Dual-localized PPTC7 limits mitophagy through proximal and dynamic interactions with BNIP3 and NIX

PPTC7 regulates BNIP3 and NIX post-transcriptionally and independent of HIF-1α

We previously reported that BNIP3 and NIX were significantly upregulated in tissues and cells derived from Pptc7^−/− mice (Niemi et al, 2019, 2023). Consistently, Pptc7 KO MEFs (Fig 1A) and PPTC7 KO 293T cells (Fig 1B) showed elevated expression of BNIP3 and NIX relative to wild-type cells. To understand the mechanisms underlying BNIP3 and NIX upregulation upon PPTC7 loss, we investigated the involvement of HIF-1α in mediating this response. BNIP3 and BNIP3L (gene name of NIX) are well-established transcriptional targets of HIF-1α in conditions of hypoxia, as well as pseudohypoxia through various pharmacological stimuli (e.g., the iron chelators deferoxamine [DFO] and deferiprone [DFP] as well as cobalt chloride [Wang & Semenza, 1993a, 1993b; Allen et al, 2013]). We thus hypothesized that the increase in BNIP3 and NIX protein levels in PPTC7 KO cells may be due to elevated HIF-1α activity. To test this, we immunoblotted for HIF-1α expression in wild-type and PPTC7 KO cells in both basal conditions as well as in the presence of bafilomycin A1 (Baf-A1), a compound previously shown to stabilize HIF-1α (Hubbi et al, 2013). These experiments revealed no differences in HIF-1α protein expression between wild-type and PPTC7 KO cells across tested conditions (Fig 1C). In addition, we found no significant changes in the abundance of select HIF-regulated proteins relative to other proteins across our previously collected proteomics datasets in Pptc7 KO mouse tissues (Niemi et al, 2019) (Fig S1A). We next tested whether BNIP3 and NIX were transcriptionally upregulated but found no significant differences in BNIP3 or BNIP3L mRNA levels between wild-type and PPTC7 KO 293T cells (Fig 1D). Consistently, we found that the transient transfection of plasmids encoding myc-BNIP3 or FLAG-NIX led to increased protein expression in PPTC7 KO 293T cells relative to wild-type 293T cells (Fig S1B and C). As these plasmids are driven by the same CMV promoter in both wild-type and PPTC7 KO cells, the elevated BNIP3 and NIX protein expression seen in PPTC7 KO likely occurs independent of transcriptional changes. Together, these data indicate that PPTC7 influences BNIP3 and NIX protein expression post-transcriptionally and is independent of HIF-1α.

The HIF-1α-independent upregulation of BNIP3 and NIX in PPTC7 KO cells suggests that HIF-1α and PPTC7 regulate BNIP3 and NIX through parallel pathways. If true, we hypothesized that loss of PPTC7 would not alter BNIP3 and BNIP3L transcriptional induction upon HIF-1α activation, but would promote an additive increase in BNIP3 and NIX protein levels. Indeed, treatment of both wild-type and PPTC7 KO cells with the iron chelator DFO induced similar fold changes in BNIP3 (Fig 1E) and NIX (Fig 1G) mRNA transcripts across genotypes. At the protein level, however, expression of BNIP3 (Fig 1F) and NIX (Fig 1H) were substantially higher in DFO-treated PPTC7 KO cells relative to other tested conditions. Notably, basal protein expression of BNIP3 and NIX in untreated PPTC7 KO cells exceeded BNIP3 and NIX induction in DFO-treated wild-type cells (Fig 1F and H), indicating the magnitude of BNIP3 and NIX upregulation in the absence of PPTC7. These experiments further indicate that neither BNIP3 nor NIX is maximally upregulated in PPTC7 KO cells in basal conditions.

The additive increase in BNIP3 and NIX expression in PPTC7 KO cells upon DFO treatment led us to hypothesize that mitophagic flux may also be additive in DFO-treated PPTC7 KO cells. To test this, we assayed wild-type and Pptc7 KO MEFs expressing mt-Keima, a pH-sensitive fluorescent mitophagy sensor (Sun et al, 2017) using flow cytometry. We first exposed wild-type or Pptc7 KO MEFs to varying concentrations of DFO and found that only the highest tested dose of DFO, 100 μM, induced mitophagy in wild-type cells (Fig S1D–F). Notably, the percentage of cells undergoing high mitophagic flux in wild-type cells treated with 100 μM DFO remained below the rates of mitophagic induction in untreated Pptc7 KO cells (Fig S1D–F). We repeated these experiments in wild-type, Pptc7 KO, and Pptc7/Bnip3/Bnip3l triple KO (TKO) MEFs and found that 100 μM DFO induced an approximate threefold increase in mitophagic flux in both wild-type and Pptc7 KO cells (Fig 1I–K). However, the absolute levels of mitophagy in DFO-treated Pptc7 KO cells significantly exceeded those of DFO-treated wild-type cells as well as untreated Pptc7 KO cells, with close to 80% of DFO-treated Pptc7 KO cells undergoing high mitophagy (Fig 1I–K). Importantly, mt-Keima-positive Pptc7/Bnip3/Nix TKO cells failed to undergo appreciable mitophagy in the presence of 100 μM DFO, demonstrating their necessity in increasing mitophagic flux in Pptc7 KO cells (Fig 1I–K). Collectively, these data demonstrate that BNIP3 and NIX are post-transcriptionally upregulated to induce mitophagy in PPTC7 KO cells, and that transcriptional activation of HIF-1α can substantially enhance BNIP3/NIX protein expression and mitophagy in the absence of PPTC7.

