Targeting autophagy with hydroxychloroquine potentiates ponatinib- and axitinib-induced cell death in BCR-ABL T315I-containing CML

Because of the dose-dependent increased risk of cardiovascular events we tried to lower the dose of ponatinib without reducing its ecacy in the treatment of BCR-ABL T315I-containing CML. Combination with hydroxychloroquine can enhance the eciency of ponatinib and axitinib through autophagy inhibition in CML cell with T315I. Cell viability, cell cycle, cellular senescence, formation of cell clones and apoptosis assay were taken to test the eciency of medicine. Lentiviral vectors containing shRNA was used to block autophagy and to verify the mechanism of the medicine. Establish tumor models in nude mice, and verify the experimental results in vivo.


Introduction
Generation of the BCR-ABL fusion gene by the reciprocal translocation of chromosomes 9 and 22 in hematopoietic stem cells causes chronic myeloid leukemia (CML) [1]. The BCR-ABL protein has constitutive tyrosine kinase activity and induces oncogenic transformation by activating a cascade of intracellular signaling pathways that lead to uncontrolled cell proliferation [2]. While the introduction of tyrosine kinase inhibitors (TKIs) has revolutionized therapy for CML, primary and secondary resistance due to point mutations in the ABL kinase domain is still great challenges in CML treatment [3][4][5]. This is especially true for the T315I mutation, which causes a threonine to isoleucine change in the kinase gatekeeper residue, resulting in an altered conformation of the kinase domain that reduces the a nity of BCR-ABL for TKIs [6]. Cells with the BCR-ABL T315I mutation are resistant to almost all TKIs (e.g., imatinib, nilotinib, dasatinib, and bosutinib), therefore therapeutic approaches for CML patients harboring T315I mutations are greatly limited [7,8].
Currently, a third generation TKI, ponatinib that was rationally designed to retain activity against the T315I substitution is used as the rst-line therapy for CML with the T315I mutation [9]. However, recent safety data on ponatinib have revealed a dose-dependent increased risk of cardiovascular events including arterial hypertension, serious arterial occlusion, and venous thromboembolism [10][11][12][13][14]. Some clinicians have tried to reduce the dose of ponatinib, but uncertainty about whether this will affect longterm survival outcomes has aroused more concern. How to lower the dose of ponatinib without reducing its e cacy is a great challenge for which combinations with other drugs may be a good option.
Targeting autophagy has been shown to enhance the anti-tumor effects of several drugs [15,16], and the autophagy inhibitor hydroxychloroquine (HCQ) can enhance the anti-CML e ciency of imatinib [17,18].
Whether HCQ is a candidate for combination with ponatinib and the possible underlying molecular mechanisms have not been studied. Additionally, it has been reported that the vascular epidermal growth factor receptor inhibitor axitinib can effectively kill CML cells with the T315I mutation in vitro [19][20][21][22]. Whether HCQ can improve the e cacy of axitinib has not been revealed.
In this study, we showed that HCQ could potentiate the e cacy of ponatinib and axitinib on apoptosis and clonality in 32Dp210-T315I cells, but not on their cell cycle progression or senescence. Both ponatinib and axitinib induce autophagy in 32Dp210-T315I cells, which can be blocked by HCQ. We further knocked-down ATG7 using a short hairpin RNA (shRNA) and con rmed that the mechanism of enhanced cell death from HCQ in ponatinib-and axitinib-treated 32Dp210-T315I cells was by inhibiting autophagy. Consistent with the in vitro results, HCQ also enhanced the anti-tumor effects of axitinib on 32Dp210-T315I cells in vivo.

Materials And Methods
Reagents Imatinib and HCQ were purchased from Selleck Chemicals (Houston, TX, USA). Axitinib, dasatinib, and ponatinib were from Sigma-Aldrich (St. Louis, MO, USA). All antibodies used in western blot analyses Brie y, 1.5-2×10 6 cells were collected, and then added to 1 mL of propidium iodide staining solution and 10 mL of cell xation/membrane solution. After incubating at room temperature for 30 min, cell cycle status was detected with ow cytometry, and cell cycle analysis software was used for analysis.

