Reduced sulfatide content in deferoxamine-induced senescent HepG2 cells

Abstract

Iron chelators, such as deferoxamine, exert an anticancer effect by altering the activity of biomolecules critical for regulation of the cell cycle, cell metabolism, and apoptotic processes. Thus, iron chelators are sometimes used in combination with radio- and/or chemotherapy in the treatment of cancer. The possibility that deferoxamine could induce a program of senescence similar to radio- and/or chemotherapy, fostering adaptation in the treatment of cancer cells, is not fully understood. Using established biochemical techniques, biomarkers linked to lipid composition, and coherent anti-Stokes Raman scattering microscopy, we demonstrated that hepatocellular carcinoma-derived HepG2 cells survive after deferoxamine treatment, acquiring phenotypic traits and representative hallmarks of senescent cells. The results support the view that deferoxamine acts in HepG2 cells to produce oxidative stress-induced senescence by triggering sequential mitochondrial and lysosomal dysfunction accompanied by autophagy blockade. We also focused on the lipidome of senescent cells after deferoxamine treatment. Using mass spectrometry, we found that the deferoxamine-induced senescent cells presented marked remodeling of the phosphoinositol, sulfatide, and cardiolipin profiles, which all play a central role in cell signaling cascades, intracellular membrane trafficking, and mitochondria functions. Detection of alterations in glycosphingolipid sulfate species suggested modifications in ceramide generation, and turnover is frequently described in cancer cell survival and resistance to chemotherapy. Blockade of ceramide generation may explain autophagic default, resistance to apoptosis, and the onset of senescence.

Keywords: CARS; Cancer cell senescence; Cardiolipin; Deferoxamine; HepG2; Lipidomics; MALDI; Mass spectrometry; Phosphoinositol; Sulfatide.

Fig. 1

DFO treatment of HepG2 cells induces either apoptosis or extensive morphological modifications, with the surviving cells losing their 3D structure and undergoing cell cycle arrest and DNA damage. (A) Representative pictures of HepG2 cells before and after DFO treatment acquired with an optical microscope (scale bar = 100 µm) and a confocal microscope after staining with DAPI (3D visualization; scale bar = 50 µm). Enlarged cells after 7 days of DFO treatment are delineated with a discontinuous black line; red arrows indicate apoptotic cells. (B) Protein expression levels of PARP, yH2AX, and survivin in untreated and treated HepG2 cells determined by immunoblotting. Graphs representing the immunoblots bands quantification are shown in Supplementary File 2, Figs. S3, S4, S7, S8. (C) Cell cycle analysis of untreated, 72 h DFO-treated, and 7-day DFO-treated HepG2 cells. (D) Representative images of control and 7-day DFO-treated HepG2 cells stained for phospho-H2AX foci (green) and DAPI (blue). Cells were observed and imaged using a confocal microscope. Scale bar = 25 µm.

Fig. 2

DFO treatment triggers the development of cellular senescence with alteration of the lysosomal compartment. (A) Representative images of untreated and 7-day DFO-treated HepG2 cells stained for β-galactosidase activity with X-Gal (blue). Cells were observed and imaged using an optical microscope. Scale bar = 100 µm. (B) Representative images of control and 7-day DFO-treated HepG2 cells stained for lysosomes (red), LC3B (green), and DAPI (blue). Cells were observed and imaged using a confocal microscope. Scale bar = 25 µm. (C) Dot plot showing the changes in the average fluorescence intensities derived from Alexa Fluor 546-bound anti-LC3B and LysoTracker Deep Red in control vs. DFO-treated HepG2 cells. The cells were visualized with a confocal microscope, and the 3D images of 20 cells for each condition were randomly chosen and analyzed using ImageJ software. The fluorescence intensities relative to lysosomes and LC3B were both significantly increased after treatment with DFO. * **p˂0.001, °°°p˂0.001 (comparing 7-day DFO-treated HepG2 cells to their corresponding controls). (D) Protein expression levels of LC3B and ferritin in untreated HepG2 cells and HepG2 cells after 7 days of DFO treatment according to immunoblotting. Graphs representing the immunoblots bands quantification are shown in Supplementary File 2, Figs. S5, S6, S8.

Fig. 3

DFO leads to altered morphology and function of the mitochondria network. (A) Representative images of control and 7-day DFO-treated HepG2 cells stained for mitochondria (red) and DAPI (blue). In gray are the details of the mitochondria network for each condition. Borders of the cells are delineated with a discontinuous white line. Cells were observed and imaged using a confocal microscope. Scale bar = 25 µm. (B) Amount of ATP produced by control cells and 72 h and 7-day DFO-treated HepG2 cells. *** p < 0.001 (comparing the control group and the two treated groups and between the two treated groups). (C) Representative images of control and 7-day DFO-treated HepG2 cells stained for mitochondria (red), LC3B (green), and DAPI (blue). Cells were observed and imaged using a confocal microscope. Scale bar = 25 µm.