PPTC7 enables BNIP3 and NIX turnover through proteasomal degradation

The elevated protein expression of BNIP3 and NIX in PPTC7 KO cells implies that PPTC7 alters the synthesis or turnover rates of these mitophagy receptors. BNIP3 and NIX turnover has emerged as a critical regulatory step in limiting basal mitophagy, as evidenced by recent studies on the E3 ubiquitin ligase FBXL4 (Cao et al, 2023; Elcocks et al, 2023; Nguyen-Dien et al, 2023). Loss of Fbxl4 phenotypically mirrors Pptc7 KO in mice (Niemi et al, 2019; Alsina et al, 2020), and PPTC7 and FBXL4 have significant and positively correlated essentiality profiles across over one-thousand cancer cell lines (Fig S2A, [Dempster et al, 2019 Preprint]). These data suggest that PPTC7 and FBXL4 influence BNIP3 and NIX via similar mechanisms, leading us to hypothesize that BNIP3 and NIX have decreased turnover rates in PPTC7 KO cells relative to wild-type cells.

To test this, we first sought to identify an experimental condition in which we could quantify endogenous BNIP3 and NIX turnover. We noted that DFO-mediated iron chelation was previously shown to decrease the protein level of select mitochondrial proteins in a manner that was reversible upon compound washout (Rensvold et al, 2013). As BNIP3 and NIX accumulate in response to DFO (Fig 1), we hypothesized that washout of DFO would induce BNIP3 turnover because of its short half-life (Schäfer et al, 2022) (Fig 2A). Indeed, treatment of wild-type 293T cells with DFO increased BNIP3 levels in a time-dependent manner, which returned to at or near baseline (i.e., untreated) levels 24 h after DFO washout (Fig 2B). Thus, DFO washout constitutes an experimental system in which we could test the effects of PPTC7 on the turnover of endogenous BNIP3 and NIX. We repeated these experiments in wild-type and PPTC7 KO 293T cells and found that BNIP3 and NIX exhibited blunted turnover in PPTC7 KO cells (Figs 2C–F and S2B–E). Modeling of BNIP3 and NIX decay rates showed that loss of PPTC7 extends the half-life of monomeric and dimeric populations of BNIP3 and NIX (Fig 2C–F). Whereas the dimer populations of BNIP3 and NIX have at least a doubling of protein half-life in PPTC7 KO cells relative to wild-type cells, the half-lives of the monomeric populations of each mitophagy receptor could not be effectively modeled in PPTC7 KO cells, consistent with substantial suppression of the turnover of these species of BNIP3 and NIX in the absence of PPTC7 (Fig 2C–F).

Given the slowed rates of BNIP3 and NIX turnover in PPTC7 KO cells, we sought to understand the pathway(s) contributing to their degradation. The phenotypic similarities between loss-of-function models of PPTC7 and FBXL4, as described above, suggest that these two proteins function in a similar pathway. In addition, previous work has shown that BNIP3 accumulates in response to the proteasomal inhibitor MG-132 (Park et al, 2013; Poole et al, 2021). If KO of PPTC7 suppresses the proteasomal degradation of BNIP3 and NIX, we hypothesized that PPTC7 KO cells would be less responsive to the proteasomal inhibitor MG-132 than matched wild-type cells. We found that BNIP3 and NIX levels increased in wild-type cells upon MG-132 treatment (Fig 2G and H) but that the level of each receptor was significantly less responsive to MG-132 in PPTC7 KO cells (Fig 2G and H). To further test this model, we exploited the DFO washout assay, predicting that if BNIP3 and NIX were turned over by proteasomal degradation upon DFO washout, MG-132 treatment would slow their turnover rates in wild-type cells but have a diminished effect in PPTC7 KO cells. We found that, upon DFO washout, MG-132 treatment caused BNIP3 and NIX accumulation in wild-type cells, whereas the levels of these proteins remained largely unchanged in PPTC7 KO cells under identical conditions (Fig 2I). Overall, these data demonstrate that PPTC7 enables the efficient turnover of BNIP3 and NIX in a manner that largely depends upon proteasomal degradation.

