A Mitochondrial Protein Helping Energy Systems May Extend Lifespan in Mice

We previously showed that COX7RP‐Tg mice exhibited a phenotype with enhanced exercise endurance and increased COX activity in muscle (Ikeda et al. 2013). In this study, we observed that male COX7RP‐Tg mice showed a significantly prolonged lifespan compared with their male WT littermates based on Kaplan–Meier survival analysis (Figure 1a). The mean lifespans for the male WT and COX7RP‐Tg mice were 126.2 and 134.5 weeks, respectively, and thus the percent increase was 6.6%. Similarly, the median lifespans for the male WT and COX7RP‐Tg mice were 129.2 and 136.4 weeks, respectively, and thus the percent increase was 5.6%. We observed a slight increase in lifespan in female COX7RP‐Tg mice compared with the WT littermates; however, the difference was not statistically significant (p = 0.0946) (Figure S1). Thus, we focused on the analyses of male mice in the current study. While body weight and tissue weights of BAT, liver, pancreas, skeletal muscle, heart, and brain of male COX7RP‐Tg mice were not substantially different from those of WT mice at 10 months old, weights of epididymal and inguinal WAT pads were significantly lower than those of WT mice (Figure 1b–e and Figure S2). Consistent with the previous observation (Ikeda et al. 2013), the expression levels of COX7RP mRNA, combining endogenous mouse Cox7rp and exogenous human COX7RP, were increased in the epididymal WAT, quadriceps femoris muscle, and liver of COX7RP‐Tg mice (Figure S3). COX7RP protein levels in COX7RP‐Tg WAT and quadriceps tissues were upregulated by ~2 and ~4 folds, respectively, compared with those in WT tissues as analyzed by Western blotting (Figure S4).

To understand the differences in WAT weights between COX7RP‐Tg and WT mice, we measured levels of serum lipids. Although no substantial difference was observed in non‐esterified fatty acid (NEFA), triglyceride (TG) and total cholesterol (TC) levels in 10‐month‐old COX7RP‐Tg mice appeared lower than those in age‐matched WT mice (Figure 2a) (Wilcox and Heimberg 1987). We next performed OGTT, ITT, and PTT in male COX7RP‐Tg and WT mice at 8 months of age (Figure 2b–e), a time point chosen to approximate middle age, as a difference in WAT weight was observed at 10 months. While the area under the curve (AUC) values for blood glucose in COX7RP‐Tg mice was similar to that of the age‐matched WT mice, a significant reduction in blood glucose levels at 120 min after oral glucose administration was observed in COX7RP‐Tg mice compared with that of WT mice, indicating a modest effect on glucose metabolism (Figure 2b). Plasma insulin levels also tended to be lower in COX7RP‐Tg mice compared with WT mice (Figure 2c). In ITT, COX7RP‐Tg mice displayed lower blood glucose levels at 0 and 120 min than WT mice (Figure 2d). PTT revealed that blood glucose levels were nearly equal in both genotype groups after pyruvate injection (Figure 2e). Overall, COX7RP‐Tg mice exhibit a phenotype with mildly altered lipid and glucose metabolism compared to the age‐matched WT mice.

We assessed respiratory states of WAT and quadriceps femoris muscle by analyzing oxygen consumption rates (OCRs) and extracellular acidification rates (ECARs), and showed that COX7RP‐Tg mice displayed elevated OCRs, particularly for basal and maximum respiration rates, in both tissues compared with WT mice (Figure 3a). Similarly, ECARs were increased in COX7RP‐Tg WAT and quadriceps femoris muscle (Figure S5). In addition, ATP synthesis and ROS levels were elevated and decreased in the mitochondria of quadriceps femoris muscle, respectively, from COX7RP‐Tg WAT compared with that of WT mice (Figure 3b,c).

