MNPs@SiO2(RITC)-induced lipid peroxidation decreases cell membrane fluidity
Here, we analyzed changes in lipid peroxidation and membrane fluidity in HEK293 cells treated with MNPs@SiO2(RITC) at 0.1 [an adequate MNPs@SiO2(RITC) concentration for cell labelling] and 1.0 μg/µl [a plateau concentration for the uptake of MNPs@SiO2(RITC)] [4, 17] for 12 h, using total internal reflection fluorescence microscopy (TIRFM). Furthermore, changes in cell membrane fluidity after MNPs@SiO2(RITC) treatment were investigated by measuring 6-dodecanoyl-2-dimethylaminonaphthalene (laurdan) generalized polarization (GP) values using TIRFM (Fig. 1a). The number of high-GP areas on the cell surface, corresponding to rigid domains, increased with MNPs@SiO2(RITC) treatment; in particular, abundantly distributed regions of MNPs@SiO2(RITC) primarily co-localized with high GP-distributed regions at a GP scale of -1.0 to 1.0 (Fig. 1b). GP frequency distribution values of treated cells were subtracted from the corresponding values of non-treated control cells to obtain frequency difference curves (Fig. 1c) and total mean GP values (Fig. 1d). A similar trend was observed in the relative levels of peroxidised lipid (Fig. 1e).
Moreover, a previous study revealed that MNPs@SiO2(RITC) induced intracellular ROS generation via mitochondria dysfunction in HEK293 cells, with ROS generation from the shell of MNPs@SiO2(RITC) and silica nanoparticles (silica NPs) rather than from the cobalt ferrite core when treating cells for 12 h [4, 17]. Thus, to determine intracellular ROS generation following MNPs@SiO2(RITC) treatment, we performed 2',7'-dichlorodihydrofluorescein diacetate staining in HEK293 cells treated with 50 nm-sized silica nanoparticles, which comprise the same material and size of the shell of MNPs@SiO2(RITC). The intracellular ROS level was increased by over 50% upon 1.0 µg/µl silica-NP treatment compared to that in non-treated control and 0.1 µg/µl silica-NP-treated cells (See Supplementary Figure 1, Additional File 1), a finding consistent with that of previous reports [4, 17]. These results indicated that the rigid regions in the plasma membrane increased through lipid peroxidation induced by ROS generated from the shell of MNPs@SiO2(RITC).
MNPs@SiO2(RITC)-treatment decreases cell polarity and spread area but increases traction force
As membrane fluidity is closely related to cell morphology as well as focal adhesion [32, 33], we investigated whether a decrease in membrane fluidity mediated by MNPs@SiO2(RITC) also affects cell morphology and focal adhesion. We then evaluated the effects of MNPs@SiO2(RITC) on cell polarity as local cell contraction is tightly associated with these activities along with changes in focal adhesion [34]. Moreover, the cell aspect ratio was measured. Images of cells and submicron pillars at 12 h were analysed after cell seeding (Fig. 2a, b). The aspect ratio of the cells at 0.1 µg/µl MNPs@SiO2(RITC) did not significantly differ from that of the non-treated control cells, although the ratio of the cells treated with 1.0 µg/µl MNPs@SiO2(RITC) was significantly smaller than that of the non-treated control cells (Fig. 2c).
Pillar deflection in the magnified images was used to measure pillar displacement (Fig. 2d) and calculate traction force (Fig. 2e, f). To calculate pillar traction force, the displacement of each pillar was multiplied with the pillar bending stiffness [27]. In particular, 1.0 µg/µl MNPs@SiO2(RITC)-treated cells showed a significant increase in pillar displacement (351 ± 65 nm; mean ± SD), which was significantly higher than that of the non-treated control cells (216 ± 46 nm). Average (8.5 ± 2 nN) and total (1112 ± 197 nN) traction forces of 1.0 µg/µl MNPs@SiO2(RITC) cells were significantly higher than those (5.2 ± 1 and 855 ± 172 nN) of the non-treated control cells, indicating that cell traction force was affected by 1.0 µg/µl MNPs@SiO2(RITC). Taken together, the increase in cell traction force results from MNPs@SiO2(RITC) treatment-induced reduction in cell polarity and spread area.
