Differential sensitivity of acute myeloid leukemia cells to daunorubicin depends on P2X7A versus P2X7B receptor expression

Acute myeloid leukemia (AML) is a common adult leukemia often arising from a preexistent myelodysplastic syndrome (MDS). High mortality rates of AML are caused by relapse and chemoresistance; therefore, we analyzed the role of P2X7 receptor (P2X7R) splice variants A and B in AML progression and response to chemotherapy. The expression of P2X7RA and P2X7RB was investigated in samples obtained from MDS and AML untreated subjects or AML patients in relapse or remission after chemotherapy. Both P2X7RA and P2X7RB were overexpressed in AML versus MDS suggesting a disease-promoting function. However, in relapsing patients, P2X7RA was downmodulated, while P2X7RB was upmodulated. Treatment with daunorubicin (DNR), one of the main chemotherapeutics for AML, upregulated P2X7RB expression while reducing P2X7RA mRNA in AML blasts. Interestingly, DNR administration also caused ATP release from AML blasts suggesting that, following chemotherapy, activation of the receptor isoforms via their agonist will be responsible for the differential survival of blasts overexpressing P2X7RA versus P2X7RB. Indeed, AML blasts expressing high levels of P2X7RA were more prone to cell death if exposed to DNR, while those overexpressing P2X7RB were more vital and even protected against DNR toxicity. These data were reproducible also in HEK-293 cells separately expressing P2X7RA and B. P2X7RA facilitation of DNR toxicity was in part due to increased uptake of the drug inside the cell that was lost upon P2X7RB expression. Finally, in an AML xenograft model administration of DNR or the P2X7R antagonist, AZ10606120 significantly reduced leukemic growth and coadministration of the drugs proved more efficacious than single treatment as it reduced both P2X7RA and P2X7RB levels and downmodulated c-myc oncogene. Taken together, our data suggest P2X7RA and P2X7RB as potential prognostic markers for AML and P2X7RB as a therapeutic target to overcome chemoresistance in AML relapsing patients.


Introduction
Acute myeloid leukemia (AML) is a clonal disorder characterized by the proliferation and the accumulation of myeloid precursors in the bone marrow (BM) that show aberrant differentiation patterns leading to hematopoietic impairment 1 . In some cases, AML can originate from an evolving MDS, i.e., a clonal disorder of hematopoietic stem cells 2 . Despite several efforts have been made to improve the clinical outcome of AML, current drugs fail to eliminate the leukemic stem cells responsible for the disease 1 . Therefore, the prognosis remains poor with most AML patients developing drug resistance. Accordingly, the identification of new progression/relapse biomarkers and potential therapeutic targets remains an urgent clinical goal in AML 3 .
The P2X7 receptor (P2X7R) emerged as a promising therapeutic target in oncology, due to its widespread expression in cancer cells and the efficacy of its antagonist in reducing cancer growth and dissemination in animal models [4][5][6] . Haemopoietic lymphoproliferative disorders, among which AML, were the first neoplasias where P2X7R oncogenic activity was demonstrated [7][8][9] . However, evidence on the role of ATP and P2X7R in leukemia etiopathogenesis remains limited [10][11][12] .
P2X7R is an ATP-gated ion channel that, when stimulated with high concentrations of agonists, triggers cell death via the opening of a large nonselective plasma membrane pore (macropore) [13][14][15] . Formation of the macropore is dependent upon the presence of the long intracellular carboxyl tail as its deletion prevents pore formation while leaving unaltered the ion channel activity 14,16 . P2X7R activated macropore opening was also shown to enhance the intracellular uptake of drugs including chemotherapeutics such as doxorubicin 17 and was therefore proposed as tumor cell-specific drug delivery system 18,19 . The structure of P2X7R C terminal was recently reported 20 and it was speculated that this region was acquired by genomic rearrangement from a P2X4R like-gene in ancient jawed vertebrates generating the actual mammalian P2X7R 21 . It is therefore not surprising that certain species such as humans still express functional P2X7R splice variants missing the C terminal domain such as P2X7RB, an isoform unable to form the macropore, and thus without cytotoxic activity, but still endowed with ion channel properties 22,23 . In recent years, others and we demonstrated a growth-promoting activity for P2X7RB in cell lines 23,24 and osteosarcoma 24 as well as its involvement in metastasis and transformation 25,26 . The tumor-promoting activity of P2X7RB is shared by the fulllength P2X7RA isoform accelerates cancer growth and dissemination and associates with poor prognosis in different malignancies 5,[27][28][29] . However, to our knowledge, the effect of in vivo P2X7R blockade or chemotherapy on P2X7R isoforms expression was never tested and an extensive analysis of the distinct behavior of P2X7RA and P2X7RB in AML patients was never performed. The present study was aimed at identifying P2X7RA and B as novel biomarkers of AML response to chemotherapy and as new therapeutic targets for the disease. We covered P2X7R isoforms expression in an MDS and AML population and their behavior in vitro and in vivo models of the pathology, following chemotherapy and P2X7R antagonist administration.