Cellular senescence assay
After being treated with TKIs, 32Dp210-T315I and 32Dp210 cells were washed and suspended in 1 mL of cell xative solution for 10-20 min. After washing with PBS, the cells were resuspended in 1 mL of βgalactosidase staining solution(Cell Signaling Technology, Danvers, MA, USA), and then incubated in a CO 2 -free incubator at 37°C for 14 h. The blue (senescent) cells were observed by light microscopy.

Formation of cell clones
First, 0.9 g of methyl cellulose was dissolved in 50 mL of RPMI-1640 medium and ltered through a 0.22μm lter to make a sterile 1.8% methyl cellulose solution. 32Dp210-T315I and 32Dp210 cells were resuspended into 5×10 2 cell/mL cell suspensions. The different drugs (40 nM axitinib, 4 nM ponatinib, 4 nM dasatinib, and 4 μM HCQ) were added into the cell suspensions and mixed. Next, 1 mL of the single cell suspensions were mixed with 1 mL of the 1.8% methyl cellulose solution to make a semisolid medium, which was transferred to 6-well cell culture plates. After incubation at 37°C for 7 d, the number of clones was observed under a microscope. The colony formation rates (%) = number of clones ÷ 500 ×100%.

Protein extraction and western blot analysis
Cells were lysed in RIPA lysis buffer (Beyotime Institute of Biotechnology, Haimen, China) containing phenylmethanesulfonyl uoride. Lysates were separated by SDS-polyacrylamide gel electrophoresis, and proteins were transferred onto nitrocellulose membranes. The membranes were blocked in 5% bovine serum albumen (BBI Life Science, Shanghai, China), and then sequentially incubated with primary and secondary antibodies (Li-Cor Biosciences, Lincoln, NE, USA). The following primary antibodies were used: anti-LC3, anti-ATG7, and anti-β-actin (Cell Signaling Technology). Immunoreactive bands were visualized using an Odyssey infrared imaging system (Li-Cor Biosciences).
The shRNA preparation and transfection Lentiviral vectors containing a shRNA against ATG7 or a corresponding control shRNA were synthesized by Gene-Pharma Inc. (Shanghai, China). Lentiviruses were produced in 293T cells by transfecting the lentiviral expression vector and packaging vectors (psPAX2 and pMD2.G [Addgene, Cambridge, MA, USA]) using the Attractenen transfection reagent (Qiagen, Hilden, Germany). After enrichment, the lentiviruses were transfected into 32Dp210-T315I and 32Dp210 cells. Transfection e ciency was estimated by evaluating GFP expression. The effects of the shRNAs on autophagy inhibition were tested by western blotting.
Establishing tumor models in nude mice Twenty-four female nude mice aged 4-5 weeks were raised in the SPF animal center. The feed and drinking water of the mice were subjected to high temperature and pressure sterilization. For inoculation, 4×10 6 32Dp210-T315I cells were subcutaneously injected into each mouse. After 2 weeks, when subcutaneous tumors with volumes of approximately 0.5cm 3 had formed, then the mice were randomly divided into 4 four groups. HCQ was dissolved into PBS, and axitinib was dissolved into carboxymethyl cellulose solution for treatment. Mice in the control group were treated with PBS intraperitoneally and carboxymethyl cellulose solution orally. Mice in the HCQ group were treated with 30 mg/kg/d HCQ intraperitoneally and carboxymethyl cellulose solution orally. Mice in the axitinib group were treated with PBS intraperitoneally and 20 mg/kg/d axitinib carboxymethyl cellulose solution orally. Mice in the combined treatment group were given 30 mg/kg/d HCQ intraperitoneally and 20 mg/kg/d axitinib carboxymethyl cellulose solution orally. After treating the mice for 10 d, they were euthanized, and tumors were isolated and weighed.

Statistical analysis
Data were from three independent experiments and expressed as mean ± SD. All data were analyzed using SPSS 7.0 software with ANOVA or two-tailed Student's t-test. P<0.05 was considered statistically signi cant Results 1. The 32Dp210-T315I cells were resistant to imatinib and dasatinib but sensitive to ponatinib and axitinib.
