Mitochondrial BMI1 maintains bioenergetic homeostasis in cells

Abstract

The polycomb complex proto-oncogene BMI1 [B lymphoma Mo-MLV insertion region 1 homolog (mouse)] is essential for self-renewal of normal and cancer stem cells. BMI1-null mice show severe defects in growth, development, and survival. Although BMI1 is known to exert its effect in the nucleus via repression of 2 potent cell-cycle regulators that are encoded by the Ink4a/Arf locus, deletion of this locus only partially rescues BMI1-null phenotypes, which is indicative of alternate mechanisms of action of BMI1. Here, we show that an extranuclear pool of BMI1 localizes to inner mitochondrial membrane and directly regulates mitochondrial RNA (mtRNA) homeostasis and bioenergetics. These mitochondrial functions of BMI1 are independent of its previously described nuclear functions because a nuclear localization-defective mutant BMI1 rescued several bioenergetic defects that we observed in BMI1-depleted cells, for example, mitochondrial respiration, cytochrome c oxidase activity, and ATP production. Mechanistically, BMI1 coprecipitated with polynucleotide phosphorylase, a ribonuclease that is responsible for decay of mtRNA transcripts. Loss of BMI1 enhanced ribonuclease activity of polynucleotide phosphorylase and reduced mtRNA stability. These findings not only establish a novel extranuclear role of BMI1 in the regulation of mitochondrial bioenergetics, but also provide new mechanistic insights into the role of this proto-oncogene in stem cell differentiation, neuronal aging, and cancer.-Banerjee Mustafi, S., Aznar, N., Dwivedi, S. K. D., Chakraborty, P. K., Basak, R., Mukherjee, P., Ghosh, P., Bhattacharya, R. Mitochondrial BMI1 maintains bioenergetic homeostasis in cells.

Keywords: PNPase; electron transport chain; mitochondrial mRNA.

Figure 1.

Subcellular fractionation demonstrates presence of BMI1 in the mitochondria. A) Relative mRNA expression of BMI1 determined by quantitative real-time PCR that is normalized to glyceraldehyde-3-phosphate dehydrogenase (GAPDH) and then compared with respect to OSE that was set to 1 (top). Expression of BMI1 at the protein level in different cell lines (bottom). B) OSE cells expressing BMI1-WT FLAG were subjected to subcellular fractionation. Total cell lysate (TCL), nuclear (Nuc), cytosoloic (Cyto), and mitochondrial (Mito) enriched fractions were isolated, and 10 μg of each fraction protein was immunoblotted for FLAG (BMI1 FLAG). Voltage-dependent anion channel (VDAC), proliferating cell nuclear antigen (PCNA), and GAPDH were used to determine enrichment of the mitochondrial, nuclear, and cytosolic fractions, respectively. C) Endogenous BMI1 levels and other fraction markers in different subcellular fractions of CP20 cells that were isolated during mitochondrial enrichment by using the sucrose gradient method (top). In a separate experiment, further purification of the mitochondrial enriched (sucrose method) fraction by the percoll gradient method was performed and an equal amount of protein for each fraction was immunoblotted for BMI1 (bottom). In addition, immunoblotting with markers that included VDAC (mitochondrial), PCNA (nuclear), GAPDH (cytosolic), syntaxin6 (golgi), and calnexin (ER) was performed. D) After mitochondrial subfractionation of CP20 cells, equal amounts of protein were loaded, and BMI1 levels were determined in the OM, IMS, IM, and MA fractions. TOM20, cytochrome c (Cyto c), COX4, and manganese-superoxide dismutase (MnSOD) were used as fraction markers for OM, IMS, IM, and MA, respectively. For details of subcellular fractionation and mitochondrial subfractionation, see Materials and Methods. *P < 0.05.

Figure 2.

