PKC epsilon is a Key Regulator of Mitochondrial Redox Homeostasis in Acute Myeloid Leukemia

Abstract

Purpose

The intracellular redox environment of acute myeloid leukemia (AML) cells is often highly oxidized compared to healthy hematopoietic progenitors and this is purported to contribute to disease pathogenesis. However, the redox regulators that allow AML cell survival in this oxidized environment remain largely unknown.

Experimental Design and Results

We show that RNA interference-mediated inhibition of the serine/threonine kinase PKC-epsilon (PKCε) reduces cell survival in a diverse panel of patient-derived AML samples and significantly delays disease onset in a genetically engineered mouse model (GEMM) of AML driven by MLL-AF9. Utilizing a combination of chemical and genetically-encoded redox sensing probes, we found that PKCε inhibition leads to the induction of multiple reactive oxygen species (ROS) including multiple mitochondrial ROS. We also show that neutralization of mitochondrial ROS with chemical anti-oxidants or co-expression of the mitochondrial ROS-buffering enzymes SOD2 and CAT, mitigate the anti-leukemia effects of PKCε inhibition. Similar to PKCε inhibition, direct inhibition of SOD2 also increases mitochondrial ROS and significantly impedes disease progression in vivo. Furthermore, we report that over-expression of PKCε protects AML cells from otherwise-lethal doses of mitochondrial ROS-inducing agents. Proteomic analysis reveals that PKCε may control mitochondrial ROS by controlling the expression of regulatory proteins of redox homeostasis, electron transport chain flux, as well as outer mitochondrial membrane potential and transport.

Conclusions

This study uncovers a previously unrecognized role for PKCε in supporting AML cell survival and disease progression by regulating mitochondrial ROS biology and positions mitochondrial redox regulators as potential therapeutic targets in AML.

INTRODUCTION

The intracellular redox environment is largely influenced by the production and clearance of reactive species such as reactive oxygen species (ROS). ROS encompass a heterogeneous class of small oxygen-containing reactive species that are produced by several intracellular processes such as aerobic glucose metabolism and enzymatic reactions (1) and contribute to physiological functions such as innate and acquired immune defense. However, aberrantly elevated ROS levels, referred to as “oxidative stress”, can cause DNA lesions, organelle dysfunction and metabolic alterations that contribute to tumorigenesis (2). Furthermore, extremely high ROS levels can promote cell death (2). Therefore, intracellular ROS levels are tightly regulated by an array of enzymatic and non-enzymatic systems (1).

Several human cancers, including AML, display perturbations in genes that encode regulators of intracellular ROS biology (3). For example, primary AML patient samples display decreased expression of the mitochondrial superoxide-neutralizing gene SOD2 and decreased glutathione metabolism (4). The activities of the ROS-generating enzymes NADPH-oxidases (NOXs) are also elevated in primary human AML samples compared to healthy controls and this is associated with increased steady-state levels of intracellular superoxides (5). Moreover, the total anti-oxidant capacity of leukemic cells from AML patients at initial diagnosis and relapse is decreased compared to healthy controls (6). Several of the oncogenic signaling molecules that are either mutated (e.g. FLT3^ITD, KRAS^G12D or BCR-ABL1) (7–12) or dysregulated (e.g. c-MYC) (13) in myeloid malignancies are known to drive ROS production. Furthermore, small nucleotide polymorphisms (SNPs) of several redox-regulatory enzymes have been identified to be associated with myeloid neoplasia susceptibility and prognosis (14). Despite these studies as well as the emerging evidence that elevated ROS levels promote the proliferation and survival of solid cancers (15, 16), the underlying molecular mechanisms that govern ROS biology in AML remain unresolved.

PKCε is a serine/threonine kinase that is expressed in numerous tissues and supports tumorigenesis in many solid cancers (17). In normal hematopoietic development, PKCε influences erythrocyte and megakaryocyte lineage commitments (18); however, in primary myelofibrosis, elevated PKCε expression antagonizes megakaryocytic differentiation (19). Furthermore, down-modulation of PKCε expression is a key event during phorbol ester-induced differentiation of primary human AML samples and increased PKCε expression protects AML cells from TRAIL-induced apoptosis (20). However, the molecular mechanism(s) by which PKCε influences AML cell fate has yet to be resolved.

Here, we show that PKCε inhibition obstructs disease progression in a GEMM of AML driven by MLL-AF9 and impairs cell survival in multiple patient-derived AML samples. Furthermore, we find that PKCε inhibition results in increased steady-state levels of multiple mitochondrial ROS and that chemical or genetic neutralization of mitochondrial ROS counteracts the anti-leukemia effects of PKCε. Furthermore, we report that over-expression of PKCε protects AML cells from otherwise lethal doses of mitochondrial ROS-inducing agents. Finally, direct inhibition of the mitochondrial ROS-neutralizing enzyme SOD2 phenocopies PKCε inhibition indicating that maintenance of mitochondrial ROS homeostasis is crucial to the maintenance of AML pathogenesis.

MATERIALS AND METHODS

Cell culture

Human AML cell lines were obtained from the American Type Culture Collection (ATCC) and the German Collection of Microorganism and Cell Cultures (DMSZ). Cells were cultured in the recommended media conditions. Murine AML cells were cultured in cytokine-enriched media (CEM): RPMI 1640 supplemented with 10% FBS and penicillin/streptomycin, 10 ng/ml mSCF (Peprotech), 6 ng/ml mIL-3 (Peprotech) and 5 ng/ml mIL-6 (Peprotech).

