BNIP3L/NIX-mediated mitophagy protects against ischemic brain injury independent of PARK2

BNIP3L is required in ischemia-reperfusion-induced mitophagy

To investigate whether BNIP3L participates in brain ischemia-reperfusion-induced mitophagy, we performed transient middle cerebral artery occlusion (tMCAO) on Bnip3l^+/+, Bnip3l^+/− and bnip3l^−/− mice. BNIP3L protein loss confirmed the gene deletion. Unlike PARK2, BNIP3L was not degraded by tMCAO (Fig. 1A). The protein levels of SQSTM1, TOMM20, and COX4I1 were examined to assess mitophagy activation. The reductions in the levels of the mitochondrial marker proteins COX4I1 and TOMM20, as well as the reduction of the autophagy substrate SQSTM1 by tMCAO, were significantly reversed in bnip3l^−/− mice, suggesting the requirement of BNIP3L in tMCAO-induced mitophagy (Fig. 1 A-D). We found the MAP1LC3B-II/LC3B-II upregulation by tMCAO was not disturbed by Bnip3l deletion (Fig. 1E), confirming the previous notion that BNIP3L does not activate the autophagy machinery.¹⁸

The primary cultured cortical neurons from Bnip3l^+/+ and bnip3l^−/− mice were subjected to oxygen and glucose deprivation reperfusion (OGD-Rep.) treatment, an in vitro model of cerebral ischemia. The loss of both TOMM20 and COX4I1 was compromised in bnip3l^−/− neurons (Fig. 1F). To evaluate mitophagy defects, the abundance of the mitochondrial-DNA encoded gene ATPase6 was normalized to the level of the genomic gene Rpl13 by RT-qPCR; this method reflects the relative amount of mitochondria. We found the reduction in the ATPase6:Rpl13 ratio that occurred with OGD-Rep was significantly reversed by Bnip3l deletion (Fig. 1G). These results implicate the involvement of BNIP3L in ischemia-reperfusion-induced mitophagy. To further confirm this assertion, the striatum of Bnip3l^+/+ and bnip3l^−/− mice were infected, respectively, with AAV-GFP-BNIP3L and AAV-GFP. The infection efficiency was confirmed (Fig. S1A). Exogenous expression of BNIP3L rescued the mitophagy impairment of bnip3l^−/− mice. We previously reported that mitophagy inhibition reinforced ischemia-reperfusion-induced neuronal apoptosis.¹ In accordance with this finding, we here found activated CASP3 in bnip3l^−/− mice with tMCAO, and this could be rescued by compensatory expression of BNIP3L (Fig. 1H). Taken together, these data suggest that BNIP3L is required in ischemia-reperfusion-induced mitophagy.

BNIP3L deficiency aggravates ischemic brain injury

Mitophagy attenuates ischemia-reperfusion-induced brain injury.^1,5 To verify the contributions of BNIP3L-mediated mitophagy to ischemic brain injury, Bnip3l^+/+, Bnip3l^+/− and bnip3l^−/− mice were subjected to transient middle cerebral artery occlusion (tMCAO). BNIP3L deficiency significantly exacerbated ischemic brain injury as revealed by increased infarct volumes and neurological deficit scores, pointing to a neuroprotective effect of BNIP3L. To further confirm the apparent benefits of BNIP3L in ischemic brains, AAV-GFP-BNIP3L was injected to the striatum of bnip3l^−/− mice to force BNIP3L expression (Fig. S1). The deleterious effects of Bnip3l deletion were completely rescued (Fig. 2A-C). Further, the contributions of BNIP3L to ischemic neuronal injury were evaluated in vitro. Compared with wild-type primary cultured neurons, bnip3l^−/− neurons showed weaker tolerance to OGD-Rep insults, and this could be rescued by AAV-GFP-BNIP3L infection. The benefits of BNIP3L are closely related to mitophagy as the rescue effects of AAV-GFP-BNIP3L were reversed by the autophagy inhibitor 3-methyladenine (3-MA; Fig. 2D).

