Arsenic trioxide promoting ETosis in acute promyelocytic leukemia through mTOR-regulated autophagy

Despite the high efficacy and safety of arsenic trioxide (ATO) in treating acute promyelocytic leukemia (APL) and eradicating APL leukemia-initiating cells (LICs), the mechanism underlying its selective cytotoxicity remains elusive. We have recently demonstrated that APL cells undergo a novel cell death program, termed ETosis, through autophagy. However, the role of ETosis in ATO-induced APL LIC eradication remains unclear. For this study, we evaluated the effects of ATO on ETosis and the contributions of drug-induced ETosis to APL LIC eradication. In NB4 cells, ATO primarily increased ETosis at moderate concentrations (0.5–0.75 μM) and stimulated apoptosis at higher doses (1.0–2.0 μM). Furthermore, ATO induced ETosis through mammalian target of rapamycin (mTOR)-dependent autophagy, which was partially regulated by reactive oxygen species. Additionally, rapamycin-enhanced ATO-induced ETosis in NB4 cells and APL cells from newly diagnosed and relapsed patients. In contrast, rapamycin had no effect on apoptosis in these cells. We also noted that PML/RARA oncoprotein was effectively cleared with this combination. Intriguingly, activation of autophagy with rapamycin-enhanced APL LIC eradication clearance by ATO in vitro and in a xenograft APL model, while inhibition of autophagy spared clonogenic cells. Our current results show that ATO exerts antileukemic effects at least partially through ETosis and targets LICs primarily through ETosis. Addition of drugs that target the ETotic pathway could be a promising therapeutic strategy to further eradicate LICs and reduce relapse.


Introduction
Acute promyelocytic leukemia (APL) is a hematological malignancy driven by a t(15;17) chromosomal translocation that generates the promyelocytic leukemia-retinoic acid receptor (PML/RARα) fusion gene 1,2 . The prognosis for patients with APL has been revolutionized by the use of all-trans retinoic acid (ATRA) and arsenic trioxide (ATO), both of which target PML/RARα for degradation 3,4 . Recently, benefits from ATO-including therapy in APL have sparked new interest in ATO. For example, patients receiving ATO plus ATRA induction therapy experienced fewer relapses and faster complete remission compared to patients receiving standard ATRA chemotherapy [5][6][7][8] . ATO induces high rates of complete hematologic remission (CR) and molecular remission (CMR) followed by a long relapse-free survival 9 . Despite the remarkable improvement in treatment outcomes in APL, refractory and relapse remain clinically significant problems 10 . Thus, further understanding of the antileukemic mechanisms of ATO when treating newly diagnosed APL and/or relapse is urgently needed.
It is known that treatment by standard chemotherapy reagents induces apoptosis while ATRA results in differentiation 3 . However, APL relapse occurs because leukemia-initiating cells (LICs) remain untouched by conventional chemotherapy and even ATRAmonotherapy 11,12 , in contrast to ATO therapy, which implies that neither apoptosis or differentiation induction is sufficient to eradicate LICs. It is attractive to speculate whether another uncovered LIC death program exists, which can be induced by ATO. Autophagy contributes to arsenic-induced PML/RARα degradation 13 , which is responsible for LIC loss in APL cells 14,15 , and it is also widely proposed to account for arsenic-induced cell death [16][17][18] . However, these studies did not fully address the questions of whether or how autophagy leads to LIC death by ATO.
First described as an alternative route of bacterial killing in 2004, the formation of neutrophil extracellular traps (NETs) (ETs) is a process of cell death distinct from apoptosis, which has since been referred to as NETosis [19][20][21] . Formed mainly by immune cells, ETs can also be released by human leukemia cells when exposed to microorganisms, reactive oxygen species (ROS) or tunicamycin 22,23 . Studies from our laboratory have shown that APL cells from patients can also undergo this novel cell death process, producing ETs through autophagy 24,25 , that has been linked to the mechanisms of ATO. More interestingly, ATRA promotes ETosis leading to procoagulant promyelocytic extracellular chromatin 25 . However, little is known about its response to ATO treatment or the role of ETosis in leukemia cell eradication.
In this study, we characterized the concentrationdependent effects of ATO exposure on ETosis in APL cells. We also continued our previous study by investigating the upstream mammalian target of rapamycin (mTOR)-mediated autophagy pathway and the role of ROS production in this process. Finally, we explored the role of ETosis in APL LIC loss, helping identify a novel pathway to target LICs and further prevent relapse in APL patients following ATO administration.