PPTC7 requires an intact active site but not a mitochondrial targeting sequence to limit BNIP3 and NIX accumulation

As loss of PPTC7 increases BNIP3 expression, we hypothesized that PPTC7 overexpression may diminish pseudohypoxia-induced BNIP3 upregulation. To test this, we overexpressed either cytosolic GFP or PPTC7-GFP in HeLa cells that were treated with the pseudohypoxia inducer cobalt chloride (CoCl₂). We fixed and immunolabeled these cells to examine endogenous BNIP3 levels relative to the general mitochondrial marker TOMM20. CoCl₂ robustly upregulated BNIP3 protein expression, and mitochondrial BNIP3 staining could be detected in nearly all cells transfected with cytosolic GFP (Fig 3A). In contrast to cytosolic GFP, PPTC7-GFP co-localized with the mitochondrial marker TOMM20 as expected (Fig 3A). Notably, mitochondrial BNIP3 signal was rarely detected in the subset of cells expressing PPTC7-GFP (Fig 3A and B). Similarly, overexpression of PPTC7 suppressed CoCl₂-induced mitophagic flux relative to identically treated wild-type cells expressing mt-Keima (Fig S3A and B). Finally, overexpression of PPTC7 in Pptc7 KO MEFs rescued basal BNIP3 protein expression to levels seen in wild-type MEFs (Fig S3C). Collectively, these data show that overexpressed PPTC7 limits BNIP3 protein expression and mitophagy induction in both wild-type and Pptc7 KO cells.

As these experiments indicated that overexpressed PPTC7 was properly localized and functional, we generated a series of mutants to understand the mechanism by which PPTC7 limits BNIP3 and NIX accumulation. PPTC7 consists of a PP2C phosphatase domain that is preceded by a short disordered N-terminus (Fig 3C), which includes a mitochondrial targeting sequence (i.e., MTS) that is processed at amino acid 14 (Calvo et al, 2017). Interestingly, when western blotting for overexpressed PPTC7, we found that the protein ran as a doublet, whereas a ΔMTS-PPTC7 mutant ran at the same molecular weight as the bottom band (Fig 3D), suggesting that wild-type PPTC7 expressed as both a full-length and processed form. Given that mitochondrial proteins are often processed upon their entrance to the mitochondrial matrix, these data suggested that PPTC7 may reside in two locations, with the full-length isoform localizing outside of mitochondria, and the processed band within the matrix. To test this, we performed a protease protection assay and found that the top band of the PPTC7 doublet was susceptible to digestion mediated by proteinase K, whereas the bottom band was protected, consistent with dual localization (Fig 3E). Given these molecular insights, we used this overexpression system to test the necessity of PPTC7 catalytic activity and mitochondrial targeting in suppressing BNIP3 and NIX induction under pseudohypoxia.

We first mutated the PP2C phosphatase domain of PPTC7 at a key catalytic residue, D78, to alanine. We previously demonstrated that recombinant PPTC7 D78A was unable to dephosphorylate BNIP3 and NIX on mitochondria isolated from Pptc7 KO MEFs (Niemi et al, 2023). Consistently, recombinant wild-type PPTC7, but not D78A PPTC7, caused a collapse in the laddering pattern of monomeric BNIP3 in mitochondria isolated from wild-type and PPTC7 KO 293T cells (Fig S3D). These data show that not only that PPTC7 D78A lacks catalytic activity, but also that the upper monomeric bands seen on BNIP3 western blots represent phosphorylated intermediates that can be directly dephosphorylated by PPTC7. As such, we predicted that a D78A mutant would be unable to suppress BNIP3 and NIX accumulation during pseudohypoxia. We overexpressed wild-type and D78A PPTC7 in HeLa FLP-IN TREx cells and found that both constructs were doxycycline-induced to similar extents and, interestingly, also expressed at higher levels in the presence of CoCl₂, similar to BNIP3 and NIX (Fig 3F). Whereas overexpression of wild-type PPTC7 decreased BNIP3 and NIX accumulation in response to CoCl₂ treatment, the D78A mutant failed to suppress the induction of these mitophagy receptors (Fig 3F), indicating that PPTC7 requires an intact catalytic motif to influence BNIP3 and NIX protein expression. Importantly, however, recent structural modeling efforts suggest that BNIP3 and NIX may interact with PPTC7 proximal to its catalytic motif (Nguyen-Dien et al, 2024 Preprint, Fig S3E). Consistently, we find that D78A significantly decreases, but does not fully ablate, the physical interaction between PPTC7 and NIX (Fig S3F), similar to these findings (Nguyen-Dien et al, 2024 Preprint).

