Patients who were scheduled for elective surgery were prospectively enrolled at the Department of Gynecology of Chungnam National University Hospital (CNUH), Daejeon, Republic of Korea. The Institutional Review Board of CNUH approved the study (approval number: 2016-07-026), and written informed consent was obtained from all participants. The study was performed in accordance with the principles of the Declaration of Helsinki. Inclusion criteria were age >19 years, and elective surgery for benign uterine and ovarian disorders. Individuals with cancer, prediabetes, diabetes mellitus, active liver disease, and/or end-stage renal disease were excluded. Finally, 48 participants were included in this study. Blood chemistry of participants was assessed, and serum insulin concentrations were measured as previously described [9].

Omental adipose tissue samples from the distal portion of the greater omentum were collected during surgery under general anesthesia. Adipose tissue specimens were snap-frozen in liquid nitrogen and stored at −70°C.

The methods that were used for RNA extraction, cDNA synthesis, and real-time polymerase chain reaction (RT-PCR) analysis have been described previously [10]. The primers for RT-PCR are listed in Supplemental Table S1. Target mRNA expression was normalized to 18S mRNA expression and determined by the 2^−ΔΔCt method. Mitochondrial DNA (mtDNA) content was determined by quantitative RT-PCR, as previously described [11].

Antibodies targeting cytochrome c oxidase subunit 1 (COX1), ubiquinol-cytochrome-c reductase complex core protein 2 (UQCRC2), DNAJA3, LONP1, and β-actin were purchased from Abcam (Cambridge, UK); antibody to NADH dehydrogenase (ubiquinone) 1 beta subcomplex subunit 8, mitochondrial (NDUFB8) was purchased from Novex (Carlsbad, CA, USA); and antibodies recognizing succinate dehydrogenase (ubiquinone) iron-sulfur subunit, mitochondrial (SDHB), transcription factor A, mitochondrial (TFAM), HSPD1, and CLPP were purchased from Santa Cruz Biotechnology (Dallas, TX, USA). Anti-rabbit secondary antibodies were obtained from Santa Cruz Biotechnology, and anti-mouse secondary antibodies were obtained from Cell Signaling Technology (Danvers, MA, USA). Western blotting images were scanned and quantified using an Odyssey imaging system and Image Studio DiGit software (LI-COR Biosciences, Lincoln, NE, USA). Lipid peroxidation was analyzed by measuring malondialdehyde concentrations with a commercial thiobarbituric acid-reactive substances assay kit (Cayman Chemical, Ann Arbor, MI, USA), according to the manufacturer’s specifications.

Statistical analyses were performed with SPSS software version 24.0 (IBM, Armonk, NY, USA). Clinical data are expressed as mean±standard deviation. mRNA levels are expressed as mean±standard error of the mean. Continuous variables were tested for normality with the Kolmogorov-Smirnov test. Clinical characteristics and mRNA levels of genes related to oxidative phosphorylation (OXPHOS) and UPR^mt were compared between groups of individuals with body mass index (BMI) <23 kg/m² and those with BMI ≥23 kg/m² by Student’s t test or the Mann-Whitney U test for data with or without a normal distribution, respectively. The strengths of the relationships between variables of interest were analyzed by calculation of Spearman’s correlation coefficients. P<0.05 was considered to represent statistical significance.

All the available Genotype-Tissue Expression (GTEx) data for VAT (n=541) were obtained from the UCSC database (http://xena.ucsc.edu/). The data used for our analyses were obtained from dbGaP, accession number: phos000424.v8.02 on April 13, 2020. Of the 541 VAT samples in the GTEx database, those with LONP1 expression in the highest (n=134) and lowest (n=134) quartiles were used for Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis. Differentially expressed genes were identified by the establishment of two groups based on LONP1 expression with the R package DESeq2 [12]. The gene-set collection of KEGG was obtained from Enrichr (https://amp.pharm.mssm.edu/Enrichr/), and the gene-set enrichment analysis was conducted with the R package Platform for Integrative Analysis of Omics data (PIANO). P values were adjusted using Benjamini-Hochberg correction for controlling false-discovery rate, and results were considered statistically significant when adjusted P values were <0.05. The full list of significantly enriched pathways is included in Supplemental Material S1.

