromatin at G₁-phase of the cell cycle (35). Given the wide-range of substrates of CRL and genetic backgrounds of tumors, it is likely that the eventual outcome is dependent on context comprised of DNA damage, DDR extent, susceptibility, and cell type etc. The MLN4924-treated WT- and T315I-BCR-ABL⁺ cells died by apoptosis, as reflected by cleavage of PARP and caspase-3. This can be explained by downregulation of Mcl-1, XIAP, Bid, and upregulation of Bim and tBid, which may favor onset of apoptosis in these MLN4924-treated CML cells.

Previous studies demonstrated that MLN4924 might induce apoptosis through p53-dependent or p53-independent manner in different cancer cells (35–37). The absence of p53 or p21 sensitized the cells to the MLN4924-mediated cellular senescence in HCT-116 cells (35). In contrast, our data reveal that p53-null K562 and KCL-22 cells are insensitive to MLN4924 when compared with KBM5, KBM5-T315I, KU812, and BV173 cells, which harbor WT-p53. Ectopic restoration of WT-p53 sensitized K562 and KCL-22 cells to MLN4924. Conversely, knockdown of p53 attenuated the effect of MLN4924-induced apoptosis in the p53-intact CML cells (e.g., KBM5-T315I). Given the instrumental functions of p53 in apoptosis, cell-cycle arrest, and DDR (38), it is plausible that MLN4924 induces apoptosis in CML cells in a p53-dependent manner. This effect of MLN4924 may have particular advantage because rare mutations in p53 are observed in patients with CML (39).

LSCs are believed a source of therapy failure in CML. In this study, we demonstrated that NAE1 was overexpressed in CML stem/progenitor cells. MLN4924 induced apoptosis and inhibited the self-renewal of primary CML CD34⁺ cells. MLN4924 also effectively reduced the survival, frequency, and self-renewal capacity of LSK cells in BCR-ABL–driven CML mice. Our results are in agreement with previous report that MLN4924 is toxic to AML and MDS stem/progenitor cells without toxicity to normal stem/progenitor cells (31). Therefore, MLN4924 might be a promising strategy and the relevant clinical trial in CML is warranted. Actually, ongoing phase I clinical trials of single-agent pevonedistat showed that the MTD can reach 83 mg/m², with C_max, maximum plasma concentration, of 2.0 μmol/L, which is much higher than the IC₅₀ values in CML cells in our study (40). Seventeen percent (4/23) of the overall complete response, partial response rate, and modest clinical activity in patients with acute myeloid leukemia and myelodysplastic syndromes at the MTD or below were observed (41). The clinical efficacy apparently correlated with the accumulation of CRL substrates CDT1 and NRF2 in biopsies of these treated patients. Despite adverse effects such as hepatotoxicity, fever, and thrombocytopenia, pevonedistat was generally well tolerated in these patients (42). Encouragingly, more clinical trials of pevonedistat in combinational regimens with azacitidine, doceraxel, gemcitabine, paclitaxel, and cisplatin in patients with solid tumors are going on (40).

Little is known about the mechanism of NAE1 inhibitor killing CSCs. Our results revealed that MLN4924 induced a prominent accumulation of p27^kip1 in nucleus of primary human CML CD34⁺ cells. The increased nucleus p27^kip1 may in turn impair maintenance and self-renewal of LSCs. Zhang and colleagues have demonstrated that the mice of p27^kip1 deficiency exhibits an accelerated leukemogenesis and a marked increase in total LSK, LT-HSC, and ST-HSC cells when induced by retroviral BCR-ABL transduction in comparison with WT-p27^kip1 CML mice (33). It is reported that BCR-ABL can elicit relocation of p27^kip1 marked by a reduced nucleus p27^kip1 by the activated AKT in the human CML CD34⁺ cells, which facilitates cell cycling and expansion of CD34⁺ pool (43). Consistently, reduced nucleus p27^kip1 caused by CaMKIIγ-dependent phosphorylation of p27^kip1 at T187 can reactivate dormant LSCs in CML (44). In this sense, the increased nucleus p27^kip1 by MLN4924 may favor restriction of LSCs.

Of note, neddylation inhibition by MLN4924 may disrupt degradation of multiple proteins (e.g., c-myc). A previous study revealed that MLN4924 inhibited cullin1 neddylation, thereby stabilizing E3 ligase Fbxw7 and degradation of its substrate c-myc (45). Considering the critical role of c-myc in the maintenance of stemness of LSCs (46), the possibility of MLN4924-mediated c-myc degradation in the elimination of LSCs cannot be excluded.

In conclusions, our findings offer a preclinical proof of concept for targeting protein neddylation as a novel therapeutic strategy to override mutational and LSC-derived imatinib resistance in CML. This study provides a rationale for clinical trials of pevonedistat as a single agent and in combination with imatinib in refractory patients with CML. However, our results suggest that MLN4924 may also cause reduced lineages of B- and T-cells in the long-term engraftment experiments, implying that intensively monitoring the immune system may be needed in patients received long-term administration with MLN4924.

No potential conflicts of interest were disclosed.

Conception and design: C. Liu, Y. Jin, J. Pan

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

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

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

Writing, review, and/or revision of the manuscript: C. Liu, Y. Jin, J. Pan

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

Study supervision: Y. Jin, J. Pan

This study was supported by grants from National Natural Science Funds (nos. U1301226 and 81373434 to J. Pan; nos. 81473247 and 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.