Given the dual localization of PPTC7, we next examined whether disrupting its mitochondrial localization would affect its ability to suppress CoCl₂-induced BNIP3 and NIX levels. We found that ΔMTS-PPTC7 fully suppressed CoCl₂-mediated NIX accumulation and partially suppressed BNIP3 accumulation (Fig 3G). This rescue of BNIP3 and NIX expression is consistent with a model in which a non-targeted, cytosolic PPTC7 influences mitophagy at sufficient (i.e., overexpressed) levels, suggesting a population of PPTC7 outside of mitochondria could mediate BNIP3 and NIX stability. To directly test this, we artificially anchored PPTC7 to the outer mitochondrial membrane (OMM) by engineering a PPTC7 construct that both lacks an MTS and is fused to OMP25, a tail-anchored protein that targets to the OMM (Horie et al, 2002). Consistent with our hypothesis, ΔMTS-PPTC7-OMP25 blunts the accumulation of BNIP3 and NIX under both basal and CoCl₂-treated conditions (Fig 3H), demonstrating that OMM-localized PPTC7 is sufficient to suppress BNIP3 and NIX protein expression. Collectively, these data show that PPTC7 requires an intact active site but not its mitochondrial targeting sequence to suppress BNIP3 and NIX expression in pseudohypoxic conditions, consistent with a role for PPTC7 in regulating these mitophagy receptors outside of mitochondria.

PPTC7 proximally and dynamically interacts with BNIP3 and NIX in cells

Our data indicate that a pool of PPTC7 resides outside of mitochondria, which may interact with BNIP3 and NIX to promote their proteasomal turnover. We sought to test this model by determining whether BNIP3 and NIX interact with PPTC7 in cells through miniTurbo-based proximity labeling experiments. We expressed PPTC7-V5-miniTurbo, as well as two control constructs, in 293T cells with or without DFO treatment. We treated half of the samples with exogenous biotin (which facilitates miniTurbo-based proximity labeling) and left the remaining cells untreated to control for potential non-specific interactions. After biotinylation, we lysed cells, performed a pulldown with streptavidin beads to enrich for biotinylated proximal interactors, and probed for interaction with BNIP3 and NIX via Western blot. Whereas PPTC7-V5-miniTurbo expressed equally across all conditions, the streptavidin pulldown revealed interactions with BNIP3 and NIX only in the biotin-treated samples, demonstrating specific proximal labeling (Fig 4A). Importantly, these interactions were also specific to PPTC7-V5-miniTurbo and were not seen in either vector-only or V5-miniTurbo control samples (Fig 4A). Furthremore, our western blotting approach resolves both dimer and monomer forms of BNIP3 and NIX, revealing that both forms of these mitophagy receptors proximally associate with PPTC7-V5-miniTurbo. These data, combined with our data demonstrating that recombinant PPTC7 can directly dephosphorylate BNIP3 in vitro (Fig S3D), strongly suggest that PPTC7 directly interacts with BNIP3 and NIX within the native cellular environment.

These data support a model in which PPTC7 is directly recruited to BNIP3 and NIX to promote their turnover. Our DFO washout experiments offer insights into the kinetics of this turnover, which allowed us to test whether the PPTC7-BNIP3/NIX interactions are dynamic throughout the turnover process. Our experiments show that BNIP3 is present at near peak levels 4 h post-DFO washout (Fig 2B and C), leading us to hypothesize that PPTC7 would have enhanced recruitment to BNIP3 and NIX during acute DFO washout to facilitate the turnover of BNIP3 and NIX. To test this, we repeated proximity labeling of PPTC7-V5-miniTurbo under conditions of acute (i.e., 4 h) DFO washout. Interestingly, whereas the PPTC7-BNIP3/NIX interactions are enriched upon 24 h of DFO treatment, similar to our previous experiment, they are further enhanced upon acute 4-h DFO washout (Fig 4B). These data demonstrate that PPTC7 is dynamically recruited to BNIP3 and NIX to promote their turnover upon resolution of DFO-mediated pseudohypoxia.

To further explore the dynamic nature of the PPTC7-BNIP3 interaction, we exploited the recent observation that BNIP3 enriches at LC3-positive punctate structures which likely represent nascent mitophagosomes (Gok et al, 2023). We hypothesized that PPTC7 would also enrich at these punctate structures under conditions of pseudohypoxia, and that this localization would be enriched upon resolution of pseudohypoxia. We overexpressed PPTC7-GFP in U2OS cells treated with the iron chelator DFP for 24 h, and then fixed and immunolabeled for BNIP3 and TOMM20. Examination of over 400 BNIP3-enriched foci in DFP-treated cells revealed that ∼14% of these structures were co-enriched for PPTC7-GFP (Fig 4C). Importantly, these foci were co-localized with the mitochondrial marker TOMM20 (Fig 4C), demonstrating these interactions occur at mitochondria and not other organelles, such the ER, where BNIP3 has been reported to localize (Zhang et al, 2009; Hanna et al, 2012). Remarkably, we found that PPTC7-GFP showed almost a threefold increase co-enrichment (∼46%) with BNIP3-enriched foci 4 h after DFP washout relative to DFP treatment alone (Fig 4D). These data, along with our proximity labeling experiments, suggest that PPTC7 is actively recruited to BNIP3 and NIX under iron chelation conditions that promote their turnover. Collectively, our data are consistent with a model in which a pool of PPTC7 dynamically localizes outside of mitochondria to associate with BNIP3 and NIX–likely through a direct interaction–to promote their ubiquitin-mediated turnover.