ATO induces ETosis and apoptosis in NB4 cells in a dosedependent manner
To distinguish the effect of ATO on ETosis and apoptosis, lactadherin and propidium iodide (PI) were used to stain NB4 cells 24,25 . In ETotic cells, the chromatin expands while the cytoplasmic membrane remains intact. PI staining can be observed in the absence of lactadherin membrane staining (green) or visible membrane blebbing. Cells undergoing ETosis could be seen releasing a single swelling bubble that stained with PI 24,25 . To investigate the effect of varying concentrations of ATO on ETosis in cultured NB4 cells, an APL cell line, cells were treated with 0, 0.1, 0.25, 0.5, 0.75, 1.0, or 2.0 μM ATO for different time points. When cultured for 48 h, concentrations of ATO over 0.5 μM caused a significant increase in the number of ETotic cells (Fig. 1a, b). When NB4 cells were treated with ATO at 1.0 μM or higher concentrations, both ETotic and apoptotic cells were visible (Fig. 1a). Using immunofluorescence, we identified that promyelocytic ET backbone as DNA-histone (Fig. 1c). ETosis% counted by DAPI/anti-histone-3 and lactadherin/PI staining were comparable ( Supplementary Fig. S1) 24 . PI staining was used to define the morphological changes leading to ETosis. Evaluating the changes in nuclear size and shape, we identified four morphologies (Supplementary Fig. S2): (i) round, (ii) budding, (iii) spread nuclei, and (iv) extracellular DNA. Because DNA material ultimately became cell-free, MPO-DNA complexes could be detected in the supernatant and used as a marker of ETosis 26 . Our results showed that ETosis dramatically increased at 0.5 μM, peaked at 0.75 μM, and decreased at higher ATO concentrations (Fig. 1d).

Moderate concentrations of ATO mainly promotes ETotic cell death in NB4 cells
To better understand the mechanism behind ATOinduced cell death, we dual stained cells with lactadherin and PI (Fig. 2a). Apoptosis was analyzed by flow cytometry, and confirmed by frequency of cells that were TUNEL-positive. Surprisingly, the results showed that the percentage of apoptotic cells (lactadherin + and PI + ) had a marked increase only at concentrations over 1.0 μM (Fig. 2a), while the viability of NB4 cells dramatically decreased at 0.75 μM ATO (Fig. 2b). Thus, 0.75 μM ATO treatment probably induced non-apoptotic cell death. To confirm the role of ETosis in non-apoptotic cell death induced by 0.75 μM ATO, NB4 cells were preincubated with various cell death inhibitors. We found that 3methyladenine (3-MA) significantly increased the number of viable cells reduced by ATO, while other cell death inhibitors had few effects. Moreover, 3-MA only decreased ETosis induced by ATO while having limited effects on apoptosis (Fig. 2c). Taken together, these results suggested that the majority of NB4 cell death induced by 0.75 μM ATO was ETosis.

