LONP1-mediated mitochondrial quality control safeguards metabolic shifts in heart development

During mammalian embryogenesis, the heart is the first organ to develop and function (Olson and Srivastava, 1996; Harvey, 2002; Bruneau, 2020). Initially, heart tissues form under hypoxic conditions before embryonic day (E) 11.5, and hypoxia inducing factor α (HIFα) drives an anaerobic glycolysis program to provide energy through transcriptional mechanisms for early heart development (Guimarães-Camboa et al., 2015; Menendez-Montes et al., 2016; Maroli and Braun, 2021). From E11.5 onward, a metabolic shift from anaerobic glycolysis to mitochondrial oxidative phosphorylation (oxidative metabolism) takes place in cardiomyocytes, which is essential for myocardial development (Lopaschuk and Jaswal, 2010; Zhao et al., 2019; Maroli and Braun, 2021). Recently, we identified the chromatin-remodeling SRCAP complex as a pivotal regulator of mitochondrial oxidative metabolism for this type of metabolic shift/transition (Xu et al., 2021). Through transcriptional regulation, the SRCAP complex controls mitochondrial maturation morphologically and functionally to meet the energy requirements of the rapidly growing myocardium.

In the mitochondria, several distinct quality control machineries monitor the proteome, which contains ∼1500 proteins to safeguard mitochondrial functions (Fischer et al., 2012; Rugarli and Langer, 2012; Song et al., 2021). AAA⁺ Lon protease (LONP1) is located in the mitochondrial matrix and functions to degrade misfolded proteins and prevent protein aggregation (Suzuki et al., 1994; Lu et al., 2013; Song et al., 2021). Under stress conditions, LONP1 can directly remove mitochondrial proteins, such as the components of the respiratory chain complexes and of the tricarboxylic acid (TCA) cycle, to sustain mitochondrial function (Song et al., 2021). There is also growing evidence to show that LONP1 acts as a chaperone to modulate mitochondrial DNA (mtDNA) stability (Chen et al., 2008; Matsushima et al., 2010; Kao et al., 2015). Thus, LONP1 plays a large role in governing mitochondrial protein folding and removing damaged proteins and, therefore, participates in mitochondrial quality control.

When cells are subjected to stresses, such as amino acid deprivation, viral infection, heme loss or endoplasmic reticulum (ER) stress, the integrated stress response (ISR) is elicited (Pakos-Zebrucka et al., 2016). Cellular stresses provoke a remarkable elevation in phosphorylated eukaryotic initiation factor 2α (P-eIF2α), further leading to an increase in translation of the master regulator of ISR, activating transcription factor 4 (ATF4) (Jousse et al., 2003; Lu et al., 2004; Dara et al., 2011). ATF4 stimulates a transcriptional program to regulate protein synthesis, the unfolded protein response (UPR), autophagy and metabolism (Vattem and Wek, 2004; Kilberg et al., 2009; B'Chir et al., 2013). A recent study indicated that mitochondrial morphological damage and/or dysfunction gives rise to mitochondrial stress and causes the accumulation of ATF4 (Quirós et al., 2017). Mitochondrial stress also promotes the decreased expression of mitochondrial ribosomal proteins (MRPs) to attenuate protein translation (Quirós et al., 2017).

Global deletion of Lonp1 in mice causes embryonic lethality at E8.5, and knockout embryos exhibit a decline in mtDNA copy number (Quiros et al., 2014). To address whether mitochondrial quality control plays a role in organogenesis and in the surveillance of mitochondrial metabolic shifts, we knocked out Lonp1 in embryonic cardiac tissues. Lonp1 deficiency caused severely defective heart development and embryonic lethality. Aggregates were evident in the mitochondria of Lonp1-deficient cardiomyocytes. Accordingly, AFT4 accumulation was shown to not only suppress the expression of Tfam- and mtDNA-encoded components of respiratory chain complexes, but also enhance the expression of glycolysis regulatory genes, including Glut1 (Slc2a1) resulting in a disrupted metabolic shift during cardiac development.