Patient-derived AML samples

Patient-derived AML samples were obtained from The Ohio State Comprehensive Cancer Center and used according to the approved IRB protocol 16-9037. Cells were plated on irradiated mono-layers of HS27 cells (21) and cultured in Stemspan (Stem Cell Technology) supplemented with 10% FBS, 100ng/ml hSCF (Peprotech), 100ng/ml hFLT3 ligand (Peprotech), 20ng/ml hIL-3 (Peprotech), 20ng/ml hIL-6 (Peprotech), 20ng/ml G-CSF (Peprotech). 500,000 cells were transduced with pLKO.1 GFP lentiviral shRNA vectors and evaluated for GFP expression every 3 days for 12 days after staining with human CD45 APC-Cy7 (BD Biosciences) and Propidium Iodide (PI) by flow cytometry.

Lentiviral transduction

500,000 human AML cells were transduced for 30 hours with recombinant pLKO.1 lentiviruses co-expressing either GFP or a puromycin resistance cassette with shRNAs from the TRC shRNA library (shRNA Pkcε: TRCN0000022759; shRNA PKCε_1: TRCN0000000848 and shRNA PKCε_2: TRCN0000000846: shRNA SOD2: TRCN0000350349; shRNA Sod2_1: TRCN0000123392; shRNA Sod2_2: TRCN0000123390. Cells transduced with recombinant lentiviruses expressing the puromycin resistance cassette were initially selected with 2μg/ml puromycin for 48 or 72 hours (MOI ~0.5–0.9) and thereafter maintained in 0.5μg/ml puromycin. To generate pLKO.1 GFP vectors, the puromycin resistance cassette was replaced with eGFP. Cells transduced with lentiviruses expressing GFP (MOI ~0.2–0.5) were purified by FACS at the indicated times following transduction.

Retroviral transduction

500,000 human or murine AML cells were subjected to spin-transduction with recombinant retroviruses (MOI ~ 0.2–0.3). The MSCV.PKCε.GFP construct was obtained by cloning murine Pkcε into the BglII and XhoI sites of the MSCV.IRES.GFP vector. Cells transduced with vectors expressing GFP were purified by FACS 96 hours after transduction. SF91-IRES-eGFP and SF91-SOD2/Catalase-IRES-eGFP vectors, were described previously (22) and provided by Dr. M. Milsom. SF91-Grx1-roGFP2, SF91-Mito-Grx1-roGFP2, SF91-roGFP2-Orp1 and SF91-Mito-roGFP2-Orp1 were described previously (23) and provided by Dr. T. Dick.

Western blot analysis

Cell pellets were lysed and resolved on 4–12% Bis Tris gels (ThermoFisher – Life technologies). Following protein transfer and blocking with 5% nonfat milk, blots were incubated with primary antibody overnight at 4 °C. Secondary antibody was incubated at room temperature for 1 hour and blots were developed with ECL prime (GE healthcare). The following antibodies were used: anti-PKCε (Cat# 2683S: Lot#: 4) and anti-Catalase (Cat#: 12980S, Lot#: 1) from Cell Signaling Technologies with the dilution 1:1000, anti- (Cat#: ADI-SOD-111, Lot#: 08021202) from Enzo Life Sciences with the dilution 1:1000, anti-α-tubulin (Cat#: T9026, Lot#: 083M4847V) from Sigma-Aldrich with the dilution 1:5000. Secondary antibodies anti-rabbit-HRP (Cat#: 7074S, Lot#: 26) and anti-mouse-HRP (Cat#: 7076S, Lot#: 31) were from Cell Signaling Technologies with the dilution 1:4000.

MTS assay

Cells were plated at the density of 100,000–200,000 cells/ml at either 72- or 96-hours post-transduction (Day 0). At Days 0, 2 and 4, 100 μl of cell suspension was incubated with 20 μl of CellTiter Aqueous One solution (Promega) for 90 minutes. Plates were read at the wavelength of 495nm. Culture media was used as blank.

Apoptosis assay

Cells were washed in PBS and stained with Annexin V and Propidium Iodide (PI) or 7AAD according to the manufacturer’s instruction (BD Biosciences). Cells were acquired and analyzed using an LSRII flow cytometer (Beckton Dickinson). All the data were analyzed using flowjo software.

CellROX and MitoSOX Staining

Human AML cell lines were washed and incubated with PBS + 5μM of CellROX DeepRed reagent (Life Technology Cat#: C10422) at 37°C for 30 minutes, or with PBS + 5μM of MitoSOX Red reagent (Life Technology Cat#: M36008) at 37°C for 10 minutes. After incubation, cells were washed 2 times in PBS and stained with Annexin V-APC according to the manufacturer’s instruction (BD Biosciences). Cells were analyzed using an LSRII flow cytometer (Beckton Dickinson).

roGFP analysis

Human AML cell lines stably-expressing the roGFP2 probes (27) were transduced with recombinant lentiviruses for 30 hours, then selected with 2μg/μl puromycin for 72h and subsequently analyzed by flow cytometry. RoGFP2 is excited at 400 nm (oxidized state) and 475–490 nm (reduced state) when fluorescence emission is monitored at 510 nm. 1mM diamide (DIA) or 1mM dithiothreitol (DTT) were used as controls to induce a complete oxidized or reduced state of the probes, respectively. Propidium Iodide staining was performed to exclude dead cells from the analysis.

