Diphenyleneiodonium triggers cell death of acute myeloid leukaemia cells by blocking the mitochondrial respiratory chain and synergizes with cytarabine

Acute myeloid leukaemia (AML) is characterized by the accumulation of undifferentiated blast cells in the bone marrow and blood. In most AMLs, relapse frequently occurs due to resistance to chemotherapy. Compelling research results indicate that drug resistance in cancer cells is highly dependent on the intracellular levels of reactive oxygen species (ROS). Modulating ROS levels is therefore a valuable strategy to overcome the chemotherapy resistance of leukemic cells. In this study, we evaluated the efficiency of diphenyleneiodonium (DPI), a well-known inhibitor of ROS production, in targeting AML cells. Results showed that although inhibiting cytoplasmic ROS production, DPI triggered an increase in the mitochondrial ROS levels caused by the disruption of the mitochondrial respiratory chain. We also demonstrated that DPI blocks the mitochondrial oxidative respiration (OxPhos) in a dose-dependent manner and that AML cells with high OxPhos status were highly sensitive to treatment with DPI, which synergizes with the chemotherapeutic agent cytarabine (Ara-C). Thus, our results suggest that targeting mitochondrial function by DPI might be exploited to target AML cells with high OxPhos status.


Introduction
Acute myeloid leukaemia (AML) is a heterogeneous clonal disorder of myeloid progenitors that accumulate due to a blockage in their differentiation, leading to death [1]. AML therapy has not much changed over the last decades, and more than 70% of AML patients relapse within 3-years after therapy [2]. AML relapse is caused by residual populations of quiescent leukemic stem cells (LSCs), associated with chemoresistant AML cells that have a high mitochondrial oxidative phosphorylation [3,4]. Altered cellular redox status with high reactive oxygen species (ROS) levels is indeed a common hallmark of AML cells. Several lines of evidence have indicated that NOX complexes that are major contributors of ROS production, including superoxide (O2 •-) and hydrogen peroxide (H2O2), are also important regulators of AML progression and drug resistance [5,6]. Thus, targeting oxidative metabolism in AML has been proposed as a promising therapeutic strategy to eradicate AML cells [7].
We recently showed that all components of NOX2, the most prominent NOX complex across AML, are highly expressed at both the transcriptional and protein levels. Surprisingly, we did not find detectable constitutive NOX activity in 24 leukemic cell lines [8]. In addition, we demonstrated that NOX2 silencing neither affected AML cell growth nor triggered cell death in vitro. Adane et al. have recently shown that, although silencing of NOX2 induces the differentiation of primary AML cells, diphenyleneiodonium (DPI) used to inhibit NOX, did not affect the differentiation but triggered apoptosis [9]. While DPI is widely used to prove NOX activity [10], it is a non-specific inhibitor of flavoproteins that can impede the activity of nitric oxide synthases (NOS), xanthine oxidases (XOS), and complexes I and III of the mitochondrial respiratory chain (MRC) [11][12][13]. Importantly, DPI has been found to trigger the inhibition of the mitochondrial oxidative metabolism (OxPhos) in breast cancer cells and to induce a chemo-quiescent phenotype that blocked the propagation of cancer stem cells [14].
Recently, it has been proposed that drug resistance of AML cells might be dependent on their OxPhos status [3]. Hence, we hypothesized that DPI could target OxPhos in AML cells rather than its canonical NOX inhibition. Thus, we examined the effects of DPI on oxidative metabolism, proliferation and resistance to chemotherapy in various AML cell lines harbouring low and high OxPhos phenotypes.

DPI reduces cytoplasmic ROS, while induces superoxide production
To study the effect of DPI on AML cells, we used eight AML cell lines, covering M0-M5 FAB stages [15] and having no endogenous NOX activity [8]. First, we measured ROS production rate using CM-H2DCFDA that detects cytoplasmic ROS (cytoROS) production, mainly H2O2, and dihydroethidium (DHE) taht detects intracellular superoxide (O2 •-). We noticed a high heterogeneity in ROS production between the cell lines ( Figure 1a). KG-1, HL-60, NB-4, and THP-1 cells clustered together, showing concomitant high production rates of O2 •-and cytoROS, suggesting steady transformation of O2 •-into H2O2. Markedly, KG-1a cells (M0), a model of immature AML, derived from KG-1, showed low production rates of superoxide (23 RFU/min) and cytoROS (59 RFU/min). This agrees with the idea that more mature cells have higher ROS levels. After treatment with 20 µM DPI, a dose sufficient to inhibit flavoproteins, all cell lines showed a substantial increase in O2 •-production, while only five of them had decreased cytoROS production rates, compared to their respective controls (Figure 1b-c).
To investigate the origin of O2 •-increase following DPI treatment, we measured mitochondrial O2 •-(mitoROS) production using MitoSOX, a DHE derivative that is specific to mitochondria. At steady state, the profile of O2 •-production detected by MitoSOX was concordant to that obtained by DHE ( Figure 1d). Noteworthy, KG-1a cells had the lowest level. Remarkably, DPI treatment triggered a strong induction of mitoROS levels in all cell lines (Figure 1e). Although KG-1a cells showed the strongest induction (40-fold) with DPI, its increased mitoROS level never reached the baseline levels of the other cell lines. Together, these data showed that DPI decreases cytoplasmic ROS production but concomitantly triggers an increase in mitochondrial ROS production. and effect of DPI on mitochondrial ROS (e) (n=4). Data are shown as mean values ± SEM. One-sample t-test was used to compare normalized rates to 1 ( * p < 0.05; * * p < 0.01; * * * p < 0.001).