ATO increase mTOR signaling-dependent autophagy that mediated ETosis
The ETotic process involves interplay between autophagy, the cytoskeleton, superoxide production, and histone citrullination 21,27 . To further unravel the molecular mechanism by which ATO primes cells for ETosis, we investigated possible signaling pathways preceding ETosis. Diphenyleneiodonium chloride (DPI) and wortmannin but not Cl-amidine were able to block ATO-triggered ETosis (Fig. 3a). When nutrient deprivation is used to induce autophagy, mTOR-mediated phosphorylation of p70 ribosomal S6 kinase (p70S6K) is prevented. Therefore, a reduced level of phosphorylated p70S6K (P-p70S6K) can be used as an indicator of mTOR-dependent autophagy 28 . ATO-treated NB4 cells exhibited increased levels of LC3-II and decreased levels of P-p70S6K by immunoblotting (Fig. 3b, c), suggesting that ATO activated autophagy via the mTOR pathway. Moreover, siRNA-mediated inhibition of the negative downstream effector of mTOR pathway, translational repressor 4E-BP1 ( Supplementary Fig. S3), reversed the effects of ATO on LC3-positive structures (autophagosomes) (Fig. 3d, e), LC3 expression (measured by western blot) and ETosis ( Supplementary Fig. S4), confirming ATO-induced ETosis was dependent on mTOR-mediated autophagy.
Rapamycin enhances the autophagy-dependent ETosis but not autophagy-independent apoptosis in response to ATO Because induction of autophagy increase ETosis 24 and the aforementioned experiments demonstrated a role of mTOR in ATO-induced autophagy, we also examined whether downregulation of mTOR activity by rapamycin would affect ETosis in NB4 cells. We found that rapamycin alone was capable of inducing a modest increase ETosis and that this drug potentiated ATO-mediated ETosis (Fig. 4a, b). However, the same combination treatment (ATO + rapamycin) of NB4 cells along with inhibition of autophagy with various autophagy inhibitors (wortmannin, hydroxychloroquine (HCQ) and bafilomycin A1 (Baf A1)) resulted in a significant decrease in ETosis (Fig. 4b). Furthermore, we knocked down autophagy-related gene Atg7 using small-interfering RNA (siRNA) to confirm the role of autophagy in ATOinduced ETosis in APL. As shown in our previous work 24 , inhibition of autophagy nearly abolished autophagy as the markedly decreased conversion of LC3B-I to LC3B-II. Knocking down of Atg7 resulted in significantly reduced ATO-induced ETosis and MPO-DNA complexes (Fig. 4c). Interestingly, rapamycin did not increase ATOinduced apoptosis, which was not affected by autophagy inhibition with wortmannin (Fig. 4d).

The role of ROS production during ETosis
We monitored intracellular ROS production in NB4 cells to investigate its role in ETosis. The results showed that intracellular ROS was increased in NB4 cells activated by ATO, and remained at baseline if inhibited by NADPH oxidase (Nox) with DPI ( Supplementary Fig. S5). Inhibition of autophagy with wortmannin or activation of autophagy with rapamycin did not alter basal ROS levels or ATOinduced ROS production ( Supplementary Fig. S6). We then assessed whether ROS was essential for ETosis. NB4 cells were stained with DAPI and anti-histone-3 to identify ETotic cells. While ATO notably increased histone staining, the addition of DPI abrogated ETosis ( Supplementary  Fig. S5). Overall, these results demonstrated that Noxdependent ROS production plays a key role as an early event of ATO-induced ETosis, independent of autophagy.

Rapamycin enhances ATO-induced ETosis in NB4 cell lines and APL cells from patients at diagnose and after relapse
For the evaluation of the combinated effects on ETosis, NB4 cell line was first treated with a fixed non-toxic concentration of ATO (0.25 μM, 48 h) and graded concentrations of rapamycin. The drug combination resulted in synergistic effects at all combination data points. When cells were treated with toxic concentration of ATO (0.5 μM and 0.75 μM, 48 h) and graded concentrations of rapamycin, lower doses of rapamycin resulted in synergistic effects while higher doses of rapamycin (10 and 20 nM) suggested additive effects (Fig. 5a). Confocal microscopy images showed that untreated NB4 cells showed little staining with either lactadherin or PI. The nuclei of ET-releasing cells spread into the extracellular space (arrow). NB4 cells undergoing early apoptosis showed ring-like staining by lactadherin but no PI staining (*). Cells in late apoptosis had nuclei that were condensed and staining by both lactadherin and PI (arrowhead). Apoptosis was determined by lactadherin/PI staining for 15 min at room temperature analyzed by flow cytometry. The data represents mean ± SD of three independent experiments done in duplicate. b Cell viability was detected by trypan blue exclusion. The data show mean ± SD of three independent experiments done in duplicate (n = 3). c NB4 cells were preincubated with corresponding cell death inhibitors (Z-VAD, Nec-1, 3-MA) for 1 h, and treated with 0.75 μM ATO for 48 h. The cell viability in untreated group was counted as 100%. Apoptosis was analyzed by flow cytometry and ETosis were counted as described in methods (n = 5). All values are mean ± SD. *P < 0.05, ns = not significant. Bars represent 10 μm in a Since resistance is the major puzzle in APL therapy, we investigated whether this combination (ATO + rapamycin) is effective on ATRA-resistant and ATO-resistant cell lines (NB4-R4 and NB4 ATO-R). Increasing doses of rapamycin were tested, starting at 5 nM and scaling up to 20 nM. Marked increase of ETosis was initially detected NB4-R4 cells using 10 nM rapamycin, whereas 20 nM rapamycin induced a significant increase in ETosis not only in NB4-R4 but also in NB4 ATO-R cells. In NB4 ATO-R cell line, less sensitive to rapamycin as single agent, the combination of 1 μM ATO with rapamycin (20 nM) resulted in a remarkable increase in the percentage of ETotic cells, as shown by lactadherin/PI staining after 48 h ( Supplementary Fig. S7).
To confirm the effect of rapamycin and ATO on ETosis, primary APL blasts from newly diagnosed APL patients (patient 1-6 in Table 1) were incubated with rapamycin (10 nM) in the presence or absence of ATO. We found that rapamycin alone was capable of inducing a modest increase (from 3.4 to 8.2%) in ETosis and that this drug potentiated ATO-mediated ETosis (from 13.3 to 20.4%) (Fig. 5b, c), suggesting that the effects of ATO and rapamycin are additive. Furthermore, addition of the mTOR inhibitor rapamycin strongly enhanced the percentage of ATO-mediated ETosis (paired P-value < 0.0001, n = 6) in APL cells taken from patients after relapse (patient 7-12 in Table 1) (Fig. 5d). Measurements of MPO-DNA complexes in the supernatant were used to indirectly quantify ETs. We found that the increase in ETotic APL cells was paralleled by an elevated abundance of MPO-DNA complexes in the supernatant (Fig. 5b, c).