Microscopic imaging demonstrates presence of BMI1 in the mitochondria. A–D) CP20 (A, B) or COS7 (C, D) cells transfected with BMI1-WT FLAG (A, C) or BMI1-NLS2 FLAG constructs (B, D) were fixed and stained with DAPI (blue), anti-FLAG (BMI1, green), and anti-COX4 (a marker for mitochondria, red) and were analyzed by confocal microscopy. A representative image of BMI1-WT FLAG–transfected CP20 cells is shown on the left. Insets displayed on the right show magnified regions within the corresponding images on the left. Yellow pixels observed in the merged panels indicate a high degree of colocalization between both BMI1-WT FLAG (A, C) or BMI1-NLS2 FLAG (B, D) in CP20 and COS7 cells. The white line in the merged insets indicates the pixels used for the RGB profile plots. E) Extranuclear localization of BMI1 was analyzed by immunogold electron microscopy in COS7 cells expressing BMI1-WT FLAG. Two representative images of different magnification are shown. Small gold of 12 nm (indicated by arrowheads) represents antimitochondrial complex IV mAB, (mt-COI; dilution 1:25). Large gold of 18 nm represents anti-FLAG (BMI1; dilution 1:100). Arrows indicate cristae. See Materials and Methods for detailed information on experimental procedures. M, mitochondria. Scale bars, 25 μm (A–D), 200 nm (E, left), 100 nm (E, right).

Figure 3.

NLS2 mutant BMI1 can rescue oxidative phosphorylation function. A) CP20 cells were transfected with scrambled (Sc) siRNA and empty vector (EV), BMI1 siRNA and EV, BMI1 siRNA and BMI1-WT FLAG, or BMI1 siRNA and BMI1-NLS2 FLAG, and the oxygen consumption rate (OCR) was measured by using the Seahorse XF analyzer. An XF trace representing means ± sem from at least 3 technical replicates per group is shown. B) Parameters depicted here are derived from panel A. Basal respiration (before oligomycin), relative ATP estimation (change in OCR before and after oligomycin), and FCCP-stimulated response were derived from the XF trace of 3 independent experiments and are represented as means ± sem. C) Immunoblot depicting expression of BMI1 in the total cellular lysate and the mitochondria-enriched fraction from CP20 cells, transfected as in panel A. D) ImageJ-based quantitation of BMI1 expression in different cellular fractions as in panel C. Intensity of BMI1 from the total cellular lysate was normalized with respect to glyceraldehyde-3-phosphate dehydrogenase (GAPDH) and that for the mitochondrial fraction was normalized with respect to cytochrome c (Cyt c). Either in the total cell lysate or in the mitochondrial fraction, BMI1-NLS2 levels were compared with respect to BMI1-WT being set to 1. NS, not significant. *P < 0.05 compared with Sc siRNA; ^@P < 0.05 compared with BMI1 siRNA + EV.

Figure 4.

BMI1 affects glucose dependence and cellular energy production. A) Relative cellular viability (mean ± sd) was determined by using ApoTox-Glo assay in CP20 cells transfected with scrambled (Sc) siRNA and empty vector (EV), BMI1 siRNA and EV, BMI1 siRNA and BMI1-WT FLAG, or BMI1 siRNA and BMI1-NLS2 FLAG. Top panel: in medium that contained 2.8 mM glucose, viability was similar among the groups, and each was thus set to 1. Bottom panel: time course of the experiment. B) ATP production was measured in CP20 cells transfected as described in A, and values are represented as mean ± sd fold change compared with Sc siRNA + EV. C) Genomic DNA isolated from Sc siRNA– and BMI1 siRNA–transfected CP20 cells was subjected to quantitative PCR using primers against the mitochondrial 16S rRNA or the nuclear 36B4. D) Citrate synthase (CS) activity was measured in the mitochondrial enriched fraction of CP20 cells and represented as mean ± sd fold change relative to Sc siRNA. E) Mitochondrial membrane potential was measured by using the fluorescent dye tetramethylrhodamine ethyl ester (TMRE) in Sc siRNA– or BMI1 siRNA–transfected CP20 cells, treated with or without FCCP. F) Activity of COX was determined to assess the function of complex IV in CP20 cells, transfected as in panel A. Data are represented as relative mean ± sd fold change, with Sc siRNA + EV set to 1. NS, not significant. All experiments were independently repeated 3 times. *P < 0.05 compared with Sc siRNA; ^@P < 0.05 compared with BMI1 siRNA + EV.