Supplies and reagents

DFO and cobalt chloride were purchased from MilliporeSigma (Burlington, MA). MG-132 and polyethylenimine was purchased from Thermo Fisher Scientific. Lipofectamine 3000 was purchased from Thermo Fisher Scientific. Fugene 6 was purchased from Promega. Multiple plasmids were ordered or generated for this study, including: myc-MmBNIP3-FL in pcDNA3.1, which was a kind gift from Joseph Gordon (#100796; Addgene, [Diehl-Jones et al, 2015]) and V5-miniTurbo-NES in pcDNA3.1, which was a kind gift from Alice Ting (#107170; Addgene, [Branon et al, 2018]). pCMV-SPORT6-MmNIX was purchased from Horizon Discoveries. pcDNA3.1-PPTC7-FLAG, pcDNA3.1-FLAG-MmNIX, pcDNA3.1-PPTC7-GFP, pcDNA3.1-PPTC7-OMP25-FLAG, pcDNA3.1-PPTC7-V5-miniTurbo, and pcDNA3.1-ΔMTS-PPTC7-FLAG were generated for this study through standard PCR and restriction-enzyme based cloning techniques. The ΔMTS-PPTC7 truncates the first 14 amino acids from the full-length construct. Plasmids for the HeLa FLP-IN TREx system were generated using Gateway cloning. Briefly, constructs were PCR amplified with primers containing attB1 or attB2 sequences. These fragments were incubated with pDONR221 (a kind gift from Julia Pagan) and BP clonase for recombination. Positive constructs were incubated with LR clonase and the pcDNA5/FRT/TO-Venus-Flag-Gateway destination vector, which was a kind gift from Jonathon Pines (#40999; Addgene, [Di Fiore & Pines, 2010]). All cloned plasmids were validated via Sanger sequencing.

Cell culture and transfection

293T cells were acquired from the American Type Culture Collection. PPTC7 KO 293T cells were generated as previously described (Meyer et al, 2020) and MEF cells were generated from wild-type and Pptc7^−/− mice embryos as previously described (Niemi et al, 2023). Wild-type, Pptc7 KO, and Pptc7/Bnip3/Bnip3l TKO MEFs were transduced with mt-Keima as previously described (Niemi et al, 2023). HeLa FLP-IN TREx cells stably expressing mt-Keima were a kind gift from Dr. Julia Pagan. Cells were cultured in growth media (DMEM supplemented with 10% heat-inactivated FBS and 1× penicillin/streptomycin). Cells were grown in a temperature-controlled incubator at 37°C and 5% CO₂. Transient plasmid transfection into 293T wild-type and PPTC7 KO cells was performed with polyethylenimine for 24–48 h. Transient plasmid transfection into U2OS and HeLa cells was performed with Lipofectamine 3000 for 5 h. Stable plasmid transfection into HeLa FLP-IN TREx cells were performed in the presence of pOG44 in a ratio of 0.5 μg pcDNA5 to 2 μg pOGG44. Transfections were performed with Fugene 6 (Promega) per manufacturer’s directions for 24–48 h before selection with 400 μg/ml hygromycin B.

