The Ubiquitin Ligase RBX2/SAG Regulates Mitochondrial Ubiquitination and Mitophagy

Abstract

BACKGROUND:

Clearance of damaged mitochondria via mitophagy is crucial for cellular homeostasis. Apart from PARKIN, little is known about additional ubiquitin (Ub) ligases that mediate mitochondrial ubiquitination and turnover, particularly in highly metabolically active organs such as the heart.

METHODS:

In this study, we have combined in silico analysis and biochemical assay to identify Cullin-RING Ub ligase 5 (CRL5) as a mitochondrial Ub ligase. We generated cardiomyocytes and mice lacking RBX2 (also known as SAG), a catalytic subunit of CRL5, to understand the effects of RBX2 depletion on mitochondrial ubiquitination, mitophagy and cardiac function. We also performed proteomics analysis and RNAseq analysis to define the impact of loss of RBX2 on proteome and transcriptome.

RESULTS:

RBX2 and Cullin 5, two core components of CRL5, localize to mitochondria. Depletion of RBX2 inhibited mitochondrial ubiquitination and turnover, impaired mitochondrial membrane potential and respiration, increased cardiomyocyte cell death, and has a global impact on the mitochondrial proteome. In vivo, deletion of the Rbx2 gene in adult mouse hearts suppressed mitophagic activity, provoked accumulation of damaged mitochondria in the myocardium, and disrupted myocardial metabolism, leading to the rapid development of dilated cardiomyopathy and heart failure. Similarly, ablation of RBX2 in the developing heart resulted in dilated cardiomyopathy and heart failure. The action of RBX2 in mitochondria is not dependent on PARKIN, and PARKIN gene deletion had no impact on the onset and progression of cardiomyopathy in RBX2-deficient hearts. Furthermore, RBX2 controls the stability of PINK1 in mitochondria.

CONCLUSIONS:

These findings identify RBX2-CRL5 as a mitochondrial Ub ligase that regulates mitophagy and cardiac homeostasis in a PARKIN-independent, PINK1-dependent manner.

Graphical Abstract

INTRODUCTION

The removal of damaged mitochondria through mitophagy, a selective form of autophagy, is critical for maintaining mitochondrial health and thus cellular hemostasis. Labeling damaged mitochondria with a phosphorylated ubiquitin (Ub) chain facilitates the recruitment of mitophagy receptors (i.e., p62, NBR1) and LC3-incorporated autophagosomes to eliminate dysfunctional mitochondria ^1–3. This process is reportedly driven by two key proteins: the E3 Ub ligase PARKIN and the kinase PINK1 ^1,2. Following mitochondrial depolarization, PINK1 is stabilized and activated on the mitochondrial outer membrane (MOM), where it phosphorylates Ub at serine 65 (pS65-Ub) and activates PARKIN^4–6. Active PARKIN, in turn, mediates the ubiquitination of other mitochondrial proteins for subsequent PINK1 phosphorylation, thereby resulting in a feed-forward loop and signal amplification.

Mitochondria play a crucial role in providing energy required for the continuously beating heart. In adult cardiomyocytes, mitochondria are constitutively turned over, as indicated by a half-life of around 6–17 days ^7,8 and a high mitophagy activity in unstressed adult mouse hearts ^9,10. Known as a master regulator of mitophagy, PARKIN is expressed at very low levels in unstressed hearts ¹¹, and germline or cardiac-specific deletion of Parkin had no impact on mitochondrial or cardiac function at baseline and during ageing^12–14. These findings suggest the existence of PARKIN-independent mitophagy in the heart, which has been linked to receptor-mediated mitophagy^15,16 and compensation via alternative Ub ligases such as HUWE1, ARIH1, MITOL and SIAH1^17–20. However, germline deletion of Bnip3 gene, or cardiac-specific ablation of Bnip3l, did not produce cardiac dysfunction ^21,22, arguing against a dominant role for receptor-mediated mitophagy in mitochondrial turnover in cardiomyocytes. While deletion of Huwe1 and Mitol in the heart caused cardiomyopathy and heart failure^23,24, neither has been directly implicated in mitochondrial ubiquitination and turnover in cardiomyocytes. More recently, the ubiquitin ligase TRAF2 was shown to regulate mitophagy and cardiac homeostasis likely through a PARKIN-dependent mechanism²⁵. Therefore, the ubiquitin ligases that control mitochondrial ubiquitination and clearance in cardiomyocytes under physiological conditions remain unidentified.

Cullin-RING Ub ligase 5 (CRL5) belongs to the Cullin-RING Ub ligase family, the largest family of Ub ligases ^26,27. CRL5 consists of four components: the scaffold protein Cul5, RING protein RBX2 (also known as SAG/ROC2/RNF7), adaptor proteins Elongin B/C and one of many substrate-recognizing receptors ²⁸. Neddylation of cullins is required for the assembly and activation of functional CRLs²⁹. As a core component of CRL5, RBX2 facilitates CUL5 neddylation and bridges Ub E2 to its substrates ²⁸, which is essential for CRL5 activity. While RBX1, the other RING finger protein, shares 56% homology with RBX2, it primarily binds to other cullins (CULLIN 1, 2, 3, 4A and 4B) to activates CRLs 1–4. RBX2-CRL5 targets various substrates for proteolysis to regulate embryonic development, inflammation, viral infection, vasculogenesis and tumorigenesis ^30–36. RBX2 is highly expressed in the heart of human and mice ³⁷, but the role of RBX2-CRL5 in mitochondrial turnover in the heart has not been explored.

In this study, we identify an association of CRLs with mitochondria. We show that RBX2 and Cul5, two key components of CRL5, localize to the mitochondria and are required for mitophagy in cardiomyocytes. Deletion of Rbx2 in mouse hearts provokes accumulation of damaged mitochondria leading to the rapid development of heart failure and premature lethality. Mechanistically, loss of RBX2 inhibits mitochondrial ubiquitination and turnover and impairs mitochondrial respiration and function in cardiomyocytes in a PARKIN-independent, PINK1-dependent manner. Our results support a pivotal role for the RBX2-CRL5 axis in maintaining mitochondrial and cardiac integrity mostly likely through directly ubiquitinating mitochondrial proteins.