Pro-oxidant treatment

Human AML cells stably expressing PKCε, PKCε-targeting shRNAs or corresponding controls were seeded at the concentration of 200,000 cells/ml in 24 well plates and treated with the indicated concentrations of Antimycin A (AA) (Sigma Aldrich A8674) and Thenoyltrifluoroacetone (TTFA, Sigma Aldrich T27006) for 24h and then analyzed by flow cytometry after Annexin V and 7AAD staining.

Bone marrow transplant leukemia model

All animal studies conducted were approved by the IUCAC of the Fox Chase Cancer Center. For recombinant viral transduction, bone marrow cells recovered from leukemia mice were cultured in CEM media overnight. The next day, cells were counted and then 500,000 cells were subjected to spin-transduction with recombinant lentiviruses (MOI ~ 0.4–0.6) or retroviruses (MOI ~ 0.2–0.3) supplemented with polybrene (5 μg/ml) in 12 well non-adherent plates. Plates were centrifuged at 2,400 rpm at 30 °C for 90 minutes. Transduced cells were then incubated overnight. The following day, viral supernatants were removed from cells and replenished with fresh CEM. For in vivo survival assays, sorted GFP⁺ cells (1,000,000 cells/mouse) were transplanted into sub-lethally irradiated (450 rad) syngenic recipient mice 48-hours post-transduction. For western blot analysis and colony formation assays, transduced cells were subjected to FACS to isolate GFP⁺ cells 48-hours post-transduction. For colony formation assays, 500 purified GFP⁺ leukemia cells were cultured in 1 ml of methylcellulose supplemented with cytokines (M3434, Stem Cell Technologies) for 5–7 days.

RESULTS

PKCε inhibition impairs in vitro and in vivo AML cell expansion and survival

To evaluate the functional role of PKCε expression in AML biology, we employed an shRNA approach in a panel of genetically distinct AML cell lines in vitro (OCI-AML3, THP-1, NOMO1 and U937). We identified two shRNA constructs, PKCε shRNA_1 and PKCε shRNA_2, that target distinct regions of the PKCε mRNA and effectively deplete PKCε protein levels (Supplemental figure 1A). Each of these PKCε-targeting shRNAs significantly reduced the expansion of the four AML cell lines compared to non-targeting shRNA controls (CTRL shRNA) (Supplemental figure 1B). This reduction in cell growth was accompanied by a significant increase in the percentage of Annexin V⁺ cells (Supplemental figure 1C), and CD11b expression (Supplemental figure 1D & E) indicating that PKCε inhibition is detrimental to survival and growth and may induce differentiation of these AML cell lines.

To assess the impact of PKCε inhibition in vivo, we utilized a GEMM of AML that is driven by the human leukemogenic fusion protein MLL-AF9 (24). Mouse MLL-AF9 leukemia cells were transduced with recombinant lentiviruses that co-express GFP in combination with either control (CTRL shRNA) or murine Pkcε-targeting shRNAs (Pkcε shRNA). The transduced cells were subsequently purified by FACS to assess protein expression growth in vitro and leukemia induction in vivo (Supplemental figures 2A & B). Depletion of Pkcε protein significantly reduced the growth of mouse MLL-AF9 leukemia cells in cytokine-enriched liquid culture (Figures 1A & B and Supplemental figures 2C–E). Furthermore, mice transplanted with FACS-purified mouse MLL-AF9 leukemia cells expressing Pkcε shRNA exhibited a significantly longer onset of disease compared to CTRL shRNA expressing cells (Figure 1C). Furthermore, depletion of Pkcε protein significantly reduced the colony forming capacity (CFC) of mouse MLL-AF9 leukemia cells in cytokine-enriched methylcellulose (Figures 1D).

Figure 1. PKCε inhibition impairs AML cell expansion and survival in vitro and in vivo.

(A) Western blot analysis of mouse MLL-AF9 leukemia cells transduced with CTRL or Pkcε shRNA. (B) In vitro competitive growth curve of mouse MLL-AF9 cells transduced with either CTRL or Pkcε shRNA GFP lentiviruses. %GFP+ cells were evaluated every two days by flow cytometry and normalized to fold change in %GFP+ at Day 3 post-transduction, which represents Day 0 in the figure. (Day 6 = Day 9 post-transduction). (C) Kaplan-Meier survival curve analysis of mice transplanted with mouse MLL-AF9 leukemia cells co-expressing GFP and either CTRL or PKCε shRNAs (p=0.0014; n=7). (D) MLL-AF9 and (E) MLL-AF9;Flt3^ITD knock in (KI) cells transduced with lentiviruses co-expressing GFP with either CTRL or Pkcε shRNA were FACS-purified and plated in M3434. (F) Dnmt3a^−/−;Tet2^−/− and (G) Dnmt3a^−/−;Tet2^−/−;FLT3^ITD cells transduced with lentiviruses co-expressing RFP with either CTRL or Pkcε shRNA were FACS-purified and plated in M3434. (D–G) Bar graph representing the number of colonies formed by mouse MLL-AF9 leukemia cells expressing CTRL or Pkcε shRNAs in methylcellulose culture. Data are represented as the mean ± SD of three technical replicates for B, D-G. (***p≤0.001, ***p≤0.0001).

