Knockdown of BNIP3L or SQSTM1 alters cellular response to mitochondria target drugs

ABSTRACT

Macroautophagy/autophagy, a pathway by which cellular components are sequestered and degraded in response to homeostatic and cell stress-related signals, is required to preserve hematopoietic stem and progenitor cell function. Loss of chromosomal regions carrying autophagy genes and decreased autophagy gene expression are characteristic of acute myeloid leukemia (AML) cells. Deficiency of autophagy proteins is also linked to an altered AML metabolic profile; altered metabolism has recently emerged as a potential druggable target in AML. Here, we sought to understand the mitochondria-specific changes that occur in leukemia cells after knockdown of BNIP3L/Nix or SQSTM1/p62, which are two autophagy genes involved in mitochondrial clearance and are downregulated in primary AML cells. Mitochondrial function, as measured by changes in endogenous levels of reactive oxygen species (ROS) and mitochondrial membrane potential, was altered in leukemia cells deficient in these autophagy genes. Further, these AML cells were increasingly sensitive to mitochondria-targeting drugs while displaying little change in sensitivity to DNA-targeting agents. These findings suggest that BNIP3L or SQSTM1 may be useful prognostic markers to identify AML patients suitable for mitochondria-targeted therapies.

Abbreviations: AML: acute myeloid leukemia; DHE: dihydroethidium; mtDNA: mitochondrial DNA; NAO: 10-N-nonyl acridine orange; PD: population doubling; R123: rhodamine 123; ROS: reactive oxygen species; TRC: transduced scramble controls

Macroautophagy/autophagy, a pathway by which cellular components are sequestered and degraded in response to homeostatic and cell stress-related signals, is required to preserve hematopoietic stem and progenitor cell function. Loss of chromosomal regions carrying autophagy genes and decreased autophagy gene expression are characteristic of acute myeloid leukemia (AML) cells [1]. Deficiency of autophagy proteins is also linked to an altered AML metabolic profile [1]; altered metabolism has recently emerged as a potential druggable target in AML. Here, we sought to understand the mitochondria-specific changes that occur in leukemia cells after knockdown of BNIP3L/Nix or SQSTM1/p62, which are 2 autophagy genes involved in mitochondrial clearance and are downregulated in primary AML cells [2,3]. Mitochondrial function, as measured by changes in endogenous levels of reactive oxygen species (ROS) and mitochondrial membrane potential, was altered in leukemia cells deficient in these autophagy genes. Further, these AML cells were increasingly sensitive to mitochondria-targeting drugs while displaying little change in sensitivity to DNA-targeting agents. These findings suggest that BNIP3L or SQSTM1 may be useful prognostic markers to identify AML patients suitable for mitochondria-targeted therapies.

Mitochondrial regulation by the autophagy machinery is central to overall cell function. Knockout of Atg5 (autophagy related 5) or Atg7, which participate in autophagosome formation, results in mice with altered hematopoiesis that eventually develops into myeloproliferative disease; Atg7 expression is also downregulated in AML patient samples [1,4]. Further, atg5^−/- and atg7^−/- mice display an altered metabolic profile characterized by decreased resting oxygen consumption and increased steady state levels of ROS [1,4]. In fact, mitochondrial changes such as elevated endogenous ROS and/or increased mitochondrial mass are characteristics of cells isolated from a variety of autophagy gene knockout mice (e.g., atg3^−/-; rb1cc1/fip200^−/-, and becn1^−/-). Altered mitochondria are also a common feature in AML. Compared to normal cells, AML cells have increased mitochondrial mass [5], a reduced ability to switch between oxidative phosphorylation and glycolysis following BCL2 inhibition [6], low spare reserve capacity [7], and a dependency on mitochondrial fatty acid oxidation for cell proliferation [8]. Loss of chromosomal regions carrying autophagy genes, decreased expression of autophagy genes, and decreased autophagic flux are characteristic of AML blasts.