METHODS

Data Availability

All supporting data and the expanded Methods section are available within the article and its online Supplemental Material. RNA sequencing data are accessible in the Gene Expression Omnibus (GSE265777). The proteomics data are available in the Center for Computational Mass Spectrometry (MSV000094583).

Animals

Mice carrying a Rbx2^Flox allele were previously reported³⁶. αMHC^Cre/+ mice (# 005657), αMHC-MerCreMer mice (#011038) and Parkin knockout mice (#006582) were obtained from the Jackson Laboratory. Tamoxifen (Sigma, Cat# T5648) was administered to 12 to 16-week-old mice by intraperitoneal injection at a dosage of 50 mg/kg/day for 5 days or 20 mg/kg/day for 10 days (5 consecutive days with a 2-day interval in between). All mice were maintained on a C57BL/6J background. Both male and female mice were used in this study. All animal experiments were approved by the Augusta University Institutional Animal Care and Use Committee.

Statistics

Statistical analyses were performed with GraphPad Prism software 10.1. All values are presented as mean±SEM. The Shapiro-Wilk normality test was used to test for normal distribution prior to data analysis. Statistical significance tests used are specified in each figure legend. Values of P<0.05 were considered statistically significant.

RESULTS

Identification of CRL5 as a mitochondria-associated Ub ligase

Using peroxidase (APEX2)-mediated proximity biotinylation, a recent study defined the mitochondrial outer membrane (MOM)-interacting proteome³⁸. Cross referencing this MOM-interactome with the E3 Ub ligase database (ESI Network ³⁹) identified 261 Ub ligase proteins (Figure 1A). Among these, 59 (~22.6%) belong to the CRL family, including RBX1, RBX2, cullin proteins, and their substrate receptors (Figure S1A–B). Several neddylation enzymes, such as NAE1, UBA3, UBE2M, and multiple subunits of deneddylase COP9 signalosome (CSN), were also identified as MOM-interacting proteins (Figure S1B). Similarly, overlap of the MOM-interactome with Ub ligases included in UbiNet 2.0 ⁴⁰ confirmed the enrichment of CRLs in mitochondria (Figure S1C).

Figure 1. RBX2 and Cul5 localize to the mitochondria.

A, Schematic of APEX2-catalyzed biotinylation at the mitochondrial outer membrane (MOM) via fusion to a mitochondrial targeted peptide derived from MAVS (mitochondrial antiviral-signaling protein). Overlap of APEX2-MOM data with the ESI (E3-substrate interaction) network reveals the association of CRLs with mitochondria. B, Representative Western blot of biotinylated proteins in neonatal rat ventricular cardiomyocytes (NRVCs) with adenoviral (Ad) expression of APEX2-MOM. NRVCs were treated with CCCP (10 μM) before H₂O₂ activation. The resultant biotinylated proteins were enriched by streptavidin beads. C, Representative Western blot of cytosol and mitochondrial fractions from NRVCs treated with CCCP (10 μM) for the indicated times. Arrowhead, neddylated CUL5. Tubulin and VDAC serve as cytosol and mitochondrial markers, respectively. D, Immuno-gold electron microscopic images showing the localization of RBX2 on mitochondrial membranes (arrowheads) in NRVCs expressing HA-RBX2. NRVCs infected with Ad-GFP and Ad-HA-OMP25 (MOM protein) serve as negative and positive controls, respectively. Scale bars, 0.5 μm. E, Representative Western blot of mitochondria with or without proteinase K (PK) treatment for 30 min on ice after isolation from NRVCs. F, Confocal images showing the colocalization (arrowhead) of RBX2 with mitochondria in CCCP-treated cardiomyocytes. NRVCs with adenoviral expression of HA-RBX2 were treated with CCCP for 3 hours and stained or immunostained as indicated. HA, green. Mitotracker, red. TOMM20, blue. Line scan co-localization analysis was done for all channels. Scale bars, 50 μm.

To validate the association of CRLs with mitochondria, neonatal rat ventricular cardiomyocytes (NRVCs) were infected with adenovirus expressing APEX2-MOM and treated with H₂O₂ to trigger biotinylation of MOM-interacting proteins. We detected the presence of MOM proteins (TOMM20, CISD1, TOMM40, SAMM50 and VDAC) and discernible levels of RBX2 in enriched biotinylated proteins at baseline (Figure 1B and Figure S1D). Interestingly, treatment with the mitochondrial uncoupler carbonyl cyanide chlorophenylhydrazone (CCCP) increased the biotinylated forms of RBX2 and CUL5, but not those of RBX1, indicating enhanced association of RBX2 and CRL5 with mitochondria (Figure 1B). Western blot following subcellular fractionation detected the presence of RBX2 and CUL5, as well as neddylation enzymes, in crude mitochondrial lysates (Figure 1C). Notably, neddylated CUL5 was predominant in mitochondria versus the cytosol, implying a role for active CRL5 in mitochondria. Immuno-gold electron microscopy detected the expression of exogenous RBX2 on mitochondrial membranes, consistent with the expression pattern of MOM protein OMP25 (Figure 1D). Moreover, in isolated mitochondria, RBX2 and CUL5, like the MOM protein TOMM20, were more sensitive to proteolysis induced by proteinase K (PK), compared with mitochondrial inner membrane proteins (ATP5a and UQCRC2) and matrix protein (SOD2) (Figure 1E). Immunostaining showed that CCCP treatment induced RBX2 recruitment to mitochondria (Figure 1F). Moreover, immunostaining of myocardium section also revealed the presence of RBX2 in mitochondria (Figure S1E). Together, these findings suggest RBX2 and CUL5 proteins partially localize to the mitochondrial outer membrane, and mitochondrial stress promotes their mitochondrial translocation.

