Tryptone-stabilized gold nanoparticles induce unipolar clustering of supernumerary centrosomes and G1 arrest in triple-negative breast cancer cells

Gold nanoparticles of different sizes, shapes, and decorations exert a variety of effects on biological systems. We report a novel mechanism of action of chemically modified, tryptone-stabilized gold nanoparticles (T-GNPs) in the triple-negative breast cancer (TNBC) cell line, MDA-MB-231. The T-GNPs, synthesized using HAuCl4.3H2O and tryptone and characterized by an assortment of spectroscopy techniques combined with high-resolution electron microscopy, demonstrated strong antiproliferative and anti-clonogenic potential against MDA-MB-231 cells, arresting them at the G1 phase of the cell cycle and promoting apoptosis. The molecular mechanism of action of these particles involved induction of unipolar clustering and hyper amplification of the supernumerary centrosomes (a distinctive feature of many tumour cells, including TNBC cells). The clustering was facilitated by microtubules with suppressed dynamicity. Mass spectrometry-assisted proteomic analysis revealed that the T-GNP-induced G1 arrest was facilitated, at least in part, by downregulation of ribosome biogenesis pathways. Due to the presence of supernumerary centrosomes in many types of tumour cells, we propose chemical induction of their unipolar clustering as a potential therapeutic strategy.

assembled, microtubules exhibit selective stability as per the need of the cell. Interphase microtubules tend to be more stable than mitotic microtubules. The stability of the microtubules is governed by several post-translation modifications. Acetylation of the tubulin subunits of the microtubules, for example, is a sign of persistently stable microtubules 12 . Due to their involvement in vital cellular functions, tubulin and microtubules are targets for several clinically-approved anticancer drugs, including taxanes, ixabepilone, and eribulin mesylate 14 . By as yet poorly understood mechanism(s), breast neoplasms have been found to respond favourably to anti-tubulin agents; all the drugs mentioned above are prescribed chiefly for breast tumours. However, in order to overcome dose-limiting and off-target toxicities that are associated with current chemotherapeutics, it is imperative to develop potential therapeutic candidates that can specifically eliminate cancer cells by exploiting their differential constitution.
Tryptone, a mixture of peptides formed by the digestion of casein by trypsin, stabilizes the gold nanoparticles in solution. The surface modifications of the nanoparticles by tryptone make them stable and reduce the tendancy of the particles to aggregate 15 . As the peptides form the coating over the gold nanoparticles, they render them biocompatible as well 16 . This study reports a novel mechanism of action of the T-GNPs that involves the supernumerary centrosomes of the triple-negative breast cancer (TNBC) cell line, MDA-MB-231.

Characterization of the T-GNPs.
The T-GNPs used in this study were synthesized from HAuCl 4 .3H 2 O and tryptone (Fig. 1A). The colour change of the solution from light yellow to reddish pink provided the first indication of the formation of nanoparticles. The formation of the T-GNPs and their possible spherical shape were indicated by the absorbance peak at 540 nm (Fig. 1B) 17 . Transmission electron microscopy (TEM) images of the nanoparticles confirmed their spherical shape and their average size distribution (~25 nm) (Fig. 1C). An energy dispersive X-ray (EDX) spectroscopy analysis of the nanoparticles revealed the presence of elemental gold (Au) (Fig. 1D). The functional groups of tryptone involved in the reduction of Au 3+ and the capping molecules on the synthesized T-GNPs were verified by Fourier-transform infrared (FTIR) analysis (Fig. 1E). The intense broad peaks at 3442 cm −1 and 3458 cm −1 for tryptone and T-GNPs, respectively, were due to amine N-H stretching vibration. For tryptone, the presence of a strong peak at 1639 cm −1 represents -C=C-stretching vibration, and the bands at 1406 cm −1 represent CH 2 bending vibration. The absorption peak at 1112 cm −1 could be attributed to C-N stretching vibrations of amine groups. Thus, it was confirmed that tryptone molecules acted as the reducing and capping agents of T-GNPs. The FTIR spectrum after the bioreduction of the T-GNPs showed a sharp peak at 1641 cm −1 , which represents string C=O stretching vibration of amide groups. The peak at 1460 cm −1 is characteristic of medium CH 2 deformation bending vibration, and the sharp peak at 848 cm −1 represents symmetrical C-N-C stretching vibration of amine groups 18 . A size-distribution histogram of the particles as determined by dynamic light scattering (DLS) exhibited hydrodynamic sizes with a mean value for the intensity distribution (Z average), 269 ± 10.0 nm ( Supplementary Fig. 1A). The polydispersity index (PdI) of 0.474 substantiated the stability of the T-GNPs ( Supplementary Fig. 1B) 19 . Further, the T-GNPs exhibited a negative zeta potential of −18.5 ± 5 mV ( Supplementary Fig. 1B), suggesting that the particles had fewer tendencies to aggregate.

