Cardiac-specific ablation of NDUFAB1 causes progressive dilated cardiomyopathy leading to heart failure

NDUFAB1 was widely expressed in different tissues and was particularly enriched in the heart (Supplementary information, Fig. S1a). Whole-body knockout of Ndufab1 was embryonic lethal (Supplementary information, Fig. S1b, c), showing that NDUFAB1 is essential for developmental viability. To explore the potential cardiac functions of NDUFAB1, we generated a cardiac-specific knockout mouse model (cKO) wherein exon 3 of Ndufab1 was flanked by loxP sites (Ndufab1, Supplementary information, Fig. S1b). The Ndufab1 mice were cross-bred with Mlc2v-Cre mice to allow cardiomyocyte-specific deletion of Ndufab1, as previously described. As shown in Fig. 1a and Supplementary information, Fig. S1d, cardiac expression of Ndufab1 was decreased by ~90% at the protein level in cKO cardiomyocytes compared to wild-type (WT) cells. The cKO mice were smaller, had a reduced body weight, and even started to lose weight precipitously after 14 weeks of age (Fig. 1b). Meanwhile, the lifespan of cKO mice was markedly shortened, with sudden death beginning at ~12 weeks of age; the maximal lifespan was <19 weeks (Fig. 1c).

We next examined the effect of NDUFAB1 ablation on cardiac morphology and functions at different ages. The cKO hearts progressively enlarged with age and developed dilated cardiomyopathy (Fig. 1d; Supplementary information, Fig. S1e). Compared to WT littermates, the heart weight/body weight ratio in cKO mice was increased by 12.8% at 6 weeks, 26.9% at 8 weeks, 51.5% at 12 weeks, and 241.7% at 16 weeks; and the heart weight/tibia length ratio was not significantly altered at 6 weeks and increased by 45.5% at 8 weeks, 80.1% at 12 weeks, and 110.9% at 16 weeks (Fig. 1e). The slight increase of heart weight/body weight ratio without change of heart weight/tibia ratio at 6 weeks is due to mild reduction of body weight (Fig. 1b). At the cellular level, the longitudinal-section area of cardiomyocytes was nearly constant in WT but increased progressively in cKO mice, with no significant change at 6 weeks and reaching a 78% increase at 16 weeks (Fig. 1f), indicative of cardiomyocyte hypertrophy. Masson staining showed that NDUFAB1 ablation induced extensive cardiac fibrosis early at 8 weeks (Fig. 1g; Supplementary information, Fig. S1f). Consistent with the changes of cardiac morphology, echocardiography revealed that the ejection fraction (EF) and fractional shortening (FS) of cKO hearts were unaltered at 6 weeks, but halved at 8 weeks; and the EF was diminished by 79% and FS by 84% at the age of 14 weeks (Fig. 1h), while the heart rate remained unchanged (Supplementary information, Fig. S2a). The thickness of both the left ventricular posterior wall and the inter-ventricular septum was decreased whereas the left ventricular diameter and volume were increased with age in cKO hearts (Supplementary information, Fig. S2b, c). These findings indicate that cardiac-specific ablation of NDUFAB1 leads to dilated cardiomyopathy accompanied by cardiomyocyte hypertrophy, interstitial fibrosis, and systolic and diastolic dysfunction, eventually resulting in heart failure and sudden death, revealing an essential role of NDUFAB1 in cardiac function.

To investigate the mechanism underlying the cardiomyopathy and heart failure induced by NDUFAB1 ablation, we next assessed changes of cardiac mitochondrial functions at different ages of cKO mice. Electron microscopic analysis revealed that cardiac mitochondria were in disarray along the sarcomeres with some exhibiting irregularities in the cristae formation in 16-week-old cKO mice, whereas mitochondrial morphology was apparently normal at 6 weeks, when the cardiac function hadn't been impaired yet (Fig. 2a). The mitochondrial DNA content and cross-sectional area of individual mitochondria didn't display detectable changes whereas the total mitochondrial volume fraction and mitochondrial number were slightly increased in 16-week-old cKO mice (Supplementary information, Fig. S3). Measuring mitochondrial membrane potential (ΔΨ) in isolated cardiomyocytes with the potential-sensitive fluorescent probe tetramethyl rhodamine methyl ester (TMRM) showed that NDUFAB1 ablation significantly decreased ΔΨ even at 6 weeks (Fig. 2b) when the heart morphology and function were normal (Fig. 1), indicating mitochondrial dysfunction is an early event in cKO mice preceding the onset of cardiomyopathy. Measurements with the fluorescent probe mitoSOX showed that mitochondrial ROS level was significantly elevated in cKO cardiomyocytes compared to WT at 10 weeks, and increased by 72% at 14-16 weeks, reaching a level comparable to that induced by 5 μg/mL antimycin A (Fig. 2c). The cellular ATP level, as measured with the luciferin luminescence assay, showed a trend of decrease in cKO mice from 6 to 8 weeks old and was significantly lowered in 14-16-week or older cKO mice (Fig. 2d). This result suggests that, while the ATP level is initially tightly safeguarded in the heart, NDUFAB1 ablation curtails the energy reserve capacity, exhaustion of which impairs ATP homeostasis. Taken together, these results indicate that NDUFAB1 is essential for mitochondrial bioenergetics and ROS metabolism.

