Repression of oxidative phosphorylation by NR2F2, MTERF3 and GDF15 in human skin under high-glucose stress

Abstract

Lifestyle factors such as a Western diet or metabolic diseases like diabetes disrupt glucose homeostasis and induce stress responses, yet their impact on skin metabolism and structural integrity remains poorly understood. Here, we performed multiomic and bioenergetic analyses of human dermal fibroblasts (HDFs), human equivalent dermis (HED), human reconstructed skin (HRS), and skin explants from diabetic patients. We found that 12 mM glucose stress represses oxidative phosphorylation (OXPHOS) through a dual mechanism: the glucose-dependent nuclear receptor NR2F2 activates mitochondrial transcription termination factor 3 (MTERF3) while inhibiting growth-differentiation factor 15 (GDF15). Promoter assays revealed that MTERF3 is regulated by NR2F2 and MYCN, whereas GDF15 is modulated by NR2F2 and FOS. Consequently, OXPHOS proteins and mitochondrial respiration were suppressed, and MTERF3 overexpression additionally interfered with collagen biosynthesis. In contrast, GDF15 supplementation fully rescued hyperglycemia-induced bioenergetic and metabolomic alterations, suggesting a pharmacological strategy to mitigate hyperglycemic damage in the skin. Finally, silencing GDF15 or TFAM impaired fibroblast haptotaxis and skin reconstruction, underscoring the crucial role of mitochondrial energetics in dermal structure and function. Collectively, these findings identify the NR2F2-MTERF3-GDF15 axis as a key mediator of OXPHOS suppression and highlight a potential therapeutic target to preserve skin integrity under hyperglycemic stress.

Keywords: GDF15; Hyperglycemia; MTERF3; Oxidative phosphorylation; Skin.

We identify the NR2F2-GDF15-MTERF3 axis as a key regulatory mechanism that represses oxidative phosphorylation in human skin under high-glucose stress. This repression profoundly impacts collagen organization and overall skin physiology. By integrating both bioenergetic and structural responses, the NR2F2-GDF15-MTERF3 axis supports skin resilience, enabling tissues to adapt to hyperglycemia. These findings reveal new insights into how skin homeostasis is maintained under metabolic stress and highlight potential therapeutic targets for improving skin integrity in hyperglycemic conditions.

Fig. 1

Dermal collagen network and human skin reconstruction are altered by 12 mM glucose stress. A) Macroscopic view of HRS cultivated 38 days in 5.55 mM or 12 mM glucose media. The images were acquired using a binocular microscope. Scale bar 0.6 cm. B) Immunofluorescence detection of collagens I and VI in HRS exposed to 5.55 mM or 12 mM glucose. C) Signal was quantified by measuring the fluorescence intensity of collagen normalized to nucleus number (as detected using DAPI staining). Collagen network morphology was analyzed by applying an internal deconvolution filter, then by counting the number of objects and relating them to the circularity parameter. Results are given as arbitrary units (A.U.). 10 images were acquired per HRS (N = 3). D, E) Masson trichrome staining of HRS exposed to 5.55 mM or 12 mM glucose. Images were acquired using a Scanner VS120 Olympus (Obj x10). Staining was performed for collagen fibers (blue-green), keratin fibers (red) and cell nuclei (dark red or purple). SC = stratum corneum; EP = epidermis; DE = dermis. F) Collagen density was quantified by measuring the staining intensity normalized to the nucleus number (DAPI staining) (N = 6). Quantification of epidermis and stratum corneum thickness was given as arbitrary units (A.U.) (N = 6) G) Immunofluorescence of corneodesmosine and involucrine in HRS exposed to 5.55 mM or 12 mM glucose. Markers expression was quantified by measuring the fluorescence intensity, as normalized by the number of nuclei (DAPI staining). 10 images were acquired per HRS (N = 3). H,I) Comparative Taqman Low Density Array (TLDA) real-time RT-PCR gene expression analyses at day 38. HRS cultivated in 12 mM glucose were compared to those grown in 5.5 mM glucose. N = 4.