We next investigated the contribution of COX7RP to mitochondrial respiratory supercomplex assembly in WAT and muscle. Immunoblotting of BN‐PAGE gel for proteins obtained from digitonin‐solubilized mitochondria showed that the amounts of supercomplexes CIII₂ + CIV and CI + CIII₂ + CIV_n were increased in COX7RP‐Tg WAT and quadriceps femoris muscle while the expression levels of representative components involved in respiratory chain complexes, NDUFA9 for CI, RISP for CIII, and COX1 for CIV, were similar between the genotypes (Figure 3d and Figure S4). Intriguingly, aging‐related biomarkers nicotinamide adenine dinucleotide (NAD⁺) levels and senescence‐associated β‐galactosidase (SA‐β‐gal) activities were increased and decreased in COX7RP‐Tg WAT and muscle, respectively, compared with those of WT mice (Figure 3e,f, and Figure S6a).

To further characterize the contribution of COX7RP to cell‐type transcriptome profiles in WAT, we performed snRNA‐seq analyses for epididymal WAT cell populations derived from four distinct groups: young (8‐week‐old) and old (2‐year‐old) mice with either WT or COX7RP‐Tg (Tg) genotypes. Cell populations from each group were classified into 10 clusters (Figure 4a–c and Figure S7). Among these cell clusters, adipocytes were defined as cell populations with high expression of adipocyte markers including Adipoq, Car3, Plin1, and Pparg. Stromal cells were classified into Pdgfra‐positive and ‐negative subclusters, both of which abundantly expressed Col1a1 and Cd34 while Dpp4 was predominantly expressed in the Pdgfra‐negative subcluster. Blood vessel endothelial cells were defined as populations abundantly expressing Pecam1 and Cdh5. Immune cells (Ptprc^high) were clustered into macrophages and lymphocytes. Macrophages highly expressed Csf1r and Cd68. On the other hand, the lymphocyte population included Cd3e‐positive or Ms4a1‐positive cells. Reproductive tissues were classified into the following categories: testicular cells as Lcn2^high/Mmp7^high, ciliated cells as Cdhr3^high/Dnah9^high/Foxj1^high/Tmem212^high, epididymis cells as Defb42^high/Mfsb42^high, and epithelial cells as Ctse^high/Aqp3^high/Krt5^high/Sptssb^high. The actual count for each cell cluster from the distinct groups is shown in Figure 4b. To dissect differentially expressed genes (DEGs) between young and old WT mice or between old WT and old Tg mice in snRNA‐seq, we depicted volcano plots for these comparisons (Figure 4d). Gene Ontology (GO) pathway analysis for upregulated DEGs revealed that fatty acid metabolism‐ and mitochondria‐related pathways were altered in old WT mice versus young WT mice and in old WT versus old Tg mice, respectively (Figure 4e). To predict upstream regulatory transcription factors (TFs) of the DEGs in adipocytes, we used Transcriptional Regulatory Relationships Unraveled by Sentence‐based Text‐mining (TRRUST) (Han et al. 2018) via Metascape (Zhou et al. 2019) (Figure 4f). TFs such as Sp1, Srf, Pparg, Myocd, and Ep300 were identified as correct upstream regulators with significant motif enrichment, likely predicting their contribution to the transcriptional regulation of DEGs in old WT mice versus young WT mice and in old WT versus old Tg mice. In the adipocyte cluster, higher expression of SASP‐associated genes was observed in old WT versus young WT mice whereas not in old COX7RP‐Tg versus young COX7RP‐Tg mice (Figure 5a,b and Figure S8). We selected prototypic SASP‐associated genes based on the gene expression data of WAT snRNA‐seq studies (Emont et al. 2022) and a list of SASP‐related genes originally proposed by Dr. Campisi Lab (Freund et al. 2010), including Il6, Cxcl1, Il1b, Tnf, Tgfb1, Egf, Mmp3, Mmp9, Igfbp5, Adam17, Ctsb, Ctsl, Ctsd, Il18, Ccl2, Ccl3, Ccl5, and Ccl6. Reverse transcriptase‐quantitative polymerase chain reaction (qRT‐PCR) verified that Ccl5 and Tnf levels were significantly decreased in COX7RP‐Tg WAT compared with those of WT mice (Figure S6b). We also examined p53 levels in COX7RP‐Tg and WT WATs, because p53 plays an important role in the development of insulin resistance and senescence‐like changes such as the expression of proinflammatory cytokines (Minamino et al. 2009). However, the expression levels of p53 mRNA were not substantially different between these genotypes (Figure S6b), indicating that p53 transcription is not primarily influenced by COX7RP overexpression. To further clarify the role of COX7RP in the adipocyte aging process, we applied the single‐cell regulatory network inference and clustering (SCENIC) algorithm (Aibar et al. 2017) to explore the potential upstream regulators for SASP‐associated genes with significant motif enrichment in adipocytes (Figure 5c and Figure S9). In particular, Cebpa, Rela, Creb3l2, Xbp1, and Zbtb7a were predicted as the top 5 upstream TFs putatively responsible for the SASP‐associated genes examined in the present study, based on the Venn diagram for the top 10 upstream TFs in comparisons between young WT versus old WT groups and between old WT versus old Tg groups (Figure 5d).