Expression of BNIP3L, which localizes to mitochondria and is ubiquitinated to signal mitochondria-specific autophagy (i.e., mitophagy) [9,10], is reduced in AML patients compared to normal controls [3]. BNIP3L also plays a role in tumor suppression and apoptosis [11]. The protein SQSTM1/p62, which participates in cell proliferation, apoptosis and facilitates phagophore formation around ubiquitinated mitochondria [12], maintains myeloid progenitor function in mice, as gene knockdown results in bone marrow loss of hematopoietic progenitors [13]. SQSTM1 knockdown also results in the development of atypical mitochondria leading to reduced viability of RAS-transformed lung cancer cells [14]; AML cells have reduced SQSTM1 expression compared to normal granulocytes [2]. Because SQSTM1 and BNIP3L are genes that participate in autophagy and mitochondrial health, and are involved in hematopoiesis and AML, we sought to investigate the direct role of these genes in AML and link gene expression to drug sensitivity.

BNIP3L and SQSTM1 were knocked down in OCI-AML2 cells using lentiviral-mediated transduction. Protein levels for BNIP3L (Figure 1(a) left) and SQSTM1 (Figure 1(b) left) were measured by immunoblotting and were less than 30 and 55%; respectively, compared to transduced scramble controls (denoted TRC). To assess the impact of gene knockdown on cell phenotype, we measured changes in cell size using a Multisizer 4 Particle Analyzer. Knockdown of BNIP3L or SQSTM1 resulted in no significant change in cell size (Figure 1, middle panels). To assess growth, viable cells were enumerated daily and population doubling (PD) was calculated using the following equation: PD = (Count on Day 2)/(Count on Day 1). There was no statistically significant difference in PD between knockdown cells and transduced controls (Figure 1, right panels). Together these results show that autophagy gene knockdown does not alter overall cell size or proliferation.

Figure 1.

Autophagy gene knockdown in OCI-AML2 cells. Immunoblots and densitometry showing protein levels of (a; left panel) BNIP3L in OCI-AML2 cells treated with 3 independent BNIP3L vectors (denoted: 1, 3 and 5) or a scramble control (denoted TRC) and (b; left panel) SQSTM1/p62 in OCI-AML2 cells treated with a scramble or 2 independent SQSTM1 vectors (denoted: SQ11 and SQ12). Arbitrary units were normalized to loading controls (GAPDH) and standardized to scramble controls. Representative blots shown. (a and b middle panel) Mean cell size was measured with a Multisizer 4 Particle Analyzer in leukemia cells deficient in BNIP3L or SQSTM1. (a and b; right panel) Population doubling was determined in autophagy gene knockdown cells. Data presented as mean ± standard deviation of 3 independent experiments.

Because BNIP3L and SQSTM1 play a role in the regulation of mitochondrial health, we next assessed whether gene knockdown was related to changes in mitochondrial function and abundance. We first measured endogenous levels of superoxide, a reactive oxygen species (ROS) generated primarily from mitochondria, using the fluorescent probe dihydroethidium (DHE). In both BNIP3L and SQSTM1 knockdown leukemia cells, basal levels of DHE were elevated compared to scramble controls (Figure 2(a): F_4,28 = 43.41; p < 0.0001; F_3,44 = 8.39; p < 0.0001; respectively). Increased mitochondrial membrane potential, which is related to the flow of electrons through the electron transport chain, can directly lead to elevated levels of ROS. Thus, we next investigated whether the observed increase in ROS was directly related to elevated membrane potential using the fluorescent dye R123, which accumulates with greater frequency in mitochondria with higher (i.e., more negative) membrane potential. Not surprisingly, membrane potential was significantly elevated in both BNIP3L and SQSTM1 knockdown cells (Figure 2(a); F_4,49 = 48.06; p < 0.0001; F_3,44 = 13.61; p < 0.0001; respectively). However, these changes in mitochondrial function did not affect mitochondrial abundance, as mitochondrial mass, which was measured by the florescent dye 10-N-nonyl acridine orange (NAO) (Figure 2(b)), and mitochondrial DNA (mtDNA), as measured by qPCR, were unchanged following BNIP3L or SQSTM1 knockdown (Figure 2(b)).

Figure 2.

Leukemia cells deficient in BNIP3L or SQSTM1 have altered mitochondrial function but not abundance. Leukemia cells deficient in BNIP3L (left panel) or SQSTM1 (right panel) were assessed for changes in (a) mitochondrial function by measuring basal levels of ROS (using DHE) or membrane potential (using R123) by flow cytometry and (b) mitochondrial abundance by measuring mitochondrial mass (using NAO) and mitochondrial DNA (mtDNA using qPCR). For NAO staining data presented as mean fluorescence normalized to TRC controls; for DHE and R123 presented as percent increase compared to TRC control; and for PCR, data presented as MT-ND1:HBB normalized to TRC controls. **p < 0.01***; p < 0.001.

