Depletion of growth differentiation factor 15 (GDF15) leads to mitochondrial dysfunction and premature senescence in human dermal fibroblasts

Abstract

Growth differentiation factor 15 (GDF15) is a stress-responsive cytokine also known as a mitokine; however, its role in mitochondrial homeostasis and cellular senescence remained elusive. We show here that knocking down GDF15 expression in human dermal fibroblasts induced mitochondrial dysfunction and premature senescence, associated with a distinct senescence-associated secretory phenotype. Fibroblast-specific loss of GDF15 expression in a model of 3D reconstructed human skin induced epidermal thinning, a hallmark of skin aging. Our results suggest GDF15 to play a so far undisclosed role in mitochondrial homeostasis to delay both the onset of cellular senescence and the appearance of age-related changes in a 3D human skin model.

Keywords: GDF15; lipofuscin; mitochondria; mitokine; senescence; skin aging.

FIGURE 1

GDF15 knockdown in HFF induces premature senescence associated with mitochondrial dysfunction. HFF were transduced with lentiviral vectors carrying GDF15 shRNA or control shRNA and grown under selection. (a) GDF15KD and control cells were cultured over an extended period of time, and cPDL were calculated. Defined measurement intervals that summarize cells across certain passages are indicated as EP = early passage, EMP = early‐middle passage, LMP = late‐middle passage, LP = late passage; (b) relative GDF15 mRNA levels were estimated by qRT‐PCR. (c) Secretion of GDF15 protein was measured using ELISA. For each time point, (d) relative cell surface area and (e) percentage of SA‐β‐Gal‐positive cells were determined. (f) Representative pictures of SA‐β‐Gal staining are shown. (g) Autofluorescence of GDF15KD and control cells was recorded by confocal live‐cell imaging. Representative pictures are shown. Scale bar = 25 μm; CM‐H2XRos (h) and JC‐1 (i) staining were used to measure mitochondrial ROS and mitochondrial membrane potential using flow cytometry, respectively. Data present mean values ± SD, n = 3; (J) OCR of GDF15KD and control cells were measured using Seahorse Flux Analyzer after the successive injection of oligomycin, FCCP, and a mixture of antimycin A and rotenone, as indicated. Data present mean values ± , n = 4; (k) cells were processed for IF using a primary antibody against Complex V. Representative pictures are shown. Scale bar = 25 μm; statistical analysis was calculated using t‐test (*p < 0.05, **p < 0.01, ***p < 0.001)

FIGURE 2

GDF15 knockdown induces a distinct SASP in HFF. (a–d) Relative mRNA levels were estimated by qRT‐PCR in control and GDF15KD HFF for (a) IL1a and several MMPs, including (b) MMP3, (c) MMP10, and (d) MMP12. Data present mean ± SD, n = 3; (e) MMP arrays were performed with supernatant from GDF15KD, and control HFF and data were summarized in a heatmap. Data present mean values, n = 4; statistical analysis was calculated using t‐test (*p < 0.05, **p < 0.01, ***p < 0.001)

FIGURE 3

GDF15 knockdown changes SASP composition of HSDF and reduces epidermal thickness in 3D skin equivalents. HSDF were transduced with lentiviral vectors carrying GDF15 shRNA or control shRNA and grown under selection. (a) The percentage of SA‐β‐Gal‐positive cells was estimated. (b) Skin equivalents containing either GDF15KD or control HSDF were stained with H&E. (c) Epidermal thickness [μm] was measured using ImageJ software. Data present mean values ± SD, n = 4; 50 to 55 epidermal sites (excluding stratum corneum) were measured per skin equivalent. N = 9 (3 different donors); (c–e) relative mRNA levels were estimated by qRT‐PCR in control and GDF15KD HFF for MMP1 (d) and IL‐6 (e). Data present mean values ± SD, n = 3; statistical analysis was calculated using t‐test (*p < 0.05, **p < 0.01, ***p < 0.001)