We examined the level of LONP1 protein in the developing heart tissue at different developmental stages, and found a slight, but significant, enhancement from E10.5 to E12.5 (Fig. S1A,B). During this period, the anaerobic glycolysis regulatory protein GLUT1 was significantly reduced and had almost disappeared by E12.5 (Fig. S1A). By contrast, the mitochondrial oxidative metabolic protein NDUFB8, a component of respiratory chain complex I (CI), was significantly increased (Fig. S1A). These results indicate a metabolic shift from anaerobic glycolysis to mitochondrial oxidative phosphorylation from E10.5 to E13.5 and suggest that LONP1 may be involved.

To determine whether LONP1 is involved in heart development, we first deleted Lonp1 in the cardiac progenitors of the second heart field, which contributes to the development of the cardiac outflow tract (OFT) and the right ventricle (RV). Lonp1-deficient mice (Mef2c-AHF-Cre; Lonp1^F/F) showed comparable heart morphology with the control at E9.5 (Fig. 1A-A″), and ISL1 staining for cardiac progenitors revealed a normal distribution of progenitor cells (Fig. 1B). Furthermore, we observed normal OFT morphogenesis and septation (Fig. S2). Thus, Lonp1 deficiency does not affect cardiac progenitor development.

However, Lonp1-deficient mice showed reduced RV size from E11.5 onward, and RV hypoplasia was more prominent from E12.5 to E18.5 (Fig. 1C,D). Interestingly, Lonp1-deficient mice showed increased LV size from E14.5 onward (Fig. 1E). Although Lonp1-deficient mice were viable at birth, all were lost by postnatal day (P) 0. The hearts of Lonp1-deficient mice showed enlargement of the two atria and a small RV at birth, indicating heart failure (Fig. 1F-F″).

TUNEL staining to detect cell death showed no obvious apoptosis signals in Mef2c-AHF-Cre; Lonp1^F/F hearts (Fig. 1G-G″). In contrast, knockout of Lonp1 led to a significant decrease in cardiac cell proliferation (Fig. 1H). Further study revealed that the cells with reduced proliferation were primarily cardiomyocytes (Fig. 1I).

Thus, overall, deletion of Lonp1 has little effect on cardiac progenitor development but impairs cardiomyocyte proliferation in the RV.

Lonp1 deletion in cardiac progenitors impeded cardiomyocyte proliferation in the RV. To confirm this phenotype, we deleted Lonp1 in embryonic cardiomyocytes using cTnT-Cre mice to generate cTnT-Cre; Lonp1^F/F mice (cardiomyocyte-specific Lonp1-deficient mice). Similar to Lonp1 deletion in cardiac progenitors, there was no significant change in OFT development or progenitor cell contribution in the hearts of cardiomyocyte-specific Lonp1-deficient mice (Fig. 2A,B). However, these mice started to demonstrate reduced ventricular volume from E11.5 and a thinned ventricular wall at E14.5 (Fig. 2C-E) and did not survive after E16.5 (Fig. 2F). The TUNEL assay did not reveal an increase in apoptotic cardiomyocytes (Fig. 2G). However, a cell proliferation analysis indicated significantly reduced cardiomyocyte proliferation during an early stage of cardiac development (Fig. 2H-H″). These results confirm the essential role of LONP1 in embryonic cardiomyocyte development.