To link loss of BNIP3L and SQSTM1 with mitochondria quality control, we isolated mitochondrial fractions at increasing time points and measured levels of the mitochondria-specific proteins SLC25A4/ANT and MT-ND1, as well as LC3B-II, a marker indicative of autophagosome formation. Consistent with our observations, there was no difference in protein levels of these 2 mitochondria-specific proteins, but levels of LC3B-II were lower over time (Figure S1). These data suggest that autophagy gene knockdown does not affect overall mitochondrial abundance, but causes a change in mitochondrial function, likely owing to a decrease in mitochondria quality. This effect, where gene knockdown was not sufficient to alter overall mitochondrial abundance, could be attributed to a lack of complete gene knockout or the redundancy of pathways within the autophagy machinery. Indeed, several studies have shown that cells lacking BNIP3 have elevated protein levels of BNIP3L highlighting a potential compensatory mechanism resulting from only single autophagy gene knockdown [15,16]. Thus, future experiments would require complete elimination of the gene (i.e. with CRISPR-Cas9) or the silencing of multiple genes within the autophagy pathway.

Because BNIP3L and SQSTM1 can be downregulated in AML patients, we next assessed whether gene expression could predict AML cell response to standard AML therapeutics. Cytarabine is an anti-metabolic agent that damages DNA during the synthesis phase of the cell cycle. Doxorubicin and daunorubicin are anthracycline agents that impart their activity by intercalating with DNA and inhibiting TOP2/topoisomerase II. These DNA damaging agents, which do not primarily target mitochondria, are added in a 7ʹ+3ʹ protocol consisting of 7 days of treatment with cytarabine followed by 3 days of treatment with an anthracycline. The cytotoxicity of these DNA damaging agents was not affected by autophagy gene knockdown (Table 1). Given the observed changes in mitochondrial function, we next assessed whether these autophagy gene knockdown cells would be more sensitive to drugs that target mitochondria. Cells deficient in BNIP3L were increasingly sensitive to treatment with FCCP and rotenone (Table 1), drugs that accumulate in mitochondria and disrupt metabolism. EC₅₀ values decreased 87–99% and 41–63% in knockdown cells compared to TRC controls following FCCP and rotenone treatment, respectively (Table 1). Similarly, knockdown of SQSTM1 resulted in leukemia cells with greater sensitivity to these drugs compared to TRC controls (Table 1). EC₅₀ values decreased 47–48% and 10–13% in knockdown cells compared to controls following FCCP and rotenone treatment, respectively (Table 1). We next profiled SQSTM1 and BNIP3L gene expression in a panel of AML cell lines. Levels of BNIP3L mRNA were significantly higher (2- to 5-fold) compared to SQSTM1 (Figure S2). Given the range of gene expression seen only with BNIP3L coupled with the low SQSTM1 expression, we stratified cell line response to rotenone and FCCP vs. BNIP3L expression. Consistent with the gene knockdown data, low levels of BNIP3L were associated with greater drug toxicity, whereas high expression was associated with higher leukemia cell viability following treatment with FCCP and rotenone (Figure 3).

Table 1.

Leukemia cells deficient in BNIP3L or SQSTM1 are increasingly sensitive to mitochondria-targeting drugs but not chemotherapeutics that target DNA replication processes. Viability was measured by the propidium iodide assay and EC₅₀ values were calculated as previously described [21]. Data presented as percent viable cells (PI⁻). CI, confidence interval.