DPI disrupts the mitochondrial membrane potential
To explain the quick mitoROS burst induced by DPI, we speculated that DPI may have induced an oxidative stress by disrupting the mitochondrial respiratory chain (MRC). We thus examined the functional impact of DPI on the mitochondrial activity of AML cells by labelling with tetramethylrhodamine ethyl ester (TMRE) as a readout to determine the effects on mitochondrial membrane potential (ΔΨm). FCCP, a common mitochondrial-depolarizing agent was used as a positive control. The basal level of ΔΨm was variable across the cell lines (Figure 2a; control black bars). This variability was mainly due to differences in the mitochondrial mass, as determined by Mitotracker Deep Red labelling ( Figure 2b and 2c). DPI notably decreased ΔΨm in all cell lines, except in KG-1a cells in which ΔΨm was negligible, in accordance with the lowest mitochondrial biomass found in this cell line (Figure 2a, b, c). Collectively, these data indicated that DPI triggers O2 •production in AML cell lines by inhibiting MRC.   (Table 1). Finally, to prove that blocking OCR may induce mitoROS, the three cell lines were treated either with DPI or a combination of rotenone and antimycin. As expected, all inhibitors triggered a similar increase in mitoROS level ( Figure 2d). Together, these data indicate that DPI blocks MRC in AML cell lines and induces an oxidative burst with a similar efficiency to standard inhibitors.

DPI reduces cell proliferation and triggers apoptosis
Since DPI induced oxidative stress by blocking mitochondrial respiration, we examined its impact on cell growth and survival. To address the effect of a chronic exposure, we used a low dose of DPI (0.2 µM) and followed the proliferation of AML cell lines for three days. The results showed that all cell lines exposed to DPI had a significant reduction in their expansion capacity compared to their corresponding controls (Figure 4a). Furthermore, to investigate whether reduced expansion resulted from proliferation slowdown or cell death induction, we quantified apoptosis at day 3 of culture, using Annexin-V and 7-AAD (Figure 4b). Following DPI treatment, most cell lines showed moderate to high levels of apoptosis that could partially explain cell growth reduction (Figure 4a-b). NB-4, THP-1, and MV-4-11 cells, which showed the highest apoptosis rates (Figure 4b), were also those with the highest maximal OCR capacities (Figure 3b). In contrast, KG-1a cells that showed low mitoROS and minimal ΔΨm (Figures 1d and 2a) also showed minimal apoptosis (Figure 4b). Together, these data suggest that DPI reduces cell growth by inhibiting cell division in an apoptosis-dependent manner and that cells with high OxPhos metabolism are more sensitive to DPI-induced apoptosis. (b) Apoptosis in DPI-treated AML cell lines. 7-AAD/Annexin V staining distinguishes between live, early apoptotic and late apoptotic cells. Data are shown as mean ± SEM (n = 3 independent experiments). Student's ttest was used to compare DPI conditions to their corresponding control counterparts ( * p < 0.05; * * p < 0.01; * * * p < 0.001).

Effect of combination therapy of DPI and cytarabin on AML cell lines
Recent findings have suggested that AML cells with high OxPhos are more resistant to therapeutic agents [3,4]. Therefore, we investigated whether DPI may synergize with Ara-C to eliminate AML cells. To address this issue, we used two representative cell lines, THP-1 and MV-4-11, with high OxPhos, and KG-1a, with the lowest OxPhos status. A dose-response matrix was designed to test 35 different combinations of doses ranging from 0 to 0.5 µM for Ara-C and 0 to 0.4 µM for DPI ( Figure  5a). Data showed that the combination of DPI and Ara-C had synergistic effect in THP-1 and MV-4-11 cells (positive Loewe scores) but not in KG-1a (a negative score) (Figure 5a-b). This suggests that only the cell lines with high OCR might be sensitized by DPI.

Discussion
This study aimed to understand the mode of action of DPI in AML cell lines. We demonstrated that DPI, although inhibiting cytoplasmic ROS production, disrupts MRC, increases mitochondrial ROS production, and triggers the apoptosis of AML cell lines, especially those with a high-OxPhos status.
In addition, we showed that DPI synergizes with Ara-C in eliminating high-OxPhos AML cells preferentially.