The 32Dp210 and 32Dp210-T315I cells stably express BCR-ABL and BCR-ABL T315I proteins, respectively. To test the sensitivity of these cells to different TKIs, we used the CCK-8 cell proliferation/toxicity detection kit to analyze cell viability after 48-h treatments with different TKIs at different concentrations, and calculated the IC 50 values. Consistent with previous studies, 32Dp210-T315I cells were resistant to imatinib and dasatinib, with IC 50 values of 11046nM and 4876 nM at 48 h, respectively, which are hundreds-and thousands-fold higher than 32Dp210 cells (IC 50 values of 58.7 nM and 1 nM for imatinib and dasatinib, respectively) (Fig. 1a). Both 32Dp210-T315I and 32Dp210 cells were sensitive to ponatinib with similar IC 50 values (4.8 nm and 4.3 nm, respectively), indicating the extensive potency of ponatinib in CML treatment (Fig. 1a, b). The IC 50 values of axitinib in 32Dp210-T315I and 32Dp210 cells were 43.1 nm and 124.5 nm, respectively, indicating that 32Dp210 cells were insensitive to axitinib. These results con rmed that 32Dp210-T315I cells were resistant to imatinib and dasatinib, but sensitive to ponatinib and axitinib.
HCQ enhanced ponatinib-and axitinib-induced apoptosis in 32Dp210-T315I cells To test whether HCQ enhanced ponatinib-and axitinib-induced apoptosis in 32Dp210-T315I cells, ow cytometry and an annexin V-7AAD apoptosis kit were used to detect apoptosis. Annexin V-positive cells were de ned as apoptotic cells. The rates of apoptosis in 32Dp210-T315I and 32Dp210 cells treated with ponatinib and axitinib with or without HCQ for 24 h and 48 h are shown in Figure 2. Compared with the control group, HCQ did not increase apoptosis levels in 32Dp210-T315I cells g. Finally, we detected the rate of apoptosis in 32Dp210 cells. Neither HCQ nor axitinib induced apoptosis in 32Dp210 cells, while both ponatinib and dasatinib could induce apoptosis. HCQ signi cantly enhanced the rate of apoptosis in 32Dp210 cells treated with ponatinib and dasatinib, but not for those treated with axitinib ( Fig. 2c,d).
HCQ enhanced the inhibitory effect of ponatinib and axitinib on the clonality of 32Dp210-T315I cells.
To test the effect of HCQ combined with either ponatinib or axitinib on cell proliferation, we inoculated the same number of 32Dp210-T315I cells (500/well) onto a semi-solid medium and observed the number of cell colonies after 7 d. As shown in Figure 3, both ponatinib and axitinib inhibited the clonality of 32Dp210-T315I cells, and the number of colonies was signi cantly lower than that of the control group ( g3a, b). HCQ alone had little effect on the clonality of 32Dp210-T315I cells, but HCQ signi cantly enhanced the effects of ponatinib and axitinib. After 7 d, almost no colonies were observed in the HCQ+ponatinib and HCQ+axitinib groups ( g3a, b).
We also performed colony formation assays for 32Dp210 cells. The results showed that HCQ alone had no effect on colony formation in wild-type BCR-ABL expressing cells. Ponatinib inhibited the clonality of 32Dp210 cells, and HCQ further enhanced this effect of ponatinib. Axitinib could not inhibit colony formation in 32Dp210 cells, and HCQ did not modulate axitinib e cacy ( g3b).
The induction of cell cycle arrest in CML cells is an important aspect of TKI treatment. We next investigated whether HCQ enhanced the cell cycle arrest induced in 32Dp210-T315I cells by ponatinib and axitinib. To this end, cellular DNA was stained with propidium iodide, and stages of the cell cycle were detected by ow cytometry. Consistent with the active proliferation of tumor cells, most of the 32Dp210-T315I cells in the control group were in S phase (60.25%±5.2%), some were in G0/G1 (32.86% ±4.24%), and a few were in G2/M phase (6.89±2.1%) (Fig. 4a, b). After 48-h treatment with ponatinib or axitinib, a large number of 32dp210-T315I cells were blocked in G0/G1 phase (60.42%±3.9% and 57.45% ±1.64%, respectively); the number of cells in S phase was also signi cantly reduced (36.15%±2.9% and 38.2%±4.28%, respectively). Compared with the control group, HCQ alone had no effect on cell cycle progression. Additionally, HCQ did not enhance the cell cycle inhibitory effects of ponatinib or axitinib on 32Dp210-T315I cells; there were no signi cant differences in the cells treated with or without HCQ (Fig.  4a, b).
