Metformin alleviates human cellular aging by upregulating the endoplasmic reticulum glutathione peroxidase 7

Abstract

Metformin, an FDA-approved antidiabetic drug, has been shown to elongate lifespan in animal models. Nevertheless, the effects of metformin on human cells remain unclear. Here, we show that low-dose metformin treatment extends the lifespan of human diploid fibroblasts and mesenchymal stem cells. We report that a low dose of metformin upregulates the endoplasmic reticulum-localized glutathione peroxidase 7 (GPx7). GP×7 expression levels are decreased in senescent human cells, and GPx7 depletion results in premature cellular senescence. We also indicate that metformin increases the nuclear accumulation of nuclear factor erythroid 2-related factor 2 (Nrf2), which binds to the antioxidant response elements in the GPX7 gene promoter to induce its expression. Moreover, the metformin-Nrf2-GPx7 pathway delays aging in worms. Our study provides mechanistic insights into the beneficial effects of metformin on human cellular aging and highlights the importance of the Nrf2-GPx7 pathway in pro-longevity signaling.

Keywords: aging; glutathione peroxidase 7; metformin; nuclear factor erythroid 2-related factor 2; oxidative stress; senescence.

Figure 1

Metformin delays replicative senescence of human diploid fibroblasts (HDFs). (a) Cumulative population doubling (CPD) analysis of HDFs proliferation in the absence (Ctrl) or presence of 100 μm metformin (Met). (b) Left: senescence‐associated β‐galactosidase (SA‐β‐Gal) staining of HDFs at late passages (LP, P40–P45). Scale bar = 100 μm. Right: statistical analysis of the percentages of SA‐β‐Gal‐positive cells. (c) Left: KI67 expression in HDFs at LP. Scale bar = 20 μm. Right: statistical analysis of the percentages of KI67‐positive cells. Data were represented as mean ± SEM from three biological replicates, n > 200 cells per condition. *p < .05, **p < .01, via two‐tailed Student's t test

Figure 2

Glutathione peroxidase 7 (GPx7) is a key regulator in the senescence of human diploid fibroblasts (HDFs). (a) Relative mRNA level of redox‐related genes in HDFs (P30) induced by 200 μm metformin (Met) for 6 hr analyzed by RT–qPCR, normalized to control cells. Data were represented as mean ± SEM from three technical replicates, *p < .05, **p < .01, via two‐tailed Student's t test. Results are representative of three independent experiments. (b) Protein expression of ER‐localized oxidoreductases in HDFs (P30) treated without or with 200 μm metformin for 24 hr. (c) Upper :GPx7 in HDFs (P30) treated with indicated concentrations of metformin for 12 hr. Lower: statistical analysis of the expression of GPx7. Data were represented as mean ± SEM from four biological replicates. *p < .05, **p < .01, via two‐tailed Student's t test. (d) GPx7 in HDFs (P36) treated without or with 100 μm metformin since P25. (e) Expression of ER‐localized peroxidases in HDFs at both early passages (EP) and LP. (f) Cumulative population doubling (CPD) analysis of HDFs proliferation transduced with lentiviral shRNA control (shCtrl) and shRNA targeting GPX7 (shGPX7). Inset: GPx7 expression in HDFs transduced with lentivirus. (g) CPD analysis of shCtrl or shGPX7 HDFs in the absence (Ctrl) or presence of 100 μm metformin. Data were represented as mean ± SEM from three biological replicates, **p < .01, n.s., not significant, via two‐way ANOVA, Tukey's multiple comparisons test. (h) Left: KI67 expression in HDFs from (g) at P37. Scale bar = 20 μm. Right: statistical analysis of the percentages of KI67‐positive cells. Data were represented as mean ± SEM from three biological replicates, n > 200 cells per condition. *p < .05, **p < .01, n.s., not significant, via two‐way ANOVA, Tukey's multiple comparisons test. ER, endoplasmic reticulum