RBX2 mediates mitochondrial ubiquitination and turnover

To define the functional role of CRL5 in mitochondria, RBX2 was silenced in NRVCs with two different siRNAs, which reduced CUL5 protein levels likely due to the disassembly and degradation of the CRL5 complex (Figure 2A and Figure S2A). Silencing of RBX2 markedly attenuated CCCP-induced pS65-Ub in both total cell lysates and mitochondrial fractions, in conjunction with reduced levels of P62 (autophagy receptor) and the accumulation of mitochondrial proteins (ATP5A, SDHB, MCQRC2, NDUFB8) (Figures 2A and 2B). Inhibition of proteasome function with Bortezomib (BZM) did not restore the levels of pS65-Ub and total ubiquitinated proteins in mitochondrial fractions in CCCP-treated, RBX2-deficient cardiomyocytes (Figure S2B), indicating that RBX2 does not target these mitochondrial proteins for proteasomal degradation. Assessment of mitolysosome numbers with cell permeable Mtphagy dye ⁴¹ revealed a significant reduction of mitophagic vesicles in RBX2-deficient cardiomyocytes at baseline and following CCCP treatment (Figure 2C and 2D). Moreover, the mitophagy flux assay showed a diminished turnover of LC3-II and P62 in mitochondria in RBX2-deficient cardiomyocytes (Figure 2E and S2C). Together, these data demonstrate that RBX2 is necessary for mitochondrial ubiquitination and turnover in cardiomyocytes.

Figure 2. RBX2 mediates mitochondrial ubiquitination and mitophagy.

A, Western blot of total cell lysates. NRVCs were transfected with indicated siRNAs, followed by CCCP (10 μM) treatment for the indicated times. B, Western blot of mitochondrial (mito) and cytosol (cyto) extracts. Cells were treated as described in A. C, Representative confocal images of live NRVCs stained with Mtphagy dye (red) and LysoTracker (blue) showing mitophagic vesicles. D, Quantification of mitophagic puncta per cell. E, Western blot of mitochondrial extracts. NRVCs were transfected with indicated siRNAs and treated with lysosomal protease inhibitors, E64d (E, 10 μg/ml) and pepstatin A (P, 10 μg/ml) for 6 hours before harvest. Fold-change of LC3-II and p62 levels between cells treated with and without lysosome inhibitors, indicative of mitophagy flux, is quantified on the right. F, Oxygen consumption rate (OCR) assessed by Seahorse analysis. Basal (Bas.) and maximal (Max.) respiration (resp.) are quantified on the right. siLuci: n=6, siRBX2: n=5 biological replicates. Data are representative of 3 repeats. G, Representative confocal images of NRVC stained with TMRM (1 μM, red) and MitoTracker (1 μM, green). The mean fluorescent intensity per view from 7–8 views (over 100 cells) per group is quantified on the right. H, Representative confocal images of NRVC stained with MitoSOX (red). The mean fluorescent intensity per view from 10 views (over 100 cells) per group is quantified on the right. Two-way ANOVA with the Tukey multiple comparisons test was used in D, the Mann-Whitney test in E-F, and unpaired t test in G-H.

Mitophagy maintains mitochondrial fitness and cellular health. Seahorse analysis demonstrated that depletion of RBX2 significantly decreased basal and maximal mitochondrial respiration and ATP production (Figure 2F). Furthermore, RBX2-deficient cardiomyocytes exhibited decreased mitochondrial membrane potential and elevated mitochondrial reactive oxygen species (Figure 2G–H). Consequently, RBX2-deficient cardiomyocytes were more susceptible to CCCP-induced cellular injury and cell death (Figure S2D–S2F). Thus, these data suggest that RBX2 plays an important role in maintaining cardiomyocyte mitochondrial integrity.

RBX2 is indispensable for the function of adult mouse heart

We next generated tamoxifen (TAM)-inducible, cardiomyocyte-specific RBX2 knockout (RBX2^iCKO, hereafter iCKO) mice by crossing Rbx2^Flox mice with αMHC^MerCreMer (MCM) mice (Figure 3A). Tamoxifen administration (50 mg/kg/d for 5 days) induced efficient cardiac RBX2 depletion and a concomitant reduction of CUL5 proteins in the iCKO hearts, with no change in CUL3 or CUL4a (Figure 3B and 3C). Nearly 40% of iCKO mice died within two weeks after tamoxifen injection (Figure 3D). Compared with MCM and F/F mice, iCKO mice promptly developed dilated cardiomyopathy and heart failure at 12 days after tamoxifen injections, as evidenced by enlarged heart size, increased ratios of heart weight to tibial length and lung weight to tibial length, left ventricular (LV) wall thinning, significant LV chamber dilatation and severely impaired cardiac contractility (Figures 3E–3H, Table S1). Consistently, deletion of RBX2 led to significantly increased cardiomyocyte cell size, marked cardiomyocyte apoptosis, upregulated expression of myocardial stress markers (Acta1, Myh7) and collagen (Col1a, Col3a), and reduced expression of Myh6 (Figure 3I–K). To control for cardiotoxicity resulting from the MCM transgene⁴², we employed a two-week regimen of tamoxifen injections at lower doses (20 mg/kg/d for 10 days with a 2-day interval after the first 5 injections). This regimen also resulted in cardiomyopathy in iCKO mice in comparison with MCM and F/F mice (Figure S3A–S3E, Table S2). Cardiac dysfunction was not as pronounced following the low-dose tamoxifen injection protocol, possibly due to less efficient deletion of Rbx2. Collectively, these data demonstrate that RBX2 is necessary to maintain normal function of the adult heart.

Figure 3. Deletion of Rbx2 in adult heart leads to heart failure and lethality.

A, Schematics of creation of tamoxifen-inducible, cardiac-specific RBX2 knockout (iCKO) mice. B, Western blot of indicated proteins in mouse hearts at 12 days after tamoxifen injection. C, Quantification of B. D, Survival curve. E, Gross morphology of mouse heart (top) and hematoxylin and eosin staining of myocardium section (bottom) at 12 days after tamoxifen injection. F, Heart weight to tibial length ratio and lung weight to tibial length ratio. F/F: n=13, MCM: n=5, iCKO: n=18 mice. G, Representative B-mode images. H, Quantification of echocardiographic parameters before (F/F: n=17, MCM: n=6, iCKO: n=23 mice) and after (F/F: n=12, MCM: n=6, iCKO: n=12 mice) tamoxifen treatment. I, Wheat germ agglutinin (WGA) staining (left) of myocardium sections and quantification of cardiomyocyte (CM) cross-sectional area (right). More than 100 cells/heart and 3 and 12 hearts from F/F and iCKO mice, respectively, were quantified. J, Terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) staining of myocardium sections (left) and quantification (right). Three fields per heart, and 3 hearts per group, were quantified. K, qPCR analysis of the indicated genes. F/F: n=4, iCKO: n=4 mice. The Kruskal-Wallis test with the Dunn multiple comparisons test was used in C, F and H, Log-rank (Mantel-Cox) test in D, the Mann-Whitney U test in I-K.