Clinical samples
Primary leukemic cells were used for mRNA extraction and obtained, as previously described 10

P2X7R activity assays
Measurement of intracellular calcium concentration with FURA-2AM and of plasma membrane permeabilization to ethidium bromide was performed as previously described 30,31 .

ATP measurement
ATP concentration was evaluated with ENLITEN rLuciferase/Luciferin reagent (Promega, Milan, Italy), according to the manufacturer's instructions as previously described 12 .

Cells viability assays
AML blasts were resuspended in RPMI medium and treated with PBS or 200 nM DNR for 6 h. HEK cells seeded in serum-free medium let adhere and subsequently treated with PBS or 200 nM DNR for 48 h. Cell numbers were assessed at time 0 and 6 (blasts) or at time 0, 24, and 48 h (HEK). The Alamar blue assay was performed as per the manufacturer's instructions after 24 h of incubation with either PBS or 200 nM DNR.

In vivo experiments
Animal procedures were approved by the University of Ferrara ethic committee and the Italian Ministry of Health. In vivo experiments were performed in the athymic nude-Foxn1nu strain acquired from Envigo (San Pietro al Natisone, Italy). A sample size of 12 animals per condition was computed a priori with the G*power software 32 based on previous data obtained with AZ10606120 33 and assuming an effect size of 1 and a power of 85%. Four-to-six weeks old female mice were subcutaneously injected with 5 × 10 6 HL-60. Animals were randomized with the randomizer software (www. randomizer.org) to receive intramass injections of sterile PBS (placebo), DNR (150 µg/ml), AZ10606120 (2 μM), or both compounds at post-inoculum days 8 and 10 as previously described 33 . The operator was blinded to the group of allocation. Tumor size was measured at postinoculum days 8, 10, and 12 and ex vivo with a caliper and volume calculated according to the following equation: volume = π/6[w1 × (w2) 2 ], where w1 = major diameter and w2 = minor diameter. Mice were euthanized 12 days after cell inoculum and excised tumors were homogenized and processed for further analysis.

Statistics
All data are shown as mean ± standard error of the mean (SEM). Except for real-time PCR data reported in Fig. 1, where Welch's correction was applied due to different sample sizes, significance was calculated assuming equal standard deviations and variance, with a two-tailed Student's t test performed with the GraphPad Prism software (La Jolla, CA, USA). Coding: *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001; ****P ≤ 0.0001.

Results
P2X7RA and P2X7RB isoforms are overexpressed in newly diagnosed AML compared to MDS and differentially expressed in AML relapsing patients P2X7R expression in AML was already reported in the past 10 , however the analysis of the differential expression of P2X7RA and B isoforms in the disease was missing. Therefore, we investigated by Real time PCR mRNA levels of P2X7RA and B in 58 patients affected with AML and 10 patients affected with MDS. AML patients were further subdivided according to the diagnostic phase and response to treatment into three groups: (1) newly diagnosed untreated subjects (de novo, n = 47), (2) relapsing subjects, with a return of the pathology after chemotherapy (relapsing, n = 6), or (3) remitting patients, with no evident reappearance of the pathology one to three months after chemotherapy (remitting, n = 5). mRNA levels of P2X7RA and B of AML de novo patients were massively increased as compared to MDS blasts, supporting the hypothesis that both receptor variants positively correlate with disease progression (Fig. 1a, b). Interestingly, AML relapsing patients were characterized by differential expression of P2X7RA and P2X7RB. Indeed, while P2X7RA expression was significantly reduced at relapse (Fig. 1c), P2X7RB mRNA substantially increased in this subset of patients (Fig. 1d) suggesting that chemotherapy would cause a decrease in P2X7RA expression and on the contrary a positive selection of P2X7RB in subjects refractory to treatment. In remitting AML patients, both P2X7RA and B expression was significantly decreased as compared to de novo diagnosed patients (Fig. 1c, d). These data suggest that individuals expressing high levels of P2X7RB might be resistant to chemotherapy and prone to relapse.