Fig. 4

DFO induces an increase in lipid droplet size. (A) Lipid droplets within control and 7-day DFO-treated HepG2 cells stained with Oil Red O. Cells were observed and imaged using an optical microscope. Scale bar = 50 µm. (B) CARS images of lipid droplets within control cells and 72 h and 7-day DFO-treated HepG2 cells. Scale bar = 10 µm. (C) Dot plot showing the percentage of lipid droplet area/cell area and lipid droplet concentrations (a.u.) in control vs. 72 h and 7-day DFO-treated HepG2 cells imaged using CARS. For % lipid droplets area/cell area: raw CARS images at 2850 cm^-1 were thresholded, and then the area above the threshold was normalized over the total cell area (from relative light transmission images). The threshold was the same for all control and DFO cells and was chosen to distinguish lipid accumulations from the cytosol signal. p = 0.01 with U-Mann Whitney test. For lipid droplet concentrations: raw CARS images at 2850 cm^-1 were thresholded, and then the square root of the mean CARS signal above the threshold was computed, which is linearly proportional to the concentration of lipid droplets in the focal volume. The threshold was the same for all control and DFO cells and was chosen to distinguish lipid accumulations form the cytosol signal (p > 0.05, U-Mann Whitney test).

Fig. 5

MALDI analysis of lipid extracts from HepG2 cells. (A) Volcano plots of changes in molecular species of lipids from control vs. 72 h (upper panel) and 7-day (lower panel) DFO-treated HepG2 cells. Grey values show lipid species with non-significant changes. Negative values (to the left) indicate upregulated lipid species in control samples (p < 0.05), whereas positive values (to the right) reflect downregulated lipids in control samples (p < 0.05). −log10 p = −log10 nominal P-value; log2FC = log2 fold changes. Lipids are colored according to the species, NA (not available) indicates lipid species that were not identified. (B) Magnification of m/z 800–1100 range acquired in negative-ion mode, including PI and ST regions (m/z 900–1100) indicated by bars.

Fig. 6

Magnification of the m/z 1300–1500 range acquired in negative-ion mode including CL species and relative identification using the LIFT method. (A) The species of CLs with the same acyl chain length were categorized in groups. CLs in HepG2 cells had three major groups: CL66, CL68, CL70. (B) Product ion spectra obtained from MS/MS of m/z 1371.0 (Δppm −51) in negative ion mode. Key product ion structures with corresponding m/z are shown. In addition to characteristic product ions of fatty acids, such as FA 16:1 (m/z 252.5) and FA 18:1 (m/z 280.5), fragments of the phosphatidyl part (PA) of CL gave information about the FA combination. In this case, the prevailing fragments m/z 642.4 (PA 16:1_16:1) and 670.4 (PA 16:1_18:1) are characteristic of CL 66:4. The peak at m/z 388.3 represents loss of FA 18:1 from the sn-2 position of the 16:1_18:1 PA structure. The fragmentation pattern is consistent with the structure 16:1_16:1/16:1_18:1. (C) Product ion spectra obtained from MS/MS of m/z 1399.0 (Δppm −42) in negative ion mode. Key product ion structures with corresponding m/z are shown. In addition to characteristic product ions of fatty acids, such as FA 16:1 (m/z 252.4) and FA 18:1 (m/z 280.3), fragments of the phosphatidyl part (PA) of CL gave information about the FA combination. In this case, the prevailing fragments m/z 670.3 and 806.2 are characteristic of CL 68:4 with the combination of FA 16:1 and FA 18:1 (PA 16:1_18:1) and diacylglyceride (DAG) moiety 16:1_18:1, respectively. The peak at m/z 388.2 represents loss of FA 18:1 from the sn-2 position of the 16:1_18:1 PA structure. The fragmentation pattern is consistent with the structure 16:1_18:1/16:1_18:1.

Fig. 7

Lipid features of DFO-treated HepG2 cells and controls. (A) PLS-DA score plots of the top most differential lipid-related m/z values obtained by the PLS-DA utility of MetaboAnalyst software for control HepG2 cells (blue) and 72 h (green) and 7-day DFO-treated cells (red). The ellipses showing the 95% confidence limits of a normal distribution for each spectra group (N = 5) have been marked in the respective colors. PLS-DA was applied to the log-transformed and Pareto scaled dataset. (B) Hierarchical clustering (heatmap) was performed on the normalized data (distance measured using Euclidean and clustering algorithm using Complete). Each colored cell on the map corresponds to an m/z intensity value, with samples in columns and m/z values in rows.

Fig. 8

Imaging of lipids using MALDI. Untreated and 72 h DFO-treated HepG2 cells were seeded on an ITO slide and imaged by MALDI-TOF in negative ion mode (spatial resolution: 50 µm). (A) TIFF images of HepG2 spheroids on the ITO slide taken prior to 9AA matrix application. (B) Distribution of PIs (m/z: 885.1, 861.2, and 909.1) with relative density plots showing the median intensities. Blue dots represent the spectra in which the intensities of the given m/z interval are between the lower and upper quartiles, and red dots represent outliers. (C) Average mass spectrum of representative SHex₂Cer species (m/z: 1036.2, 1052.4, 1066.3). (D) CLs (m/z 1393.2, 1421.2, 1448.2, 1473.1). Total ion current (TIC) normalization was used. Color bars: 0–100% relative intensity. Scale bar = 500 µm.