SDS–PAGE and immunoblotting

Cells were lysed with radioimmunoprecipitation buffer (RIPA; 0.5% wt/vol sodium deoxycholate, 150 mM sodium chloride, 1.0% vol/vol IGEPAL CA-630, 1.0% SDS, 50 mM Tris pH 8.0. 1 mM EDTA pH 8.0 in water) supplemented with 1x protease inhibitor cocktail (0.5 μg/ml pepstatin A, chymostatin, antipain, leupeptin, and aprotinin) and 1x phosphatase inhibitor cocktail (0.5 mM imidazole, 0.25 mM sodium fluoride, 0.3 mM sodium molybdate, 0.25 mM sodium orthovanadate, and 1 mM sodium tartrate) unless otherwise specified. After generating cell lysates, samples were clarified by centrifugation (21,100g) at 4°C for 10 min, snap frozen in liquid nitrogen, and stored at −80°C before use. All samples were quantified with the bicinchoninic acid assay kit (Thermo Fisher Scientific). Lysates were mixed with 5x sample buffer (312 mM Tris-Base, 25% wt/vol sucrose, 5% wt/vol SDS, 0.05% wt/vol bromophenol blue, 5% vol/vol β-mercaptoethanol, pH 6.8) and boiled at 95°C for 10 min. Lysates (20–40 μg) were run on SDS–PAGE gels with Precision All-Blue Protein Standards (Bio-Rad) before being transferred onto nitrocellulose membranes. Membranes were incubated with primary antibodies with 2% BSA or 3% nonfat diary milk in TBS-T for periods as indicated above. Primary antibodies used in immunoblotting include: anti-human BNIP3 (catalog #44060, dilution 1:1,000, 48 h incubation at 4°C; Cell Signaling Technology [CST]), anti-rodent BNIP3 (catalog #3769, dilution 1:1,000, 48 h incubation at 4°C; CST), anti-NIX (catalog #12396, dilution 1:1,000, 48 h incubation at 4°C; CST), anti-PPTC7 (catalog #NBP190654, dilution 1:1,000, 48 h incubation at 4°C; Novus), anti-HIF-1α (catalog #36169, dilution 1:1,000, overnight incubation at 4°C; CST), anti-β-actin (catalog #3700, dilution 1:1,000; CST; catalog #4970, dilution 1:1,000; CST; and catalog #ab170325, dilution 1:1,000; overnight incubation at 4°C; Abcam), anti-FLAG (catalog #F1804, dilution 1:2,000, overnight incubation at 4°C; Sigma-Aldrich), anti-V5 (catalog #PIMA515253, dilution 1:1,000, overnight incubation at 4°C; Thermo Fisher Scientific), and anti-myc (catalog #MA121316, dilution 1:1,000, overnight incubation at 4°C; Thermo Fisher Scientific). Membranes were washed 2–3x with TBS-T for 5 min per wash and incubated with corresponding fluorophore-conjugated antibodies for 30 min at room temperature. Anti-680 or anti-800 conjugated mouse or rabbit antibodies (LiCOR) were used for detection. Membranes were then washed 2–3x with TBS-T for 5 min per wash and were imaged with a LiCOR OdysseyFC instrument using Image Studio software (LiCOR; version 5.2).

Co-immunoprecipitation assay

HeLa FLP-IN TREx mt-Keima-positive cells exogenously expressing doxycycline-inducible FLAG-tagged PPTC7 WT or PPTC7 D78A were treated with 10 μM doxycycline for 48 h and CoCl₂ for 16 h before harvesting. Cell pellets were lysed in a Tris-Triton lysis buffer (50 mM Tris-Cl pH 7.5, 150 mM NaCl, 1 mM EDTA, 1 mM EGTA, 5 mM MgCl₂, 1 mM β-glycerophosphate and 1% Triton) containing a protease inhibitor cocktail (PI78437; Thermo Fisher Scientific) and phosphatase inhibitors (0.5 mM imidazole, 0.25 mM sodium fluoride, 0.3 mM sodium molybdate, 0.25 mM sodium orthovanadate, and 1 mM sodium tartrate) on ice for 30 min. Cell lysates were separated by centrifugation at 20,000g for 10 min at 4°C. To precipitate FLAG-tagged PPTC7 WT and D78A, cell lysates were incubated on a rotator for 2 h at 4°C with Anti-FLAG M2 Affinity Gel beads (A2220; Sigma-Aldrich). The bound species were washed with Tris-Triton lysis buffer four times, with 1-min centrifugations at 2,000g between each wash, before elution with 15 μl 2x sample buffer at 95°C for 10 min. The samples were then run on an SDS–PAGE gel and immunoblotted for NIX and FLAG.

RNA extraction and qRT-PCR

For RNA extraction, the collected cell pellets were processed using Monarch RNA extraction kit (New England BioLabs). The RNA samples were then quantified and normalized before cDNA synthesis (Lambda Biotechnologies). Quantitative PCR analyses with SYBR Green PCR Master Mix (Applied Biosystems) were performed on a Bio-Rad CFX-96 Touch Real-Time PCR Detection System controlled by CFX Maestro (ver2.2) on a computer. Primers used were: GAPDH forward 5′-TTCGCTCTCTGCTCCTCCTGTT-3′, GAPDH reverse 5′-GCCCAATACGACCAAATCCGTTGA-3′, BNIP3 forward 5′-GCCCACCTCGCTCGCAGACAC-3′, BNIP3 reverse 5′-CAATCCGATGGCCAGCAAATGAGA-3′, NIX forward 5′-CTACCCATGAACAGCAGCAA-3′, and NIX reverse 5′-ATCTGCCCATCTTCTTGTGG-3′.

Flow cytometry

For all experiments, wild-type MEFs or HeLa cells (i.e., with no mt-Keima expressed) were used as a negative control for gating and were plated at the same density as other cells in the corresponding experiment. Cells were grown in standard DMEM media (25 mM glucose, 2 mM glutamine, 10% heat-inactivated FBS, 1x penicillin/streptomycin) for 48 h. Cells were harvested by trypsinization and were resuspended in FluoroBrite media with 0.8% heat-inactivated FBS in 5 ml polystyrene round-bottom tubes (Falcon) immediately before the flow cytometry experiments. The LSR-Fortessa (BD Biosciences) flow cytometer was used and was controlled by BDFACSDiva software (version 9.0). Channels used include: FSC (488 nm), SSC (excitation 488 nm, emission 488 nm), QDot 605 (excitation 405 nm, emission 610 nm), and PE-TexasRed (excitation 585 nm, emission 610 nm). The laser intensities for Qdot 605 and PE-TexasRed were changed based on the emission profile of wild-type MEF cells for each experiment and were kept constant throughout the experiment. Cells were gated to select for live cells, single cells, and mt-Keima positive cells sequentially. Once gates were established, they were used for the duration of that experiment. Flow cytometry data were processed with FlowJo (version 10.2.2), and the high mitophagy gate was drawn during analysis.