Preprints (www.preprints.org) | NOT PEER-REVIEWED | Posted: 30 November 2021
First, we showed that DPI reduces cytoROS production in five of the eight tested cell lines. This decrease was previously shown in three leukemic cell lines (KU-812, MOLM-13 and HEL) treated with DPI [16]. DPI had also been found to reduce the cytoROS production, thereby impacting the proliferation of prostate cancer cells [17]. Moreover, our results showed that DPI can inhibit the mitochondrial respiration, induce a rise in mitoROS level and induce apoptosis. This is in line with the data showing that DPI induces cell cycle arrest and decreases the mitochondrial potential of prostate cancer cell lines [18]. Furthermore, Ozsvari et al. demonstrated that the treatment of breast cancer cell lines with DPI inhibited mitochondrial oxidative metabolism (OxPhos), thereby reducing mitochondrial ATP production by more than 90% [14]. However, DPI did not trigger the production of mitochondrial ROS in these cells, which could be explained by the use of a very low dose (10 nM).
Collectively, these data showed that DPI can efficiently inhibit mitochondrial respiration in different types of cancer cells even at very low doses.
Our data indicated that DPI can readily impede the mitochondrial respiration in OxPhos-high AML cells, independently of NOX inhibition, causing a strong burst of superoxide production. Moreover, we recently reported that another NOX inhibitor, VAS3947, induces the apoptosis of AML cells through cysteine thiol alkylation, independently of NOX inhibition [19]. Several studies indicated that NOX complexes are important regulators of AML progression and drug resistance [9,[20][21][22][23][24][25], but many of these have used non-specific inhibitors, including DPI and VAS3947, to prove NOX activity or to study the functional impact of NOX inhibition on leukemic cells. Our findings suggest that the use of such inhibitors to study the role of NOX in oxidative metabolism can be misleading and highlight the need to develop more specific NOX inhibitors.
We showed for the first time that DPI acts synergistically with Ara-C to induce apoptosis in AML cells having a high-OxPhos metabolism (THP-1 and MV-4-11), but not in low-OxPhos cells (KG-1a).
The combined treatment of DPI with FLT3-ITD inhibitors, midostaurin or sorafenib, has been found to synergistically inhibit the proliferation of AML cell lines harbouring FLT3-ITD, and its combination with the tyrosine kinase inhibitor imatinib has been showed to synergistically increase apoptosis in chronic myeloid leukemia (CML) cells in vivo [25]. Remarkably, the authors have demonstrated that the viability of healthy CD34-positive cells has not been affected by DPI, suggesting that this compound might be safely used in the treatment of myeloid leukaemias. Recent reports revealed that chemoresistance and relapse may arise from cells bearing high-OxPhos metabolism [3,4]. Notably, Ara-C resistant AML populations exhibit metabolic characteristics and gene signatures compatible with a high-OxPhos status [3]. In these cells, targeting the mitochondrial metabolism induced an energetic shift towards low-OxPhos and enhanced anti-leukaemic effects of Ara-C [3]. Altogether, these data support our findings that targeting the high-OxPhos status of AML cells might help to overcome the resistance to chemotherapy.
In summary, this work reports that DPI antileukemic activity is caused by the inhibition of MRC and OxPhos disruption. We also found that DPI can synergize with Ara-C in targeting high-OxPhos AML cells. Thus, our data pave the development of therapies that specifically target mitochondrial respiration in myeloid leukaemias in the future.

Mitochondrial respiration
Oxygen consumption rate (OCR) was quantified using a Seahorse XFe96 Analyzer (Agilent Technologies), as described previously with slight modifications [26]. Briefly, cells were plated at rotenone/antimycin A (0.5 µM; Sigma) were used to determine the main respiratory parameters, in particular the acute response to DPI and the impact of the latter on maximal respiration. A modified strategy was used to calculate IC50 values for DPI, rotenone and antimycine A, as described in Figure   2d. The effect of DPI, or standard inhibitors, concentrations on the OCR variation were compared with the effect of the rotenone antimycin A mix, used as a positive control for inhibition of the respiratory chain. This allowed to define the IC50 of the respiratory activity for each compound.

Apoptosis assay
Cells were cultured alone or in the presence of DPI (0.2 µM). Three days following drug addition, cells were harvested and washed with cold PBS and then resuspended in Annexin V Binding Buffer (BioLegend, London, UK). Next, the cells were stained with APC-conjugated Annexin V (BioLegend) and 7-AAD (Sigma-Aldrich) according to BioLegend's instructions and then analysed using C6 Accuri® flow cytometer and FlowJo® software.

Drug combination assay
Cells were seeded in 160 µL media at a density of 4x10 3 cells/well and incubated overnight at 37°C.
They were then exposed to various concentrations of DPI and Cytarabine (Ara-C) (Sandoz France Levallois-Perret, France) in a final volume of 200 µL. Proliferation assay was followed after 72 h of treatment through resazurin fluorescence assay. Resazurin (0.1 mg/mL) was added at 20 µL/well and incubated for 4 h at 37°C in the dark, then fluorescence was (λex = 529.5-19 nm, λem = 582-36 nm) measured using ClarioStar microplate reader. Synergy analyses was done in R environment using the Synergyfinder package [27].

Conclusions
This work 1) demonstrates that DPI affects AML proliferation in absence of NOX activity, 2) confirms its inhibitory effect on MRC, and 3) shows that combining conventional chemotherapy with a MRC inhibitor may help to eradicate the chemotherapy resistance of leukemic cells.