When studying 32Dp210 cells, we found that ponatinib signi cantly blocked cell cycle progression. The proportion of cells in G0/G1 increased from 31.8%±8.86% to 60.68%±5%. Additionally, the ratio of S phase cells decreased to 17.09% from 56.03%±7.7%. Axitinib had no effect on the cell cycle progression of 32Dp210 cells. HCQ did not enhance the e cacy of either ponatinib or axitinib on 32Dp210 cells (Fig.  4b).
Inducing senescence is the primary mechanism of many anti-tumor drugs. Whether ponatinib and axitinib can induce senescence in 32Dp210-T315I cells, and whether HCQ can further enhance the cell senescence induced by these TKIs have not been evaluated. We assayed for cell senescence by staining with β-galactosidase. Similar to many other tumor cells, senescent cells account for a low proportion of all 32Dp210-T315I and 32Dp210 cells; we found that the rates of senescence in control cells were 1.2% ±0.9% and 1.775%±0.56%, respectively(Fig5). While ponatinib and axitinib signi cantly increased the levels of senescent cells, HCQ did not in uence their effects. After treatment with ponatinib and axitinib for 48 h, the proportions of senescent 32Dp210-T315I cells increased from 1.2%±0.9% to 4.97%±1.07% and 5.6%±1.2%, respectively. The proportions of senescent 32Dp210-T315I cells treated with ponatinib and axitinib combined with HCQ were 5.9%±1.5% and 5.7%±1.03%, respectively. These results con rmed HCQ did not increase ponatinib-and axitinib-induced the senescence in 32Dp210-T315I cells (Fig5) .
Many studies have revealed that HCQ potentiates the anti-tumor effects of other drugs by inhibiting autophagy. To further understand the mechanism through which HCQ achieves the previously de ned effects in 32Dp210-T315I cells, we must answer two questions: (1) do ponatinib and axitinib induce autophagy in 32Dp210-T315I cells; and (2) does HCQ inhibit ponatinib-and axitinib-induced autophagy in 32dp210-T315I cells?
Ponatinib and axitinib induced autophagy in 32Dp210-T315I cells LC3II is speci cally expressed on autophagic vacuole membranes, and the quantity of LC3II protein re ects the number of autophagosomes. We tested the expression of LC3II protein in 32Dp210-T315I cells by western blot after 12-h treatment with ponatinib and axitinib. The results showed that compared with the control group, LC3II levels were signi cantly increased in the ponatinib and axitinib groups (Fig6a). We also labeled LC3 protein with a uorescent antibody and observed the number of autophagosomes under confocal uorescence microscopy. This showed that while autophagosomes were rarely seen in 32Dp210-T315I cells in the control group, there were a signi cantly increased number of autophagosomes in the ponatinib group and axitinib groups (Fig. 6b).
The increased levels of LC3II protein expression and number of autophagosomes do not demonstrate induction of autophagy, because autophagy is a dynamic process. Conditions of either autophagy induction or blocking autophagosome degradation will cause LC3II levels to increase. To further identify whether ponatinib and axitinib induced autophagy or blocked autophagic ux, we assayed for LC3II expression in the control group after blocking autophagic ux with HCQ as well as in the ponatinib and axitinib groups. LC3II levels should be unchanged between the three groups if ponatinib and axitinib block autophagosome degradation, whereas if ponatinib and axitinib induce autophagy, LC3II expression should be signi cantly higher in the treated cells compared with the control group. The results shown in Figure 6c demonstrated that LC3II expression was higher in the ponatinib and axitinib groups than in the control group. These results con rmed that both ponatinib and axitinib induced autophagy in 32Dp210-T315I cells.
We also found that LC3II expression was increased in 32Dp210 cells in the ponatinib group and dasatinib group compared with the control group, but this was not true for the axitinib group (Fig. 6a). After autophagosome degradation was blocked with HCQ, we found that ponatinib and dasatinib could still induce autophagy in 32Dp210 cells, while axitinib could not (Fig. 6c).