Immunofluorescence assay and confocal microscopy

For immunofluorescence, cells grown on glass-bottom cover dishes (CellVis) were fixed in 4% PFA solution in PBS (15 min, room temperature). Fixed cells were permeabilized (0.1% Triton X-100 in PBS), blocked (10% FBS and 0.1% Triton X-100 in PBS), and then incubated overnight at 4°C with the indicated primary antibodies in blocking buffer. Primary antibodies used in immunofluorescence assays include anti-TOMM20 (catalog #56783, dilution 1:400; Abcam) and anti-human BNIP3 (catalog #44060, dilution 1:100; CST). After three washes (5 min each) in PBS, cells were incubated with secondary antibodies in blocking buffer for 30 min. The secondary antibodies used in immunofluorescence assays include donkey anti-mouse Alexa Fluor 647 (catalog #A-31571, dilution 1:400; Thermo Fisher Scientific) and donkey anti-rabbit Alexa Fluor 555 (catalog #A-31572, dilution 1:400; Thermo Fisher Scientific). Cells were subsequently washed three times in PBS before imaging. Images were acquired on a Nikon Ti2 microscope equipped with Yokogawa CSU-W1 spinning disk confocal and SoRa modules, a Hamamatsu Orca-Fusion sCMOS camera and a Nikon 60x objective (for Fig 3A and B) or Nikon 100x 1.45 NA (for Fig 4C and D) objective. All images were acquired using a 0.2-μm step size with the spinning disk module, and image adjustments were made with ImageJ/Fiji.

DFO-induced iron chelation and washout

For immunoblotting experiments to visualize changes of BNIP3 and NIX protein levels upon iron chelation, 293T wild-type and PPTC7 KO cells were treated with 100 μM DFO for 0–24 h as indicated in the results section. After the indicated incubation time, the cells were harvested by cell scraping in PBS before cell lysis and immunoblotting analysis. For qRT-PCR experiments to quantify changes in BNIP3 and BNIP3L transcript level, 293T WT and PPTC7 KO cells were treated with 100 μM DFO for 24 h, and the cells were harvested by cell scraping in PBS before RNA extracton and qRT-PCR analysis. For DFO washout experiments, after treating 293T WT and PPTC7 KO cells with 100 μΜ DFO for 24 h, cells were washed with PBS and was switched to fresh DMEM media (for Fig 2B–F) or with DMEM containing 0.1% EtOH (as vehicle control) or 10 μM MG-132 (for Fig 2I) for 0–24 h as indicated in the results section. After the indicated incubation time, the cells were harvested by cell scraping in PBS before cell lysis and immunoblotting analysis. For flow cytometry experiments to determine the change of mitophagic flux with DFO dosage, MEF mt-Keima wild-type and Pptc7 KO cells were treated with 0–100 μM DFO as indicated in the results section for 24 h before downstream processing for flow cytometry. To understand whether the DFO-induced mitophagy is BNIP3- and NIX-mediated, MEF mt-Keima wild-type, Pptc7 KO, and Pptc7/Bnip3/Bnip3l TKO cells were treated with 100 μM DFO for 24 h before downstream processing for flow cytometry.

MG-132 proteasomal inhibition time course

For MG-132 proteasomal inhibition time course experiment performed in Fig 2G and H, 293T WT and PPTC7 KO cells were treated with 0.1% EtOH (as vehicle control) or 10 μM MG-132 for 0–6 h as indicated in the results section. After the indicated incubation time, the cells were harvested by cell scraping in PBS before cell lysis and immunoblotting analysis.

Protease protection assay

HeLa FLP-IN TREx PPTC7-FLAG mt-Keima-positive cells were seeded at 1 × 10⁶ in a 15-cm² plate. After 48 h, cells were treated with 10 μM doxycycline to induce PPTC7-FLAG expression for 40 h. Cells were then harvested by scraping in PBS and spun down at 600g for 10 min at 4°C. To obtain a mitochondrial-enriched fraction, cells were resuspended in mitochondrial isolation buffer (0.1 M Tris-MOPS pH7.4, 0.1 M EGTA, 1 M sucrose in water) and homogenized at 1,000 rpm with a Potter-Elvehjem tissue homogenizer at 4°C. The resulting mixture was then centrifuged at 600g for 10 min at 4°C, and the supernatant was then kept and recentrifuged at 7,000g for 10 min at 4°C. The resulting pellet was the washed and resuspended with mitochondrial isolation buffer and quantified by using the Pierce 660 nm Protein Assay Kit (Thermo Fisher Scientific). After the mitochondrial-enriched fraction was obtained, 15 μg of crude mitochondria were used per condition. Proteinase K (10 μg/ml final concentration, NEB) and/or SDS (1% final, in 50 mM Tris–HCl, pH 8.0) were added as necessary per condition. The reaction mixtures were incubated on ice for 30 min, quenched by the addition of 5x sample buffer, and incubated at 95°C for 10 min. The samples were then loaded onto SDS–PAGE gels for downstream immunoblotting analyses.