T-GNPs inhibited cell viability, clonogenicity and cell cycle progression, and induced cell death.
The T-GNPs-treated cells showed concentration-dependent inhibition of cell viability ( Fig. 2A). For example, the T-GNPs at 100 µg/mL, 300 µg/mL, and 500 µg/mL inhibited the cell viability by 15%, 59%, and 75%, respectively, yielding a half-maximal inhibitory concentration (IC 50 ) of 260 ± 5 µg/mL. The nanoparticles also inhibited the clonogenic propagation of the cells in a concentration-dependent manner (Fig. 2B). For example, 130 µg/mL and 260 µg/mL of the particles inhibited the clonogenicity by 56% and 89%, respectively, whereas 520 µg/mL of the particle nearly completely inhibited the clonogenicity. Vinblastine, used as a control, reduced the number of colonies by 95%. As demonstrated by flow cytometry, the cells treated with T-GNPs for 24 h arrested the cells at G 0 /G 1 phase (Fig. 2C). Specifically, cells treated with the IC 50 (260 µg/mL) of T-GNPs for cell viability showed a 31% increase in the G 0 /G 1 population (from 49% to 64%), compared to the control cells. When the cells were exposed to twice this concentration (520 µg/mL) for 24 h, the sub-G 1 phase increased substantially, suggesting the accumulation of degraded DNA associated with cell death (Fig. 2C). A 48-h treatment with the T-GNPs showed the accumulation of cells in the sub-G 1 phase with little or no cells present in other phases ( Supplementary Fig. 2), indicating time-dependent increase in cell death. Thus, the T-GNPs are capable of inducing robust G 1 arrest and promoting their death. Vinblastine, as expected, showed strong G 2 /M arrest (55% cell in G 2 /M, compared to the control). Cell death was confirmed using the DNA-intercalating fluorescent dyes acridine orange and ethidium bromide (AO/EtBr staining). AO can enter the cells through the intact cell membrane and stain the DNA, whereas EtBr can only stain the DNA of cells with compromised membrane integrity 20 .
The T-GNPs induced concentration-dependent induction of cell death. The following observations were made: (1) untreated cells were mostly stained green, indicating the predominance of healthy cells, (2) cells in the early stages of apoptosis showed greenish yellow staining, (3) cells in late apoptosis showed orange staining, and (4) dead cells showed dark orange to red fluorescence (Fig. 2D).

Induction of unipolar clustering of the supernumerary centrosomes by the T-GNPs.
Immunostaining of the cells with anti-gamma tubulin antibodies revealed the presence of supernumerary centrosomes in interphase cells (Fig. 3A), which presented as scattered dots, as reported 21 . An average of 5-6 centrosomes per cell were found in untreated cells (Fig. 3Ai). The cells treated with the T-GNPs (260 µg/mL) showed clustering of the centrosomes to one pole (Fig. 3Ai). The centrosomes thus clustered also showed further amplification (Fig. 3Aii).