DNR toxicity is increased by P2X7RA and reduced by P2X7RB expression
All of the relapsing and remitting patients tested underwent first-line therapy including treatment with anthracyclines (DNR and/or idarubicin), this class of drugs is known to cause an increase of ATP in the TME 11,34 . Accordingly, DNR treatment caused the release of ATP from AML blasts (Fig. 2a) and, interestingly, caused downregulation of P2X7RA and upregulation of P2X7RB as seen in AML relapsing patients (Fig. 2b, c). When analyzing P2X7R activity as calcium channel or macropore we could clearly distinguish de novo patients expressing high levels of P2X7RA mRNA (HIGH-P2X7RA), which were characterized by high intracellular calcium and ethidium uptake (Fig. 2d, e) from those expressing high levels of P2X7RB (HIGH-P2X7RB) that only showed ion channel activity (Fig. 2d, e). Interestingly, HIGH-P2X7RB AML blasts when in colture were more vital and if treated with DNR less prone to cell death than Both P2X7RA and P2X7RB isoforms mRNA are overexpressed in de novo AML as compared to MDS patients. a mRNA was extracted from primary leukemic cells from MDS and de novo AML patients and expression of P2X7R isoform A was compared (MDS, white, n = 10, de novo AML, black, n = 47). b mRNA was extracted from primary leukemic cells from MDS and de novo AML patients and expression of P2X7R isoform B was compared (MDS, white, n = 10, de novo AML, black, n = 47). c, d Within the AML group P2X7R isoforms are differently expressed in relapsing patients as compared to remitting patients. c Expression of P2X7R isoform A within the AML patient's group by subdividing into de novo (black), relapsing (red), remitting (green) (de novo n = 47, relapsing n = 6, remitting n = 5). d Expression of P2X7R isoform B within the AML patient's group by subdividing into de novo (black), relapsing (red) and remitting (green). (de novo n = 47, relapsing n = 6, remitting n = 5). Data are represented as mean ± SEM. *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001.
HIGH-P2X7RA blasts (Fig. 3a). To further dissect the different roles played by P2X7RA versus P2X7RB in response to DNR we analyzed the effect of this drug on HEK-293 cells separately expressing P2X7RA, P2X7RB, or an empty vector (HEK P2X7RA, HEK P2X7RB, HEK MOCK) 23 . Interestingly, DNR toxicity was strongly increased in HEK P2X7RA cells as compared to mock controls, while expression of P2X7RB not only did not affect viability but even protected the cells from DNRdependent death (Fig. 3b, c). These data were reproducible by both cell counts (Fig. 3b) and vitality assays with the Alamar Blue reagent (Fig. 3c). DNR treatment induced an ATP release from all tested cell lines (Fig. 3d) with a stronger effect on HEK P2X7RA, possibly due to increased cell death and consequent ATP release.
P2X7RA facilitates cellular uptake of DNR P2X7RA-mediated pore opening is known to mediate cell permeabilization to large molecules 14 , therefore we tested whether P2X7RA can further facilitate DNR-dependent cell death by favoring its cellular uptake. To test this hypothesis we took advantage of the natural red fluorescence emitted by DNR to measure its loading inside cells treated with 3 mM ATP (Fig. 4). As expected the only isoform able to promote DNR entry, possibly via macropore, was P2X7RA that significantly increased DNR uptake as compared to both control and P2X7RB expressing HEK cells (Fig. 4a). Similar results were obtained in HIGH-P2X7RA AML blasts, which, when treated with ATP, uploaded higher amounts of DNR than HIGH-P2X7RB patients cells (Fig. 4b-j).