We next addressed whether HCQ could inhibit ponatinib-and axitinib-induced autophagy in 32Dp210-T315I cells. ponatinib or axitinib was added with or without HCQ, and then LC3II expression were tested after 6 h. Compared with the ponatinib group, LC3II expression was signi cantly increased in the ponatinib+HCQ 32Dp210-T315I cells. Similarly, LC3II expression was signi cantly increased in the axitinib+HCQ group compared with the axitinib group (Fig.7). These results suggested that HCQ inhibited ponatinib-and axitinib-induced autophagy in 32Dp210-T315I cells. We also con rmed HCQ inhibited ponatinib-and dasatinib-induced autophagy in 32Dp210 cells. Fig7 .
Thus far, we have con rmed that HCQ inhibits ponatinib-and axitinib-induced autophagy, but it remained unclear whether inhibiting autophagy was the mechanism by which HCQ enhanced the killing effect ponatinib and axitinib on 32Dp210-T315I cells. ATG7 is an important autophagy regulator, and it has been reported that knocking down ATG7 can speci cally block autophagy. We knocked down ATG7 using a shRNA to inhibit autophagy, and then analyzed whether inhibiting autophagy enhanced the killing effect of ponatinib and axitinib on 32Dp210-T315I cells.
We designed and synthesized a lentiviral plasmid vector with shRNA-ATG7 and GFP to knockdown ATG7 protein expression in 32Dp210-T315I and 32Dp210 cells. Five days after lentiviral infection, the proportion of GFP-positive cells was detected by ow cytometry, and the results showed that the proportion of GFP positive cells was >90% (Fig. 8a). Cells were also collected to detect ATG7 protein expression by western blot. This revealed that shRNA-ATG7-1 and shRNA-ATG7-2 signi cantly inhibited ATG7 protein expression compared with the control shRNA (Fig. 8b ). After knocking-down the expression of ATG7 in 32Dp210-T315I cells, we tested whether ponatinib and axitinib could still induce autophagy. As seen in Figure8b, after ponatinib and axitinib treatment, LC3II expression was much higher in the control group than in the ATG7-knockdown groups, with almost undetectable LC3II levels in ATG7knockdown cells. This result showed that knocking down ATG7 blocked ponatinib-and axitinib-induced autophagy in 32Dp210-T315I cells. Similar results were seen in 32Dp210 cells (Fig.8b). After treatment with ponatinib, LC3II expression was lower in ATG7-knockdown 32Dp210 cells compared with control cells (Fig.8b). Thus, knocking down ATG7 blocked ponatinib-induced autophagy in 32Dp210 cells.
Knocking down ATG7 was nontoxic to 32Dp210-T315I and 32Dp210 cells We next examined the effects of knocking down ATG7 on apoptosis, cell cycle progression, and colony formation. Annexin V and 7AAD were used to detect apoptosis 15d after lentiviral infection. Annexin V positive celles were thought apoptosis. As shown in Figur9a, the apoptosis rate in ATG7-knockdown cells did not increase compared with controls. The cell cycle analysis showed that there was no signi cant difference in cell cycle distribution between the ATG7-knockdown group and the control group, with most cells in S and G1 phase, and few in G2/M phase (Fig. 9b). We also found that ATG7 knockdown had no effect on colony formation in 32Dp210-T315I and 32Dp210 cells. After in vitro culture for 7 d, the colony formation rates in the control group were 89.8±14.2%, and 86.9%±17.4% and 90.2%±18% in ATG7knockdown 32Dp210-T315I cells (Fig.9c). In 32Dp210 cells, the colony formation rates of control and ATG7 knockout groups were 91.8±10.4%, 88.8±13.1% and 87.4±9.6%, respectively (Fig9c). Together, these results suggested that ATG7 knockdown was nontoxic to 32Dp210-T315I and 32Dp210 cells.
ATG7 knockdown did not in uence the ponatinib-and axitinib-induced cell cycle arrest of 32Dp210-T315I cells.