PARK2 is dispensable in BNIP3L-mediated mitophagy evoked by ischemia-reperfusion

We have previously reported the involvement of PARK2 in ischemia-reperfusion-evoked mitophagy in brains.^1,5 To establish whether BNIP3L-mediated mitophagy requires PARK2, we conducted experiments with park2^−/− mice. The colocalization of BNIP3L and LC3B was observed in both Park2^+/+ and park2^−/− neurons subjected to OGD-Rep (Fig. 3A and B), indicating that the ability of BNIP3L to bind with LC3B is not influenced by PARK2. Additionally, mitophagy was still observed in park2^−/− mice subjected to tMCAO, suggesting the presence of alternative mitophagy mechanisms. Moreover exogenous expression of GFP-BNIP3L in the striatum reinforced ischemia-reperfusion-induced mitophagy in park2^−/− mice (Fig. 3C). Consistently, forced BNIP3L expression also increased mitophagy in primary cultures of park2^−/− neurons (Fig. 3D), indicating the dispensability of PARK2 in BNIP3L-mediated mitophagy. We tested the neuroprotection effects of BNIP3L-mediated mitophagy in park2^−/− mice and found that the extent of brain injury was significantly reduced by GFP-BNIP3L overexpression (Fig. 3E). Similar results were obtained in cultured park2^−/− neurons (Fig. 3F). These results further support the assertion that PARK2 is dispensable for BNIP3L-mediated mitophagy in ischemic brains.

Given the involvement of PARK2 in ischemia-reperfusion-induced mitophagy, we were curious about whether PARK2-mediated mitophagy, conversely, depends on BNIP3L. We thus examined the recruitment of PARK2 to mitochondria, which has been demonstrated as a prerequisite for its subsequent biological functions.²⁷ By immunostaining PARK2 and TOMM20 in primary cultured neurons, we found the translocation of PARK2 to mitochondria increased along with reperfusion within 3 h, and decreased afterward, suggesting mitophagic degradation. Interestingly, Bnip3l deletion did not disturb the observed trends of PARK2 recruitment to mitochondria (Fig. 4A and B). These results suggest that BNIP3L is indeed dispensable for PARK2-dependent mitophagy in ischemic brains. This supposition was further confirmed by the fact that forced expression of PARK2 significantly attenuated infarct volumes in bnip3l^−/− mice (Fig. 4C). Taken together, these results strongly suggest that BNIP3L and PARK2 function in mutually independent pathways to mediate mitophagy in ischemic brains.

Mitophagy in ischemic brains is synergistically impaired in bnip3l and park2 double-knockout mice

To further study the association of BNIP3L with PARK2 in mitophagy induction in ischemic brains, we crossed bnip3l^−/− and park2^−/− mice to generate bnip3l and park2 double-knockout (DKO, Fig. S1B) mice. Primary cultured neurons from mice of various germlines were previously transfected with plasmids encoding GFP-LC3B and Mito-DsRed, and the extent of mitophagy was examined by the ratio of LC3 puncta on mitochondria:total LC3 puncta. After OGD-Rep treatment, the ratio was significantly decreased in bnip3l^−/− and park2^−/− neurons as compared with wild-type neurons. Of note, the ratio in DKO neurons was reduced to a further extent relative to that observed in either bnip3l or park2 single deletion cells (Fig. 5A and B), indicating a synergistic reduction in mitophagy. We next subjected brain corticostriatal slices from the bnip3l^−/−, park2^−/− and DKO mice to OGD-Rep., and assessed the SQSTM1, TOMM20, and COX4I1 levels. Although bnip3l and park2 deletion alone partly compromised OGD-Rep-induced mitophagy, mitophagy was further diminished in DKO slices (Fig. 5C). However, as revealed by monitoring the number of GFP-LC3B puncta, autophagic flux was not blocked in these mutant neurons (Fig. 5B), indicating specific impairment of mitophagy. Consistently, DKO corticostriatal slices showed a greater extent of ischemic injury as compared with the injury in slices of the single-gene deletion genotypes, as determined by 2, 3, 5-triphenyltetrazolium hydrochloride (TTC) conversion assay (Fig. 5D), a redox indicator that measures metabolic activity. Importantly, exogenous expression of either BNIP3L or PARK2 was sufficient to rescue the ischemic brain injury in DKO mice (Fig. 5E). Taken together, these data further confirm the hypothesis that BNIP3L and PARK2 function in mutually-independent pathways to mediate mitophagy in ischemic tissues, and establish that the neuroprotective effects of BNIP3L-mediated mitophagy are functionally independent of PARK2 in the protection of ischemic brains.