Next, we analyzed the transcriptome of left ventricles from cKO and WT mice by RNA sequencing (RNA-seq). Among 11494 genes analyzed, 430 were downregulated and 1081 were upregulated (Supplementary information, Tables S1 and S3). Interestingly, the top 5 Gene Ontology terms of the downregulated genes were mainly associated with mitochondrial metabolism, including metabolic process, fatty acid metabolism, lipid homeostasis, oxidation-reduction process (Fig. 2e; Supplementary information, Table S2), in good agreement with the bioenergetic defects in the cKO heart (Fig. 2b-d). The downregulated genes were involved in the regulation of muscle contraction, consistent with the phenotype of cardiomyopathy (Fig. 2e). Meanwhile, the top enriched Gene Ontology terms of upregulated genes were mostly involved in immune system process, inflammatory response, cell adhesion, and chemotaxis (Fig. 2e; Supplementary information, Table S4). We deduce that these upregulated processes might reflect compensatory as well as maladaptive responses to cardiac damage due to mitochondrial dysfunctions.

To further understand the mechanism underlying these mitochondrial energy and ROS metabolism dysfunctions induced by NDUFAB1 ablation, we tested whether NDUFAB1 deficiency affected lipoic acid synthesis in FAS II pathway. We measured pyruvate dehydrogenase activity as lipoic acid is its obligate cofactor, and found similar activities in WT and cKO hearts (Supplementary information, Fig. S4a). Further, the protein lipoylation status was not significantly altered in cKO hearts (Supplementary information, Fig. S4b). These results revealed NDUFAB1 ablation did not affect FAS II pathway in the heart. Given the pivotal role of the mitochondrial ETC in energy metabolism and ROS production, we next investigated whether and how NDUFAB1 ablation impacted on the assembly and activity of individual ETC complexes in the heart. First, we measured the respiratory activity of isolated mitochondria from 6-week-old mice when the cardiac morphology and function were normal and 10-week-old mice when cardiomyopathy was already developed. In the presence of substrates of either complex I, complex II or complex III, the oxygen consumption rates (OCRs) of cKO mitochondria were significantly decreased at both ages (Fig. 3a; Supplementary information, Fig. S5a-c), whereas the OCR supported by complex IV substrate remained unchanged (Fig. 3a; Supplementary information, Fig. S5d). These data suggest that, rather than being merely a complex I subunit, NDUFAB1 alters the ETC function at multiple sites.

Further, we assessed the assembly of ETC complexes using blue native polyacrylamide gel electrophoresis (BN-PAGE). Consistent with previous reports, complexes I, III, and IV were assembled into SCs, as evidenced by immunoblot analysis following BN-PAGE using antibodies against NDUFB8 (complex I), SDHA (complex II), UQCRFS1 or UQCRC1 (complex III), COXIV (complex IV), and ATPB (complex V), while complexes II and V mainly remained as individual entities in both WT and cKO groups (Fig. 3c). Moreover, NDUFAB1 ablation decreased the SCs comprised of complexes I, III, and IV in the hearts at both ages (Fig. 3c; Supplementary information, Fig. S6b). We also assessed the abundance of individual complexes and found that complexes I, II, and III were significantly diminished in cKO mitochondria at both ages (Fig. 3c; Supplementary information, Fig. S6b), consistent with the aforementioned functional data (Fig. 3a). The level of individual complex IV was increased in cKO mitochondria due to decreased SC assembly (Fig. 3c; Supplementary information, Fig. S6b). Further analysis showed the percentages of complex III- and complex IV-containing SCs were decreased in cKO mitochondria at both ages (Supplementary information, Fig. S6c). The percentage of complex I-containing SCs was not significantly changed because both the free and complexed complex I were proportionally diminished in cKO mitochondria (Supplementary information, Fig. S6b, c). In-gel activity analysis in the presence of complex I substrate revealed that the activity of either complex I or SCs was markedly lowered in the absence of NDUFAB1 (Fig. 3b; Supplementary information, Fig. S6a). Moreover, we measured the abundance of individual complexes and SCs in the hearts of newborn mice (born within 24 h) and found that complexes I, II, III and SCs were diminished while complex IV was not significantly changed in cKO mitochondria (Supplementary information, Fig. S7b, d). Likewise, in-gel activity analysis in the presence of complex I substrate showed decreased activities of complex I and SCs in cKO mitochondria of newborn mice (Supplementary information, Fig. S7a, c). Meanwhile, knocking down Ndufab1 in cultured neonatal rat cardiomyocytes also decreased the abundance of complex I and SCs (Supplementary information, Fig. S8). All these results implicate that NDUFAB1 is essential for the assembly of complexes I, II, III, and SCs in the heart. Furthermore, the early onset of impaired mitochondrial energetics induced by disrupted assembly of ETC complexes and SCs indicates a causative role of this NDUFAB1 ablation-induced ETC defect in heart failure.