To analyze murine metabolic phenotypes in relation to expression of Lonp1, GeneNetwork (www.genenetwork.org) was used for multi-omics data for BXD recombinant inbred (RI) strains [13]. Expression of Lonp1 in subcutaneous WAT (sWAT) of BXD mice that were fed either chow or high-fat diets (GN accession numbers: GN779 and GN778, respectively) was used for the analysis. Metabolic phenotypes were compared between mice in the upper (Lonp1-high) and lower (Lonp1-low) quartiles with respect to WAT Lonp1 expression (n=9–10 mice per group). Phenotypic data recorded for the mice were body weight in males aged 8, 16, and 28 weeks old (GN record ID: 17557-17562); blood glucose measured during the oral glucose-tolerance test (OGTT) at 17 weeks old (GN record ID: 17643-17660); insulin measured during OGTT at 17 weeks old (GN record ID: 17665-17670); and lipid profiles measured in fasted state at 29 weeks old (GN record ID:17801-17812) [14,15]. Metabolic phenotypes and Lonp1 mRNA levels in sWAT were compared between Lonp1-low and Lonp1-high groups with Student’s t test. A paired-sample comparison of Lonp1 expression in the sWAT of mice fed chow versus high-fat diets was conducted in 35 mouse strains.

Among 48 patients, 11 were obese (≥25 kg/m²), 11 were overweight (23 to 24.9 kg/m²), and 26 were of normal or underweight (<22.9 kg/m²), according to the World Health Organization Asia-Pacific Obesity Classification [16]. Clinical characteristics of the participants stratified by BMI (<23 kg/m² vs. ≥23 kg/m²) are summarized in Table 1. BMI, waist circumference, fasting blood glucose, fasting blood insulin, homeostatic model assessment of insulin resistance (HOMA-IR), and alanine aminotransferase values were significantly higher in the high-BMI than in the low-BMI group, whereas age and other variables did not differ significantly between the two groups. Among OXPHOS-complex genes (NADH:ubiquinone oxidoreductase subunit A9 [NDUFA9], SDHB, UQCRC2, COX4, and ATP synthase F1 subunit alpha [ATP5A1A]), UQCRC2 (from OXPHOS complex III) was only significantly lower in the high-BMI group than in the low-BMI group (Fig. 1A). The mitochondrial biogenesis-related genes, PPARGC1A (which encodes PPARG coactivator 1α), a key activator of mitochondrial transcription, TFAM expression and mtDNA content did not differ significantly between the groups (Fig. 1B, C). Although excess fat accumulation increases mitochondrial production of reactive oxygen species to cause adipocyte mitochondrial dysfunction [2], cellular concentrations of malondialdehyde, a naturally occurring product of lipid peroxidation, did not differ significantly between the groups (Fig. 1D). In correlation analyses, expression of these genes including UQCRC2, VAT mtDNA content, and cellular malondialdehyde concentrations did not correlate significantly with BMI (data not shown).

We compared the expression of genes encoding mitochondrial chaperones (HSPD1 and DNAJA3), mitochondrial matrix proteases (CLPP and LONP1), inner membrane protease (YME1 like 1 ATPase [YME1L1]), intramembrane protease (HtrA serine peptidase 2 [HTRA2]), outer membrane protease (ubiquitin specific peptidase 30 [USP30]) in VAT in relation to BMI. Among these UPR^mt-related genes, LONP1 expression was around twice as high in the high-BMI group as in the low-BMI group (Fig. 1E). LONP1 gene expression was positively correlated with BMI (ρ=0.308, P=0.050) (Fig. 1F). LONP1 expression was also positively correlated with expression of HSPD1 and DNAJA3 (ρ=0.622, P<0.001 and ρ=0.409, P= 0.012, respectively), but was not correlated with expression of OXPHOS genes, PPARGC1A or TFAM, VAT mtDNA content or malondialdehyde concentration, participant age, serum glucose, lipid profile or HOMA-IR (data not shown). This finding indicates that transcriptional activation of LONP1 is a signature feature occurred in the course of human visceral obesity.