MNPs@SiO2(RITC)-treatment impairs cell movement
We previously analysed the effect of MNPs@SiO2(RITC) treatment on the migratory activity of human bone marrow-derived mesenchymal stem cells (hBM-MSCs). The impairment of migratory activity was observed in hBM-MSCs using conventional assays, such as wound healing and invasion assays [35]. To evaluate the biophysical changes related to the biological functions of HEK293, we analysed the effect of 0.1 or 1.0 µg/µl MNPs@SiO2(RITC) on HEK293 cell movement using conventional assays. In the wound healing assay, no difference was observed in terms of the migratory activity between MNPs@SiO2(RITC)-treated HEK293 cells and non-treated control cells (Fig. 3a). Similarly, no difference was observed in invasion ability, analysed using transwell invasion assay, between MNPs@SiO2(RITC)-treated HEK293 cells and non-treated control cells (Fig. 3b).
The aforementioned results of the wound healing and invasion assays do not exclude the cell growth effect and treatment of growth-arrest agents such as mitomycin C, as the assays are highly toxic to HEK293 cells [36]. Thus, we analysed individual cells’ movement by tracking cells on pillars for 24 h after treating them with MNPs@SiO2(RITC) for 12 h on a dish (Additional File 2: Movie S1, Movie S2, and Movie S3). The distances traveled by the cells were significantly decreased to a greater extent for MNPs@SiO2(RITC)-treated HEK293 cells than for non-treated control at 6 and 24 h (Fig. 3c). Similar result was observed for movement speeds at 6 h and 24 h (Fig. 3d).
MNPs@SiO2(RITC)-treatment decreases intracellular ATP level
To evaluate the changes in intracellular ATP level of MNPs@SiO2(RITC)-treated cells, HEK293 cells were treated with MNPs@SiO2(RITC) at 0 to 2.0 µg/µl for 6, 12, and 24 h (Fig. 4). Intracellular ATP levels were decreased in a dose-dependent manner in MNPs@SiO2(RITC)-treated cells starting from 0.13 µg/µl dose. Moreover, the decrement pattern was similar at 6, 12, and 24 h treatment.
Metabotranscriptomic network of MNPs@SiO2(RITC)-treated HEK293 cells
To analyse the dispersed phenomena of MNPs@SiO2(RITC)-treated cells, a co-expression network of gene and metabolites was constructed using a transcriptome generated from microarray analysis and the metabolome derived from amino acid and organic acid profiling using Ingenuity Pathway Analysis (IPA, http://www.ingenuity.com) [17]. Microarray expression analysis showed that the levels of 21 genes associated with lipid peroxidation and those of 31 genes related to focal adhesion formation were altered in MNPs@SiO2(RITC)-treated cells (Fig. 5a, b). In the transcriptomic network with changes determined using a 3-fold cut-off only, more pronounced changes were detected in 1.0 µg/µl MNPs@SiO2(RITC)-treated cells than in 0.1 µg/µl MNPs@SiO2(RITC)-treated cells and control cells; these changes were related to cell movement (Fig. 5c, See Supplementary Figures 2, Additional File 1). The in-silico prediction of the network revealed activation of lipid peroxidation as well as suppression of focal adhesion and cell movement in 1.0 µg/µl MNPs@SiO2(RITC)-treated HEK293 cells (Fig. 5d).
Although transcriptomics provided comprehensive information regarding MNPs@SiO2(RITC)-treated cells, the data were qualitative. Thus, for evaluating network-based analysis, we used a combination of transcriptome and metabolome networks, termed the metabotranscriptomic network, through the substitution of amino acid and organic acid profiles (cut-off ± 20% change), as described in our previous report [17]. In the group treated with 1.0 µg/µl MNPs@SiO2(RITC), amino acid, organic acid, and free fatty acid profiles showed increased levels of tyrosine, pyruvic acid, glutamic acid, lysine and lignoceric acid; in contrast, decreased levels of cysteine, asparagine, 3-hydroxybutyric acid, glutamine, serine, aspartic acid, oxaloacetic acid, glycine, and acetoacetic acid were observed [37]. The combined network provided more reliable information than the transcriptomic network alone, with more pronounced changes also being detected in 1.0 µg/µl than in 0.1 µg/µl MNPs@SiO2(RITC)-treated cells and control cells (Fig. 6a; Supplementary Figures 3, Supplementary Table 1, Additional File 1).µl The in-silico prediction of the integrated network also showed similar trend of transcriptome network (See Supplementary Figure 4, Additional File 1).
The expression levels of genes associated with the network were subsequently quantified using semiquantitative reverse transcription PCR (Fig. 6b) and quantitative (q) PCR (Fig. 6c). In particular, the expression levels of superoxide dismutase 2 (SOD2) and LIM zinc finger domain-containing 1 (LIMS1) were increased, whereas those of NCK adaptor protein 1 (NCK1) and complement C3a receptor 1 (C3AR1) were decreased in the 1.0 µg/µl MNPs@SiO2(RITC)-treated cells relative to those in the non-treated control cells.