Fig. 2

High glucose stress represses mitochondrial proteome, metabolome and bioenergetics. A) Label free unbiased differential proteomic study of HDF exposed 48H to 6.5 mM glucose as compared to 5.55 mM. The repressed proteins are shown on the left side of the Volcano Plot and the over-represented proteins on the right. The X axis gives the quantitative change expressed of log2 of the fold change (6.5 mM versus 5.55 mM). The Y axis provides the significancy of the changes expressed as the -log10 of the Adjusted P value (N = 4). B) Similar study was performed between the 12 mM and the 5.55 mM glucose conditions. C) Quantification of RNA transcripts by Taqman quantitative RTPCR of mt-RNR1 (Mitochondrially Encoded 12S RRNA) and TFAM (Transcription Factor A, Mitochondrial) in HDF grown in 5.55 mM or 12 mM glucose. Normalization of the data was performed to GusB (β-glucuronidase) (N = 3). D) Metabolomic analysis of HDF grown 48H in DMEM with 5.55 mM or 12 mM of glucose (N = 3). The following metabolites were quantified: Glucose-6-phosphate (Glc6P), Fructose-6-phosphate (Fru6P), Fructose-1.6-bisphosphate (FruBP), 3-phosphoglycerate and 2-phosphoglycerate (2PG/3 PG), phosphoenolpyruvate (PEP), pentose-5-phosphate (Pent5P), Mannone-6-phosphate, 6-phosphogluconate (6-PG), ribulose-1,5-bisphosphate (RibuBP), sedoheptulose-7-phosphate (Sed7P), pyruvate, lactate, citrate and isocitrate (Cit/Isocit), Succinate and Malate. The metabolite content was expressed as % of the 5.55 mM control conditions. E) Oxygen consumption rate (OCR) was measured using the Seahorse XFe96. Routine respiration, non-phosphorylating respiration (Oligomycin) and uncoupled respiration (CCCP) were determined in human skin fibroblasts grown during 48h in DMEM with 5.55 mM or 12 mM of glucose. Analysis was performed using the Wave 2.6 software (N = 6). The cell passage numbers used in the different experiments was between P6 and P8. F) Determination by Simple WES of AMPK and phospho-AMPK^Thr172 protein content in HDF grown during 48H in DMEM with 5.55 mM or 12 mM of glucose. The ratio of phospho/total AMPK is given in panel F. Protein expression was normalized to total protein content (N = 3). G) Growth curves of HDF grown in DMEM with 5.55 mM or 12 mM of glucose (N = 12). All data are expressed as the mean ± SEM. ∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001. Unpaired t-test was used for panels A,B,C,E and F. Regarding panel D, unpaired t-test was used for each metabolite, while comparing 5.55 mM and 12 mM glucose.

Fig. 3

GDF15 biosynthesis is indispensable for human skin reconstruction. A) The effect of 100 nM GDF15 supplementation was determined using comparative proteomics on HDFs exposed to 12 mM glucose for 48H. The pathway analysis is shown as a bubble volcano plot (significant pathways with -logAdjPvalue>1.3 are shown. The pathways with blue dots are inhibited while pathways with orange dots are activated. The number of proteins detected for each pathway is represented by the diameter of each dot. Activation or inhibition was determined using the Z-score calculated by IPA Qiagen. B) Proteins of the Wound Healing Signaling, AMPK Signaling, Oxidative Phosphorylation or Protein Kinase A Signaling altered by the 12 mM glucose treatment are shown. C) Human skin reconstruction was performed using fiboblasts expressing a shGDF15, wild-type fibroblasts exposed to 12 mM glucose or wild-type fibroblasts exposed to 12 mM glucose and supplemented with 2 nM GDF15. development. Macroscopic view is shown with a scale bar of 0.6 cm, D) Immunofluorescence study of HRS using DAPI marker (blue), Collagen I (yellow) and MKi67 (pink). Scale bar. 50 μM, N = 3. E,F) Migration assay of HDF cultivated in 5.55 mM, 12 mM or 25 mM glucose and HDF cultivated in 12 mM glucose supplemented with 100 nM gdf15, HDF transfected with esiGDF15 or esiTFAM (N = 15). All data were expressed as the mean ± SEM. ∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001. Ordinary one-way ANOVA with Dunett's test correction was used for panel F.