Phosphorylation of BNIP3L Ser81 is required for BNIP3L-mediated mitophagy

The mechanisms through which BNIP3L induces mitophagy are not fully understood. It was reported that phosphorylation of key mitophagy-related proteins, including AMBRA1, FUNDC1, and BNIP3, are critical for the regulation of mitophagy.^28-30 We therefore hypothesized that BNIP3L phosphorylation may also play a role in ischemia-induced mitophagy. To test this, we assessed BNIP3L phosphorylation via Phos-tag assays. In line with our hypothesis, we detected the appearance of a shifted phosphorylated BNIP3L band following tMCAO treatment in mice brains. Interestingly, the level of this phosphorylated BNIP3L increased along with reperfusion, and increased following the inhibition of autophagy with 3-MA (Fig. 6A), suggesting that BNIP3L phosphorylation appears to be related to mitophagy. To explore the phosphorylation sites of BNIP3L, we screened the phosphorylated serine residues of BNIP3L revealed by phosphoproteome studies.^31-34 We constructed plasmids expressing nonphosphorylated BNIP3L mutations by replacing the candidate serine residues with alanine. A screen of these mutants revealed that serine 81 was required for BNIP3L-mediated mitophagy (Fig. S3), and the alignment analysis showed it was conserved among species (Fig. 6B). We transfected COS7 cells with a plasmid encoding BNIP3L^S81A and found that this mutation reversed BNIP3L-induced mitophagy. Similar results were reproduced in HeLa cells lacking PARK2, again supporting the independence of PARK2 in BNIP3L-mediated mitophagy (Fig. 6C). To specifically detect the phosphorylation of BNIP3L (Ser81), we developed an antibody against phosphorylated Ser81 of BNIP3L. The specificity of this antibody was confirmed because only wild-type but not the mutant BNIP3L^S81A in HeLa cells could be detected (Fig. 6D, IB: Phos-BNIP3L [S81]). The phosphorylation of BNIP3L (Ser81) decreased with CCCP treatment in a time-dependent manner, while the total BNIP3L level remained intact (Fig. 6D, IB: Flag). These data indicated the selective degradation of phosphorylated BNIP3L (Ser81), suggesting its association with mitophagy. Phosphorylation of BNIP3L (Ser81) was also observed in both Park2^+/+ and park2^−/− brains. Importantly, phospho-BNIP3L (Ser81), rather than total BNIP3L was reduced after tMCAO, which further confirmed the close link of phospho-BNIP3L (Ser81) with mitophagy in ischemic brains (Fig. 6E). Previous studies have demonstrated that the binding of BNIP3L with LC3A and LC3B mediate the recruitment of phagophores to mitochondria.²⁰ Here, we found that the BNIP3L^S81A mutant protein could hardly bind with MYC-LC3A and MYC-LC3B whereas the phosphomimic mutant BNIP3L^S81E showed stronger interaction with both GFP-LC3A and GFP-LC3B (Fig. 6F). In addition, in HeLa cells, recruitment of phagophores (marked by GFP-LC3B) to BNIP3L-labeled mitochondria (marked by Flag-BNIP3L and TOMM20) was evident, but this could not be detected in cells transfected with the BNIP3L^S81A mutant protein (Fig. 6G). Together, these results support the characterization of Ser81 as a novel phosphorylation site of BNIP3L and confirm that phosphorylation at this site is required for BNIP3L-mediated mitophagy.