Perinatal deletion of RBX2 results in cardiomyopathy during ageing

To determine the functional importance of RBX2 in the developing heart, we created αMHC^Cre-mediated RBX2 knockout (RBX2^CKO, hereafter CKO) mice (Figure S4A). Deletion of Rbx2 in postnatal hearts reduced CUL5 expression but increased the levels of RBX1 protein (Figure S4B–S4D). While endothelial-specific deletion of RBX2 causes embryonic lethality ³⁵, CKO mice were viable and morphologically indistinguishable from their littermate controls. Serial echocardiography showed that compared with F/F and αMHC^Cre mice, RBX2^CKO mice exhibited discernible cardiac dysfunction (EF%: 65.9% of CKO vs 73.9% of Rbx2^F/F) as early as 1 month of age and severe deterioration by 8 months of age (Figure S4E). αMHC^Cre mice develop cardiomyopathy during ageing^43,44. At 8 months of age, CKO hearts were enlarged compared with the hearts from F/F and αMHC^Cre mice and exhibited more pronounced LV wall thinning and systolic dysfunction (Figure S4E–H, Table S3). Moreover, CKO hearts exhibited more myocardial interstitial fibrosis and heightened expression of cardiac stress maker genes such as Nppa, Nppb, Mhy7 and Acta1, and downregulation of Atp2a2 expression (Figure S4I–J), indicative of adverse cardiac remodeling. These data support a crucial role of RBX2 in maintenance of cardiac homeostasis.

RBX2 deficiency has a major impact on the mitochondrial proteome

To gain mechanistic insight into how RBX2 regulates cardiomyocyte function, we performed tandem mass tag (TMT)-based quantitative proteomics analyses on total cell lysates from control (CTL) and RBX2-deficient (KD) cardiomyocytes in the absence and presence of CCCP treatment (Figure 4A). Principal component analysis (PCA) of protein abundances showed distinct separation between the different groups (Figure 4B). Mass spectrometry (MS) analysis did not identify any differentially expressed proteins (DEPs) between CTL and KD groups, between CTL+CCCP and CTL groups, and between KD+CCCP and KD groups (FDR<0.1, FC>1.3 or <0.77) (Figure 4C). These data indicate a negligible impact of RBX2 deficiency and CCCP treatment (for 6 hours) on the global proteome. In contrast, following CCCP treatment, ~18.5% of the quantitively detected proteins were dysregulated (180 downregulated and 269 upregulated) in RBX2-deficient cardiomyocytes compared with CCCP-treated CTL cardiomyocytes (Figure 4C and 4D, Table S4). Gene ontology analysis identified the enrichment of upregulated proteins in the pathways relevant to oxidative phosphorylation, diabetic cardiomyopathy, cardiac muscle contraction, electron transportation chain and in mitochondrial subcompartments such as matrix and outer membrane, whereas the downregulated proteins are enriched in pathways related to actin cytoskeleton organization and the sarcomere (Figure S5A–5B). Notably, 127 (47.2%) of the 269 upregulated proteins are annotated in MitoCarta 3.0 (Figure 4E), whereas only 10 (5.6%) of the 180 downregulated proteins are in mitochondria. These dysregulated mitochondrial proteins localize in different mitochondrial subcompartments (Figure S5C and 5D). These data are consistent with a critical role for RBX2 in the regulation of mitochondrial turnover.

Figure 4. RBX2-regulated mitochondrial proteome.

A, Scheme of procedures for identification Rbx2-regulated proteome in NRVCs. NRVCs were transfected with indicated siRNAs, followed by CCCP (10 μM) treatment for 6 hours. Cell lysates were collected for trypsin digestion. The resultant peptides were labeled by TMT before mass spectrometry analysis. B, Principal component analysis (PCA) of normalized protein expression in whole cell proteome. CTL, siLuci. KD, siRBX2. C, Analysis of the percentage of differential expression of proteins (DEPs) in each group. D, Volcano plot of differentially expressed proteins (Sig., blue and red) in CCCP-treated RBX2-deficient CMs (KD) compared with CCCP-treated CTL. E, Venn diagram showing overlap of RBX2-regulated proteome and mitochondrial proteins annotated in MitoCarta 3.0. F, Venn diagram showing the identification of RBX2-regulated MOMs. G, Heatmap showing the relative expression of MOM proteins amongst the groups of cells. H, Western blot of MOM proteins in control and RBX2-deficient cardiomyocytes. N=4 biological replicates per group. Multiple unpaired t test with the Holm-Šídák correction was used. I, Western blot of WT and RBX2KO Hela cells treated with cycloheximide (CHX, 100 nM) for indicated times. RBX2 was deleted in Hela cells via CRISPR/Cas9-mediated gene editing. J, WT and RBX2KO Hela cells were transfected with plasmids expressing HA-Ub (pCDNA3-HA-Ub), treated with a proteasome inhibitor Bortezomib (BZM, 100 nM) for 6 hours, and subjected to immunoprecipitation followed by Western blot. The Mann-Whitney U test was used in I.

Crossing the mitochondrial DEPs with MOMs annotated in MitoCarta 3.0 and the APEX2-MOM study ³⁸ identified 21 MOMs as potential RBX2 substrates (Figure 4F). All of these MOMs except ITGA3 were upregulated in the KD+CCCP group compared with the CTL+CCCP group and showed a trend towards upregulation, though not statistically significant, in the KD group compared with the CTL group (Figure 4G). Furthermore, among 112 MOM proteins annotated in MitoCarta 3.0, 37 were detected by our assay and 10 were significantly upregulated in RBX2-deficient cardiomyocytes (Figure S5E). Immunoblotting confirmed the upregulation of these MOM proteins (TOMM20, TOMM40, VDAC1, CISD1, RHOT2, SAMM50, ACSL1) and MIM proteins (HSP60, AGK, CYC1, NUDUFS5), without discernable changes in their transcript levels in RBX2-deficient cardiomyocytes (Figure 4H, Figure S5F–5G). Cycloheximide-based pulse chase assays showed reduced degradation rate of ASCL1, VDAC, SAM50 and TOMM40 in RBX2-deficient Hela cells (Figure 4I). Moreover, immunoprecipitation showed that RBX2 deficiency inhibits the ubiquitination of SAM50 and ASCL1 (Figure 4J). Together, these data suggest that RBX2 controls the ubiquitination and degradation of a subset of MOMs.