In addition to MLL-AF9, we observed that PKCε supports the growth of mouse hematopoietic stem and progenitor cells (HSPCs) by expressing alterations in genes commonly mutated in AML. Briefly, leukemia cells co-expressing MLL-AF9 and an internal tandem duplication (ITD) of the murine Flt3 gene (MLL-AF9;Flt3^ITD) were transduced with lentiviruses expressing GFP in combination with either CTRL or Pkcε shRNAs. HSPCs null for the combination of Dnmt3a and Tet2 deletion (Dnmt3a−/−;Tet2−/−) as well as Dnmt3a−/−;Tet2−/−HSPCs co-expressing human FLT3-ITD and GFP (Dnmt3a−/−;Tet2−/−;FLT3^ITD) were transduced with lentiviruses expressing RFP in combination with either CTRL or Pkcε shRNAs. Following stable transduction, cells from each condition were purified by FACS and plated separately in cytokine-enriched methycellulose. In all of the models analyzed, shRNA-mediated inhibition of PKCε expression significantly hindered the CFC of leukemic cells compared to CTRL shRNA-expressing cells (Figures 1E–G).

PKCε supports the survival of patient-derived AML samples

To investigate the functional role of PKCε in patient-derived AML samples, we employed an shRNA approach to inhibit the expression of PKCε in 10 sub-type diverse patient-derived AML samples (Supplemental Table 1 and Supplemental figure 3A). Briefly, cryopreserved patient-derived samples were thawed and plated on the HS-27 supportive stroma in cytokine-enriched media. After a short recovery period, cells were transduced with lentiviruses co-expressing GFP and either CTRL shRNA or PKCε shRNA_1 and subsequently re-plated on fresh HS-27 stroma cells. Three days following transduction, cells were assessed by flow cytometry to determine the percentage of live hCD45⁺, GFP⁺ cells (represented as time point day 0). Cells were co-cultured for an additional 9 days and then assessed for the percentage of live hCD45⁺, GFP⁺ cells (time point day 9). We then determined the fold change in the percentage of hCD45⁺, GFP⁺ cells by dividing the day 9 percentages of hCD45⁺, GFP⁺ divided by the percentages of hCD45⁺, GFP⁺ at day 0. From this analysis, we observed that 8 of the 10 patient samples expressing PKCε-targeting shRNAs displayed a significantly lower fold change in the percentage of hCD45⁺, GFP⁺ cells compared to those samples expressing CTRL shRNAs (Figure 2 and Supplemental figure 3B). These data indicate that PKCε supports leukemia growth in a genetically diverse subset of patient-derived AML samples.

Figure 2. PKCε inhibition impedes Patient-derived–AML growth.

Ex vivo growth curve analysis of 10 patient-derived AML samples transduced with lentiviruses co-expressing GFP and CTRL, PKCε shRNA_1 and then analyzed by flow cytometry for the GFP⁺, hCD45⁺, PI^- cells every 3 days for a total of 12 days, post-transduction (P.T.). AML samples displayed maximal GFP mean fluorescence intensity (MFI) at three days P.T. and thus represents analysis point, Day 0. The data presented are expressed as the fold change of GFP⁺ cells at day 9 (day 12 P.T.) versus day 0, relative to the mean of the CTRL shRNA (*p≤0.05; **p≤0.01;***p≤0.001; ****p<0.0001). Data shown are the end point analyses (day 12 P.T.) and are represented as mean ± SD of three technical replicates.

PKCε regulates the intracellular redox environment of AML cells

Given the emerging recognition that ROS influences both normal and malignant HSPCs (4–6) and the previous connections of PKCε to redox biology (25–28), we investigated the relationship between PKCε and ROS biology in AML cells. As an initial assessment of whether shRNA-mediated inhibition of PKCε impacted intracellular ROS levels, we stained CTRL and PKCε-shRNA expressing cells with the fluorogenic probe CellROX^™, which detects multiple types of ROS. From this analysis, we found that inhibition of PKCε resulted in an increase of CellROX staining in OCI-AML3, NOMO1 and THP-1 cells (Figure 3A and Supplemental figures 3C and D).

Figure 3. PKCε regulates intracellular ROS biology in AML.

(A) OCI-AML3 cells stably expressing CTRL shRNA or PKCε shRNAs were stained with CellROX, 5-days post-transduction. The left panel is a representative histogram plot of a single experiment and the right panel shows the average MFI of CellROX. Data are represented as the mean ± SD of three technical replicates. (B) OCI-AML3 cells were stably transduced with the indicated roGFP2 probes to evaluate glutathione or H₂O₂ redox potential in cytoplasm or mitochondria. Cells were transduced with lentiviruses expressing CTRL shRNA or PKCε shRNA_1 or PKCε shRNA_2 and analyzed by flow cytometry 5 days later. Bar graph represents the percentage of live (PI^-) oxidized cells in the cytoplasm (Cyto) or mitochondria (Mito). Data represents the mean ± SD of three independent experiments. (C) OCI-AML3 were transduced with CTRL shRNA or PKCε shRNA and stained after 4 days with MitoSOX and Annexin V (to exclude dead cells). The left panel is a representative histogram plot of a single experiment and the right panel shows the average MitoSOX MFI of live cells (Annexin V^-) expressed as the mean ± SD of three independent experiments. (D) MLL-AF9 leukemia cells co-expressing GFP and either CTRL or PKCε shRNAs were stained and evaluated for MitoSOX levels by flow cytometry, 5 days after transduction. Bar graph represents the MitoSOX MFI of live cells (Annexin V^-) expressed as the mean ± SD of three independent experiments. (*p≤0.05, **p≤0.01; ***p≤0.001; ****p<0.0001).