	FCCP		Rotenone		Cytarabine		Doxorubicin		Daunorubicin
	EC50 (μM)	95% CI	EC50 (μM)	95% CI	EC50 (μM)	95% CI	EC50 (μM)	95% CI	EC50 (μM)	95% CI
TRC	12.8	10.3 to 15.8	3.2	2.7 to 3.8	4.4	4.1 to 4.7	8.03E-02	0.074 to 0.087	5.83E-02	0.051 to 0.066
BNIP3L1	1.3	0.3 to 4.4	1.5	1.1 to 2.1	5.1	4.8 to 5.4	6.33E-02	0.059 to 0.067	5.38E-02	0.048 to 0.060
BNIP3L3	1.6	0.7 to 1.6	1.9	1.6 to 2.3	4.5	4.2 to 4.8	8.07E-02	0.078 to 0.083	6.34E-02	0.056 to 0.072
BNIP3L5	0.004	0.5E-05 to 0.4	1.2	1.1 to 1.4	3.8	3.6 to 4.0	8.30E-02	0.081 to 0.085	6.16E-02	0.054 to 0.071
TRC	68.44	46.3 to 101.3	4.0	3.4 to 4.9	6.7	6.2 to 7.2	8.03E-02	0.075 to 0.086	5.83E-02	0.051 to 0.067
SQ 11	36.41	30.4 to 43.6	3.2	2.4 to 4.2	6.8	6.3 to 7.4	6.17E-02	0.059 to 0.065	5.41E-02	0.049 to 0.059
SQ 12	35.37	28.5 to 43.9	3.7	2.9 to 4.7	5.4	4.48 to 6.46	7.01E-02	0.068 to 0.072	6.28E-02	0.056 to 0.070

Figure 3.

BNIP3L expression is a predictor of drug response. AML cells were exposed to rotenone (5 µM, 72 h); FCCP (5 µM 72 h), cytarabine (2.5 µM, 72 h) or daunorubicin (0.03 µM, 72 h) and viability was measured with 7AAD. Data presented as percent viability relative to untreated controls (Untx) and stratified by decreasing BNIP3L expression.

Knockdown of BNIP3L or SQSTM1 sensitized leukemia cells to death following treatment with drugs that target mitochondria, but not chemotherapeutics that target DNA. This points toward a mechanism by which knockdown of BNIP3L or SQSTM1 alters mitochondria, rendering leukemia cells sensitive to drugs targeting this organelle. Indeed, several studies have shown that mitochondria -targeting drugs induce preferential AML cell toxicity. Inhibition of (i) BCL2 with ABT-263 [6]; (ii) mitochondria protein translation with tigecycline [5]; or (iii) fatty acid oxidation with avocatin B [17] all impart selective AML cell death. Moreover, leukemia cells with low mitochondrial spare reserve capacity are increasingly susceptible to oxidative stress-induced apoptosis [7]. Rotenone and FCCP can induce apoptosis by generation of ROS [18,19]. Given that cells with depleted BNIP3L and SQSTM1 expression have elevated basal ROS, we hypothesized that the increased sensitivity towards mitochondria-targeting drugs is linked to the inability of the cell to mitigate the oxidative stress generated by FCCP and rotenone. Accordingly, treatment with mitochondria-targeting drugs resulted in higher levels of mtROS in knockdown cell lines relative to the TRC control (Figure 4(a)). Furthermore, treatment with the anti-oxidants NAC or α-tocopherol was able to rescue the viability of knockdown cells treated with FCCP or rotenone (Figure 4(b)). In contrast, loss of viability due to daunorubicin and cytarabine treatment was not rescued to the same extent by the antioxidants (Figure 4(c)).

Figure 4.

Higher mitochondrial ROS is observed in leukemia cells with low BNIP3L and SQSTM1 relative to transduced control. (a) The level of mitochondria-specific ROS was determined (using MitoSOX) in the absence or presence of antioxidants (2 mM NAC or α-tocopherol) and is presented as fold-increase relative to the untreated controls. (b) BNIP3L or SQSTM1 knockdown cells were treated with FCCP (25 µM, 48 h) or rotenone (2.5 µM, 48 h) in the absence or presence of antioxidants (NAC, 1 mM or α-tocopherol, 2 mM). Cell viability was measured using 7AAD and data presented as percent viability relative to untreated controls. (c) BNIP3L or SQSTM1 knockdown cells were treated with cytarabine (10 µM, 48 h) or daunorubicin (0.06 µM, 48 h) in the absence or presence of antioxidants (NAC or α-tocopherol at 2 mM). Cell viability was measured using 7AAD and data presented as percent viability relative to untreated controls. *p < 0.05.