Figure 5.

BMI1 regulates stability of mitochondrial mRNA transcripts. A) Flow chart describing the experimental time course for determining the stability of the mitochondrial transcripts. B) CP20 cells transfected with scrambled (Sc) siRNA or BMI1 siRNA were treated with 0.04 μg/ml EtBr (96 h post-transfection) for indicated time points, and quantitative real-time PCR was performed on RNA that was extracted from these treated cells. EtBr differentially blocks mitochondrial transcription but not nuclear transcription. The 18S rRNA transcript level was used to normalize gene expression of mitochondrial transcripts at each time point, which are presented as mean fold change ± sd relative to 0 h. Bar graph represents efficient knockdown of BMI1 at all relevant time points. All experiments were independently repeated 3 times. ND, mitochondrially encoded NADH dehydrogenase subunits 1–6; CO, mitochondrially encoded cytochrome c oxidase subunits I–III; CYTB, mitochondrially encoded cytochrome b. *P < 0.05.

Figure 6.

BMI1 protects mitochondrial transcripts against degradation by PNPase. A) Endogenous BMI1 was immunoprecipitated from CP20 lysates and probed with antibodies against BMI1 and PNPase. B) PNPase was immunoprecipitated from scrambled siRNA (Sc siRNA)– or BMI1 siRNA–transfected CP20 cells with agarose A/G beads (confirmed by immunoblotting; right) and incubated with total cellular RNA. RNA was isolated, reverse transcribed, and subjected to quantitative PCR with mitochondrial gene specific primers. Quantitative real-time PCR was performed to determine relative mitochondrial mRNA level, which was normalized to 18S rRNA, and then Sc siRNA was set as 1 (left). C) CP20 cells were transfected with control Sc siRNA, BMI1 siRNA, PNPase siRNA, or both BMI1 and PNPase siRNA [dual knockdown (dual KD)], and, at 96 h post-transfection, were treated with or without 0.04 μg/ml EtBr for 3 h, and RNA was isolated and quantitative real-time PCR was performed. 18S rRNA was used to normalize the expression of the mitochondrial transcripts, BMI1 and PNPase, which are presented as mean ± sd fold change with respect to Sc siRNA 3 h EtBr treatment (set to 1). GAPDH, glyceraldehyde-3-phosphate dehydrogenase. All experiments were independently repeated 3 times. ND, mitochondrially encoded NADH dehydrogenase subunits 1–6; CO, mitochondrially encoded cytochrome c oxidase subunits I–III; CYTB, mitochondrially encoded cytochrome b. *P < 0.05 compared with Sc siRNA; ^@P < 0.05 compared with BMI1 siRNA.

Figure 7.

Model summarizing the mitochondrial and nuclear functions of BMI1. Previous work has shown that the BMI1 complex rapidly recruits to double-stranded DNA breaks (DSBs) where it monoubiquitinates H2A (lysine 119), blocks transcriptional elongation by excluding RNA Pol-II, and contributes to homologous recombination repair. In addition, BMI1 acts as a transcriptional repressor of the INK4a/ARF locus that encodes the cell-cycle inhibitors p16^Ink4a and p19^Arf, thereby promoting cellular proliferation and self-renewal of stem cells. Our current work indicates that another distinct pool of BMI1 localizes to the IM, possibly via interactions with other mitochondrial proteins (?). This mitochondrial BMI1 interacts with and inhibits mitochondrial mRNA (mt-mRNA) decay that is mediated by PNPase. Mitochondrial BMI1, likely through stabilization of the mt-mRNA, influence functioning of the electron transport chain (ETC) that enhances oxidative phosphorylation and ATP synthesis and prevents aberrant ROS production.