Stabilization and bundling of cellular microtubules by the T-GNPs. The cells treated with T-GNPs
(260 µg/mL or 520 µg/mL) for 24 h showed extensive bundling of the microtubules (Fig. 3Bi). Vinblastine, that induced mitotic arrest, depolymerized the cellular microtubules. Persistent stabilization of the microtubules was further revealed by immunostaining with anti-α-acetyl tubulin antibodies (Fig. 3Bii) and immunoblotting of the treated cells' proteins ( Fig. 3Biii,iv). In both cases, the control used, Taxol, also showed elevated levels of acetylated tubulin. Taxol revealed extensive acetylation of the microtubules. Total α-tubulin was also estimated by immunostaining with anti-α-tubulin antibodies ( Supplementary Fig. 3).
In order to further understand the T-GNPs' mechanism of action on the TNBC cells, proteomic analyses were performed. Protein identification in untreated and treated cells was performed as per established criteria 22 . Fasta files from the Human Proteome database (downloaded from the UniProt) were used for protein searches, and a list of identified proteins was generated. The changes in protein expression levels in the treated cells compared to the control cells were quantified using label-free quantification methods and are given in Supplementary Table 1. R 2 values for two repeats were more than 0.99 in the untreated and treated groups (one-way analysis of variance). The interaction network of the 155 proteins that were downregulated by the T-GNPs were analyzed using the Search Tool for the Retrieval of Interacting Genes/Proteins (STRING) database ( Supplementary Fig. 4A). The proteins that were thus differentially expressed were subjected to gene ontology analysis (GO), which identified 134 biological processes (BP), 81 cellular components (CC), and 68 molecular functions (MF) that were enriched for the dataset. Among these, 56 MFs, 61 CCs, and 98 BPs showed p < 0.05 ( Supplementary Fig. 4B-D). Kyoto Encyclopaedia of Genes and Genomes (KEGG) pathway analysis showed that the ribosome biogenesis pathway was the most distinctly downregulated, with 12 proteins (Fig. 4A,B). The spliceosome pathway showed downregulation with eight proteins, and carbon metabolism with seven proteins (Fig. 4A).
www.nature.com/scientificreports www.nature.com/scientificreports/ T-GNPs elevated ROS production and induced loss of mitochondrial membrane potential. T-GNPs significantly (p < 0.01) elevated the levels of ROS inside the cells. Specifically, compared to the untreated cells, the cells treated with 130 µg/mL and 260 µg/mL of the T-GNPs increased the ROS level by 2.6-fold and 3-fold, respectively, as measured by their relative fluorescence intensities ( Supplementary Fig. 5A). To confirm whether the increased ROS production observed in the treated cells was mediated by the T-GNPs, we pre-treated the cells with N-acetyl cysteine (NAC) and then with the T-GNPs. As shown ( Supplementary Fig. 5A), the increase in ROS production in the T-GNPs-treated cells was significantly quenched when the cells were pre-treated with NAC. H 2 O 2, a known inducer of ROS, was used as the positive control which increased the ROS by 5-fold ( Supplementary Fig. 5A). As the increased levels of ROS can adversely affect the integrity of the mitochondrial membrane potential, we next examined the effect of the particles on the potential (Δψm) using the membrane permeable, a lipophilic dye, Rhodamine 123. Rhodamine 123 exhibited a heterogeneous staining pattern in the untreated cells with both red and green fluorescence coexisting in the same cell ( Supplementary Fig. 5B). Cells treated with 130 µg/mL of T-GNPs showed a considerable loss of the membrane potential as evidenced by the marked reduction in red fluorescence and the increase in green fluorescence. When www.nature.com/scientificreports www.nature.com/scientificreports/ the concentration was doubled, a near-complete absence of the red fluorescence was observed, indicating substantial loss of the mitochondrial membrane potential.