Combined rapamycin and ATO treatment synergistically decreases LIC production
Primary APL blasts were isolated from APL patients (patient 13-20 in Table 1)  . Immunofluorescence images showed inhibition of the negative downstream effector of mTOR pathway by knocking down 4E-BP1 reduced the aggregation of LC3 (green). e The LC3 puncta per cell in ATO-treated NB4 cells measured by immunofluorescence (n = 3). f The levels of LC3B-I and LC3B-II were detected by western blotting (n = 3). All values are mean ± SD. *P < 0.05 vs. untreated group, # P < 0.05 vs. control e, f. Bars represent 20 μm in d. DPI diphenyleneiodonium chloride, Wort Wortmannin, Rap rapamycin (mTOR inhibitor), PAD4 peptidylarginine deiminase 4, UT untreated group, Ctrl, control methylcellulose, a semisolid medium, in the presence of rapamycin, ATO, or both. ATO resulted in a striking decrease in colony-forming units when in combination with rapamycin compared with either drug alone (Fig. 6a). Furthermore, when cultured for an additional week, treated colonies had a reduction in the number of secondary colonies (Fig. 6b). Thus, rapamycin and ATO synergistically diminished the clonogenicity of APL cells. Meanwhile, inhibition of ETosis with wortmannin blocked the effects of both ATO and rapamycin (Fig. 6a,  b). To confirm our results obtained from primary APL cells, we investigated the effect of ATO-induced ETosis on the side population of NB4 cells, a model of APLderived LICs. We observed that rapamycin exhibits synergism with ATO (0.75 μM, 48 h) for efficient side population clearance (Fig. 6c) and PML/RARα degradation in NB4 cells (Fig. 6d). then performed secondary transplantation of the bone marrow cells of leukemic mice harvested on day 28 after treatment (Fig. 7a). We observed that ATO alone or ATO + rapamycin produced significantly longer survival of secondary recipients. Mice receiving bone marrow treated with ATO + rapamycin had a median survival time of 27 days compared to 16.5 days with placebo or 20 days with ATO alone (Fig. 7b). Altogether, these data suggest that this combination is able to significantly decrease the LIC compartment in a mouse model of APL.