To test whether inhibiting autophagy promoted ponatinib-and axitinib-induced cell cycle arrest in 32Dp210-T315I cells, cell cycle distributions were evaluated in ATG7-knockdown and control 32Dp210-T315I cells after treatment with these drugs for 48 h. Both ponatinib and axitinib had similar effects on the ATG7-knockdown and control groups; most cells were blocked in G0/G1 phase, and there were few cells in S phase and in G2/M phase (Fig. 11a, b). There were no statistical difference between ATG7knockdown group and control groups. After ponatinib, dasatinib and axitinib treatment, the cell cycle distributions of ATG7-knockdown and control 32dp210 cells were similar (Fig. 11 b).

HCQ enhanced the killing effect of axitinib on 32Dp210-T315I cells in vivo.
Currently, studys on the toxicity of axitinib to BCR-ABL expressing cells with the T315I mutation were limited to only in vitro experiments. Therefore, we next tested whether axitinib could kill cells with the T315I mutation and whether HCQ could enhance this activity using an in vivo model. To answer these questions we established a model of subcutaneous neoplasia with 32Dp210-T315I cells in nude mice.
Twenty-four female nude mice aged 5-weeks-old were subcutaneously injected with 4×10 6 32Dp210-T315I cells .After 2 weeks, tumors had formed under the skin, and the volumes were 0.49-0.52 cm 3 (Fig.  12a, b). Nude mice were randomly divided into four groups with six mice in each group: the control group; HCQ group; axitinib group; and HCQ+axitinib group. There was no statistical difference in tumor volume among groups before treatment (Fig. 12b). Ten days after treatment, the mice were sacri ced, and tumors were removed to compare weights. The mean tumor weights were similar in the HCQ and control groups, and were the highest among the four groups. The mean tumor weight in the HCQ+axitinib group was the lowest, while that in the group treated with axitinib alone was the second lowest (Fig 12c). These results con rmed that axitinib could kill 32Dp210-T315I cells in vivo and were consistent with the in vitro results that showed HCQ could further enhance the anti-tumor effects of axitinib.

Discussion
Ponatinib is now the rst-line therapy for CML with the T315I mutation; however, reducing the side effects of ponatinib without reducing its e cacy is a great challenge [9,23]. Amino acid 315 is normally a threonine that serves as the "gatekeeper" residue for the BCR-ABL kinase, and mutations at this site often seriously affect the binding of TKIs with the kinase, which is the primary cause of multi-drug resistance [23,24]. The substitution of threonine with isoleucine at this position gives rise to the T315I mutation, which is resistant to rst and second generation TKIs. For a long time, the prognosis of CML patients with T315I mutations was very poor due to the lack of an effective drug [25]. The third generation TKI ponatinib was developed on the basis of molecular structure and can overcome steric hindrance to bind to BCR-ABL T315I protein and effectively inhibit its kinase activity. Ponatinib is potent in the treatment of CML with the T315I mutation, regardless of whether patients are in the chronic, advanced, or blastic phases [23]. A retrospective study showed that ponatinib in chronic phase-CML with the T315I mutation was associated with signi cantly longer overall survival than stem cell transplantation [26]. In patients with advanced CML, ponatinib can be used as a bridge to allogeneic stem cell transplantation and can induce a secondary chronic phase [27]. In non-transplantable patients at high risk and in relapsed patients after transplantation, ponatinib signi cantly improves long-term survival [28]. Our study also con rmed that 32Dp210-T315I cells were sensitive to ponatinib but resistant to imatinib and dasatinib.
Ponatinib is the rst-line therapy for CML and ALL harboring the T315I mutation; however the potential risk and toxicity of ponatinib treatment must be taken seriously. Several studies have revealed a dosedependent increase in risk of serious cardiovascular events, including severe arterial hypertension, serious arterial occlusive events (AOEs), and venous thromboembolic events (VTEs) [29][30][31]. The Phase II PACE trial showed after 5 years of follow-up, the cumulative incidence of AOEs in the chronic phase-CML population for whom the initial dose of ponatinib was 45 mg daily was as high as 31%, with severe AOEs accounting for 26%. Except for commonly recognized cardiovascular risk factors (hypertension, hypercholesterolemia, diabetes, and obesity), AOEs were ponatinib dose-related [32]. Thus reducing ponatinib doses without reducing its e cacy is crucial, for which combinatorial treatment with other drugs is a good option. In the past few years, a large number of clinical trials have con rmed that targeting autophagy with HCQ can enhance the e cacy of both radiotherapy and chemotherapy in various hematological diseases and solid tumors [33,34]. Our study revealed that HCQ signi cantly enhanced the e cacy of ponatinib on 32Dp210-T315I cells with regards to apoptosis and colony formation, while it did not affect cell cycle arrest or senescence. These data suggested that HCQ+ponatinib may be a new strategy for treating CML with the T315I mutation that makes it is possible to reduce the ponatinib dose and its side effects without compromising e cacy.