To determine the localization and further define the specific types of ROS regulated by PKCε, OCI-AML3 cells were engineered to express the redox-sensitive roGFP2 protein genetically fused with glutaredoxin (Grx1) or the hydrogen peroxide (H₂O₂)-neutralizing yeast peroxidase Orp1. The Grx1-roGFP2 probe measures the redox potential of GSH:GSSG redox couples, where roGFP2-Orp1 probe measures changes in H₂O₂ levels. Both probes are expressed in the cytoplasm, however, we also engineered cells to express versions of each probe that are tagged with mitochondrial localization signals (Grx1-roGFP2–mito and roGFP2-Orp1–mito) to evaluate changes in mitochondria ROS biology (Supplemental figures 4A & B). roGFP2 has an emission of 510nm as well as excitation peaks at 400nm and 490nm. Upon oxidation, the intensity of the 400nm roGFP2 peak increases while the amplitude of the 490nm peak decreases resulting in the ratio of peak intensities shifting towards 400nm (23). OCI-AML3 cells were treated with either the oxidizing agent diamide (DIA) or the reducing agent dithiothreitol (DTT) to establish the oxidized and reduced gates (data not shown). Using these probes, we observed that the expression of PKCε shRNA_1 or shRNA_2 resulted in a significantly higher percentage of oxidized OCI-AML3 cells expressing either cytoplasmic or mitochondrial Grx1-roGFP2, compared to control shRNAs. We also observed that PKCε inhibition led to a significant increase in the percentage of cells expressing oxidized roGFP2-Orp1 in the mitochondria (Figure 3B & Supplemental figures 4A & B).

The most well-defined ROS generated by mitochondria are superoxides, which are either neutralized by glutathione or converted to H₂O₂ by superoxide-dimutase (SOD) enzymes. Given that PKCε inhibition consistently altered mitochondrial levels of glutathione and H₂O₂, we next examined how PKCε inhibition impacted mitochondrial superoxide levels using the fluorogenic probe MitoSOX^™. From this analysis, we observed that shRNA-mediated inhibition of PKCε led to a significant increase in MitoSOX staining in both human (OCI-AML3, Figure 3C & Supplemental figures 5A and THP-1, Supplemental figure 5B) and mouse (MLL-AF9, Figure 3D) AML cells. We also observed that shRNA-mediated inhibition of PKCε in patient derived AML cells expressing MLL-AF9 also display elevated MitoSOX levels (Supplemental figure 5C). Collectively, these findings reveal that PKCε regulates steady-state levels of mitochondrial ROS and possibly certain cytoplasmic ROS in human and mouse AML cells.

Reducing mitochondrial ROS partially reverses the anti-leukemia effects of PKCε inhibition

To determine whether changes in the intracellular redox biology contribute to the anti-leukemia effects of PKCε inhibition, we assessed how various chemical anti-oxidants impacted the growth and survival of AML cells expressing CTRL and PKCε-targeting shRNAs. Twice daily treatment with either N-acetyl-L-cysteine (NAC) or glutathione was unable to reverse the anti-leukemia effects of PKCε inhibition in both human and mouse cells despite being able to effectively block the cytotoxicity of the glutathione-depleting pro-oxidant, Menadione (Supplemental figures 6A & B and data not shown). However, administration of either butylated hydroxyanisole (BHA) or MitoTEMPO, both compounds that neutralize mitochondrial ROS, was able to significantly reduce the cell death of human and mouse AML cells mediated by PKCε inhibition (Figures 4A & B and Supplemental figure 6C).

Figure 4. Neutralization of ROS partially reverses the anti-leukemia effects of PKCε inhibition.

(A & B) OCI-AML3 cells were transduced with CTRL shRNA, PKCε shRNA_1 or PKCε shRNA_2 and then administered 25μM BHA (A) or 100nM MitoTEMPO (B). Ninety-six hours post-transduction, cells from each condition were assessed for Annexin V staining by flow cytometry. Data are represented as the mean ± SD of two independent experiments. (C) OCI-AML3 cells stably transduced with retroviruses co-expressing SOD2-IRES-Catalase (SOD2-Catalase) and GFP or control vector (CTRL) were transduced with CTRL shRNA, PKCε shRNA_1 or PKCε shRNA_2. Four days after transduction, cells were analyzed for cell death evaluated as percent of Annexin V⁺ cells by flow cytometry. Data are represented as the mean ± SD of three independent experiments for each panel. (*p≤0.05;**p≤0.01). (D) SOD2-Catalase or CTRL-expressing mouse MLL-AF9 were transduced with CTRL shRNA or Pkcε shRNA, selected with puromycin and grown in methylcellulose. After 5 days of culture, colonies were enumerated and data represents the mean ± SD of two independent experiments. (*p≤0.05, **p≤0.01; ***p≤0.001; ****p<0.0001).