LONP1 is primarily located in the mitochondrial matrix. To determine whether deletion of Lonp1 affects mitochondrial morphology, we isolated cardiomyocytes from the hearts of control and cTnT-Cre; Lonp1^F/F mice and cultured them in vitro. We then examined the mitochondria and found a dramatic morphological difference between control and Lonp1-deficient cardiomyocytes. Whereas the mitochondria in control cardiomyocytes showed a slender lamellar network, those in Lonp1-deficient cardiomyocytes contained aggregated lamellar structures (Fig. S3). Next, we performed transmission electron microscopy (TEM) analysis to study the mitochondria in the hearts of control and Lonp1-deficient mice. The results showed accumulation of highly electron-dense substances in the cardiomyocytes from cTnT-Cre; Lonp1^F/F mice (Fig. 3A-F). In addition, mitochondria in these cardiomyocytes were swollen and cristae could not be seen (Fig. 3D-F). Compared with the control group, the damage to mitochondria increased significantly in cardiomyocytes from cTnT-Cre; Lonp1^F/F mice [Fig. 3G; Ctrl: 0.01167±0.008731 versus conditional knockout (cKO): 0.1947±0.04107].

Severe mitochondrial abnormalities could trigger the increased production of reactive oxygen species (ROS). Significantly increased levels of ROS were observed in Lonp1-deficient cardiomyocytes (Fig. 3H-H″,I). High levels of ROS suppress cell cycle progression, and we found that the mRNA levels of Cdkn1a and Ddit3 were significantly increased in cKO hearts (Fig. 3J), suggesting inhibition of the cell cycle. In addition, we observed a significantly increased level of γH2A.X, a DNA damage marker in Lonp1-deficient cardiomyocytes (Fig. 3K,L).

Taken together, these results demonstrate that Lonp1 deletion causes mitochondrial abnormalities, increased ROS production and DNA damage.

Deletion of Lonp1 impaired heart development starting at E11.5 (Fig. 2C), a developmental stage at which a metabolic shift takes place from anaerobic glycolysis to mitochondrial oxidative phosphorylation. To determine whether Lonp1 deficiency affects this metabolic shift, we examined the metabolic pattern in the hearts of control and Lonp1-deficient mice. Western blot analysis showed that protein levels of the CI component NDUFB8 and the complex IV (CIV) component mitochondrial (mt)-COX2 were reduced significantly compared with those of the control as early as E11.5 (Fig. S4A,B). At E12.5, the protein levels of the CIV component COX4 and the complex V (CV) component ATP5a1 were substantially decreased compared with those of the control (Fig. 4A,B). Immunofluorescence (IF) staining also confirmed a significant reduction in NDUFB8 in Lonp1-deficient cardiomyocytes (Fig. S4C).

In contrast, we observed markedly enhanced protein levels of GLUT1 and GLUT4 in the hearts of Lonp1-deficient mice (Fig. 4C-E; Fig. S4D). In addition, the transcript levels of Glut1, Glut4 (Slc2a4) and Hk2 were also elevated in a HIF1-independent manner in the hearts of Lonp1-deficient mice compared with control mice (Fig. S4E-I). Phosphorylation of AMPK can represent the tissue energy balance; we found a substantial increase in phosphor-AMPK levels in the Lonp1-deficient hearts compared with control hearts, indicating an energy shortage (Fig. 4F,G).

Taken together, these results demonstrate that a relatively high level of glycolysis takes place in the Lonp1-deficient heart, suggesting a disrupted metabolic shift.

We performed RNA-sequencing (RNA-seq) using RNA isolated from control and Lonp1-deficient heart tissues. By comprehensive alignment analysis of the altered genes and Gene Ontology (GO) analysis, we found that Atf4 was highly expressed in Lonp1-deficient hearts (Fig. 5A). Quantitative real-time PCR analysis confirmed significantly elevated mRNA levels of Atf4 in Lonp1-deficient hearts compared with control hearts at both E10.5 and E12.5 (Fig. 5B). Western blot analysis also showed a marked increase in AFT4 levels in Lonp1-deficient hearts (Fig. 5C,D). IF staining clearly showed increased expression of ATF4 in both the nucleus and cytoplasm from E10.5 to E12.5 (Fig. 5E; Fig. S5A). Accordingly, there was significantly increased expression of the downstream target genes of ATF4 (Pck2, Phgdh, Shmt2, Pc, etc.) (Fig. S5B-D). In addition, there were elevated phosphorylation levels of eIF2α in Lonp1-deficient hearts (Fig. 5F,G).