Axitinib is also potential drug for treating CML and ALL with the T315I mutation [35]. Axitinib is a vascular epidermal growth factor receptor inhibitor that also has TKI activity. Two in vitro studies have con rmed that axitinib speci cally binds to the BCR-ABL-T315I fusion protein and inhibits kinase activity.
However, it is noteworthy that axitinib does not bind to wide-type BCR-ABL protein and has no therapeutic effect on CML and ALL without the T315I mutation [35]. After toxicity was detected using the CCK8 kit, our study con rmed that axitinib effectively killed 32Dp210-T315I cells, but did not show toxicity towards 32Dp210 cells with the wild-type BCR-ABL fusion protein ( Fig. 1.2). Meanwhile HCQ enhanced the killing effects of axitinib on 32Dp210-T315I cells but not on 32Dp210 cells. Further experiments in nude mice con rmed these results in vivo, which imply HCQ combined with axitinib also is potentinal therapy for CML with T315I mutation.
Similar to wild-type BCR-ABL protein, mutant BCR-ABL-T315I protein inhibits autophagy in CML cells. As a tyrosine kinase the BCR-ABL fusion protein can activate a variety of intracellular signaling pathways, among which, activation of the PI3K/AKT pathway can not only promote excessive proliferation of malignant cells, but can also inhibit autophagy. The mTOR is an important autophagy regulator that inhibits autophagy by blocking formation of the ULK1/2-ATG13-FIP2100 complex [36]. BCR-ABL protein can activate the PI3K/AKT pathway, upregulate ATF5 expression, and active mTOR, thus inhibiting autophagy [37]. Targeting BCR-ABL with imatinib simultaneously induces apoptosis and autophagy in CML cells [38,39]. Crkl protein is a substrate of the fusion kinase, and levels of phosphorylated Crkl can re ect the kinase activity of the fusion protein. Consistent with previous data, our study have con rmed that ponatinib and dasatinib induce autophagy while inhibiting the kinase activity of BCR-ABL protein in 32Dp210 cells (Fig6). Furthermore, after treatment with ponatinib or axitinib for 12 h, phosphorylated Crkl levels were decreased in 32Dp210-T315I cells, which implied the inhibition BCR-ABL-T315I kinase activity.
Meanwhile, LC3II expression was increased, as were the numbers of autophagosomes (Fig. 6). As the negative control dasatinib did not inhibit the kinase activity of the BCR-ABL-T315I protein, LC3II expression and the number of autophagosomes were unchanged. These results suggested that the molecular conformation of the BCR-ABL-T315I fusion protein was different from that of wild-type BCR-ABL protein, but it can also inhibit autophagy in 32Dp210-T315I cells. Thus, targeting the fusion protein with ponatinib and axitinib can not only induce apoptosis in 32Dp210-T315I cells, but can also induce autophagy. The induction of autophagy is related to the inhibited kinase activity of BCR-ABL-T315I by ponatinib and axitinib in 32Dp210-T315I cells, but whether the PI3K/AKI/ATF5 pathway is the key pathway for autophagy inhibition remains to be further studied.