Discussion and Conclusion
The unique effects of gold nanoparticles, when exposed to the cellular milieu, are just beginning to be fully comprehended. This study used T-GNPs (~25 nm) to unravel a novel mechanism of action these particles against one of the most aggressively metastatic breast cancer cell lines, MDA-MB-231. The mechanism involves unipolar clustering and hyper amplification of supernumerary centrosomes and robust G 1 arrest, leading to cell death.
We began our investigation by assessing the antiproliferative potential of the T-GNPs against the cells. After verifying that the T-GNPs at microgram quantities inhibit cell viability by retarding the cells' proliferative, clonogenic, and cycling potential (Fig. 2), we set out to investigate the particles' molecular mechanism of action. One characteristic feature of TNBC cells is the presence of supernumerary centrosomes 21 . This numerical abundance of centrosomes has been thought to assist the cells in reorganizing their cytoskeleton, facilitating survival and metastasis 4 . Therefore, we first examined if the nanoparticles have any effect on these supernumerary centrosomes. Using anti-gamma tubulin antibodies, we identified a novel reorganization of these centrosomes in the treated cells. Unlike the untreated TNBC cells that displayed scattered centrosomes (Fig. 3A), the treated cells exhibited unipolar clustering and amplification of the centrosomes (Fig. 3A). Since the positioning of centrosomes is governed chiefly by the forces transmitted by microtubules 13 , we next examined whether the particles altered the structure or stability of the microtubule network. As revealed by fluorescence microscopy, the T-GNPs induced bundling of the microtubules (Fig. 3Bi). Further, the particles-treated microtubules displayed enhanced acetylation indicating suppression of their dynamicity (Fig. 3Bii) 23 . Specifically, alpha-tubulin can get acetylated at its lysine 40. Whether acetylation is the cause or consequence of microtubule stabilization is still unclear. Nevertheless, enhanced acetylation pattern indicate the presence of persistently stable microtubules 23 . Next, we examined whether the particles interacted directly with tubulin (the building block protein of microtubules) to rule out the possibility that the observed bundling of the microtubules was a secondary, indirect www.nature.com/scientificreports www.nature.com/scientificreports/ effect of the T-GNPs on microtubules. The direct binding of the particles to tubulin was verified using a tryptophan fluorescence-quenching assay (Supplementary Fig. 6). Next to be answered was how the stabilization and bundling of the microtubules induced unipolar clustering of the scattered centrosomes. In an elegant study on ciliary functions 24 , Pitavel and colleagues demonstrated that hyperstable microtubules exert pushing forces that can propel centrosomes towards one pole. Therefore, it can assumed that the particles propelled and clustered the centrosomes to one pole by interfering with the structure and dynamics of the microtubules and thereby, inducing their hyper stabilization (Fig. 3B). The centrosomes thus clustered were apparently amplified as well (Fig. 3Aii), which could be due to the oxidative stress exerted by the particles (Supplementary Fig. 5A) 25 complimented via loss of mitochondrial membrane potential ( Supplementary Fig. 5B) 26 . Finally, how the cells lost their cycling potential at the G 1 stage was investigated. Although the loss of mitochondrial function ( Supplementary  Fig. 5B) can promote G 1 arrest 27 , we sought to uncover additional mechanisms that might have facilitated the arrest with the help of mass spectrometry-assisted proteomics analysis. The analyses suggested that the downregulation of the ribosomal biogenesis pathway (Fig. 4A) played a key role in inducing G 1 arrest 28,29 . Drugs that target ribosome biogenesis pathway have been found to be effective in retarding cancer cell proliferation by inhibiting pre-rRNA processing, rRNA synthesis and interfering with translation. Several chemotherapeutic agents target ribosomes biogenesis as an effective antiproliferative strategy 30,31 . Here, as revealed by KEGG pathway analysis, at least twelve ribosomal proteins with close interactions to each other were downregulated in cells that were exposed to the T-GNPs (Fig. 4A,B). The proteins were the ribosomal P-complex proteins (RPLP0 [for ribosomal protein lateral stalk subunit P0], RPLP1, RPLP2), RPL4, RPL7A, RPL22, RPS4X, RPS5, RPS7, RPS8, RPS10, and www.nature.com/scientificreports www.nature.com/scientificreports/ RPS16. Notably, downregulation of P-complex proteins, such as RPLP0, is known to promote G 1 arrest 32 . In fact, depletion of ribosomal proteins from either ribosomal subunit can elicit G1 arrest 33 . Other ribosomal proteins that were downregulated by the T-GNPs also hold therapeutic significance for cancer. For example, depletion of the RPS4X protein in SK-OV-3 cells can strongly retard their proliferative potential 34 . In addition, there is a relation between elevated levels of ROS (Supplementary Fig. 5A) and suppression of ribosomal function. Specifically, oxidative stress is known to retard global protein synthesis, caused largely by a slower rate of ribosomal runoff due to its inhibitory effect on translation elongation or termination 35 . Further, an elegant study by Willi and colleagues showed that oxidative stress damages rRNA inside the ribosome 36 .
In summary, this study elucidated a novel mechanism of action of gold nanoparticles in cancer cells, which is characterized by unipolar clustering of supernumerary centrosomes. The clustering was facilitated by loss of microtubule dynamics. The clustering, coupled with robust G 1 arrest promoted by the downregulation of ribosomal biogenesis pathway, contributed to the cell death.