Analysis of PPTC7 localization relative to BNIP3

U2OS cells were transiently transfected with PPTC7-GFP plasmid, and then passaged into glass-bottom dishes. Cells were allowed to adhere for 12 h and subsequently treated with freshly prepared DFP (1 mM; Sigma-Aldrich) for 24 h before fixation. Where indicated, cells were then washed with fresh media, and fixed after an additional 4 h of incubation. Immunofluorescence staining and confocal microscopy were then performed on cells. To determine the co-enrichment of PPTC7 with BNIP3, enlarged foci enriched for BNIP3 signal relative to TOMM20 were manually counted in Fiji by examining single plane images throughout z-series of individual cells blinded to the corresponding PPTC7 image, followed by assessment of whether PPTC7 was co-enriched.

Analysis of the effect of PPTC7 overexpression on BNIP3 immunofluorescence

For analysis of BNIP3 staining by immunofluorescence, HeLa cells were transiently transfected with cytosolic GFP or PPTC7-GFP plasmids. Cells were allowed to adhere to glass-bottom dishes for 12 h and treated with freshly prepared CoCl₂ (500 μM; Sigma-Aldrich) 12 h before fixation. Cells were imaged with identical imaging conditions between experimental replicates. Cells that had GFP signal above an arbitrary threshold that was consistently maintained between experiments were blindly identified and manually categorized as having mitochondrial or diffuse non-mitochondrial BNIP3 signal in Fiji by examining maximum z-projections.

Analysis of the effect of PPTC7 mutant expression on BNIP3 and NIX total protein levels

HeLa FLP-IN TREx cells expressing wild-type PPTC7 or various mutants (e.g., D78A) were generated as described above. To induce expression of these constructs, cells were treated with 10 μM doxycycline for 24–32 h, after which BNIP3 and NIX expression was induced by treating cells with 200 μM CoCl₂ for 16 h. Cells were collected, lysed in RIPA buffer, run via SDS–PAGE and western blotted for each protein as described above (see the “SDS–PAGE and immunoblotting” section).

Proximity labeling

Wild-type 293T cells were transiently transfected with pcDNA3.1-MCS (empty vector), pcDNA3.1-V5-miniTurbo-NES, or pcDNA3.1-PPTC7-V5-miniTurbo before cells were treated with either vehicle only or 100 μM DFO (supplemented in growth media) for another 24 h. For labeling, DMSO or 250 μM biotin were added to cells, and cells were incubated at 37°C for 30 min. Labeling reactions were quenched by washing the cells three times with ice-cold PBS and incubating the cells at 4°C. Cells were then harvested in ice-cold PBS before lysed with RIPA, clarified, and quantified via bicinchoninic acid. The lysates were then incubated with 80 μl streptavidin magnetic beads (New England Biolabs) at 4°C overnight. The beads were then washed twice with RIPA buffer, once with Wash Buffer 2 (500 mM sodium chloride, 0.1% wt/vol deoxycholate, 1% Triton X-100, 1 mM EDTA, 50 mM HEPES, pH 7.5), once with Wash Buffer 3 (250 mM sodium chloride, 0.5% Triton X-100, 0.5% wt/vol deoxycholate, 1 mM EDTA, 50 mM HEPES, pH 8.1), once with Wash Buffer 4 (150 mM sodium chloride, 50 mM HEPES, pH 7.4), and once with Wash buffer 5 (50 mM ammonium bicarbonate in MS-grade water). The bound species were eluted by adding 20 μl 5x sample buffer, boiling at 95°C for 10 min, vortexing for 10 min, and boiling at 95°C for 10 min. The samples were then run on SDS–PAGE gels, and immunoblotting was performed to probe for BNIP3 and NIX.

Data/statistical analysis and figure generation

Statistical analysis was performed using Microsoft Excel and/or Prism software (GraphPad, version 10). Fig S1A was generated using Seaborn library on Python. A selected list of HIF target genes were curated (Dengler et al, 2014) as mined from datasets in mouse Pptc7 KO heart and liver proteomics (Niemi et al, 2019). Fig 2C was generated using AlphaFold2. Fig S3E was generated using AlphaFold multimer mode; images were generated using PyMOL (Version 2.5.2). Fig 3A was generated using BioRender with an appropriate license. All figures were generated using Adobe Illustrator.