Gold nanoclusters elicit homeostatic perturbations in glioblastoma cells and adaptive changes of lysosomes

Unique physicochemical features place gold nanoclusters at the forefront of nanotechnology for biological and biomedical applications. To date, information on the interactions of gold nanoclusters with biological macromolecules is limited and restricts their use in living cells. Methods: Our multidisciplinary study begins to fill the current knowledge gap by focusing on lysosomes and associated biological pathways in U251N human glioblastoma cells. We concentrated on lysosomes, because they are the intracellular destination for many nanoparticles, regulate cellular homeostasis and control cell survival. Results: Quantitative data presented here show that gold nanoclusters (with 15 and 25 gold atoms), surface-modified with glutathione or PEG, did not diminish cell viability at concentrations ≤1 µM. However, even at sublethal concentrations, gold nanoclusters modulated the abundance, positioning, pH and enzymatic activities of lysosomes. Gold nanoclusters also affected other aspects of cellular homeostasis. Specifically, they stimulated the transient nuclear accumulation of TFEB and Nrf2, transcription factors that promote lysosome biogenesis and stress responses. Moreover, gold nanoclusters also altered the formation of protein aggregates in the cytoplasm. The cellular responses elicited by gold nanoclusters were largely reversible within a 24-hour period. Conclusions: Taken together, this study explores the subcellular and molecular effects induced by gold nanoclusters and shows their effectiveness to regulate lysosome biology. Our results indicate that gold nanoclusters cause homeostatic perturbations without marked cell loss. Notably, cells adapt to the challenge inflicted by gold nanoclusters. These new insights provide a framework for the further development of gold nanocluster-based applications in biological sciences.

. Hydrodynamic size measurements and zeta potential for Au 15 NCs and Au 25 NCs.
Hydrodynamic diameters were determined with time-resolved fluorescence anisotropy, according to our published protocols [75]. The zeta potential was measured with a Malvern Zetasizer Nano ZS; it was negative for all AuNCs studied here.

AuNC
Hydrodynamic  Figure S1. Effect of AuNCs on the metabolic activity of U251N and HEK293 cells.
U251N cells incubated for 72 hours with AuNCs at the final concentrations indicated, using the methods described for Figure 2. The MTT assay assessed metabolic activities for at least two independent experiments, with triplicate samples for each data set. Results are shown as average + SEM; *, p<0.05; **, p<0.01; ***, p<0.001. Figure S2. Effect of AuNCs on LysoTracker® Red DND-99 fluorescence.
The fluorescence intensity of 100 nM LysoTracker Red in PBS (pH 7.2) or 100 mM sodium acetate buffer (pH 4.5) was measured in the presence of different AuNC concentrations as indicated. Excitation was at 576 nm, emission spectra were measured with a Cary Eclipse Fluorescence Spectrophotometer (Agilent Technologies) between 586 to 700 nm using an excitation slit of 10 nm and an emission slit of 10 nm with a photomultiplier tube voltage of 600 V. The area under the curve (AUC) was obtained by integrating the emission spectra between 586 and 700 nm. AuNCs did not quench LysoTracker Red DND-99 fluorescence at concentrations up to 10 μM. AU; arbitrary units.      Immunocytochemistry located Nrf2 in U251N cells treated for 24 hours with vehicle (196 cells), 10 µM Au 15 SG 13 (179 cells) or 300 µM hydrogen peroxide (165 cells). DAPI demarcated nuclei; scale bar is 20 µm. Pixel intensities in the nucleus/area (Nuc), cytoplasm/area (Cyt) and the ratios of nuclear/cytoplasmic fluorescence (N/C) were determined for one representative experiment. Results normalized to the vehicle control are shown as average ± SEM. Student's ttest identified significant differences between vehicle and AuNC-treated samples; ***, p<0.001. Figure S9. AuNC-dependent production of reactive oxygen species (ROS).
U251N cells were treated with vehicle, 1 µM Au 15 SG 13 or 1 µM Au 15 PEG for 24 hours, as in Figure 8. CellRox® Green or H 2 -DCFDA was used to monitor changes in ROS abundance. Microscopic images were acquired and pixel intensities were quantified per nuclear area (CellRox® Green) or per cell area (H 2 -DCFDA). ROS production was evaluated for a typical experiment with CellRox® Green (192 to 215 cells per condition) or H 2 -DCFDA (30 to 47 cells for each condition). Bars depict average +SEM. Student's t-test revealed significant differences between the vehicle control and AuNC-treated cells (***, p<0.001), or between cells treated with 1 µM Au 15 SG 13 and 1 µM Au 15 PEG. Note that CellRox® Green and H 2 -DCFDA differ in the signal-to-noise ratio and stability. Figure S10. Lack of stress granule formation in response to AuNC treatment.
U251N cells were incubated with vehicle, 10 µM Au 15 SG 13 or 10 µM Au 15 PEG. At the times indicated, samples were fixed and processed for immunocytochemistry for the detection of HuR and importin-α1. Nuclei were stained with DAPI. Scale bar is 20 µm. Figure S11. AuNC-induced changes in stress granule formation.
U251N cells were incubated with vehicle, 10 µM Au 15 SG 13 (Au 15 ) or 10 µM Au 25 SG 18 (Au 25 ) for 24 hours and kept under non-stress control conditions (control) or exposed to oxidative stress (0.5 mM sodium arsenite, 2 hours). (A) The stress granule components G3BP1 and importin-α1 were detected by immunocytochemistry. Nuclei were demarcated with DAPI; scale bar is 20 µm. (B) Fluorescence intensities were measured for G3BP1 and importin-α1. Results were normalized to vehicle controls (V). Fluorescence signals/area were quantified in the nuclear and cytoplasmic compartments for 162 to 203 non-stressed cells (control, no sodium arsenite) per experiment and for each treatment. Nuclear pixel intensities/area were determined for 112 to 149 arsenite-treated cells for each of two independent experiments and per condition. Between 1448 and 2422 stress granules were assessed per experiment for each treatment. One-way ANOVA combined with Bonferroni posthoc analysis was performed for statistical evaluation; *p<0.05. (C) The stress granule size distribution was determined for the combined results of two independent experiments.