We measured the expression of mitochondrial chaperones, proteases, OXPHOS-complex proteins, and proteins involved in mitochondrial biogenesis in VAT of the five participants with the lowest BMI (mean 18.6 kg/m², range 17.4 to 19.3 kg/m²) and the five with the highest BMI (mean 29.6 kg/m², range 27.5 to 31.8 kg/m²). Among the UPR^mt proteins, only LONP1 showed a difference in expression, which was significantly higher in the VAT of participants with high BMI (Fig. 2A). Among the OXPHOS-complex proteins, participants with high BMI had significantly lower expression of the OXPHOS complex I protein NDUFB8 than those with low BMI; no other differences in expression were observed (Fig. 2B). Expression of the mitochondrial biogenesis-related protein TFAM was similar in participants with low BMI and those with high BMI (Fig. 2C). LONP1 was therefore the only differentially expressed protein for which a consistent change in mRNA levels occurred in the UPR^mt in the VAT of participants with high BMI, suggesting that LONP1 is a marker for nutritional stress in the mitochondria of human VAT.

To gain further insight into the relationship between energy metabolism and LONP1 expression in human VAT, gene-enrichment analysis was performed with VAT expression data from the GTEx database, with reference to levels of LONP1 expression. Of the 541 VAT samples in the GTEx database, samples with LONP1 expression levels in the highest quartile (n=134) and the lowest quartile (n=134) were used for GO and KEGG pathway analyses (Fig. 3A). In these samples, 6,455 genes were upregulated in the high-LONP1 group compared with the low-LONP1 group (Fig. 3B). In a GO–biological-process analysis, 1,119 processes were upregulated in the high-LONP1 group compared with the low-LONP1 group (Fig. 3C). Among the significantly upregulated processes were cellular response to nutrient level, lipid homeostasis, carbohydrate and glucose homeostasis, and response to insulin (Fig. 3D). In addition, genes involved in regulation of the lipid metabolic process, fatty acid metabolic process, and fatty acid oxidation were upregulated in the VAT of individuals with higher LONP1 expression (Fig. 3E), as were genes involved in the glucose metabolic process and response to glucose (Fig. 3F).

In the KEGG analysis, 133 pathways were upregulated in the VAT of the high-LONP1 group compared with the low-LONP1 group (Fig. 3C), including pathways related to the citrate cycle, OXPHOS, insulin signaling pathway, regulation of lipolysis in adipocytes, and insulin resistance (Fig. 3G). These results indicate that LONP1 expression in human VAT is associated with metabolic regulation involving processes such as the TCA cycle, and glucose and lipid metabolism.

To attempt to validate the physiological role of LONP1, we analyzed phenotypic data in the BXD RI mouse strain database according to Lonp1 expression in the sWAT (data for VAT were not available). Lonp1 expression was significantly higher in the sWAT of mice that were fed a high-fat diet than in the corresponding mice that were fed a chow diet (Fig. 4A), which was consistent with the high LONP1 expression that we observed in the VAT of individuals with high BMI. In comparisons between chow-fed BXD strains in the highest quartile for Lonp1 expression in sWAT (Lonp1-high) and those in the lowest quartile (Lonp1-low), body weight was significantly lower in the Lonp1-high group at 16 weeks old, and blood glucose was significantly lower in the Lonp1-high group during OGTT at 17 weeks old, although insulin levels during OGTT did not differ significantly (Fig. 4B–E). In comparisons between high-fat-diet-fed mice in the highest quartile for Lonp1 expression in sWAT (Lonp1-high-HFD) and those in the lowest quartile (Lonp1-low-HFD), body weight and blood glucose did not differ between the groups (Fig. 4F–H), but the insulin level during OGTT was more than twofold higher in the Lonp1-high-HFD group than in Lonp1-low-HFD mice (Fig. 4I). Therefore, Lonp1 expression in murine sWAT was related to systemic glucose metabolism, which was consistent with the results of the human VAT GTEx gene analysis. On the other hands, the lipid profiles including free fatty acids, high-density lipoprotein, low-density lipoprotein, triglycerides and cholesterol, were not significantly different according to Lonp1-expression level in the both chow and high fat diet BXD RI mouse strain (data not shown).