Phosphorylation of BNIP3L on Ser81 is responsible for ischemia-reperfusion-induced mitophagy and its neuroprotective effects

We next investigated the involvement of the phosphorylation of BNIP3L (Ser81) in BNIP3L-mediated mitophagy in ischemic scenarios. Cultured bnip3l^−/− neurons transfected with wild-type BNIP3L enhanced OGD-Rep-induced mitophagy, as revealed by the reductions in the levels of TOMM20 and COX4I1. However, these effects were not observed following transfection with the BNIP3L^S81A mutant protein (Fig. 7A). These data clearly indicated the requirement of BNIP3L (Ser81) phosphorylation for BNIP3L-mediated mitophagy in ischemic neurons. We next explored this in ischemic mice brains and found that the results of experiments with GFP-BNIP3L, rather than the GFP-BNIP3L^S81A, expression in bnip3l^−/− mice reinforced the idea of tMCAO-induced mitophagy (Fig. 7C). Finally, the forced expression of GFP-BNIP3L, but not the forced expression GFP-BNIP3L^S81A, reversed both neuronal cell viability and brain infarct volumes (Fig. 7B and D). These results indicate that phosphorylation of BNIP3L at Ser81 is required for its mitophagy activity and neuroprotective effects in ischemic brains.

We next asked whether the phosphomimetic BNIP3L mutant showed additional neuroprotective effects. We transfected the wild-type neurons with Flag, Flag-BNIP3L, Flag-BNIP3L^S81A and Flag-BNIP3L^S81E plasmids. Compared to vector, BNIP3L and Flag-BNIP3L^S81E reinforced the OGD-Rep-induced mitophagy to a similar extent, but failed to show further neuroprotective effects, while Flag-BNIP3L^S81A reversed BNIP3L-mediated mitophagy and aggravated injury (Fig. 7E and F). These results indicated that BNIP3L-mediated mitophagy is required for but not sufficient to protect ischemic neurons.

Animals

Male C57BL/6 mice and male BALB/C mice weighing 22–25 g were used. The park2^−/− mice were kindly provided by Prof Zhuohua Zhang from Central South China University; the bnip3l^−/- mice were kindly provided by Prof. Ney Paul from St. Jude Children's Research Hospital. We then crossed park2^−/− and bnip3l^−/− mice to generate the park2 bnip3l double-knockout mice. For the culture of primary cortical neurons, pregnant mice with embryonic (E17) fetuses were used. All experiments were approved by and conducted in accordance with the ethical guidelines of the Zhejiang University Animal Experimentation Committee.

Transient MCAO mouse models

Mice were anesthetized by inhalation of isoflurane for surgery. Cerebral blood flow (CBF) was evaluated in the area of the middle cerebral artery (MCA). Transient focal cerebral ischemia was induced by MCAO, as described previously with minor modifications.⁴⁵ Briefly, a 6–0 nylon monofilament suture was inserted 10 mm into the internal carotid to occlude the origin of the MCA. Animals with less than 80% reduction in CBF were excluded from the study. Body temperature was maintained at 37°C during surgery and for 2 h after the start of reperfusion. To measure infarct volume, coronal mice brain slices, at 2-mm intervals, were stained with 0.25% TTC (Sigma, T8877). Infarcted areas were analyzed using Image Pro Plus 7.0 and measured by the indirect method, which corrects for edema. Neurological deficit scores were evaluated after 24 h of reperfusion.