Figure 3

Metformin upregulates glutathione peroxidase 7 (GPx7) through activation of nuclear factor erythroid 2‐related factor 2 (Nrf2). (a) Left: immunofluorescence analysis of the translocation of Nrf2 to nucleus in human diploid fibroblasts (HDFs) treated with or without 100 μm metformin (Met) for 4 hr. Scale bar = 10 μm. The nuclei were outlined in dotted white circles. Right: statistical analysis of the fluorescence intensity of Nrf2 localized in the nucleus. Data were represented as mean ± SEM, n = 7 nuclei per condition. **p < .01, via two‐tailed Student's t test. Results are representative of two independent experiments. (b) Cytosolic GPx7 and nuclear Nrf2 in HDFs treated with indicated concentration of metformin for 6 hr. α‐Tubulin and Lamin B1 were used as cytoplasmic and nuclear loading controls, respectively. (c) Left :GPx7 expression in HDFs transduced with shCtrl or shNRF2 lentivirus, treated with or without 100 μm metformin for 12 hr; right: the efficiency of NRF2 depletion. (d) Left :GPx7 protein levels in HDFs transduced with GFP (Ctrl), wild‐type NRF2 (WT), or constitutively activated NRF2 (CA) lentivirus; Right :GPx7 protein levels in HDFs stimulated without or with 200 μm tBHQ for 12 hr. (e) Luciferase reporter assay measuring GPX7 promoter activity. HEK 293T cells were transfected with luciferase reporter construct (pGL3‐Basic) containing antioxidant response elements (ARE) of HO‐1 or a series of 5′ fragments (F1–F6) of GPX7 promoter as indicated, in combination with either pcDNA3.1‐NRF2 or empty vector (EV). The relative luciferase activities were normalized relative to EV transfection. Data were represented as mean ± SEM from three technical replicates, **p < .01 via two‐tailed Student's t test. Results are representative of three independent experiments. (f) ChIP‐qPCR analysis of the Nrf2 occupancy on GPX7 promoter. Upper: localization of the ARE‐containing sites (S1) and one nonspecific site (S2) at the GPX7 promoter. Lower: ChIP‐qPCR analysis was performed with IgG or anti‐Nrf2 antibody in HDFs with or without 100 μm metformin treatment for 36 hr. Enrichment values were normalized to input and shown as the fold changes relative to IgG group. Data were represented as mean ± SEM from six technical replicates, **p < .01, n.s., not significant, via two‐way ANOVA, Tukey's multiple comparisons test. (g) Electrophoretic mobility shift assays (EMSA) analysis of the binding between Nrf2 and GPX7‐ARE. 5′‐biotin‐labeled GPX7‐ARE from S1 was incubated with HDFs nuclear extracts (NE) in the absence or presence of fivefold or 50‐fold excess of unlabeled HO‐1‐ARE or mutated HO‐1‐ARE (HO‐1‐mARE). The open arrow indicates the complex of endogenous Nrf2 with GPX7‐ARE, and the filled arrow indicates the ternary complex of Nrf2, GPX7‐ARE, and anti‐Nrf2 antibody.

Figure 4

Glutathione peroxidase 7 (GPx7) plays a key role in defencing the oxidative stress in human diploid fibroblasts (HDFs). (a, b) Cell viability analysis of the proliferation of HDFs at P35 (a) or at P36 (b) in the absence or presence of 1 mm paraquat (PQ) for 24 hr. The HDFs were stably transfected with shCtrl, shGPX7, GFP, GPX7 lentivirus as indicated. Data were represented as mean ± SEM from three technical replicates, **p < .01 via two‐way ANOVA, Tukey's multiple comparisons test. Results are representative of two independent experiments. (c) Western blotting to detect the efficiency of knockdown of nuclear factor erythroid 2‐related factor 2 (Nrf2) and overexpression of GPx7 by lentivirus as indicated in HDFs.(d, e) Left: SA‐β‐Gal staining (d) and KI67 expression (e) of HDFs at P28. Scale bar = 50 and 20 μm in (d) and (e), respectively. Right: Statistical analysis of the percentages of SA‐β‐Gal‐positive cells and KI67‐positive cells is illustrated. Data were represented as mean ± SEM from three biological replicates, n > 200 cells per condition,**p < .01 via two‐way ANOVA, Tukey's multiple comparisons test