To examine how specific neutralization of mitochondrial ROS impacts the anti-leukemia effects of PKCε inhibition, we engineered OCI-AML3 cells to constitutively co-express SOD2 and Catalase (SOD2-Catalase), which neutralize mitochondrial superoxides and H₂O₂, respectively. Control and SOD2-Catalase expressing OCI-AML3 cells were then transduced with CTRL or PKCε shRNAs and subsequently analyzed for MitoSOX levels and cell death. SOD2-Catalase expression significantly blocked the induction of MitoSOX staining mediated by PKCε inhibition (Supplemental figure 6D) and significantly reduced cell death induced by PKCε-targeting shRNAs (Figure 4C). We also observed that SOD2-Catalase expression restored the CFC of mouse MLL-AF9 leukemia cells expressing Pkcε shRNA (Figure 4D) further indicating that AML cells rely on PKCε to maintain mitochondrial ROS levels and survival.

The mitochondrial ROS-neutralizing enzyme, SOD2 supports AML in vitro and in vivo

To examine how a specific induction of mitochondrial superoxides impacts AML cell growth, we evaluated how shRNAs directly targeting SOD2 impacted human AML cell biology (Supplemental figure 7A). Similar to PKCε inhibition, shRNA-mediated depletion of SOD2 led to increased steady-state levels of MitoSOX and reduced AML cell growth and survival (Supplemental figures 7B–D). Furthermore, direct inhibition of SOD2 in mouse MLL-AF9 leukemia cells (Figure 5A) led to increased levels of MitoSOX staining (Figure 5B), reduced CFC (Supplemental figures 7E) and significantly delayed the time of disease onset in vivo (Figures 5C). These results show that similar to the anti-leukemia effects of PKCε inhibition, SOD2 inhibition leads to an accumulation of mitochondrial ROS and diminishes AML cell growth and survival.

Figure 5. SOD2 inhibition phenocopies the anti-leukemic effects of PKCε inhibition.

(A) Mouse MLL-AF9 leukemia cells were transduced with CTRL shRNA or SOD2 shRNA, sorted 48 hours later and analyzed for SOD2 expression by western blot. (B) MitoSOX staining of mouse MLL-AF9 leukemia cells evaluated by flow cytometry 5 days after transduction with CRTL, Sod2 shRNA_1 or Sod2 shRNA_2 (The bar graph shows the MitoSOX MFI of live cells (Annexin V^-) (Ctrl shRNA vs. shRNA sod2_1 and _2. *p≤0.05). Data represents the mean ± SD of three technical replicates. (C) Kaplan-Meier survival curve analysis of mice transplanted with mouse MLL-AF9 leukemia cells expressing either CTRL or Sod2 shRNA_2 (p=0.0042; n=7).

PKCε regulates the expression of proteins that regulate mitochondrial biology

To identify potential downstream effectors of PKCε in AML, we performed nanoscale liquid chromatography coupled to tandem mass spectrometry (nanoLC-MS/MS) on OCI-AML3 cells expressing CTRL shRNA, PKCε shRNA_1 or PKCε shRNA_2. From this analysis, 3192 peptides were captured and using a statistical cut-off of p<0.05, we found that 707 peptides were differently expressed between CTRL shRNA and PKCε shRNA_1-expressing cells and 641 peptides were differently expressed between CTRL shRNA and PKCε shRNA_2-expressing cells. Furthermore, we observed that 226 peptides were similarly differentially expressed between OCI-AML3 cells expressing CTRL shRNA versus PKCε shRNA_1 and CTRL shRNA versus PKCε shRNA_2. Integrated pathway analysis did not reveal any significant enrichment of particular molecular pathways or processes however, we did observe that approximately 15% (34 proteins) of these differentially expressed proteins between CTRL shRNA and either PKCε shRNA_1 or PKCε shRNA_2 were related to mitochondrial biology. Aberrant mitochondrial ROS arise from multiple disruptions in mitochondrial biology such as reduced expression of anti-oxidant systems, perturbations in the activities of the complexes that regulate electron flux through the electron transport chain (ETC), as well as alterations in outer mitochondrial membrane (OMM) potential and transport. Many of the proteins whose expression is altered by PKCε inhibition are components of ETC complexes involved in the regulation of OMM potential (UQCR10, ATPC1, ATP5H, COX6C, SDHAF2 and NDUFB10) or proteins involved in mitochondrial membrane transport (VDAC1, VDAC3, TOMM22, SLC25A1, SLC25A11 and SLC25A12) (Figure 6A). We also observed that two anti-oxidant proteins, GSS and TXN were significantly reduced by PKCε inhibition as well as a variety of other anti-oxidant proteins that were reduced but not statistically significant (Supplemental figure 7F).

Figure 6. PKCε protects AML cells against agents that induce mitochondrial dysfunction and mitochondrial ROS-oxidative stress.