Collectively, these results demonstrate a substantial increase in ATF4 levels and suggest that this is related to elevated phosphorylation of eIF2α.

Mitochondrial respiratory chain complex components are coordinately encoded by nuclear DNA and mtDNA. Gene expression analysis revealed no significant change in the mRNA levels of mitochondrial respiratory chain complex components encoded by nuclear DNA (Fig. 6A). However, the mRNA levels of respiratory chain complex components encoded by mtDNA (mt-Cox2 and mt-Atp6) were significantly decreased in Lonp1-deficient hearts compared with control hearts (Fig. 6A; Fig. S6A). A previous study detected a decreased mtDNA copy number in global Lonp1-deficient mice (Quiros et al., 2014). In the current study, there was a substantial reduction in the total amount of mtDNA in Lonp1-deficient hearts (Fig. 6B). As the most abundant mtDNA-binding protein, mitochondria-associated transcription factor A (TFAM) regulates mtDNA replication and transcription (Matsushima et al., 2010). Examination of the level of TFAM showed a profound decrease in Lonp1-deficient hearts (Fig. 6C-C″,D). Accordingly, the protein levels of mtDNA-encoded CI ND1 (CI-mt-ND1), CIV-mt-COX2 and CV-mt-ATP6 were all significantly decreased in Lonp1-deficient hearts compared with control hearts (Fig. 6D,E).

Furthermore, reductions in TFAM and mtDNA-encoded respiratory chain complex components in the Lonp1-deficient heart occurred at E11.5 rather than at E10.5 (Fig. S6B-E).

Thus, Lonp1 deficiency results in reduced protein levels of TFAM and mtDNA-encoded components of respiratory chain complexes.

Next, we constructed inducible Lonp1-deficient mouse embryonic fibroblasts (MEFs) derived from ERt2; Lonp1^F/F mice. Upon 4-OH tamoxifen administration, MEFs without Lonp1 showed abnormal mitochondrial morphological alterations (Fig. S7A), which was consistent with the results in Lonp1-deficient cardiomyocytes (Fig. S3). MEFs lacking Lonp1 also showed a decrease in the expression level of respiratory chain complex components CI-NDUFB8 and CIV-mt-COX2 but an increase in glycolysis regulatory proteins (Fig. 7A,B). In addition, there was an increase in ATF4 protein levels and a decrease in TFAM protein levels in MEFs lacking Lonp1 (Fig. 7B). Quantitative real-time PCR results showed that the changes in the transcription levels of these genes were consistent with those in Lonp1-deficient hearts (Fig. 7C). Accordingly, decreased mtDNA copy number was also observed in Lonp1-deficient MEFs (Fig. 7D), whereas the p-eIF2α level was enhanced in Lonp1-deficient MEFs (Fig. 7E). Similar alterations were observed in human 293T cells (Fig. 7F) and rat H9C2 cells (Fig. S7B), suggesting conserved regulation across species and cell types.

Furthermore, we overexpressed ATF4 (ATF4 OE) in several cell lines. This did not affect the expression level of LONP1 but was sufficient to reduce that of TFAM (Fig. 7G; Fig. S7C) and led to decreased levels of respiratory chain complex components (Fig. 7G). Accordingly, the ATF4 OE groups showed a lower mtDNA copy number (Fig. 7H). In addition, there was a decrease in the Tfam transcription level but an increase in the Glut1 transcription level in Lonp1-knockdown cells and ATF4 OE cells (Fig. 7I). ATF4 positively or negatively regulates target gene expression by binding to the cAMP response element (CRE) sequences proximal to the promoter regions of the target gene (Koyanagi et al., 2011). We identified the CRE sequences in the promoters of both Tfam and Glut1 and constructed luciferase reporters. A luciferase reporter assay revealed that ATF4 activated the promoter activity of Glut1 but repressed the promoter activity of Tfam (Fig. 7J,K). Collectively, these results demonstrate that ATF4 directly represses the transcription of Tfam but activates the transcription of Glut1.