By measuring the abundance of individual subunits of complexes I-III, we found that the FeS-containing subunits, SDHB (succinate dehydrogenase subunit B) of complex II and UQCRFS1 (ubiquinol-cytochrome c reductase, Rieske iron-sulfur polypeptide 1) of complex III, were both significantly decreased, whereas the other subunits examined were unaltered in the absence of NDUFAB1 (Fig. 3e, f; Supplementary information, Fig. S6e). This indicates that NDUFAB1 ablation selectively disrupts FeS-containing subunits of complexes II and III. Concomitantly, a complex III assembly intermediate lacking UQCRFS1, which is incorporated into complex III at the last step, accumulated in cKO mitochondria (Fig. 3c; Supplementary information, Fig. S7b). These findings suggest that NDUFAB1 coordinates the assembly of ETC complexes and SCs through regulating FeS cluster biogenesis as reported previously in yeast cells lacking complex I. In this regard, we found that ISCU and NFS1, two important components of the FeS biogenesis complex, were down-regulated in cKO cardiomyocytes (Supplementary information, Fig. S9). For complex I, however, all subunits examined, regardless of whether they contain (NDUFS1, NDUFS4, and NDUFS7) or don't contain (NDUFS6 and NDUFB8) FeS cluster(s), were dramatically decreased in cKO mitochondria (Fig. 3d; Supplementary information, Fig. S6d), indicating the involvement of additional mechanism by which NDUFAB1 regulates its assembly (see Discussion). Since the mRNA levels of the above subunits of complexes I-III were not significantly altered (Supplementary information, Fig. S10), the NDUFAB1-mediated regulation appears to occur at the protein rather than the mRNA level, presumably by affecting protein stability. Therefore, NDUFAB1 coordinates the assembly of complexes I-III and SCs through its dual roles both as a regulator of FeS biogenesis and as a complex I subunit.

Different from the effect of NDUFAB1 ablation on respiratory complexes, knockout of Ndufs4, another accessory subunit of complex I, induced complex I deficiency but showed little effect on other complexes. Meanwhile, NDUFS4 ablation led to decreased NAD/NADH ratio and increased protein acetylation, however, we found that neither the NAD/NADH ratio nor protein acetylation levels were significantly altered in Ndufab1 cKO cardiomyocytes (Supplementary information, Fig. S11). These results suggest a specific role of NDUFAB1 in the assembly of respiratory complexes and SCs. In addition, mitochondrial cardiolipin content was not altered in Ndufab1 cKO mitochondria (Supplementary information, Fig. S12). All these results substantiate that NDUFAB1 ablation-induced cardiomyopathy is mainly caused by impaired assembly of ETC complexes and SCs.