All animal experiments were approved by the Animal Care and Use Committee of Saitama Medical University, and conducted in accordance with the Institutional Guidelines and Regulations for the Care and Use of Experimental Animals. Mice were maintained in a temperature‐controlled room (23°C) with a 12‐h light/dark schedule and fed a standard diet (ce2, CLEA Japan), with free access to water. COX7RP‐Tg mice expressing Flag‐tagged human COX7RP under CAG (CMV‐IE enhancer, chicken β‐actin promoter, rabbit β‐globin genomic DNA) promoter were generated using a microinjection of fertilized eggs from C57BL/6 mice as described previously (Ikeda et al. 2013).

Serum TG, TC, and NEFA levels were measured using the Wako TG E test kit, TC E test kit, and NEFA C test kit, respectively (Wako Pure Chemical).

Male COX7RP‐Tg and WT mice at 8 months old (n = 6 for each group) were fasted for 16 h before OGTT as described (Shiba et al. 2014). Mice were challenged with orally administered d‐(+)‐glucose (2 g kg⁻¹). Blood samples were collected from the saphenous vein at 0, 30, 60, and 120 min after glucose administration. Blood glucose levels were measured using a glucose analyzer (Sanwa Kagaku Kenkyusho). Plasma insulin levels were measured using Mouse Insulin ELISA KIT (S‐type) (Shibayagi).

Male COX7RP‐Tg and WT mice at 8 months old (n = 6 for each group) were fasted for 1 h before ITT as described (Shiba et al. 2014). Mice were injected intraperitoneally with insulin (0.5 U kg⁻¹). Blood samples were collected from the saphenous vein at 0, 30, 60, and 120 min after injection of insulin, and blood glucose levels were measured.

Male COX7RP‐Tg and WT mice at 8 months old (n = 5 for each group) were fasted for 16 h before PTT as described (Shiba et al. 2017). Mice were injected intraperitoneally with sodium pyruvate (2 g per kg⁻¹). Blood samples were collected from the saphenous vein at 0, 10, 30, 60, 90, and 120 min after injection of pyruvate, and blood glucose levels were measured.

WAT and quadriceps femoris muscle were dissected from male COX7RP‐Tg and WT mice at 2 years old (n = 6 for each group). The freshly isolated tissues were cut into sections, and a 1 mg piece was placed in each well of XF 24 Islet Capture Microplate (Seahorse Bioscience). The OCR and ECAR were measured using Seahorse XF24 analyzer (Seahorse Bioscience) with Cell Mito Stress Test Kit (Seahorse Bioscience) in which oligomycin (60 μM), FCCP (10 μM), and antimycin A (15 μM) and rotenone (5 μM) were sequentially injected. Basal and maximal OCR were calculated by subtracting the OCR in the presence of antimycin A/rotenone from those before the injection of oligomycin and in the presence of FCCP, respectively.