To further understand the relationship between Lonp1 deficiency and AFT4 activation, we administered CDDO, a well-established LONP1 inhibitor, to cells to mimic Lonp1 deletion (Bernstein et al., 2012; Zurita Rendon and Shoubridge, 2018; Shin et al., 2021). We first tested the dosage effect of CDDO on cell proliferation and found that, at doses ranging from 1.5 to 2.5 µM, CDDO significantly decreased the proliferative capacity of the cells (Fig. S8A-C); CDDO treatment also substantially reduced the protein levels of respiratory chain complex components (Fig. 8A; Fig. S8D). We also observed the time course-dependent accumulation of ATF4 but a reduction in TFAM, CIV-COX4 and CV-mt-APT6 (Fig. 8B). Treatment of cells with CDDO (2.5 µM) for 10 h substantially enhanced the mRNA and protein levels of ATF4 but remarkably decreased the mRNA levels of Tfam (Fig. 8C-E). CDDO administration was also sufficient to change the mitochondrial morphology and to augment the level of ROS (Fig. S8E).

Although mitochondrial abnormalities can cause heart developmental defects, the importance of mitochondrial quality control for prenatal heart development has not been well documented. Our study shows that the mitochondrial matrix protease LONP1, as a key element in controlling mitochondrial quality control, has a crucial role in embryonic heart development. In combination with mouse genetics and pharmacological treatment, we demonstrated that LONP1-mediated mitochondrial quality control safeguarded the metabolic shift during heart development to meet the energy demands of the growing myocardium and embryonic body.

We identified AFT4 as an important mediator of abolished mitochondrial quality control. Disruption of mitochondrial quality control resulted in a profound accumulation of ATF4, probably as a consequence of activation of the ISR-p-eIF2α signaling pathway, which is consistent with the results of a previous study (Quirós et al., 2017).

We also determined that Tfam is a pivotal target gene of ATF4. A CRE was discovered in the promoter region of Tfam, and a biochemical study demonstrated that ATF4 negatively regulates Tfam transcription. Consequently, the expression of several mtDNA-encoded respiratory chain complex components was significantly suppressed after ATF4 activation, which in turn hindered mitochondrial oxidative metabolism. In addition, we found that the key anaerobic glycolysis gene Glut1 is another target gene of ATF4. Through direct regulation, ATF4 triggers Glut1 transcription and augments glycolysis. Thus, ATF4 exerts bilateral effects on metabolic regulation. On the one hand, ATF4 suppresses mitochondrial oxidative metabolism through inhibition of TFAM activity and, on the other hand, ATF4 boosts glycolysis by activating Glut1 expression. The detailed working model of LONP1-mediated mitochondrial quality control in safeguarding metabolic shifts is presented in Fig. 9.

From this study, two important conclusions could be drawn: (1) mitochondrial quality control is essential for heart development; and (2) mitochondrial quality control safeguards metabolic shifts during heart development. Thus, LONP1 connects mitochondrial quality control with metabolic shifts to ensure an energy supply for embryonic heart development.

We used Lonp1-floxed, Mef2c-AHF-Cre, cTnT-Cre, ctTA-Cre and ERt2-Cre mice (from The Jackson Laboratory, Stock no: 008463) in this study. All mouse strains were maintained on a C57BL/6 genetic background. Mice were group-housed in accordance with the regulations on mouse welfare and ethics of Nanjing University, with 12 h/12 h light–dark cycles and ad libitum access to food and water. The Institutional Animal Care and Use Committee (IACUC) of the Model Animal Research Center of Nanjing University approved all animal procedures used in this study.

Embryos and hearts fixed in 4% paraformaldehyde (PFA) were dehydrated and embedded in paraffin. Eight-micrometer-thick sections were cut using a Leica RM2016 microtome and then deparaffinized in xylene. Decreasing concentrations of ethanol were used to rehydrate the sections, which were then stained with Hematoxylin and Eosin (H&E; Uteambio, H0006-5G; SHJ0594).