The results thus far reveal that NDUFAB1 plays an essential role in coordinating the dynamic assembly of individual ETC complexes and SCs and thereby regulates mitochondrial bioenergetics and ROS metabolism. We reckoned that augmenting the abundance of NDUFAB1 might provide an effective strategy to build a more robust and efficient powerhouse to benefit the heart, providing that endogenous NDUFAB1 is present at a sub-saturating level. To test this possibility, we generated an Ndufab1 transgenic mouse model (TG) using a β-actin promoter (Supplementary information, Fig. S13a) in which NDUFAB1 was overexpressed ~8-fold in the heart (Fig. 4a; Supplementary information, Fig. S13b). Compared with WT littermates, TG mice grew normally and had similar lifespans under our experimental conditions (Supplementary information, Fig. S13c, d). Cardiac morphology and functional performance determined by echocardiography (Supplementary information, Fig. S13e-g), mitochondrial morphology revealed by electron microscopy (Supplementary information, Fig. S13h-j), and mitochondrial DNA content (Supplementary information, Fig. S13l) were all comparable in WT and TG hearts. Nonetheless, the mitochondrial ROS level was markedly decreased in TG cardiomyocytes (Fig. 4b), suggesting an alleviation of basal oxidative stress. The ΔΨ was significantly increased (Fig. 4c), and the maximal OCR with either glucose/pyruvate or palmitate as substrate measured by Seahorse assay was higher in TG than in WT cardiomyocytes (Fig. 4d-g). These results imply an enhanced reserve capacity for ATP production, although the homeostatic ATP level was unaltered (Supplementary information, Fig. S13k). The respiratory control ratio was significantly improved in the presence of substrates of complex I, II, or III in TG cardiac mitochondria (Fig. 4h), indicating greater efficiency of mitochondrial energy metabolism. Moreover, the electron flow assay showed enhanced activities of complexes I, II and III in TG mitochondria (Supplementary information, Fig. S14), indicating that NDUFAB1 overexpression augments the ETC activity.

At the molecular level, in-gel activity analysis showed that activity of both complex I and SCs were significantly enhanced in TG hearts (Fig. 4i, j), and BN-PAGE analysis showed that NDUFAB1 overexpression significantly increased the contents of complex I and SCs (Fig. 4k, l; Supplementary information, Fig. S15a), whereas the expression of all subunits examined, including those for complexes I-III, was unchanged (Supplementary information, Fig. S15b, c). Importantly, overexpression of NDUFAB1 in cultured neonatal cardiomyocytes enhanced the abundance of complex I and SCs without significant effect on the other complexes (Supplementary information, Fig. S16). In addition, NDUFAB1 overexpression did not impact on FAS II pathway in the heart, because neither pyruvate dehydrogenase activity nor protein lipoylation status was significantly altered in TG mitochondria (Supplementary information, Fig. S17). The NAD/NADH ratio and protein acetylation levels were also comparable in the WT and TG hearts (Supplementary information, Fig. S18). Taken together, NDUFAB1 overexpression enhances the abundance of complex I and SCs, conferring on mitochondria greater capacity and efficiency of energy metabolism and less ROS emission.

The cardiac performance as well as the lifespan of TG mice was similar to that of WT littermates under unchallenged conditions (Supplementary information, Fig. S13). However, TG mice might be more resistant to cardiac injury because of augmented capacity and efficiency of mitochondrial bioenergetics along with attenuated ROS production. Next, we subjected the WT and TG hearts to IR injury to unmask this potential cardio-protective role of NDUFAB1. First, we applied an ex vivo IR protocol based on Langendorff perfusion with 30-min ischemia followed by 30-min reperfusion (Supplementary information, Fig. S19a). Since excessive ROS production during IR is a major contributor to IR injury, we tracked the ROS production with indicator 5-(and-6)-chloromethyl-2',7'-dichlorodihydrofluorescein diacetate acetyl ester (DCF) (Fig. 5a). In WT hearts, there was a trend of increasing ROS during ischemia followed by a prominent ROS burst immediately after reperfusion (Fig. 5a, b). Remarkably, the ROS burst after IR was ameliorated by 29% in TG hearts (Fig. 5a, b). Furthermore, we measured HO production in the presence of succinate in isolated cardiac mitochondria using the Amplex Red assay and found that the rotenone-sensitive HO production was significantly attenuated in TG mitochondria (Fig. 5c).

Because loss of ΔΨ not only manifests a defective bioenergetics but also is an early event of cell death, we further assessed cardiac damage by monitoring loss of ΔΨ with TMRM and membrane integrity with evans blue dye (EBD). Our results showed that cells that both lost TMRM and were EBD-positive were significantly fewer in TG than WT hearts (Fig. 5d-f). Moreover, we adopted an in vivo IR experimental protocol in which a 30-min ischemia was followed by 24-h reperfusion (Supplementary information, Fig. S19b). In agreement with the ex vivo IR results, the area of infarction in TG was significantly smaller than that in WT hearts (Fig. 5g-i). Meanwhile, lactate dehydrogenase (LDH) release induced by IR injury was significantly attenuated in TG mice (Fig. 5j). Echocardiography revealed that TG mice exhibited significantly improved heart function at 48 h after IR injury compared to WT mice (Fig. 5k). Altogether, these findings demonstrate that increasing the abundance of mitochondrial NDUFAB1 attenuates ROS production and protects the heart against IR injury.