RBX2 deficiency inhibits mitochondrial ubiquitination and mitophagic activity in the heart

We next examined the impact of RBX2 deficiency on cardiac mitochondrial turnover in iCKO hearts. Detection of phosphorylated ubiquitin in tissues is challenging⁴⁵. Indeed, we did detect the myocardial levels of phosphorylated ubiquitin by western blot, likely due to a relatively low percentage of mitochondria undergoing mitophagy in the normal heart at a given time point. Immunostaining of pUb on myocardium sections from MCM detected pUb-positive mitochondria, which were significantly reduced in iCKO hearts (Figure 5A and 5B). Western blot showed a trend of reduction in mitochondrial P62, an adaptor protein that delivers mitochondria to autophagosomes for degradation⁴⁶, but not those in cytosol, in iCKO hearts (Figure 5C and 5D). Consistently, immunostaining also detected diminished co-localization of p62 with mitochondria in iCKO hearts (Figure S6A). We further performed autophagy flux assay in mouse hearts by treating MCM and iCKO mice with a lysosome inhibitor bafilomycin A1 (BFA). Western blot showed that loss of RBX2 suppressed LC3-II turnover in mitochondrial fraction (Figure 5E and 5F). Examination of mitophagic activity via AAV9-mediated transduction of mt-Keima reporter⁴⁷ demonstrated marked suppression of mitophagy in iCKO hearts (Figure 5G and 5H). Assessment of mitophagic activity in 2-month-old CKO hearts, in which cardiac function is largely maintained, also demonstrated decreased mitophagy activity (Figure S6B–S6C). In isolated adult cardiomyocytes, RBX2 deficiency reduced the levels of phosphorylated ubiquitin (Figure 5I) and CCCP-induced mitophagic vesicles (Figure S6D–S6E). Together, these in vivo and ex vivo data suggest that RBX2 is required for the ubiquitination of mitochondria proteins and mitochondrial turnover in the heart.

Figure 5. Impaired mitochondrial ubiquitination and mitophagy in RBX2-deficient hearts.

Adult MCM and RBX2^iCKO (iCKO) mice were administered with tamoxifen (50 mg/kg/d for 5 days). Tissues were collected for indicated analyses (A-H) at 12 days after tamoxifen injections. A, Representative confocal images (left) of MCM and RBX2^iCKO myocardium sections immunostained with pUb (green), HSP60 (red, mitochondrial marker) and DAPI. Scale bars, 10 μm. B, Quantification of pUb+ foci normalized by mitochondria (HSP60+) area. A total of 16 views from two hearts per group were quantified. C-D, Western blot (C) and quantification (D) of mitochondrial (mito) and cytosolic (cyto) P62 in mouse hearts. E-F, Western blot (E) and quantification (F) of mitochondrial (mito) and cytosolic (cyto) LC3-II in mouse hearts. Mice at 12 days after tamoxifen administration were intraperitoneally injected with bafilomyocin A1 (BFA, 3 μmol/kg) for 3 hours before tissue harvest. 4 hearts per group were quantified. G, Representative confocal images of mt-Keima at 488 nm and 568 nm, respectively, in epicardial cardiomyocytes and the derived heatmaps. Neonatal MCM and RBX2^iCKO mice were transduced with AAV9-mt-Keima (1X10¹¹ GC/pup). At 10 weeks of age, mice were treated with tamoxifen and intact mouse hearts were excised 12 days later and scanned for mt-Keima signals in epicardial cardiomyocytes in situ with confocal microscope. Scale bars, 20 μm. H, Quantification of relative 568/488 ratio. 20–40 views per heart, 3 hearts per group were quantified. I, Western blot of indicated proteins in adult cardiomyocytes isolated from 2-month-old CTL (RBX2^F/F) or RBX2^CKO (CKO) mouse hearts. Results from two different batches of cells are shown. The Mann-Whitney test was used in B, D and F, and nested unpaired t test in H.

Loss of RBX2 disrupts metabolic pathways and impairs mitochondrial homeostasis

To gain additional insights into how RBX2 regulates cardiac function, we performed transcriptomic analyses of CTL (F/F) and iCKO hearts. A total of 453 downregulated and 881 upregulated genes were identified in RBX2-deficient hearts (FC > or < 1.5, P_adj < 0.05. Figure S7A–S7B, Table S5). Pathway enrichment analysis revealed enrichment of downregulated genes in fatty acid degradation, PPAR signaling and cardiac muscle contraction, while upregulated genes were enriched in extracellular matrix remodeling, relaxin signaling and PI3K-Akt signaling, among others (Figure 6A, Figure S7C). Specifically, many mitochondrial genes were downregulated (Figure 6B, Figure S7D). Transmission electron microscopy demonstrated ultrastructural abnormalities in RBX2-deficient cardiomyocytes, including focal myofibrillar lysis, enlarged and swollen mitochondria with disrupted cristae associated with lysosomes, massive mitophagic vesicles, and dilated endoplasmic reticulum (ER) (Figure 6C). Immunostaining of myocardium sections revealed a substantial increase of LAMP1 (a lysosome marker)-positive mitochondria in iCKO hearts compared with MCM hearts (Figure 6D), indicative of accumulation of damaged mitochondria. Despite the diminished mitochondrial ubiquitination, the increased mitophagic vesicles in RBX2-deficient hearts suggest the possible contribution of additional Ub ligases and receptor-mediated mitophagy to mitochondrial clearance (Figure S7E). Seahorse analysis further confirmed compromised mitochondrial bioenergetics in RBX2-deficient hearts (Figure 6E). Consistent with the transcriptomics analysis, immunoblotting showed that loss of RBX2 resulted in substantial downregulation of mitochondrial proteins (ATP5A, UQCRC2, SDHB, NDUFB8) in the heart (Figure 6F). Moreover, Parkin and PINK1 were upregulated and downregulated in RBX2-deficient hearts, respectively (Figure 6F). Collectively, these in vivo data, coupled with our in vitro findings, are consistent with the critical role for RBX2 in maintaining mitochondrial homeostasis and function in the heart.

Figure 6. Alterations in metabolic pathways and mitochondrial homeostasis in RBX2-deficient hearts.