Fig. 4

High glucose stress activates the mitochondrial repressor MTERF3. A) Quantification of the MTERF3 (Mitochondrial Transcription Termination Factor 3) RNA transcript by taqman quantitative PCR in HDF grown in 5.55 mM or 12 mM glucose. Normalization of the data was performed using GusB (β-glucuronidase) (N = 3). B) Determination by Simple WES of MTERF3 protein expression in HDF cultivated 48H in DMEM with 5.55 mM or 12 mM glucose. Total protein was used for normalization. (N = 3). C) Human Reconstructed Skin (HRS) generation process. HDF were seeded on bovine collagen matrix. After 21 days, Human Equivalent Dermis (HED) was produced. Human Epidermal Keratinocytes were seeded on HED. After 38 days, HRS was produced. D) Immunofluorescence analysis of MTERF3 expression in HRS epidermis grown in 5.55 mM or 12 mM glucose. Signal intensity was quantified by measuring MTERF3 fluorescence (purple color) normalized to the number of cells, as determined by the number of nuclei (DAPI staining; blue color). 10 images were acquired per HRS (N = 3). E) Similar analysis was performed in HRS dermis. F) Determination by Simple WES of the protein expression level of various mitochondrial respiratory chain subunits in human dermal fibroblasts (HDF): complex I (NADH:Ubiquinone Oxidoreductase Subunit B8), complex II (Succinate dehydrogenase [ubiquinone] iron-sulfur subunit), complex III (Ubiquinol-Cyt C Reductase Core Protein 2), complex IV (Cyt C Oxidase Subunit 4) and complex V (ATP synthase alpha subunit) in HDF cultivated 48 h in DMEM with 5.55 mM or 12 mM of glucose. Total protein loading was used for normalization (N = 3). Similar analysis was performed in G) Human Equivalent Dermis (HED, N = 3) and H) Human Reconstructed Skin (HRS, N = 3). All data are expressed as the mean ± SEM. ∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001. Unpaired t-test was for all panels.

Fig. 5

MTERF3 ectopic overexpression represses OXPHOS. A) Proteomic study of HDF overexpressing human MTERF3 (MTERF3OE) as compared to wild-type control expressing the empty plasmid p-lenti (N = 4). B) Schematic representation of the mitochondrial proteins reduced in expression (with Adjp<0.05) as a result of the MTERF3 overexpression (MTERF3OE) in HDF. C) Ingenuity Pathway Analysis (Z-score) of the proteins significantly altered in expression, in response to MTERF3 overexpression. D) Oxygen consumption rate (OCR) was measured using the Seahorse XFe96. Routine respiration, non-phosphorylating respiration (oligomycin) and uncoupled respiration (CCCP) were determined in human skin fibroblasts with MTERF3 overexpression, as compared to control cells grown in 5.55 mM or 12 mM of glucose (N = 6). E) Mitochondrial morphology study of HDF overexpressing MTERF3 and control cells grown in DMEM with 5.55 mM. Mitochondrial staining was performed using 50 nM of MitoTracker Red (N = 15). Three parameters of the mitochondrial network were analyzed using Image J: particles count, tubules length and interconnections. F) Quantification of TFAM (Transcription Factor A, Mitochondrial) RNA transcript by Taqman quantitative PCR in HDF overexpressing MTERF3, as compared to plenti-control cells. Normalization of the RT-QPCR data was performed to GusB (β-glucuronidase), N = 3. G) Growth curves of HDF lenti-control and MTERF3 overexpression grown during 7 days in DMEM with 5.55 mM or 12 mM of glucose (N = 12). H) Determination by Simple WES of the protein expression level of mitochondrial respiratory chain subunits: complex I (NADH:Ubiquinone Oxidoreductase Subunit B8), complex II (Succinate dehydrogenase [ubiquinone] iron-sulfur subunit), complex III (Ubiquinol-Cyt C Reductase Core Protein 2), complex IV (Cyt C Oxidase Subunit 4) and complex V (ATP synthase alpha subunit) in HDF control and MTERF3 overexpression cultivated 48 h in DMEM with 5.55 mM or 12 mM of glucose. Protein normalization was performed using total proteins (N = 3). All data are expressed as the mean ± SEM. ∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001. Unpaired t-test was used for panels A,E,F and H.