(A) Heatmap analysis displaying the expression of mitochondrial-regulatory proteins that were significantly differentially expressed between CTRL shRNA versus PKCε shRNA_1 or versus PKCε shRNA_2 transduced OCI-AML3 cells. (B & C) OCI-AML3 cells were transduced with retroviral vectors that constitutively express PKCε and GFP (PKCε) or just GFP-expressing control (Ctrl) viruses. Four days after transduction, GFP⁺ cells were isolated by FACS and treated with 100μM of (B) AA or (C) TTFA. The percentage of Annexin V⁺ cells was evaluated 24 hours after treatment by flow cytometry. Data are represented as the mean ± SD of three independent experiments. (D) Ctrl and PKCε-expressing mouse MLL-AF9 leukemia cells were generated as described in panel B & C. GFP⁺ cells were isolated by FACS and seeded in methylcellulose with or vehicle, 100μM AA or 200 μM TTFA. The bar graph shows the number of colonies formed in methylcellulose after 5 days of culture. The data represent the mean ± SD of three technical replicates. (**p≤0.01;***p≤0.001)

PKCε protects AML cells against agents that promote mitochondrial dysfunction and induce superoxides

To assess the role PKCε in superoxide-induced oxidative stress, we examined how modulating PKCε expression impacted the survival of AML cells challenged with either of the mitochondrial superoxide inducing agents, 2-thenoyltrifluoroacetone (TTFA) or antimycin A (AA). To examine whether increased expression of PKCε is able to protect AML cells from the cytotoxic effects of TTFA or AA, we generated OCI-AML3, THP-1 and mouse MLL-AF9 expressing AML cells that constitutively express PKCε (Supplemental figure 8A and data not shown). Constitutive PKCε expression did not impact OCI-AML3 growth or the CFC of mouse MLL-AF9 leukemia cells (Supplemental figures 8B & C). However, PKCε-expressing OCI-AML3 or THP-1 cells treated with TTFA or AA displayed significantly lower percentages of cell death compared to similarly-treated control cells (Figures 6B & C and Supplemental figures 8E & F). Additionally, constitutive PKCε expression significantly improved the CFC of mouse MLL-AF9 leukemia cells challenged with TTFA or AA compared to similarly treated control cells (Figure 6D). Of all the human AML cell lines we tested, U937 cells were the least impacted by PKCε inhibition, cell viability-wise (Supplemental figure 1B, bottom panel). Therefore, we evaluated whether PKCε inhibition rendered U937 cells more sensitive to TTFA and/or AA treatment. From this analysis, we found that shRNA-mediated inhibition of PKCε exacerbated the cytotoxic effects of TTFA and AA in comparison with PKCε inhibition or pro-oxidant treatment alone (Supplemental figure 8G & H). Collectively, these results suggest that PKCε protects AML cells from agents that perturb mitochondrial function and induce oxidative stress.

DISCUSSION

PKCε has been implicated as an oncogenic kinase in several human cancers including prostate, breast, colon, lung and certain forms of squamous cell carcinoma (29–34). In AML, down-modulation of PKCε is necessary for the pro-differentiating actions of phorbol esters (20, 35), insinuating that PKCε supports leukemia growth and expansion. However, the impact of blocking PKCε expression on AML growth and survival has not been comprehensively investigated.

Here, we demonstrate that shRNA-mediated reduction of PKCε significantly reduces the survival of human AML cells in vitro and significantly impedes disease progression in a GEMM of AML driven by MLL-AF9. Additionally, inhibition of PKCε reduced the in vitro growth properties of multiple, genetically diverse patient-derived AML samples confirming that the pro-leukemia roles of PKCε is not restricted to a particular genetic sub-type of AML.

At the molecular level, we have discovered that PKCε is a key regulator of intracellular redox homeostasis in several mouse and human AML models. Using multiple ROS-detection strategies we have found that shRNA-mediated inhibition of PKCε increases the steady-state levels of multiple ROS, including several mitochondrial ROS. Based on these observations, we postulated that increased/excess production of mitochondrial ROS antagonizes leukemia cell viability and that management of mitochondrial ROS levels is a key pro-leukemia function of PKCε. We have made three central observations that support these hypotheses. First, chemical anti-oxidants that specifically neutralize mitochondrial ROS, such as BHA (36) and MitoTEMPO (37) are able to significantly and consistently blunt cell death mediated by PKCε inhibition, whereas indiscriminant anti-oxidants such as NAC and glutathione are not. Moreover, reconstitution of mitochondrial ROS-neutralizing enzymes SOD2 and Catalase, partially reverses mitochondrial superoxide induction and cell death mediated by PKCε inhibition. Second, similar to PKCε inhibition, shRNA-mediated inhibition of SOD2 increased mitochondrial ROS, limited disease progression in vivo and suppressed the growth of patient-derived AML samples. Third, elevated expression of PKCε is able to largely protect AML cells from otherwise toxic doses of agents that drive mitochondrial ROS production. Moreover, impeding PKCε expression rendered leukemia cells more sensitive to these pro-oxidants. Collectively, these results support that certain sub-types of AML rely on PKCε for proper management of mitochondrial ROS biology and survival.