A-B, Bulk RNA sequencing of mouse hearts at 12 days after tamoxifen (50 mg/kg/day for 5 days) treatment. A, KEGG pathways enriched in downregulated (top) and upregulated (bottom) genes in iCKO hearts. B, Chord plot showing downregulated genes involved in the indicated metabolic processes. C, Transmission electron microscopic images showing myofibril lysis (*), degenerating mitochondria (yellow arrowheads) surrounding by lysosomes (arrows), and abundant mitophagic vesicles (red arrowheads) in mutant cardiomyocytes. The number of mitophagic vesicles was quantified. Over 50 cardiomyocytes per group were quantified. D, Representative confocal images of myocardium sections immunostained with LAMP1 (green), TOMM20 (red) and DAPI (blue). Quantification of LAMP1+ puncta in mitochondria is shown. 5–6 views per heart, 2 hearts per group were quantified. E, Seahorse analysis of mitochondria isolated from mice at 8 days after tamoxifen injections. n=4 technical replicates/heart, 3 hearts/group were analyzed. Basal and maximal oxygen consumption rates are shown. F, Western blot of mitochondrial proteins in mouse hearts. *, cleaved form. Mann-Whitney test was used in C and D, and nested t test in E.

PARKIN is dispensable in RBX2-mediated mitophagy

Given the central role of PARKIN in mitochondrial ubiquitination, we asked whether RBX2 acts through PARKIN to mediate mitophagy. Silencing of RBX2 did not alter Parkin transcriptional levels (Figure S8A). PARKIN protein expression is very low in neonatal cardiomyocytes. In cardiomyocytes infected with Ad-PARKIN, RBX2 depletion insignificantly reduced PARKIN protein levels (Figure 7A). Moreover, deletion of RBX2 did not alter endogenous PARKIN expression at baseline but resulted in accumulation of PARKIN protein following CCCP treatment in SH-SY5Y cells (Figure S8B). Furthermore, we detected increased PARKIN protein levels in RBX2-deficient mouse hearts (Figure 6G). Thus, loss of RBX2 does not lead to reduced PARKIN expression. In cardiomyocytes isolated from neonatal wild-type and PARKIN null mice ⁴⁸, deletion of Rbx2 exhibited similar inhibitory effects on CCCP-induced pS65-Ub in wild-type and PARKIN null cardiomyocytes and cardiac fibroblasts (Figure 7B and S8C). Similarly, depletion of RBX2 exerted a greater repressive effect on pS65-Ub than depletion of PARKIN via siRNA and remained effective in PARKIN-depleted cardiomyocytes (Figure 7C and S8D). Moreover, PARKIN overexpression dose-dependently promoted pS65-Ub following CCCP treatment, which was attenuated by siRNA-mediated depletion of RBX2 (Figure 7D), suggesting that RBX2 is required for PARKIN-mediated mitochondrial protein ubiquitination. We compared the RBX2-regulated 137 mitochondrial proteins with 47 previously identified PARKIN mitochondrial substrates in cancer cell lines ⁴⁹. We detected only 9 overlapping proteins (Figure S8E), suggesting that RBX2 and PARKIN have distinct mitochondrial substrates. These data suggest that RBX2 does not require PARKIN to mediate mitochondrial ubiquitination.

Figure 7. The role of Parkin in RBX2-regulated mitochondrial ubiquitination and cardiac homeostasis.

A, Western blot of Parkin in NRVCs. Cells were infected with Ad-Parkin and transfected with indicated siRNAs. B, Representative Western blots of three independent repeats. Neonatal mouse ventricular CMs (NMVCs) were isolated from WT or Parkin^−/− mouse hearts, transfected with siRNA, and treated with CCCP (10 μM) for 12 hours. C, Western blots of cell lysates from NRVCs transfected with siRNAs and treated with CCCP (10 μM). D, Western blots of cell lysates from NRVCs infected with Ad-Parkin, transfected with siRNAs, and treated with CCCP (10 μM). E, Schematics of generation of RBX2 and Parkin double knockout (RBX2^CKO/Parkin^−/−) mice. F, Ejection fraction and fractional shortening at 5 (Parkin^−/−: n=5, RBX2^Het/Parkin^+/−: n=8, RBX2^CKO: n=12. RBX2^CKO/Parkin^−/−, n=8 mice) and 8 (Parkin^−/−: n=7, RBX2^Het/Parkin^+/−: n=7, RBX2^CKO: n=11. RBX2^CKO/Parkin^−/−, n=8 mice) months of age. G, Survival curves of indicated mice. Mann-Whitney test was used in A, one-way ANOVA followed by post hoc Tukey test in F, and Log-rank (Mantel-Cox) test in G.

To examine potential interactions between RBX2 and PARKIN in the heart, we generated PARKIN and RBX2 double knockout (DKO) mice (Figure 7E). Consistent with previous reports^12–14, PARKIN deficiency did not impair cardiac contractility in 8-month-old mice. Importantly, deletion of Parkin in CKO mice did not alter the onset or progression of dilated cardiomyopathy as demonstrated by temporal echocardiography, nor did it affect the survival rate of the CKO mice (Figure 7F, 7G, and S8F, Table S6). Together, these data suggest that RBX2 regulates cardiac homeostasis in PARKIN-independent mechanism.

RBX2 stabilizes PINK1

Phosphorylation of Ub is coordinated by PINK1 and phosphatases such as PPEF2, PGAM5 and PTEN-L 9–11. PINK1 is readily degraded and was barely detected in cultured cardiomyocytes infected with adenovirus expressing PINK1 under basal condition (Figure 8A). Depletion of RBX2 significantly inhibited CCCP-induced PINK1 stabilization without affecting the expression of PPEF2, PGAM5 and PTEN (Figure 8A). Moreover, deletion of RBX2 in Hela cells via CRISPR/Cas9 also reduced CCCP-induced expression of endogenous PINK1 (Figure 8B). RBX2 deficiency did not alter the levels of PINK1 transcripts, and the reduction of PINK1 in RBX2-deficient cardiomyocytes was attenuated by the proteasome inhibitor bortezomib (Figure 8C and 8D). To determine whether RBX2 regulates PINK1 degradation, cardiomyocytes were treated with CCCP for 3 hours to promote PINK1 accumulation, and following removal of CCCP, PINK1 degradation was assessed by cycloheximide-based pulse chase assay. Under these conditions, loss of RBX2 shortened the half-life of PINK1 (Figure 8E). Moreover, RBX2 depletion increased the ubiquitinated forms of PINK1, which was blunted by proteasome inhibition and abolished CCCP-induced PARKIN phosphorylation (Figure 8F and 8G), indicating enhanced PINK1 ubiquitination and diminished PINK1 activity. Consistently, PINK1 expression was reduced in RBX2iCKO hearts (Figure 6G). Together, these data suggest that RBX2 is required for PINK1 stabilization upon mitochondrial depolarization, which in turn enhances phosphorylation of Ub on mitochondria and thus mitochondrial ubiquitination.