Fig. 6

NR2F2/Coup-TFII glucose-dependent transcription factor inhibition upregulates MTERF3. A) Promoter region of the human MTERF3 gene and regulatory sites for transcription factors binding. Identification of NR2F2 binding site using Swiss Regulon (Expasy). B) Quantification of NR2F2 (Nuclear Receptor Subfamily 2 Group F Member 2) RNA transcript by taqman quantitative PCR in HDF cultivated in DMEM with 5.55 mM, 12 mM or 25 mM of glucose. Normalization of the data was performed to GusB (β-glucuronidase) (N = 3). C) MTERF3 promoter reporter activity assay using a Gaussia-luciferase (Gluc) lentiviral approach. D) Determination by bioluminescence of the MTERF3 promoter activity in HDF with lentiviral transduction of the reporter. Fibroblasts were grown 24 h in DMEM with 5.55 mM,12 mM or 25 mM of glucose, or 5.55 mM of galactose (N = 4). Data were normalized to the number of cells. E) Dose-dependent relationships between MTERF3 promoter activation and glucose concentration in the cell culture medium. F) Quantification of NR2F2 mRNA transcript by taqman quantitative PCR in HDF transfected with esiControl and esiNR2F2. Normalization of the data to GusB (β-glucuronidase), N = 3. G) MTERF3 promoter activity determination in HDF transfected with esiNR2F2 (N = 4). H) Quantification of MTERF3 mRNA transcript by taqman quantitative PCR in HDF transfected with esiNR2R2. Normalization of data to GusB (β-glucuronidase), N = 3. I) Determination by Simple WES of MTERF3 protein expression level in HDF transfected with esiNR2F2. Protein normalization was performed to total protein loading (N = 3). J) Schematic representation of lentiviral CRISPR guide RNA for the generation of NR2F2 knock-out. K) Quantification of NR2F2 mRNA transcripts by taqman quantitative PCR in stable HDF using sgControl, sgRNA 1 targeting NR2F2 and sgRNA2 targeting NR2F2. Normalization of the data to GusB (β-glucuronidase), N = 3. L) Oxygen consumption rate (OCR) was measured using the Seahorse XFe96. Routine respiration, non-phosphorylating respiration (oligo) and uncoupled respiration (CCCP) were determined in HDF expressing sgControl, sgRNA 1 targeting NR2F2 and sgRNA2 targeting NR2F2. M) Quantification of RNA transcripts by taqman quantitative PCR of COQ9 (coenzyme Q9), TFAM (Transcription Factor A, Mitochondrial) and MT-RNR1 (Mitochondrially Encoded 12S RRNA), in stable HDF sgControl, sgRNA 1 targeting NR2F2 and sgRNA2 targeting NR2F2. Normalization of data to GusB (β-glucuronidase), N = 3. N) Mitochondrial morphology of HDF expressing sgControl, sgRNA 1 targeting NR2F2 and sgRNA2 targeting NR2F2. Imaging was performed using 50 nM of MitoTracker Red (N = 40). O) Schematic representation of the impact of high(12 mM)-glucose medium on NR2F2, MYCN, MTERF3 and OXPHOS. P) Quantification of MTERF3 mRNA transcript by taqman quantitative PCR in HDF transfected with esiMYCN. Normalization of the data to GusB (β-glucuronidase), N = 3. Q) MTERF3 promoter activity assay in HDF transfected with esiMYCN (N = 4). R) Determination by Simple WES of the MTERF3 protein expression level in HDF transfected with esiMYCN (N = 4). All data were expressed as the mean ± SEM. ∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001. Ordinary one-way ANOVA with Dunett's test correction was used for panel B, D, K, L, M and N. Unpaired t-test was used for panels F, G, H, I, P, Q and R.

Fig. 7

cFOS and NR2F2 transcription factors mediate glucose-dependent repression of GDF15 in human dermis. A) Determination by Simple WES of the protein expression level of proGDF15 in wild-type HDF expressing shcontrol or shGDF15 (N = 3). B) Expression of GDF15 in the skin from human protein expression atlas from EMBL-EBI (https://www.ebi.ac.uk) which includes RNA-seq analyses from tissue samples of 122 human individuals, representing 32 different tissues. The results are expressed as TPM (Transcripts Per Kilobase Million). C) Quantification of GDF15 mRNA transcripts by taqman quantitative PCR in HDF, A549 and HEPG2 (N = 3). D) Quantification of GDF15 mRNA transcripts by taqman quantitative PCR in HDF, 786-O and SN005 cells (N = 3). E) ELISA-Based Quantification of GDF15 secretion in HDFs. HDFs were cultured under conditions of normal (5.5 mM) and high (12 mM) glucose concentrations during 48h. GDF15 levels in the culture supernatants were quantified using an enzyme-linked immunosorbent assay (ELISA) following the manufacturer's instructions. F) Quantification of GDF15 mRNA transcripts by taqman quantitative PCR in HDF grown in 5.55 mM, 12 mM or 25 mM glucose (N = 3). G) GDF15 promoter activity in HDF grown 24 h in DMEM with 5.55 mM,12 mM or 25 mM of glucose or 5.55 mM of galactose (N = 4). H) Dose-dependent relationship between GDF15 promoter activation and glucose concentration in the medium. I) GDF15 gene promoter sequence with identification of the binding site for the FOS transcription factor (Swiss Regulon Expasy). J) Quantification of FOS mRNA transcript by taqman quantitative PCR in HDF cultivated in 5.55 mM or 12 mM glucose (N = 3). K) Determination by Simple WES of FOS protein expression level in HDF cultivated in 5.55 mM or 12 mM glucose (N = 3). L-M) Determination by Simple WES of GDF15 protein expression level in HDF transfected with esiFOS (N = 3). N) Quantification of GDF15 mRNA transcript by taqman quantitative PCR in HDF transfected with esiNR2F2 (N = 3). O) Quantification of GDF15 mRNA transcript by taqman quantitative PCR in HDF expressing sgControl, sgRNA 1 targeting NR2F2 and sgRNA2 targeting NR2F2. Normalization of the data to GusB (β-glucuronidase), N = 3. P–S) Quantification of ATF3, ATF4, CHOP and P53 mRNA transcripts by taqman quantitative PCR in HDF expressing siCTRL and esiNR2F2 in 5.5 mM glucose or 12 mM glucose growth medium. Normalization of the data to GusB (β-glucuronidase), N = 3. T) Schematic representation of the NR2F2-MTERF3-GDF15 axis and its control on OXPHOS function in response to glucose stress. All data are expressed as the mean ± SEM. ∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001. Ordinary one-way ANOVA with Dunett's test correction was used for panel C, E and N. Unpaired t-test was used for panels A, B, H, I, J, K, L and M.