ATP synthesis rate was quantified using mitochondria isolated from WAT and quadriceps femoris muscle in 2‐year‐old COX7RP‐Tg and WT mice as described previously (Ikeda et al. 2013). Briefly, mitochondria were resuspended in a buffer containing 25 mM Tris–HCl, pH 7.4, 150 mM KCl, 2 mM EDTA, 10 mM potassium phosphate, 0.1 mM MgCl₂, 0.1% bovine serum albumin, and 50 mg mL⁻¹ digitonin, and then energized with 1 mM malate and 1 mM pyruvate in the presence of 1 mM ADP and 0.15 mM adenylate kinase inhibitor P1, P5‐di (adenosine‐50) pentaphosphate for 10 min at 37°C. Parallel incubations were also carried out in the presence of 2 mg mL⁻¹ oligomycin to measure mitochondria‐specific ATP synthesis. ATP concentrations were measured using the MicroLumat Plus luminometer (Berthold Technologies) with ATP luminescent reagent (TOYO B‐Net).

ROS production was quantified in WAT and quadriceps femoris muscle, both of which were obtained from 2‐year‐old COX7RP‐Tg and WT male mice. Mitochondria were purified from these tissues by homogenization in a buffer (20 mM Tris–HCl, pH 7.4, 150 mM KCl, 0.5 mM EDTA, 1 mM MgCl₂, 5 mM glucose, and 0.5 mM octanoic acid) followed by centrifugation at 10,000 g for 5 min. ROS production in mitochondria was measured using the OxiSelect In Vitro ROS/RNS Assay Kit (CELL BIOLABS) based on a dichlorofluorescin probe.

BN‐PAGE and Western blot analysis of mitochondria extracted from WAT and quadriceps femoris muscle were performed as previously described (Ikeda et al. 2013, 2019). Briefly, tissues were homogenized with a glass‐teflon homogenizer in a buffer containing 10 mM HEPES‐KOH (pH 7.4), 0.22 M mannitol, 0.07 M sucrose, and 0.1 mM EDTA and the mitochondrial fraction was obtained by differential centrifugation. Mitochondria were resuspended in 10 μL of a buffer containing 50 mM Bis‐Tris and 1 M 6‐aminocaproic acid, and subsequently solubilized with digitonin (digitonin/protein ratio of 4 g g⁻¹). The solubilized proteins were supplemented with 1 μL of sample buffer (5% Coomassie Brilliant Blue G‐250 in 0.5 M 6‐aminocaproic acid) and separated on BN‐PAGE. Western blotting was performed according to standard protocols (Ikeda et al. 2013) and the blots were probed with anti‐COX1 (1:5000 dilution, ab14705, Abcam) and Fp70 (1:5000 dilution, 459200, Invitrogen).

The amount of NAD⁺ was quantified using the NAD/NADH Assay Kit‐WST (Dojindo Laboratories) according to the manufacturer's instructions. Briefly, WAT and quadriceps femoris muscle tissues were homogenized in the NAD/NADH extraction buffer and subjected to the enzymatic reaction at 37°C for 1 h. Then, the absorbance of the samples was read on a microplate reader at a wavelength of 450 nm. The NAD⁺ amount was normalized with the protein concentrations.

SA‐β‐gal activity was measured using the Senescence β‐galactosidase Activity Assay Kit (Cell Signaling Technology) according to the manufacturer's instructions. Briefly, WAT and gastrocnemius muscle tissues were homogenized in Senescence Cell Lysis Buffer and, then, the cytosolic fractions were plated for fluorescent reading with excitation at 355 nm and emission at 460 nm. For whole tissue staining, freshly isolated WAT was incubated for 12 h at 37°C in a β‐gal staining solution containing 1 mg mL⁻¹ 5‐bromo‐4‐chloro‐3‐indolyl β‐D‐galactopyranoside (X‐gal), 5 mmol L⁻¹ potassium ferrocyanide, 5 mmol L⁻¹ potassium fericyanide, 150 mmol L⁻¹ NaCl, 2 mmol L⁻¹ MgCl₂, 0.01% sodium deoxycholate, and 0.02% Nonidet‐40.