DNR and the P2X7R antagonist AZ10606120 when coadministered are more efficacious than separately in reducing leukemic growth in an in vivo xenograft model To analyze the role of P2X7R isoforms in an AML in vivo model, we took advantage of HL-60 human promyelocytic cell line that expresses both P2X7RA and B at the mRNA and protein level (Fig. 5a, b), shows functional activity of the receptor as an ion channel and macropore showing an increase of intracellular calcium following stimulation with 500 µM BzATP of HIGH-P2X7RA AML patient blasts (gray) and HIGH-P2X7RB AML patient blasts (blue). e Representative traces showing ethidium bromide uptake following stimulation with 500 µM BzATP of HIGH-P2X7RA AML patient blasts (gray) and HIGH-P2X7RB AML patient blasts (blue). Data are represented as mean ± SEM. *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001; ****P ≤ 0.0001. (Fig. 5c, d), and releases ATP upon DNR treatment (Fig. 5e). We subcutaneously injected HL-60 cells in athymic nude mice. Tumor-bearing animals were then treated with intramass injections of DNR, the P2X7R antagonist AZ10606120, or both compounds together. Both drugs significantly reduced experimental leukemia growth (Fig. 6a-c). Moreover, coadministration proved to be more efficacious than single-agent treatment (Fig. 6a-c). DNR effect on P2X7R isoforms expression in HL-60 derived tumors was similar to patients' data as P2X7RB expression was significantly increased (Fig. 6e), while P2X7RA showed a tendency to reduction (Fig. 6d).
Interestingly, coadministration of DNR with AZ10606120 reduced both P2X7RA and P2X7RB (Fig. 6d-g). We had previously demonstrated that P2X7R antagonism reduced N-myc levels in neuroblastoma murine models 35 , therefore we asked whether AZ10606120 could similarly affect the expression of c-myc that is a well-known oncogene in AML and MDS 36 . Figure 6h, i shows a reduction of c-myc expression associated with P2X7R blockade in HL-60 xenografts. Interestingly, while DNR did not affect c-myc expression, P2X7R blockade alone tended to reduce and AZ10606120 in combination with DNR strongly diminished c-myc levels. These data are not depicted in the figure to increase clarity. Reported significance is relative to the comparison among different cell types all treated with DNR **P ≤ 0.01; ****P ≤ 0.0001. c Cell viability was assessed also by Alamar Blue assay, see "Materials, subjects, and methods." 20,000 cell/100 µl were seeded in a 96-well plate and treated with vehicle (PBS, black) or DNR (200 nM, red) for 24 h. n = 18 for each group. Data are represented as mean ± SEM. *P ≤ 0.05; **P ≤ 0.01. d Extracellular ATP (pM) was measured in the culture supernatants as described in "Materials, subjects, and methods." HEK MOCK, HEK P2X7RA, and HEK P2X7RB cells were treated for 24 h with vehicle (PBS black) or DNR (200 nM red). Data are represented as mean ± SEM. n = 6 for each group, *P ≤ 0.05; ***P ≤ 0.001,****P ≤ 0.0001.