Viral delivery

Under sodium pentobarbital anesthesia (50 mg/kg, intraperitoneal), mice were mounted in a stereotaxic apparatus and adeno-associated virus (AAV, designed, produced and identified by Obio Technology Corp., Ltd., Shanghai, China) including AAV-GFP, AAV-GFP-PARK2, AAV-GFP-BNIP3L, and AAV-GFP-BNIP3L_S81A, were injected into the corpus striatum (AP +0.5 mm; L -2.0 mm; V -3.0 mm). The viruses were infected for a minimum of 2 wk before further experiments to allow time for sufficient protein accumulation.

Preparation of brain slices and oxygen-glucose deprivation

Acute brain slices were prepared from adult male mice. The oxygen–glucose deprivation (OGD) procedures were performed as we described previously.⁴⁶ Briefly, corticostriatal slices (400-mm thick) were immersed in ice-cold oxygenated artificial cerebrospinal fluid (ACSF) containing 124 mM NaCl, 5 mM KCl, 1.25 mM KH₂PO₄, 2 mM MgSO₄, 26 mM NaHCO₃, 2 mM CaCl₂, 10 mM glucose (pH 7.4, all chemicals were purchased from Sigma) for one h and then incubated at 37°C for 15 min before further experiments. For OGD, slices were transferred into glucose-free ACSF bubbled with 5% CO₂ and 95% N₂ for 15 min and then returned to oxygenated ACSF for one h. Slice viability was determined by 0.25% TTC (Sigma, T8877) staining as described previously.⁴⁷ Formazan extracted was measured at 490 nm and normalized to the dry weight of the slice.

Cell culture, OGD procedures, and cell viability

The primary cortical neuronal culture experiments were performed as described previously.⁴⁵ Briefly, the dissected cortex from E17 fetal mice was digested with 0.25% trypsin (Invitrogen, 25200–056). Approximate 2 × 10⁵ cells/cm² were seeded onto poly-L-lysine (10 µg/ml)-coated plates (Sigma, P1399) and dishes. The neurons were grown in Neurobasal medium (Invitrogen, 21103–049) supplemented with 2% B27 (Invitrogen, 17504–044) and 0.5 mmol/L glutamine (Invitrogen, 21051–024). Cultures were maintained for 8 d before treatment. Human HeLa cells and COS7 cells were routinely cultured in DMEM containing 10% fetal bovine serum (Invitrogen, 10099141).

For OGD treatment, primary neurons were refreshed with O₂- and glucose-free DMEM. Cells were then immediately placed in a sealed chamber (Billups-Rothenburg, MIC-101) loaded with mixed gas containing 5% CO₂ and 95% N₂. For reperfusion, neurons were refreshed with normal culture medium. 3-MA (5 mM; dissolved in normal culture medium) was added to the cells at the onset of reperfusion.

Transfection

Primary cultured neurons were transfected with AAVs containing GFP-LC3B (Hanbio, China). For Mito-DsRed delivery, the Mito-DsRed DNA sequence was cloned from pDsRed2-Mito (Clontech, 632421) and inserted into the pLVX-puro plasmid (Clontech, 632164) and then transfected into HEK293T cells with a set of plasmids for lenti-virus packaging (Clontech, 631247). The lentivirus and/or the AAVs were added to neurons at 6 d in vitro.

BNIP3L, BNIP3L_S81A, and BNIP3L_S81D or S81E were ligated into the p3 × FLAG-CMV-7.1 vector (Sigma, E7533) and transfected into primary cultured neurons using Amaxa electroporation (VPG-1001, Lonza, Switzerland) according to the manufacturer's protocol. Briefly, after digestion of the cortex, 5 × 10⁶ cells were collected and resuspended in 100 μl room temperature Nucleofector Solution (Lonza, VPG-1001) containing 3 μg DNA per sample. The cell/DNA suspension was then transferred into a cuvette. The cuvette with cell/DNA suspension was inserted into the Nucleofector Cuvette Holder. After being electroporated using program O-005, the cells were resuspended in the medium mentioned above and then transferred into the prepared culture dishes. The plasmids were transfected into COS7 or HeLa cells according to the jetPRIME (n114–15, Polyplus, France) protocol.