Discussion
This is the first study to show that ATO significantly increases the amount of ETosis in APL cells. ATO has dose-dependent antileukemic effects on NB4 cells, preferentially inducing ETosis at moderate concentration (0.5-0.75 μM) and switching to apoptosis at relatively high doses (1.0-2.0 μM). Further investigation indicated that ATO-induced ETosis requires both mTOR-mediated autophagy and Nox-dependent ROS production. More importantly, we found that rapamycin promoted ATOinduced ETosis leading to decreased LIC activity, suggesting that ETosis facilitates LIC loss in APL cells. These results further clarify the cellular and molecular mechanisms of ATO treatment. Our discovery suggests that therapy-triggered ETosis could be used to target LICs and, therefore, reduce relapse risk in patients.
We have recently found that ATRA promotes a novel cell death process, named ETosis, in APL cells 24,25 . Since ATO exerts dose-dependent cellular effects 16,17,29 , we investigated its dosage effect on ETosis. In concordance with the finding that low doses of ATO cause only minimal cell death 29 , ETotic cell death started and peaked at median doses of ATO. ATO at high concentrations triggered the apoptosis pathway 29 , which is thought to block the ETosis pathway 20 . As expected, we observed that ETosis decreased at higher concentrations. We performed all our experiments in the presence APL serum containing cytokines, which mimicked the conditions at inflammatory sites as a two-hit model to favor ET formation 30 . It is not surprising, however, that ATO can induce this novel form of cell death. Indeed, previous reports have shown that ATO is a potent inducer of apoptotic and non-apoptotic cell death 16,17 . While our work strongly supports a role for ATO in priming ETosis, the fluctuating levels of circulating APL-derived ETs in the plasma of hospitalized patients treated with ATO need to be explored. An assay of patient plasma for MPO-DNA complexes, a marker of ETosis, could potentially be used to predict drug sensitivity and/or risk of relapse.
The process of ETosis is distinct from apoptosis, in that there is no caspase activation or phosphatidylserine (PS) exposure before cell death, and the morphological characteristics of the two forms of active cell death are different 24,25 . Beyond apoptosis, ATO has at least a partial tendency to favor ETosis which results in lower overall PS exposure on APL cells. PS exposure is a major mechanism through which APL cells support procoagulant activity 31 . Therefore, our results explain the mechanism by which ATO treatment results in rapid correction of patient coagulopathy preceding disappearance of the circulating leukemic promyelocytes 32 . Thus, this alternative cell death process may represent a therapeutic target in the treatment of resistant cases and attenuating apoptosisassociated complications. NET-like structures are also generated in other myeloid cell lines 23 , and ATO is known to exert anticancer effects on other hematopoietic tumor cells 17,18 . Further studies are needed to determined whether ATO induces ETosis in other hematological malignancies.
ATO induces autophagic cell death in several types of tumor cells 16,17 , and degrades PML/RARα protein via mTOR pathway-mediated autophagy in APL cells 13 . In this study, we have identified the pivotal role for the mTOR pathway in ETosis via regulation of autophagy. From a novel cell death perspective, our data explains the results obtained by Altman et al. 33 that pharmacological inhibition of mTOR enhances the suppressive effects of ATO on leukemic progenitor colony formation. Beyond the prior reports, we found that mTOR-mediated signals regulated the ETotic response to ATO implying that  6). b, c APL cells from newly diagnosed APL patients were treated with vehicle or rapamycin (10 nM) in the presence or absence of ATO (0.75 μM) for 48 h. b APL cells were stained with lactadherin (green) and PI (red) and imaged with confocal microscopy. ETotic cells showed nuclei (stained with PI) that had lost shape and expanded to fill most of the cytoplasm, but did not stain with lactadherin (*). In contrast, cells undergoing early apoptosis showed diffuse rim staining by lactadherin but no PI staining (arrowhead). Late apoptotic cells had condensed nuclei and were stained by both lactadherin and PI (arrow). c Cells undergoing ETosis or apoptosis were counted as described in methods (n = 6). MPO-DNA complexes levels were also measured (n = 6), *P < 0.05. d Primary APL cells from patients at relapse were treated with vehicle or rapamycin (10 nM   HCR complete hematologic remission, PR partial remission, ED early death mTOR could be a potential pathway to suppress LIC activity. Although the role of the PI3K-Akt-mTOR pathway in NET formation is contradictory, reports of fMLPinduced NETosis are consistent with our results showing that this pathway inhibits autophagy 34 . Our results reveal a new facet of this pathway, suggesting that ETosis might be directly linked to, or be a variant of, the autophagic cell death pathway. This preclinical data suggests the need for a clinical investigation of ATO in combination with other agents interfering with mTOR-mediated autophagy. Combined therapy could contribute to the killing of APL LICs through the induction of ETosis and thereby improve remission rates. Oxidative production is one of the proposed mechanisms by which ATO affects cellular behavior. NB4 cells exposed to ATO show increased Nox-derived ROS production [35][36][37] , which shapes a critical component of NETosis 27 . In accordance with other studies, our results show that ATO enhanced ET formation via a Noxdependent mechanism, which is supported by the finding that Nox is involved in ATO-induced cytotoxicity 35 . Evidence has emerged that autophagy and activated Nox are both required in ETosis where they work together to inactivate caspases and prevent cells from entering apoptosis 26,38,39 . This explains our observation that blocking either autophagy or ROS generation decreases ETosis in NB4 cells. Based on the current findings, it seems that ROS regulate autophagy that mediates ATOinduced ETosis 40 .
Since ATO-based therapy is strongly correlated with APL LIC eradication 12,41 , we tested whether ATOinduced ETosis plays a role in LIC loss. Rapamycin as a classic autophagy inducer augmented ATO-induced cytotoxicity in NB4 cells 42 . Since LICs remain untouched by apoptosis-inducing conventional chemotherapy 11,43 and rapamycin did not promotes apoptosis by ATO, it is convincing that activation of the ETotic rather than apoptotic cell death pathway contributes to rapamycin-enhanced LIC loss by ATO. In addition, we confirmed the role of ETosis in ATO-induced APL LIC At the molecular level, a previous study suggested that autophagy contributes to ATO-induced degradation of the PML/RARα 13 , whose degradation is necessary for LIC eradication 14,43 . By monitoring APL LIC activity-linked oncoprotein PML/RARα, we observed that enhanced ETosis significantly correlated with oncoprotein degradation. Thus, it seems that autophagic degradation of PML/RARA leads LICs to ETosis. This study goes beyond the prior reports, correlating the cellular fate of LICs with ETotic cell death. Besides loss of self-renewal, differentiation and apoptosis, our finding adds ETosis to the potential cellular mechanisms of LIC clearance, though there is not yet direct evidence of ETotic LICs.
Since prompt treatment with ATO has been the most important step in treating and preventing relapse, future therapeutic strategies could focus on combined application of rapamycin or autophagy inducer with ATO to eradicate LICs and further decrease risk of relapse, especially in high-risk APL patients. In brief, our study uncovers the role of ETosis in LIC eradication, representing a novel mechanism for ATO anticancer activity. Thus, enhancing ATO-induced ETosis could be a potential strategy to eradicate LICs.