Embryos and hearts fixed in 4% PFA were dehydrated and embedded in OCT medium (Sukura, 4583). Eight-micrometer-thick sections were cut using a Leica CM1950 automated cryostat. For IF staining, sections were placed at room temperature for 20 min and washed in PBS; goat serum (Beyotime, C0265) was used to block the sections, which were then incubated with primary antibodies overnight at 4°C. The next day, the sections were washed with PBS and incubated with secondary antibodies for 2 h. DAPI stains the nuclei. Finally, the sections were washed again in PBS and sealed with 50% glycerol before confocal imaging. Details of the antibodies used are shown in Table S1.

Following a previously described protocol (Vaccaro et al., 2020), in situ ROS detection was performed using 2′,7′-dichlorofluorescein (H2DCF; Invitrogen, D399) in fresh E11.5 hearts. Briefly, fresh hearts were washed with PBS and then immediately incubated in the dark with 10 μM H2DCF and DAPI for 15 min at room temperature. Before mounting, tissues were washed with PBS three times for 5 min each time.

Lonp1^F/F and cTnT-Cre; Lonp1^F/F primary cardiomyocytes were derived from E13.5 hearts. Under aseptic manipulation, each heart was washed in cold PBS three times. The hearts were then digested using enzyme buffer [0.1% trypsin (Biochannel, BC-CE-006) and 0.25% collagenase type II (Gibco, 17101015)] concomitant with shaking at 37°C in a water bath. Digestion was stopped by adding fetal bovine serum (FBS; Gibco, 10099) every 10 min and pipetting the solution. The operation was repeated until no obvious heart tissue was seen. The collected solution was centrifuged at 1000 g for 5 min, and 1 ml of complete medium (high-glucose Dulbecco's Modified Eagle Medium; Gibco, 12800017) was added, followed by transfer to a 6 cm Petri dish. After 2-3 h, nonadherent cardiomyocytes were transferred to new 6 cm Petri dishes. The cultured cardiomyocytes were used to perform IF staining, as described above.

Lonp1^F/F and ERt2-Cre; Lonp1^F/F MEFs were derived from E13.5 embryos. After removing the head, limbs and internal organs, the remaining tissues were washed with ice-cold PBS, digested with 1 ml 0.25% trypsin and cut into pieces using microscissors. After digestion at 37°C for 15 min in a 6 cm Petri dish, the cells were blown with pipette to disperse them, and then complete medium was added in addition to 0.5 mM 4-OH-tamoxifen (Sigma-Aldrich, H7904) to induce Cre expression over the following 3 days. The cells were then collected for western blot, quantitative real-time PCR and measurement of mtDNA copy number.

Adherent cells were cultured in complete medium (high-glucose Dulbecco's Modified Eagle Medium; Gibco, 12800017) supplemented with 10% FBS (Excell, FSP500) and 100 U/ml penicillin-streptomycin (Gibco, 15140122) at 37°C and 5% CO₂.

For RNAi, a small interfering RNA (siRNA) duplex, shown below, was designed to knockdown Lonp1. siLonp1-1: sense, 5′-GGGACAUCAUUGCCUUGAATT-3′; antisense, 5′-UUCAAG-GCAAUGAUGUCCCTT-3′; siLonp1-2: sense, 5′-CCGAGAACAAGAAGGACUUTT-3′; antisense, 5′-AAGUCCUUCUUGUUCUCGGTT-3′. Mixed siLonp1 was used to treat cells to achieve higher knockdown efficiency. According to the protocol, transfection was performed using GP-Transfect mate (GenePharma, G04009). Cells were harvested for western blot and quantitative real-time PCR analysis.