Figure 8. RBX2 stabilizes PINK1.

A, Western blots of indicated proteins in NRVCs. Cells were infected with Ad-PINK1, transfected with indicated siRNAs and treated with or without CCCP (10 μM) for 12 hours. B, Western blots. RBX2 was deleted in Hela cells via CRISPR/Cas9 using a single guided RNA against RBX2 (gRBX2). WT and RBX2KO cells with treated with CCCP (10 μM) for indicated times before harvest. C, Analysis of Pink1 transcript levels in NRVCs transfected with indicated siRNAs by qPCR. D, Western blots. NRVCs were infected with Ad-PINK1, transfected with siRNAs, and treated with Bortezomib (BZM, 100 nM) for 6 hours. E, Cycloheximide-based pulse chase assay. NRVCs were transfected with siRNAs, treated with CCCP (10 μM) for 3 hours, followed by removal of CCCP, and then chased for the indicated time in the presence of cycloheximide (CHX, 100 nM). F, Immunoprecipitation of PINK1, followed by Western blots. NRVCs were transfected with indicated siRNAs and treated with or without Bortezomib (BZM, 100 nM) for 6 hours. G, Western blots of cell lysates from NRVCs transfected with indicated siRNAs and treated with CCCP (10 μM) for 3 hours. H, A proposed model showing the role of RBX2-CRL5 in regulation of physiological mitophagy and cardiac homeostasis. Mann-Whitney test was used in C.

DISCUSSION

In summary, our work identifies RBX2-CRL5 ubiquitin ligase as a novel regulator of mitophagy and cardiac homeostasis. Our evidence supports a model (Figure 8H) in which RBX2 translocates to dysfunctional mitochondria of cardiomyocytes, where it mediates the ubiquitination of mitochondrial outer membrane proteins and the clearance of damaged mitochondria in a PARKIN-independent manner. Meanwhile, RBX2 also enhances the stability of PINK1 to promote phosphorylation of Ub on depolarized mitochondria, thereby amplifying mitochondrial ubiquitination and mitophagy. Consequently, loss of RBX2 in mouse hearts impairs physiological mitochondrial turnover and provokes severe mitochondrial stress, leading to rapid development of dilated cardiomyopathy and heart failure. Our findings provide a new mechanism regulating mitochondrial quality control through targeting both mitochondrial ubiquitination and phosphorylation of ubiquitin on mitochondria. These findings may also explain why PARKIN is dispensable for physiological mitochondrial turnover in the heart.

RBX2-CRL5 is a mitochondrial Ub ligase

The role of RBX2-CRL5 in mitochondria has not been explored previously. Our studies identify RBX2-CRL5 as a mitochondrial Ub ligase. This is supported by the translocation of RBX2 and CUL5 to the mitochondrial outer mitochondrial membrane following mitochondrial depolarization, the necessity of RBX2 in mitochondrial ubiquitination and turnover, the global impact of RBX2 deficiency in mitochondrial proteome, and a crucial role of RBX2 in maintenance of mitochondrial integrity and function (Figure 1–2, Figure S1–2). Retention of neddylated CUL5, an active form of CRL5 (Figure 1C) further supports its role in mitochondrial. Interestingly, CCCP treatment induced pronounced accumulation RBX2 and CUL5 in mitochondria CCCP-treated in APEX2-MOM pulldown experiments but not in crude subcellular mitochondrial fractions (Figure 1B–1C). The discrepancy may be due to the weak, transient interactions of RBX2-CRL5 with mitochondria that are more effectively captured by biotinylation. RBX2 and CUL5 do not contain a mitochondrial targeting sequence. Such non-covalent binding of CRL5 to mitochondria may offer flexibility for CRL5 to recognize and ubiquitinate different MOM proteins in damaged mitochondria by pairing with different substrate-recognizing receptors. RBX2 was identified to be a redox sensitive protein⁵⁰. Whether increased mitochondrial ROS induces translocation of RBX2 to the mitochondria is unknown and worthy of future investigation. In line with the involvement of CRLs in mitophagy, two CUL3-associating adaptors, Keap1 and Klhl10, and the core component of CRL1–4, RBX1, were reported to associate with mitochondria^51,52. Therefore, it is possible that other CRLs may also regulate mitophagy in cardiomyocytes.

Essential role of RBX2 in regulation of physiological mitochondrial turnover and cardiac homeostasis

Mounting evidence demonstrates an essential role of mitophagy in the heart under the conditions of stress ^53,54. In contrast, much less is known about how physiological mitophagy (mitochondrial clearance under steady-state conditions) is initiated and regulated in the heart. We noted that in normal hearts, only a small fraction of mitochondria undergo mitophagy at any given time (Figure 5A, 5G, 6D, and S6), indicating a relatively low mitophagy activity in homeostatic hearts, possibly as a safeguard to maintain overall mitochondrial homeostasis. However, accumulation of damaged mitochondria over time due to defective mitophagy can have catastrophic effects on mitochondrial and cardiac function, as seen in VPS34-, ATG5-, and PINK1-deficient hearts ^55–57. In agreement with its pivotal role in mitophagy, loss of RBX2 in the postnatal and adult heart causes cardiac dysfunction and heart failure (Figure 3, S3 and S4). The severe cardiac phenotype is attributable to mitochondria dysfunction arising from defective mitochondrial turnover (Figure 5–6). Notably, RBX2 deficiency results in a rapid development of cardiomyopathy with much more pronounced cardiac dysfunction in adult hearts compared with developing hearts. The difference of disease progression rates in the two models could be due to a higher demand for energy production and thus mitochondrial homeostasis in adult hearts, the plasticity in adapting to mitochondrial dysfunction in young hearts, and perhaps tamoxifen-induced cardiotoxicity. Indeed, as seen in the cases of two ubiquitin ligases involved in mitophagy, TRAF2 and MITOL, and the key regulator ATG5, loss of these mitophagy regulators in adult hearts has a more profound impact on cardiac function, and typically leads to higher mortality, in adult mice than in developing mice^{24,56,58 59}. Notably, RBX1, the other RING-box family protein, was upregulated by 3-fold in CKO hearts but not in iCKO hearts (Figure 3B and S4B), suggesting the possibility of compensation by RBX1 in the CKO heart. Collectively, our data identify RBX2-CRL5 as an important regulator of physiological mitophagy and cardiac integrity, although we cannot rule out the involvement of other RBX2-regulated substrates and pathways ²⁸.