Exploiting mitochondrial vulnerabilities appears a viable anti-AML strategy and our results highlight a relationship between BNIP3L and SQSTM1 status in leukemia cells and sensitivity to mitochondria-targeting compounds. We cannot exclude the possibility that knockdown of these autophagy genes results in a reduced cell survival response, which would result in rapid cell death compared to cells with an intact cell-survival pathway. In fact, a prevailing hypothesis in studies showing autophagy morphology in dead cells is simply that the cell death machinery overcame the cell survival response [20]. While our results showing the ineffectiveness of DNA damaging agents support our conclusions, future studies are needed to delineate cell death from pro-survival responses. In closing, several studies show that AML cells possess altered mitochondria phenotypes compared to normal cells, and that drugs targeting mitochondria selectively eliminate AML cells. Our results highlight BNIP3L and SQSTM1 as potential novel prognostic markers to identify AML cells with altered mitochondria phenotypes that would respond to novel mitochondria-targeted drug therapy.

Materials and methods

Cell culture, assessment of size and viability

Cell lines (OCI-AML2, OCI-AML3, HL60, KG1a, NB4) were cultured at 5% CO₂ at 37°C in Iscove’s Modified Dulbecco’s Medium (IMDM; Fisher, SH3022801) plus 10% fetal bovine serum (FBS; VWR, 97068-085). TEX cells were cultured in IMDM plus 15% (v:v) FBS and 20 ng/mL stem cell factor (Peprotech, 300-07), 2 ng/mL IL3 (Peprotech, 200-03), and 2 mM L-glutamine (Sigma Chemical, G7513). Cell viability was measured using ANXA5/annexin V (Biovision, 1001) and propidium iodide (Sigma, P4170) staining, according to the manufacturer’s protocol and as previously described [17]. Average cell size for each cell line was determined in triplicate on separate days using the Multisizer 4 Particle Analyzer (Beckman Coulter). Approximately 1-10 × 10⁵cells were diluted 1:20 in PBS (to a final volume of 10 mL) and then loaded for average cell size determination.

Cytarabine (Tocris Bioscience, 4520); daunorubicin (Tocris,1467); doxorubicin (Sigma, D1515); FCCP (Sigma Chemical, C2920); and rotenone (Sigma, R8875) were purchased from commercial suppliers and solubilized as per the manufacture’s protocol.

Immunoblotting

Western blotting was performed as previously described [21]. Briefly, whole cell lysates were collected from treated cells, denatured for 5 min at 95°C, and subjected to gel electrophoresis. Gels were then transferred to a PVDF membrane and blocked with 5% bovine serum albumin (BSA; Sigma, A3059) in Tris-buffered saline-tween (TBS-T) for 1 h. The membrane was incubated overnight at 4°C with the primary antibody, SQSTM1/p62 (1:700; Cell Signaling Technology, 5114), BNIP3L/Nix (1:700; Cell Signaling Technology, 12396), LC3B (1:1000; Cell Signaling Technology, 3868), ANT (Santa Cruz Biotechnology, sc-9299), ND1 (Santa Cruz Biotechnology, sc-20493) and GAPDH (1:15000; Invitrogen, MA5-15738-HRP) or TUBA/α-tubulin (1:1000; Abcam, ab7291). Membranes were then washed and incubated in the appropriate secondary antibody (1:10000) for 1 h at room temperature. Enhanced chemiluminescence was used to detect proteins according to the manufacturer’s instructions (Clarity Western ECL Substrate; Bio-Rad, 1705061) and luminescence was captured using the Kodak Image Station 4000MM Pro and analyzed with Kodak Molecular Imaging Software Version 5.0.1.27. Densitometry was determined using the imaging software and arbitrary units were calculated by dividing band intensity by its loading control.

Measurement of mitochondrial phenotype

Mitochondrial mass was measured, as previously described [17], using 10-N-nonyl acridine orange (NAO; Enzo Life Sciences, ENZ-52306), which binds to cardiolipin, a component of the inner mitochondrial membrane, giving an approximate quantitative measure of mitochondrial mass [22]. Here, 0.35 µM NAO was added to cells seeded at a concentration of 1 × 10⁶ cells/mL and mass was measured by flow cytometry [22]. To determine changes in mitochondrial membrane potential rhodamine 123 (R123; Enzo Life Sciences, ENZ-52307) was used. R123 is a cationic dye readily sequestered by mitochondria in a relationship proportional to the negative (hyperpolarized) mitochondrial membrane potential [23]. Here, 0.6 µM R123 was added to cells (1 × 10⁶ cells/mL) and membrane potential was measured by flow cytometry. Finally, mitochondria-derived superoxide (O₂^·−), was measured using dihydroethidium (DHE; Sigma, 37291). Here, 3 µM DHE was added to cells seeded at a concentration of 1 × 10⁶ cells/mL and superoxide was measured by flow cytometry. In all analyses, forward vs side scatter plots were gated to exclude debris (e.g., dead cells), as cell death is associated with increased ROS and altered membrane potentials [24].