Our observations that the anti-leukemia effects of PKCε inhibition could not by fully rescued by SOD2-Catalase expression, MitoTEMPO or BHA suggest that PKCε may regulate additional molecular processes to support AML cell survival. Our proteomic analysis shows that, in addition to multiple key ROS-buffering enzymes, such as TXN and GSS, several proteins related to mitochondrial biology and function are impacted by PKCε inhibition. Therefore, it remains possible that the increase in mitochondrial ROS mediated by PKCε inhibition is due to mitochondrial dysfunction either in addition to or in place of the observed decrease in ROS-regulating enzymes. Consistent with this idea, several studies have shown that the mitochondrial redox state of AML cells is directly related to their metabolic needs. Specifically, compared to bulk AML cells, leukemia-initiating cells from multiple AML patients display a lower oxidized redox environment that is associated with lower rates of oxidative phosphorylation (OXPHOS) (38). Furthermore, cytarabine-treated AML cells display high levels of mitochondrial ROS and OXPHOS and this altered mitochondrial state may contribute to cytarabine resistance (39). Also, multiple AML patients display increased mitochondrial mass compared to healthy HSPCs and as a result are more sensitive to OXPHOS-induced oxidative stress (40).

Though our results indicate that a key pro-leukemia function of PKCε is to manage mitochondrial ROS biology, they do not exclude the possibility that PKCε regulates additional redox-related (that are mitochondrial-independent) and/or redox-independent mechanisms to support AML cell survival. For example, our proteomic analysis shows that PKCε inhibition, although not statistically significant, impacts multiple redox-regulatory systems. For example, protein levels of both PRDX2 and PRDX4, which have been implicated as growth suppressors in AML and acute promyelocytic leukemia (APL), respectively, increase upon PKCε inhibition (41, 42). PKCε inhibition also altered the expression of various glutathione-regulatory components and Pei et al. (4) have shown that pharmacological inhibition of glutathione metabolic enzymes, such as GPX1 and GCLC, antagonize primitive human leukemia cell survival. In fact, the relationship of PKCε and redox biology varies among distinct biological settings. PKCε activation in neuronal and cardiac tissues correlates with the induction of ROS mediated by ischemia, hypoxia or pro-oxidants such as buthionine sulfoximine (BSO) or AA. However, other studies have shown that PKCε activation promotes ROS generation in smooth muscle and immortalized epithelial cells and that hepatocytes void of PKCε display enhanced stress-induced ROS formation (28, 43). These divergent observations may result from distinct tissue- or ROS-specific roles of PKCε; however, in the context of AML, our results establish that PKCε works to suppress mitochondrial ROS and possibly other types of ROS.

Several studies suggest that tumor cells, including leukemia cells, maintain high levels of ROS to drive cell growth and survival and therefore targeting redox regulators may be a viable anti-cancer therapeutic strategy (2, 34, 44). In our models, reducing steady-state levels of mitochondrial superoxides through the combined over-expression of SOD2 and Catalase does not impede colony formation suggesting that elevated mitochondrial ROS are not a central driver of AML cell growth in this model. However, increasing mitochondrial ROS, by inhibiting PKCε or SOD2 or by administering chemical ROS-inducing agents diminishes AML cell survival. Collectively, these results suggest that strategies for increasing, rather than decreasing, mitochondrial ROS may carry a significant therapeutic potential. Consistent with this concept, the efficacy of arsenic trioxide, which is commonly used to treat APL, works primarily by inducing ROS (45). Moreover, high doses of Vitamin C, which have been previously shown to induce toxic ROS levels in certain types of cancer cells (46), were recently reported to selectively eliminate AML cells carrying TET2 (47, 48) or IDH (49) mutations. However, it should be noted that the proposed mechanism of action of Vitamin C in AML is to activate other TET family members (47, 48).

The intracellular redox environment of AML cells is often distinct from their normal counterparts. Therefore, identifying and defining the molecular regulators of redox biology, such as PKCε and SOD2, may provide key insights into the etiology and pathogenesis of AML as well as possibly contribute to the design of more effective anti-leukemia therapies. However, AML encompasses a wide variety of genetic sub-types and individual tumors often display complex clonal heterogeneity (50) and it remains unclear which genetic sub-types are susceptible to redox imbalances. Our loss-of-function studies in patient-derived AML cells show that not all AML samples rely on PKCε for growth and survival. Thus, it is possible that certain AML cells utilize PKCε- and/or SOD2-independent mechanisms to regulate mitochondrial superoxide biology or that certain genetic sub-types or clones are insensitive to increases in mitochondrial superoxide levels. Therefore, future studies defining the genetic sub-types that are sensitive to changes in redox homeostasis or PKCε/SOD2 inhibition as well as the role of PKCε/SOD2 in healthy HSPC biology will be needed to fully gauge the therapeutic potential of targeting these pathways in AML.

STATEMENT OF TRANSLATIONAL RELEVANCE.

Patient derived AML cells often display significantly higher levels of intracellular ROS compared to their normal counterparts. Elevated ROS levels are often associated with increased DNA damage, dysfunctional organelle biology, oncogenic signaling and altered cellular metabolism and thus are purported to contribute to disease pathogenesis and therapeutic responses in a variety of human tumor settings. However, excess ROS can also promote tumor cell death and therefore intracellular ROS are tightly maintained by various regulatory mechanisms. Therefore, regulators of redox biology may provide opportunities for therapeutic intervention. We have revealed that PKCε suppresses mitochondrial ROS to support AML and that blocking PKCε or enzymes that directly neutralize mitochondrial ROS, such as SOD2 diminishes AML cell survival by inducing lethal ROS levels.