Fig. 8

GDF15 inhibition by hyperglycemia or shRNA alters mitochondrial biogenesis. A) Metabolomic profile of HDF grown 48H in DMEM with 5.55 mM or 12 mM + 100 nM GDF15 of glucose and HDF expressing a shGDF15 cultivated in 5.55 mM glucose (N = 3). B) Oxygen consumption rate (OCR) was measured using the Seahorse XFe96. Routine respiration and uncoupled respiration (CCCP) were determined in HDF grown in 5.55 mM glucose and HDF expressing shGDF15 grown in 5.55 mM glucose or 5.55 mM glucose supplemented with 100 nM gdf15. C) Oxygen consumption rate (OCR) was measured using the Seahorse XFe96. Routine respiration and uncoupled respiration (CCCP) were determined in HDF grown in 5.55 mM glucose and supplemented with low doses of gdf15: 20pM, 80pM, 1 nM, 10 nM and 100 nM. D) Mitochondrial respiratory chain proteins (gene loci) specifically activated at the level of chromatin accessibility by GDF15 100 nM. E-H) Quantification of mRNA transcripts by taqman quantitative PCR for NR2F2, GDF15, MTERF3 and TFAM in HDF cultivated with 5.55 mM glucose or 5.55 mM glucose supplemented with 100 nM gdf15. Normalization of the data was performed to GusB (β-glucuronidase), N = 3. I-L) Quantification of mRNA transcripts by taqman quantitative PCR for GDF15, MAPK1, MAPK3, PGC1α (Peroxisome proliferator-activated receptor-gamma coactivator 1 alpha) in HDF cultivated with 5.55 mM glucose supplemented with low doses of gdf15. Normalization of the data was performed to GusB (β-glucuronidase), N = 3 M) Quantification of TFAM mRNA transcripts by taqman quantitative PCR in HDF cultivated with 5.55 mM, 12 mM or 25 mM glucose or in HDF expressing shGDF15 grown in 5.55 mM glucose. Normalization of data was performed to GusB (β-glucuronidase), N = 3. N) Quantification of PGC1α mRNA transcript by taqman quantitative PCR in HDF cultivated with 5.55 mM, 12 mM or 25 mM glucose or in HDF expressing shGDF15 grown in 5.55 mM glucose. Normalization of the data to GusB (β-glucuronidase), N = 3. O) Quantification of PGC1α mRNA transcript by taqman quantitative PCR in HDF transfected with esiNR2F2. Normalization of data to GusB (β-glucuronidase), N = 3. P,Q) Quantification of the total Coenzyme Q10 (oxidized and reduced forms) in HDF cultivated with 5.55 mM, 12 mM or 12 mM glucose medium supplemented with 100 nM gdf15. Analysis was also performed in HDF expressing shGDF15 in 5.55 mM glucose. R) Quantification of TFAM in HDF cultivated with 5.55 mM glucose medium or medium supplemented with 20pM, 100 pM and 100 nM rGDF15. S) Summary of the regulatory network linking TFAM, PGC1α, COQ9 and COQ10. All data are expressed as the mean ± SEM. ∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001. Ordinary one-way ANOVA with Dunett's test correction was used for panel A, B, F, G, I-M, P–R. Unpaired t-test was used for panels E, H and O.