For overexpression, different plasmids were constructed and transfected into 293T cells using Lipo2000 (Invitrogen, 11668019) according to the manufacturer's protocol. The primers used are detailed in Table S3. Cells were collected for western blot, quantitative real-time PCR and measurement of mtDNA copy number. For CDDO treatment, cells cultured normally to 80% density were supplemented with CDDO. The concentration of CDDO and the time of treatment varied depending on the experimental requirements (optimal working concentration and time course). Cells were harvested for western blot and quantitative real-time PCR analysis.

The harvested heart tissues and cells were washed with PBS and then lysed on ice with RIPA buffer (Beyotime Biotechnology, P0013B) with a protease inhibitor cocktail (Roche, 4693116001) and PhosSTOP (Roche, 4906837001) for 30 min. The solution was then centrifuged at 13,500 g at 4°C for 15 min. Subsequently, the protein concentration in the collected supernatant was quantified by a BCA Protein Assay Kit (Beyotime, P0012). Proteins were isolated on SDS-PAGE gels (10%) and then transferred onto PVDF membranes (Millipore, IPVH00010). The proteins were then blocked with 5% bovine serum albumin (BSA; BioFroxx, 4240GR100) in Tris-buffered saline-Tween 20 (TBST; CST, 9997) and then incubated with different antibodies (Table S1). Given the small size of the embryonic hearts, two to six hearts were used per group, and the PVDF membranes were cut so that each section could be incubated with different antibodies to make optimal use of the samples available. All quantifications were performed by ImageJ.

Total RNA was extracted from the cardiac ventricles of (without atria and outflow tract) control, cTnT-Cre; Lonp1^F/F and Mef2C-AHF-Cre; Lonp1^F/F hearts and different cell lines by using TRIzol reagent (Invitrogen, 15596026). RNA reverse transcription was performed using a HiScript II Q Select RT Supermix for qPCR (+gDNA wiper) kit (Vazyme Biotech, R223-01). cDNA was used to generate a MicroAmp Optical 96-Well Reaction Plate (Applied Biosystems, N8010560) using an AceQ Universal SYBR Master Mix Kit (Vazyme Biotech, Q511-02) and an Applied Biosystems QuantStudio 5 Real-Time PCR System. Data were processed using the ΔΔCt analytical method with controls normalized to 1 to reveal the fold up- or downregulation. The primers used are detailed in Table S2.

The mtDNA copy number of different tissues and cells was measured as previously described (Quiros et al., 2014). Total DNA was extracted using a FastPure DNA Isolation Kit (Vazyme Biotech, DC112-01). RNase was added to ensure no RNA interference. The total DNA was added to a MicroAmp Optical 96-Well Reaction Plate using an AceQ Universal SYBR Master Mix Kit and an Applied Biosystems QuantStudio 5 Real-Time PCR System. Primers used are detailed in Table S2.

The full-length fragment of the mouse ATF4 coding sequence (CDS) was cloned into the pCMV7.1-3×FLAG vector. Based on the results predicted by JASPAR (https://jaspar.genereg.net/), the ∼2 kb Glut1 promoter fragment and an ∼2 kb Tfam promoter fragment were cloned into the pGL3-Basic Vector (Promega, E1751) to generate the Glut1-2K Luc and Tfam-2K Luc vectors, respectively. The primers used are detailed in Table S3. Plasmids were transfected into cells using Lipo2000 (Invitrogen, 11668019). Luciferase reporter assays in 293T cells were performed using the Dual Luciferase Assay System (Promega, E1910) following a standard protocol (Jiang et al., 2021) after 24-36 h of transfection.

All data are reported as mean±s.e.m. Differences between mean values were compared by two-tailed unpaired Student's t-tests or one-way and two-way ANOVAs using GraphPad Prism 8.0 software. Differences were significant at P <0.05.

We thank Dr Bin Zhou (Albert Einstein College of Medicine, New York) for providing the cTnT-Cre mice.

The peer review history is available online at https://journals.biologists.com/dev/article-lookup/doi/10.1242/dev.200458