Tissues were prepared in sample buffer for sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS‐PAGE), heated at 100°C for 15 min, and subjected to SDS‐PAGE followed by western blot analysis using antibodies against Cox7rp (1:1000 dilution), Risp (1:5000 dilution, ab14746, Abcam), Cox1 (1:5000 dilution, ab14705, Abcam), Ndufa9 (1:5000 dilution, 459100, Invitrogen), and β‐Actin (1:4000 dilution, AC‐74, Sigma‐Aldrich). The signal intensity of each protein band was quantified densitometrically using ImageJ software (https://imagej.nih.gov/ij/↗). The densitometric value of each protein was normalized to the corresponding β‐Actin signal and presented as a relative value. Mouse monoclonal anti‐COX7RP antibody was generated using a polypeptide containing amino acids 101–114 of the human COX7RP protein as described elsewhere (Sato et al. 2021).

Total RNAs were extracted from the tissues in COX7RP‐Tg and WT mice using ISOGEN reagent (Nippon Gene) and then subjected to cDNA synthesis using reverse transcriptase SuperScript II (Invitrogen). qPCR was performed with StepOnePlus (Applied Biosystems) using KAPA SYBR Fast qPCR Kit (Nippon Genetics) to determine the expression levels of human and mouse COX7RP, Ccl5, and Tnf. The relative expression levels of genes were calculated by normalization with Gapdh expression as an internal control. The sequences of PCR primers are as follows: common primers for human and mouse COX7RP forward, 5′‐GTTTCCACAGAAGCACCACCTATC‐3′; human and mouse COX7RP reverse, Ccl5 forward, 5′‐CCCTCACCATCATCCTCACTG‐3′; Ccl5 reverse, 5′‐GAGAGGTAGGCAAAGCAGCAG‐3′; Tnf forward, 5′‐ACCGTCAGCCGATTTGCTAT‐3′; Tnf reverse, 5′‐CTTGACGGCAGAGAGGAGGTT‐3′; Gapdh forward, 5′‐GCATGGCCTTCCGTGTTC‐3′; and Gapdh reverse, 5′‐TGTCATCATACTTGGCAGGTTTCT‐3′.

Single‐cell suspensions were prepared from the WAT of COX7RP‐Tg and WT mice at 8 weeks old and 2 years old. Single cell nuclei were isolated using Rosen's protocol (Emont et al. 2022). Each flash‐frozen adipose tissue sample was placed into a gentleMACS C tube (Miltenyi Biotec) with 2 mL freshly prepared TST buffer (0.03% Tween 20 [Bio‐Rad], 0.01% Molecular Grade BSA [New England Biolabs], 146 mM NaCl [ThermoFisher Scientific], 1 mM CaCl₂ [VWR International], 21 mM MgCl₂ [Sigma Aldrich], and 10 mM Tris–HCl pH 7.5 [ThermoFisher Scientific] in Ultrapure water [ThermoFisher Scientific]) with or without 0.2 U μL⁻¹ of Protector RNase Inhibitor (Sigma Aldrich). gentleMACS C tubes were then placed on the gentleMACS Dissociator (Miltenyi Biotec) and tissue was dissociated by running the program mr_adipose_01 twice, and then incubated on ice for 10 min. Lysate was passed through a 40‐μm nylon filter (CellTreat) and collected into a 50‐mL conical tube (Corning). The filter was rinsed with 3 mL of freshly prepared ST buffer (146 mM NaCl, 1 mM CaCl₂, 21 mM MgCl₂; 10 mM Tris–HCl pH 7.5) with or without 0.2 U μL⁻¹ RNase Inhibitor, and collected into the same tube. Flow‐through was centrifuged at 500 g for 5 min at 4°C with the brake set to low. Following centrifugation, the supernatant was removed, and the nuclear pellet was resuspended in 50–200 μL PBS pH 7.4 (ThermoFisher Scientific) with 0.02% BSA, with or without 0.2 U μL⁻¹ RNase inhibitor. Nuclei from each sample were stained with ViaStain AO/PI Staining Solution (Nexcelom Biosciences) and counted in a hemocytometer using fluorescence to identify intact nuclei. The snRNA‐seq libraries were prepared with a Fixed RNA profiling reagent kit according to 10× Genomics specifications. After hybridization, nuclei from each sample were stained with ViaStain AO/PI Staining Solution (Nexcelom Biosciences) and counted in a hemocytometer again. For each sample, bc001, bc002, bc003, and bc004 barcodes were added. Samples were pooled to achieve an expected cell count of 10,000 cells each (10× Genomics), then loaded in one channel of a Chromium Chip (10× Genomics). The indexed libraries were subjected to a second round of double‐sided size selection, and the libraries were then quantified and quality‐assessed with Agilent TapeStation. The libraries were clustered using NovaSeq 6000 on a paired‐end read flow cell. Sequencing was performed with a 100 cycle kit and the following read structure: Read1: 28, index1(i7): 10, index2(i5): 10, Read2: 90. The 10× Genomics Cell Ranger Single Cell Software was used for sample demultiplexing, alignment to the mouse reference genome (mm10), filtering, UMI counting, single‐cell 3′‐end gene counting, and quality control according to the manufacturer's parameters.