Materials and methods
See Supplementary files for a full description of materials and methods (available on the Cell Death and Disease Web site). ATO stock solutions were made at the concentration of 1 mM with phosphate-buffered saline (PBS) and diluted to working concentrations before use.

Patients
Twenty newly diagnosed APL patients admitted to the First and Second Affiliated Hospital of Harbin Medical University from May 2014-February 2016 agreed to participate in the study after informed consent. The diagnosis was based on clinical data, morphology, cytochemistry, immunology, cytogenetics, and molecular pathology testing or alternatively confirmation of the presence of the t(15;17) (PML-RARα) fusion gene 31 . Compassionate use of ATO was initiated in all the 20 previously untreated APL patients, and this treatment was very well tolerated. This study was approved by Ethics Committee of Harbin Medical University and conducted in accordance with the Declaration of Helsinki.

Cell lines and primary cells
Human APL NB4 cell line and ATRA-resistant cell line, NB4-R4, were gifts from Dr. James O'kelly (Los Angeles, CA). While the ATO-resistant cell line, NB4-ATO-R, was established by culturing in increasing concentrations of ATO, with initial concentration of 200 nM, until the cell line survived a concentration of 1 μM. The resistant cell line was then maintained in complete RPMI 1640 medium containing 1 μM ATO. NB4 cells transiently transfected with Atg7 siRNA were obtained in the same way as our previous work 24 . Bone marrow samples from APL patients were collected before intravenous ATO (0.1 mg/kg per day) treatment and at hematological relapse after obtaining a written informed consent. APL cells were harvested from the bone marrow of 14 newly diagnosed APL patients for further experiment. APL cells from the BM of another six APL patients at hematological relapse were treated with vehicle or rapamycin (10 nM) in ATO (0.75 μM) for 48 h.