Discussion
The P2X7R was previously reported to be expressed and functional in AML 37 and its targeting reduced leukemic growth in AML experimental models 12 . In vitro experiments also showed that high doses of ATP selectively disrupted the growth of leukemic blasts versus normal haemopoietic stem cells 10 . However, these studies focused on P2X7RA isoform that is known to mediate both cell proliferation upon stimulation by low doses of agonist and cell death through the opening of a large unselective pore in the presence of millimolar concentrations of ATP 38 . Humans also express another splice variant of the receptor named P2X7RB 22 , which while retaining the proliferation-promoting activity loses the ability to mediate the opening of the cytotoxic pore 23 . Limited evidence is available on the physio-pathological role of P2X7RB, we know that it is expressed mainly in the central and peripheral nervous system and in leukemic cells 23 and that, its expression associates to stem cell differentiation 39 . In solid cancers, P2X7RB was hypothesized to participate in proliferation 24 , matrix invasion, 26 and tumor spreading 25 . To better understand the different roles of P2X7RA and P2X7RB splice variants in leukemias, we investigated the expression of these isoforms in patients affected by MDS or AML and in AML relapsing and remitting patients. Both P2X7RA and P2X7RB mRNA levels were strongly increased in firstly diagnosed AML as compared to MDS, supporting the hypothesis that both receptor variants positively correlate with disease progression. Interestingly, AML relapsing patients after chemotherapy were characterized by showing an increment of intracellular calcium following stimulation with 500 µM BzATP alone (black) or applied after 5 min pretreatment with AZ10606120 (green) or following stimulation with 200 nM DNR alone (red). e Representative traces showing ethidium bromide uptake following stimulation with 500 µM BzATP alone (black) or applied after 5 min pretreatment with AZ10606120 (green) or following stimulation with 200 nM DNR alone (red). f Extracellular ATP (nM) was measured in the culture supernatants as described in "Materials, subjects, and methods." HL-60 cells were plated at 5 × 10 5 cells per ml and treated for 24 h with vehicle (PBS, black) and DNR (200 nM, red) (n = 8 for each group, vehicle versus DNR ****P < 0.0001). differential expression of P2X7RA and P2X7RB as compared to de novo patients as while P2X7RA expression was significantly reduced, P2X7RB expression significantly increased. On the contrary, in remitting AML patients both P2X7RA and B expression was significantly decreased. These data suggest that chemotherapy may cause a decrease in P2X7RA expression while positively modulating P2X7RB and therefore that individuals expressing high levels of P2X7RB could be resistant to chemotherapy, prone to relapse and refractory to treatment.
We recently demonstrated that the chemotherapeutic DNR can cause an increase of ATP in the leukemic microenvironment 11 and confirmed these data in all cell types tested in the present study. Based on these findings, we hypothesized that chemotherapy could cause a release of ATP from dying leukemic cells, which will, on one hand, kill P2X7RA overexpressing blasts through macropore formation and, on the other facilitate the proliferation of P2X7RB expressing blasts that do not undergo cell death but on the contrary exploit ATP-dependent proliferation (see Fig. 7).
(see figure on previous page) Fig. 6 DNR and AZ10606120 combined treatment is more effective in reducing HL-60 growth in nude mice in comparison to their single administration, affecting P2X7RA, P2X7RB, and c-myc expression. a-i Female nude mice were subcutaneously inoculated into the fat of the right limb with 5 × 10 6 HL-60 cells. DNR (0.75 mg/kg), AZ10606120 (5 mg/kg), and both drugs combined were i.m. administered to mice at postinoculum days 8 and 10. a, b Placebo depicted in black, DNR depicted in red, AZ10606120 depicted in green, DNR + AZ10606120 depicted in blue, n = 12 for each group. a At the indicated time points, tumor volume was in vivo assessed by a caliper. Data are represented as mean ± SEM. Significance is not depicted in the figure to increase clarity. Placebo versus DNR + AZ10606120 at day 10 P ≤ 0.0001, at day 11 P ≤ 0.01, at day 12 P ≤ 0.001. DNR versus DNR + AZ10606120 at day 10 P ≤ 0.05. AZ10606120 versus DNR + AZ10606120 at day 10 P ≤ 0.01. b Ex vivo tumor volume assessed at day 12 by a caliper. Data are represented as mean ± SEM. *P ≤ 0.05; **P ≤ 0.01; ****P ≤ 0.0001. c Representative picture of tumors from treated mice at post-inoculum day 12. d, e P2X7R isoforms A and B mRNA tumor levels were evaluated by real-time PCR as described in "Materials, subjects, and