Quantitative PCR was performed in triplicate on a 96-well plate probing for the mitochondrial gene MT-ND1 (mitochondrially encoded NADH:ubiuinone oxidoreductase core subunit 1) alongside the nuclear control gene HBB (hemoglobin subunit beta). Genomic DNA samples were isolated using the Purelink Genomic DNA Mini Kit (Invitrogen, K1820-01) according to the manufacturer’s protocol (eluted with 150 µL elution buffer). The concentration of gDNA was measured using a Nanodrop 200c (Thermo Scientific). To each well, 10 µL Maxima SYBR Green/ROX qPCR Master Mix (Thermo Scientific, K0221), 6 µL nuclease-free water (Thermo Scientific, 4387936), 1 µL PrimePCR Assay HBB, Hsa (Bio-Rad, 10025636), or 1 µL PrimePCR Assay MT-ND1, Hsa (Bio-Rad, 10025636) and 3 µL of 25 ng/µL template DNA were added. For negative controls, 3 µL of nuclease-free water was added instead of DNA template. For positive controls, 3 µL of PrimePCR Template (Bio-Rad) was added instead of DNA template. The assay was run on the StepOnePlus Real-Time PCR System (Applied Bioscience). Initial gDNA content was determined using the ΔΔCT method.

RNAi knockdown of BNIP3L and SQSTM1

Lentiviral transductions were performed as previously described [21]. Briefly, OCI-AML 2 cells (5 × 10⁶) were centrifuged and resuspended in 5mL media containing protamine sulfate (5 μg/mL; Sigma, P3369) and 2 mL of virus cocktail (containing the short hairpin RNA [shRNA] sequence and a puromycin antibiotic resistance gene) and incubated overnight at 37°C. Next, the virus was removed via centrifugation and the cells were washed and resuspended in fresh media containing puromycin (1 μg/ml; Sigma, P9620). After 2 days, puromycin-resistant live cells were plated for viability and growth assays. The shRNA coding sequences were: SQSTM1/p62 (SQ11) (Accession No. NM_003900.4-783s21c1) shRNA: 5ʹ-GAGGATCCGAGTGTGAATTTC-3ʹ, SQSTM1/p62 (SQ12) (Accession No. NM_003900.2-325s1c1) shRNA: 5ʹ-CCGAATCTACATTAAAGAGAA-3ʹ, BNIP3L1 (Accession No. NM_004331.2-831s21c1) shRNA: 5ʹ-TATTGTCACAGTAGCTTATTT-3ʹ, BNIP3L3 (Accession No. NM_004331.1-688s1c1) shRNA: 5ʹ-GCTAGGCATCTATATTGGA-3ʹ, BNIP3L5 (Accession No. NM_004331.1-562s1c1) shRNA: 5ʹ- CCCTAAACGTTCTGTGTCTTT-3ʹ.

Statistical analysis

Unless otherwise stated, the results are presented as mean ± SD. Data were analyzed using GraphPad Prism 4.0 (GraphPad Software, La Jolla, CA). p ≤ 0.05 was accepted as being statistically significant.

Funding Statement

This work was supported by the Leukemia and Lymphoma Society of Canada; Leukemia Research Foundation; Government of Canada | Natural Sciences and Engineering Research Council of Canada (NSERC).

Acknowledgments

This work was supported by grants to PAS by the Leukemia and Lymphoma Society of Canada, Canadian Hematology Society, Leukemia Research Foundation, Canadian Foundation for Innovation, the Ontario Research Fund and NSERC.

Disclosure statement

No potential conflict of interest was reported by the authors.

Supplementary material

Supplemental data for this article can be accessed here.