Targeting autophagy is the primary activity of HCQ investigated in our research. HCQ is a multi-functional drug that inhibits antigen delivery, prostaglandin and cytokine synthesis, regulates toll-like receptors, and affects serum metalloproteinase levels [40,41]. As a classic anti-malarial and anti-rheumatic drug, HCQ has been used in clinical settings for over 70 years. Recently HCQ has attracted much attention as an autophagy inhibitor. Through inhibiting the fusion of autophagosomes with lysosomes, HCQ effectively blocks autophagosome ux and causes the aggregation of autophagosomes in cells [42]. After con rmed HCQ can potentiat the effects of ponatinib and axitinib on 32Dp210-T315I cells, Our work showed that HCQ could effectively block ponatinib-and axitinib-induced autophagy in 32Dp210-T315I cells ( Fig. 1.4).
Furthermore inhibiting autophagy with a shRNA targeting ATG7 also enhanced the apoptosis induced by ponatinib and axitinib in 32Dp210-T315I cells. These results proved that HCQ enhanced the killing effect of ponatinib and axitinib in 32Dp210-T315I cells by inhibiting autophagy.
Autophagy allows cell survival in unfavorable conditions; thus, inhibiting autophagy has become a hotspot for anti-tumor therapy. Our study con rmed that ponatinib and axitinib induce autophagy in cells with the T315I mutation; thus, HCQ enhanced the killing effects of ponatinib and axitinib by inhibiting autophagy. However, the mechanism by which autophagy promoted cell survival under TKI treatment is unclear. Activation of apoptosis-related proteins by cytochrome C, which is released from mitochondria, is the key step to initiate apoptosis. Some researchers have suggested that autophagosomes engulf damaged mitochondria, blocking the release of cytochrome C from mitochondria, and thus blocking the major apoptotic pathway [43,44]. Our study suggested that by inducing autophagy, 32Dp210-T315I cells evaded the apoptosis induced by ponatinib and axitinib to some extent, but didnot evade the cell cycle arrest and senescence induced by ponatinib and axitinib. Further research focused on apoptosis-related proteins that are regulated by autophagy will help reveal the key mechanism of how autophagy promotes survival.
In conclusion, ponatinib and axitinib killed 32Dp210-T315I cells as well as inducing autophagy, which promoted their survival under pressure from TKIs. By inhibiting autophagy, HCQ enhanced the killing effect of ponatinib and axitinib on 32Dp210-T315I cells. This suggested that HCQ combined with ponatinib may be a new strategy for treating CML and ALL that harbor the T315I mutation. Thus, this combination may make it possible to reduce the dose of ponatinib and reduce its side effects without compromising e cacy.
autophagy preserve survival after apoptotic cytochrome c release in the absence of caspase activation.  Figure 1 The sensitivity of 32Dp210-T315I and 32Dp210 to TKIs. a and b After 48-h treated with TKIs (imatinib dasatinib ponatinib and axitinib) the survival of 32Dp210-T315I cells and 32Dp210 cells.  HCQ did not increase ponatinib-and axitinib-induced senescence in 32Dp210-T315I cells.    Knocking down ATG7 was nontoxic to 32Dp210-T315I and 32Dp210 cells. a Compared with controls, ATG-7 knockdown did not affect the cell apoptosis in 32Dp210-T315I and 32Dp210 cells. b The cell cycle distribution between ATG7-knockdown group and control group had no signi cant difference. c After in vitro culture for 7d, the colony formation rates in the control group and ATG7-knockdown group were similar.

Figure 10
ATG7 knockdown enhanced ponatinib-and axitinib-induced apoptosis in 32Dp210-T315I cells. a Typical gure of cell apoptosis in ATG7-knockdown 32Dp210-T315I cells after treated with ponatinib and axitinib. b Compared to control cells, ATG7-knockdown 32Dp210-T315I cells were more sensitive to ponatinib and axitinib left and ATG7-knockdown 32Dp210 cells were more sensitive to ponatinib and dasatinib(right).

Figure 11
ATG7 knockdown did not in uence the ponatinib-and axitinib-induced cell cycle arrest of 32Dp210-T315I cells. a Typical gure of cell cycle distribution in ATG7-knockdown 32Dp210-T315I cells after treated with ponatinib and axitinib. b After treated with ponatinib and axitinib , the cell cycle distributions have no statistical difference between control group and ATG7-knockdown 32Dp210-T315I (upper). Compared to the control group ATG7 knockdown did not affect the cell cycle distributions in 32Dp210 cells after treated with ponatinib axitinib and dasatinib(blow).