Immunoblotting and co-immunoprecipitation

The brain tissues and cells were homogenized in RIPA buffer (Sangon Biotech, C500005). A 40-µg aliquot of protein from each sample was separated using SDS-PAGE. The following primary antibodies were used: phos-BNIP3L (S81, 1:2000; Abcam, ab208190), LC3 (1:1,000; Sigma, L7543), SQSTM1 (1:1,000; Cell Signaling Technology, 5114), cleaved CASP3 (1:1,000; Cell Signaling Technology, 9661), COX4I1 (1:1,000; Cell Signaling Technology, 4850), TOMM20 (1:1,000; Anbo Biotechnology, c16678), PARK2 (1:1,000; Sigma, P5748), or GAPDH (1:3,000; KangChen, KC-5G4). Secondary antibodies conjugated with HRP against either rabbit or mouse IgG (1:5,000; Cell Signaling Technology, 7071 and 7072) were applied. Digital images were quantified using densitometric measurement with Quantity-One software (Bio-Rad).

For co-immunoprecipitation, HeLa cells were transiently transfected. At 24 h post transfection, the cells were lysed and immunoprecipitated using a FLAG Immunoprecipitation Kit (Sigma, FLAGIPT1). Thereafter, the precipitants were washed 3 times with lysis buffer and the immune complexes were eluted with sample buffer containing 1% SDS (AMRESCO, 0227) for 5 min at 100°C and analyzed by SDS-PAGE.

Detection of phosphorylated BNIP3L by using Phos-tag SDS-PAGE was performed according to the manufacturer's protocol (Wako Pure Chemical Industries, Japan). Briefly, a 50-µg aliquot of protein from each sample was separated using 8.5% polyacrylamide gels containing 100 µM Phos-tag acrylamide (Wako Pure Chemical Industries, AAL-107) and 200 µM MnCl₂ (Sigma, 244589). Gels were washed with transfer buffer containing 1 mM EDTA (Sigma, E6758) for 10 min at room temperature and then with transfer buffer without EDTA for another 10 min. Proteins were transferred to PVDF membranes (Millipore, IPVH00010) and immunoblotting performed using a chemiluminescent digital imaging system (6100, Tanon, China).

Immunocytochemistry and confocal microscopy

For immunostaining, cells were incubated with antibodies against Flag (1:200; Cell Signaling Technology, 2368S), PARK2 (1:200; Sigma, P5748), or TOMM20 (1:100; Abcam, ab56783). Secondary antibodies labbled with Alex Fluor 488, 596 or 647 (Invitrogen, A32731, A21203 and A32733) were subsequently added to the cells. Coverslips were observed on a confocal microscope (Fluoview FV1000, Olympus, Japan). Mander's overlap efficiency was measured and analyzed with Image Pro-Plus 7.0 software. Five randomly selected fields from one coverslip were included to calculate an average, and experiments were repeated independently at least 3 times.

Real-time PCR

After 6 h of OGD-reperfusion, total cellular DNA was extracted with a DNeasy Blood & Tissue kit (Qiagen, DP304–03). Aliquots of 20 ng total DNA were analyzed via real-time PCR to evaluate the expression of the mitochondrial gene mt-Atp6 and the genomic gene Rpl13, as described previously.¹ The copy numbers of Atp6 and Rpl13 were calculated and normalized by the standard curve. Relative expression was presented as mt-Atp6:Rpl13.

Statistical analysis

All data were collected and analyzed in a blind manner. Data are presented as mean ± SD. One-way ANOVA (analysis of variance) with Dunnett's T3 post-hoc test was applied for multiple comparisons. p < 0.05 was considered statistically significant.