methods." d P2X7RA mRNA expression. Data are represented as mean ± SEM. **P ≤ 0.01; ***P ≤ 0.001. e P2X7RB mRNA expression. Data are represented as mean ± SEM. **P ≤ 0.01. f P2X7RA protein (~70 kDa) expression in excised tumors was evaluated by western blot with an antibody directed against P2X7R C terminal domain. Gel loading control Skp-1 (~20 kDa). g P2X7RB protein (~50 kDa) expression in excised tumors was evaluated by western blot with an antibody directed against P2X7R extracellular domain. Gel loading control Skp-1 (~20 kDa). h Representative Immunoblot showing c-myc expression (~50 KDa) in excised tumors. Gel loading control Skp-1 (~20 kDa). i Densitometric analysis of c-myc expression in HL-60 derived tumors normalized on Skp-1 loading control. Data were obtained in three independent experiments and are relative to three different tumor masses per group. Data are represented as mean ± SEM. *P ≤ 0.05.001. Fig. 7 Chemotherapy with anthracyclines induces ATP release in the tumor microenvironment. P2X7RA expressed by AML blasts in the presence of high ATP concentration mediates the opening of a large nonselective pore facilitating intracellular uptake of anticancer drugs that induce cell death. On the other hand, the increase of ATP allows the proliferation of P2X7RB expressing blasts, unable to form the cytotoxic pore but still able to activate the channel function of the receptor that protect cells from chemotherapy-dependent death and favors the relapse of AML. This hypothesis was confirmed by the finding that DNR toxicity was increased in HIGH-P2X7RA AML blasts, while the overexpression of P2X7RB protected these cells from DNR-dependent death, and further supported by similar data obtained in HEK-293 cells separately transfected with either P2X7RA or P2X7RB. These findings are in line with a differential role of the two isoforms in response to chemotherapy (see Fig. 7) and with P2X7RB upregulation in AML relapsing patients. To further dissect the mechanisms underlying P2X7RA facilitation of DNR toxicity we asked whether P2X7RA-mediated pore opening could facilitate DNR uptake by tumor cells, as previously demonstrated for the uptake of its analog doxorubicin in macrophages 17 . Indeed, this proved to be the case as DNR uptake by both HEK P2X7RA and HIGH-P2X7RA AML blasts stimulated with 3 mM ATP was higher than in HEK mock controls, HEK P2X7RB cells, and HIGH-P2X7RB AML blasts, respectively.
To validate patients' data in an in vivo model of leukemia we took advantage of HL-60 human promyelocytic cell line, which expresses both P2X7R isoforms, can form the cytolytic pore and releases ATP upon DNR stimulation. HL-60-derived leukemia-bearing mice were treated with DNR and with a P2X7R antagonist, alone or in coadministration. Both compounds significantly reduced leukemia growth, but their combination was more efficient than single-drug treatment. DNR effect on P2X7R isoforms expression was similar to what was seen in relapsing patients as P2X7RB expression was significantly increased by DNR, while P2X7RA showed a tendency to decrease. Interestingly, coadministration of DNR with P2X7R antagonist normalized both P2X7RA and B levels and reduced intratumor levels of the c-myc oncogene. These data are in line with our previous demonstration that P2X7R blockade decreases expression of the N-myc oncogene, another member of the myc family 35 , and suggest a mechanism for P2X7R-mediated transformation in AML. c-myc is commonly overexpressed in both AML and MDS 36,40 as it is a central transcription factor in the maturation of hematopoietic stem cells, thus exerting a critical function in hematopoietic malignancies 41 . Moreover, c-myc is involved in AML cells resistance to chemotherapy and therefore identifying new drugs able to downmodulate c-myc levels might greatly improve the therapeutic approach of AML relapsing patients 42,43 . AML relapse is the main cause of patients' death and efforts to improve their survival is an important goal for onco-hematologists to ensure patients' recovery. Our study suggests that P2X7R isoforms could be exploited as valuable prognostic markers to predict response to chemotherapy. Moreover, P2X7R could be an interesting target for a new combined therapy in AML. Based on our data, we suggest a therapeutic approach with DNR administered as first-line, to eliminate P2X7RA-expressing leukemic cells. P2X7R antagonists could be administered in a second phase to those that developed chemotherapy resistance and whose blasts are positive for P2X7RB expression. Of interest, several P2X7R targeting drugs are under clinical trials for other pathologies and proved to be well tolerated with limited side effects thus favoring their use also in oncologic patients 5,44,45 .