The R package Seurat (v4.0.1) was used to cluster cells in a merged matrix. Several quality control steps were performed based on the plot pattern of the expression gene number and ratio of mitochondrial RNA in each sample. First, cells expressing fewer than 200 genes or > 5% mitochondrial genes (low quality) were excluded from further analysis. The gene counts for each cell were divided by the total gene counts for each cell and multiplied by a scale factor of 10,000, after which a natural log transformation was applied. The FindVariableFeatures function was used to select variable genes with default parameters. The ScaleData function was used to scale and center the counts in the dataset. To define the cell types of all cells, we performed principal component analysis on the variable genes, and 15 principal components were used for cell clustering (resolution = 0.5) and UMAP dimensional reduction. The cluster markers were found using the FindAllMarkers function, and cell types were manually annotated based on the cluster markers. DEGs between two groups were defined by FindMarkers with Welch t‐test. The single‐cell gene expression data were visualized with UMAP overlays, dot heatmaps, and violin plots. A volcano plot was constructed in R. Gene Ontology analysis was performed using enrichGO in clusterProfiler. All of the subontologies (Molecular Function, Biological Process, and Cellular Component) were included in the list, and the method for adjusting the p value was Benjamini and Hochberg.

We selected major SASP‐associated genes based on the gene expression information in the WAT snRNA‐seq (Emont et al. 2022) and a list of SASP‐related genes (Freund et al. 2010). The list included Il6, Cxcl1, Il1b, Tnf, Tgfb1, Egf, Mmp3, Mmp9, Igfbp5, Adam17, Ctsb, Ctsl, Ctsd, Il18, Ccl2, Ccl3, Ccl5, and Ccl6. To calculate the overall SASP induction, the R package of calculate_signature_scores was used. The result calculated from the sum of Z‐scores within SASP‐associated genes was presented as the gene.score. A significance test was conducted using the one‐way ANOVA followed by the Tukey method, and p values were calculated for each group comparison.

SCENIC was used to identify the gene regulatory networks (GRNs) and the upstream TF from snRNA‐seq data in R (Aibar et al. 2017). To identify TF binding motifs, the database (mm10_refseq‐r80_10kb_up_and_down_tss.mc9nr.feather/mm10_refseq‐r80_500bp_up_ and_100bp_down_tss.mc9nr.feather) was downloaded as cisTarget_database in the RcisTarget package. For running GENIE3, the parameter of ntree in the random forest model was 500. Among the scored GRNs, the extended regulons were selected as predicted upstream TFs. TF identification was performed in two pairs of comparisons: young WT (8‐week‐old) versus old WT (2‐year‐old), and old WT versus old COX7RP‐Tg. The regulon specificity score (RSS) was calculated by the calcRSS package and plotted by the plotRSS package with Z score normalization (Suo et al. 2018) across the samples between the pairs, individually.

Significance of differences between two groups was analyzed by a two‐tailed unpaired Student's t‐test. The details of statistical analysis in snRNA‐seq data analysis, SASP gene scoring, and SCRNIC analysis were described above.