Figure 5

Metformin‐Nrf2‐GPx7 pathway functions in human mesenchymal stem cells (HMSCs). (a) Protein expression of nuclear factor erythroid 2‐related factor 2 (Nrf2) and glutathione peroxidase 7 (GP×7) in human diploid fibroblasts (HDFs) and HMSCs at EP and LP, and in WRN ^+/+ and WRN ^−/− HMSCs at P5. For HMSCs, P3–P5 are taken as EP and P9–P12 as LP. (b) Nrf2 and GPx7 in wild‐type HMSCs (NRF2 ^+/+) and HMSCs genetically bearing an endogenous NRF2 (A254G) variation (NRF2 ^{AG / AG}). (c, f) Cumulative population doubling (CPD) analysis of HMSCs proliferation. Wild‐type HMSCs from P3 to P12 were continuously cultured in the absence or presence of 100 μm metformin (Met) (c). Inset: the expression of GPx7 at P7. HMSCs transduced with shCtrl and shGPX7 lentivirus were continuously cultured from P4 to P9 (f). Inset :GPx7 expression in HMSCs transduced with lentivirus. (d, g) Left :SA‐β‐Gal staining of wild‐type HMSCs at LP (d) and lentivirus‐transduced HSMCs at P7 (g), respectively. Scale bar = 100 μm. Right: statistical analysis of the percentages of SA‐β‐Gal‐positive cells. (e, h) Left :KI67 expression in wild‐type HMSCs at LP (e) and lentivirus‐transduced HMSCs at P7 (h), respectively. Scale bar = 20 μm. Right: statistical analysis of the percentages of KI67‐positive cells. Data were represented as mean ± SEM from three biological replicates, n > 200 cells per condition. *p < .05, **p < .01, via two‐tailed Student's t test. (i, j) Measurement of the in vivo retention of transplanted HMSCs by in vivo imaging system. Left: Conditioned HMSCs with serial metformin administration (i) or transduced with shRNA lentivirus (j) overexpressing luciferase were implanted into the left and right tibialis anterior muscles of immune‐deficient mice, respectively. Photon flux was captured on the fifth or sixth day after implantation. Right: Statistical analysis of each mouse implanted with HMSCs with the relative luminescence intensities normalized to log₂ fold. Data were presented as scatter dot plots displaying the mean ± SEM, from eleven (i) or four (j) biological replicates. *p < .05, **p < .01, via two‐tailed Student's t test.

Figure 6

The Metformin‐SKN‐1‐GPX‐6 pathway functions in Caenorhabditis elegans. (a) Left: mcherry was translationally fused to the CDS encoding gpx‐6 and gfp was translationally fused to the genomic DNA encoding tram‐1 or mitochondrial signal peptide (mito), respectively. The expression was driven by the y37a1b.5 promoter. Right: Transgenic worms simultaneously expressed GPX‐6 [Py37a1b.5::gpx‐6::mcherry] (red channel) and TRAM‐1 (upper) [Py37a1b.5::tram‐1::gfp] (green channel) or Mito‐GFP (lower) [Py37a1b.5::mito::gfp]. Scale bar = 5 μm. (b) Human glutathione peroxidase 7 (GPx7) and C. elegans GPX‐6 peroxidase activities were measured by the decrease in absorbance at 340 nm due to NADPH consumption. Inset: purification of the recombinant human PDI, human GPx7, and C. elegans GPX‐6. (c) Upper: gfp was translationally fused to the full‐length genomic DNA of gpx‐6 including the 2,876‐bp promoter regions. Lower: Images and quantification of the worms expressing GPX‐6::GFP subjected to vehicle (Ctrl) or 50 mm metformin (Met) for 24 hr post‐L4 larval stage. Scale bars = 100 μm. Data were represented as mean ± SEM, n ≥ 14 worms per condition. **p < .01, via one‐way ANOVA, Tukey's multiple comparisons test. (d) Worms expressing GPX‐6::GFP without or with skn‐1 RNAi were subjected to 50 mm metformin for 48 hr post‐L4 larval stage, imaged, and quantified. Scale bars = 100 μm. Data were represented as mean ± SEM, n > 15 worms per condition. **p < .01 via two‐way ANOVA, Tukey's multiple comparisons test. Results are representative of two independent experiments. (e) Survival curves of WT or gpx‐6 RNAi worms treated without or with 50 mm metformin. Results are representative of four independent experiments. The mean lifespan is 16.1 and 18.7 days for vector fed worms raised on 0 and 50 mm metformin, respectively, and the survival curves of the two groups are significantly different (p = .0012 via the log‐rank test). The mean lifespan is 13.8 and 14.6 days for gpx‐6 RNAi fed worms raised on 0 and 50 mm metformin, respectively, and the survival curves of the two groups are not significant (p = .2455 via the log‐rank test) (see also Table S1).

Figure 7

A model illustrating metformin delays aging through the Nrf2‐GPx7 pathway. Nuclear factor erythroid 2‐related factor 2 (Nrf2) is a master transcription factor for modulating cellular antioxidant responses. Through binding onto the antioxidant response elements (ARE), Nrf2 stimulates the expression of a wide arrays of antioxidant enzymes, among which glutathione peroxidase 7 (GPx7) is a unique ER‐localized peroxidase. In young cells, sufficient Nrf2 transcriptionally induces GPx7 expression to defend oxidative stress. In old cells, Nrf2 and GPx7 expression decreases and oxidative stress accumulates. Low‐dose metformin can promote the nuclear translocation of Nrf2 to upregulate the expression of GPx7 to alleviate cellular aging. ER, endoplasmic reticulum