Cell culture
Freshly isolated APL blasts were obtained from bone marrow specimens by centrifugation through Ficoll-Hypaque and were used for experiments immediately. These cells and NB4 cells (5 × 10 5 /ml) were resuspended in complete RPMI 1640 medium supplemented with 10% FBS, 10% APL serum, 2 mM L-glutamine and 1% penicillin-streptomycin solution at 37°C in a 5% CO 2 humidified atmosphere.

Confocal microscopy
To characterize cell death, smears of APL or NB4 cells (5 × 10 5 ) were incubated with DAPI (100 ng/ml) 44 or the indicated concentration of PI and FITC-labeled lactadherin 45 . Cells were washed to remove unbound proteins and analyzed immediately on Zeiss LSM 510 Meta confocal microscope (Carl Zeiss Jena GmbH, Jena, Germany). The samples were excited with the 488 nm emission line of a krypton-argon laser, and the emission wavelength overlap was restricted with narrow bandpass filters.

Determination of ETosis and apoptosis
Cells undergoing ETosis were characterized by rounded morphology, PI staining, and the presence of nuclear content diffused throughout the cell 24,25 . When cells were stained with DAPI/anti-histone-3, ETosis was identified by histone-3-stained cells 24 . When cells were stained with lactadherin and PI, cells in early apoptosis showed diffuse rim staining by lactadherin but no PI staining. Late apoptosis was characterized by condensed nuclei and stained by both lactadherin and PI. 45 Apoptosis with PI staining was defined by the presence of membrane blebbing and nuclei fragmentation. Cells were counted from six random fields in triplicate wells for each condition and expressed as percentage of total number of cells in the field 46 .

Quantification of ETosis
To quantify ETs in cell culture supernatant, a capture ELISA to measure myeloperoxidase (MPO) associated with DNA was performed as described previously 26,47 . For the capture antibody, Mouse MPO ELISA kit (Hycult biotech, HK210-01) was used according to the manufacturer's directions. A peroxidase-labeled anti-DNA mAb (component No.2, Cell Death ELISAPLUS, Roche; Cat. No: 11774424001) was used.

Cell apoptosis assay
For the apoptosis assay, cells were washed with PBS, resuspended in binding buffer from BD Biosciences and stained with Lactadherin-FITC and PI for 15 min. Fluorescence was measured with a flow cytometer, and the membrane integrity of the cells was simultaneously assessed by the PI exclusion method. Apoptosis was determined using the Lactadherin-FITC apoptosis detection kit (BD Pharmingen, San Diego, USA), according to the manufacturer's instructions. Prepared cells were analyzed with FACScan flow cytometer and CELLQuest software (Becton Dickinson, Franklin Lakes, USA). The TUNEL (terminal deoxynucleotidyl transferasemediated deoxyuridine triphosphate nick end labeling) assay was performed using the in situ cell death detection kit (Roche, Philadelphia, USA), according to the manufacturer's instructions.

Assays of primary hematopoietic progenitors from APL patients
To evaluate drug effects on LIC activity in APL cells, APL cells were cultured (10,000 cells per well) in the methylcellulose medium, supplemented with recombinant cytokines (GF M3434, StemCell Technologies) in the presence or absence of the drugs. Rapamycin (10 nM) was used alone or in combination with ATO (0.75 μM). After a week of treatment, colonies were counted and cells were harvested, washed and, when specified, plated in the methylcellulose medium for a further week without drugs.

Side population of NB4 cells
In cell lines, the stem cell population is contained in the so-called side population determined by staining exclusion in the presence or absence of ABC transporter inhibitors such as verapamil 49,50 .

Secondary transplantation experiments
In all, 5 × 10 6 NB4 cells were injected into 6-8-weeksold SCID mice followed by treatment with ATO with or without rapamycin or no drug treatment (placebo) from day 15-day 28. On day 28, the mice were sacrificed and 1 × 10 7 bone marrow cells were transplanted into a new batch of 6-8-weeks-old mice (no treatment post-transplant) and followed up to their death. Details of methodology used for "experimental animals", "in vivo xenograft model", and "secondary transplantation experimental design" are provided in "Supplemental Methods".

Statistical analysis
The results are expressed as mean ± standard deviation (SD), and all statistical analyses were performed using Student's t-test (two-tailed, unpaired). A P-value of 0.05 or less was considered significant. In cell viability assays, the data are normalized to control cells and expressed as the percentage of live cells relative to total cells.