Distinct actions of RBX2 and PARKIN in mitochondrial quality control

Despite its well-established role in mitophagy, PARKIN appears non-essential for mitochondrial turnover in normal hearts^11–14. Conceivably, multiple ubiquitin ligases may cooperatively survey the fitness of highly abundant mitochondrial in the heart. Among the reported Ub ligases involved in mitophagy, TRAF2 and MITOL have been shown to facilitate PARKIN recruitment to mitochondria and regulate mitophagy in the heart^25,60. We presented evidence demonstrating that RBX2 does not require PARKIN to mediate mitochondrial ubiquitination and possibly even acts downstream of PARKIN (Figure 7 and S8). Moreover, loss of PARKIN does not impact cardiomyopathy induced by RBX2 deficiency, supporting a key role of RBX2 in physiological mitophagy in parallel to PARKIN. Several reports have also implicated PARKIN-independent roles for other CRLs in mitophagy. ARIH1 was shown to mediate mitophagy in PARKIN-deficient cancer cells ¹⁹. CUL9 ubiquitin ligase regulates mitochondrial quality control⁶¹, and loss of CUL9 and PARKIN together does not exaggerate mitochondria damage induced by CUL9 deficiency in neuronal cells ⁶². Given the well-documented role of PARKIN under stress conditions, it is possible that RBX2 and PARKIN may recognize different substrate pools (Figure S8E) to regulate mitophagy in cardiomyocytes under physiological and pathological conditions, respectively.

RBX2 is a new regulator of PINK1 expression

An interesting finding of this study is that RBX2 is required for PINK1 stabilization following mitochondrial depolarization (Figure 8). PINK1 is imported into mitochondria and cleaved by the mitochondrial protease PARL⁶³. Cleaved PINK1 is then degraded by the proteasome through the ubiquitin ligases UBR1/UBR2/UBR4⁶⁴. It will be interesting in the future to explore whether RBX2 acts as a ubiquitin ligase to control the degradation of PARL or UBR1/2/4, thereby increasing PINK1 stability in damaged mitochondria. Loss of PINK1 promotes mitochondrial dysfunction in the heart eventually leading to cardiac dysfunction and heart failure⁵⁷. The dual role of RBX2 in regulation of mitochondrial ubiquitination and PINK1 expression explains the severe mitochondrial and cardiac phenotypes observed in adult RBX2-deficient hearts.

NOVELTY AND SIGNIFICANCE:

What Is Known?

• Defects in mitochondrial quality control are associated with heart failure and cardiac disease.

• PARKIN mediates mitochondrial ubiquitination but is not essential for mitochondrial turnover in homeostatic hearts.

• RBX2 is the catalytic subunit of Cullin-RING ubiquitin ligase 5 known to regulate tumorigenesis and viral infection.

What New Information Does This Article Contribute?

• RBX2 localizes to the mitochondria.

• RBX2 regulates the ubiquitination of mitochondrial outer membrane proteins and mitophagy.

• RBX2 is indispensable for the maintenance of mitochondrial and cardiac function in adult hearts.

• RBX2 requires PINK1, but not PARKIN, to mediate mitophagy.

Which ubiquitin ligases mediate mitochondrial turnover in the heart remains obscure. The current study identified RBX2 as a mitochondrial ubiquitin ligase that has a dual role in regulation of mitochondrial ubiquitination and PINK1 expression. Using mice and cardiomyocytes deficient of RBX2, we demonstrated that RBX2 is crucial for mitochondrial quality control and the normal function of adult heart, potentially providing a new target to prevent and treat cardiac diseases associated with mitochondrial dysfunction.

Non-standard abbreviations and acronyms:

RBX2

RING-Box Protein 2

SAG

Sensitive to Apoptosis Gene

Ub

Ubiquitin

pS65-Ub

phosphorylated Ub at serine 65

MAVS

mitochondrial antiviral-signaling protein

AAV

adeno-associated virus

AV

adenovirus

siRNA

Small interfering RNA

GFP

green fluorescent protein

CUL

cullin

RING

Really Interesting New Gene

CRLs

cullin-RING ligases

CSN

COP9 signalosome

APEX2

ascorbate peroxidase 2

mito

mitochondrial; cyto, cytosolic

MOM

mitochondrial outer membrane

CCCP

Carbonyl Cyanide Chlorophenylhydrazone

OMP25

Outer membrane protein 25

PK

proteinase K

HA

hemagglutinin

TMRM

Tetramethylrhodamine methyl ester perchlorate

αMHC

α-myosin heavy chain

CKO

cardiomyocyte-specific knockout

TAM

tamoxifen

TMT

tandem mass tag

KD

knockdown

CTL

control

MCM

MerCreMer

iCKO

inducible cardiomyocyte-specific knockout

BFA

bafilomycin A1

PCA

principle component analysis

MS

Mass spectrometry

DEPs

differentially expressed proteins

FC

fold change

FDR

False Discovery Rate

KEGG

Kyoto encyclopedia of genes and genomes

ER

endoplasmic reticulum

DKO

double knockout

CM

cardiomyocyte

cTnT

cardiac troponin T

NRVCs

neonatal rat ventricular cardiomyocytes

NRVMs

neonatal mouse ventricular cardiomyocytes

NMVFs

neonatal mouse ventricular fibroblasts

HF

heart failure; KO, knockout

MF

Molecular Functions

CC

Cellular Components

BP

Biological Process

TUNEL

Terminal deoxynucleotidyl transferase dUTP nick end labeling

SCF

Skp1-Cullin 1-F-box