Materials and Methods
Materials. Gold  cell culture. MDA-MB-231 (triple-negative breast cancer) cells were obtained from American Type Culture Collection (ATCC, Manassas, VA). The cells, at their 21 st passage, were cultured in DMEM supplemented with 10% heat-inactivated FBS, and penicillin (100 U/mL)/streptomycin (0.1 mg/mL) solution. The cells were maintained in a humidified atmosphere at 37 °C and 5% CO 2 in a Forma SteriCycle incubator (Thermo Scientific, Waltham, MA). The cells were of low passage number. By using a MycoAlert mycoplasma detection kit (Lonza, Basel, Switzerland), they were determined to be free of mycoplasma. cell viability and colony-formation assays. The effect of the T-GNPs on the cell viability was assessed using an MTT assay 38 . In brief, the cells were seeded at a density of 5000 cells/well in a 96-well plate and incubated for 24 h. Next day, the cells were treated with different concentrations of the T-GNPs (100 µg/mL-1000 µg/ mL) for 24 h at 37 °C. After the specified time point, MTT (5 mg/mL) in phosphate-buffered saline (PBS) was added to each well and incubated for 4 h at 37 °C. The formazan crystals formed by the viable cells were dissolved in DMSO. Absorbance measurements were carried in a microplate reader (570 nm; TECAN infinite 200 PRO; Tecan, Switzerland). The experiment was performed three times in triplicates. For the colony-formation assay, the cells were plated in each well of a 6-well plate at 1000 cells/mL and were allowed to adhere for 24 h. They were then treated with 130 µg/mL, 260 µg/mL, and 520 µg/mL of the T-GNPs for 24 h. After the specified time point, the particles-containing media was replaced with fresh, complete media and the cells were grown in it for eight days with one media change on the fourth day. The colonies formed were fixed with 3.7% formaldehyde (37 °C; 15 min) and stained with crystal violet (0.5% (w/v); 1 h, 25 °C). The wells were then washed with distilled water, air-dried, and the colonies were enumerated using Image J software, (National Institutes of Health, USA). The experiment was repeated at least two times. www.nature.com/scientificreports www.nature.com/scientificreports/ incubated with propidium iodide (50 µg/mL; dissolved in 0.1% Triton X-100). The samples analyzed in a BD FACS Aria (BD Bioscience, San Jose, CA) equipped with a FACS DIVA software. The percentage of cells in each phase of the cycle was determined using FlowJo software (BD Bioscience). Further, the number of cells present in each phase of the cycle (G 0 /G 1 , S and G 2 /M) was determined using Cyflogic software (version 1.2.1, Cyflo Ltd., Turku, Finland).

Detection of apoptosis.
Induction of cell death was studied as reported earlier 39 . Briefly, the cells were seeded in 12-well plates and grown for 24 h. The next day, they were treated with the T-GNPs (130 µg/mL or 260 µg/mL) for an additional 24 h. The cells were collected by trypsinization and were washed with 1X PBS. The cell suspension was then incubated with 1 µL of AO/EtBr dye (100 mg/mL each) in PBS. The stained cells were visualized under a Nikon Eclipse 90i fluorescence microscope (Nikon, Tokyo, Japan).
Visualization of centrosomes, microtubules, and acetylated microtubules. For immunofluorescence visualization of the centrosomes, the cells were stained with anti-gamma tubulin antibodies. Briefly, the cells (5 × 10 4 cells/mL) were seeded on surface-treated coverslips and grown overnight in 12-well plates. The cells were then treated with the T-GNPs (0 µg/mL or 260 µg/mL) for 24 h and fixed in 3.7% formaldehyde (37 °C, 20 min). The cells were then permeabilized using absolute, chilled methanol at (4 °C, 15 min). Non-specific binding sites were blocked using 5% horse serum (25 °C, 1 h) in a humidified chamber. The cells were stained subsequently with FITC-conjugated γ-tubulin antibodies (Biorbyt, CA, USA; 1:200 dilution; 25 °C, 2 h). After incubating with the antibodies, they were washed with 1X PBS and incubated with Hoechst 33342 (Molecular Probes, Eugene, OR; 1:1000 dilution; 25 °C, 10 min,) to visualize the DNA. To examine the effect of T-GNPs on the cellular microtubule, the cells grown as mentioned above were treated with the T-GNPs (260 µg/mL, 520 µg/ mL i.e., IC 50 and 2 × IC 50 for the cell viability, respectively) for 24 h, fixed, permeabilized, and the non-specific binding sites were blocked, as mentioned. They were then stained first with anti-α-tubulin antibodies (Sigma, 1:300 dilution; 25 °C, 1 h) and then with Alexa-568-conjugated goat anti-mouse antibodies (Molecular Probes) for the same duration and the temperature. To observe acetylation patterns of the treated microtubules, the cells grown and treated with the nanoparticles (260 µg/mL, 520 µg/mL) were stained with anti-acetyl-α-tubulin Mass spectrometry. For mass-spectrometry-assisted proteomics analysis, the cells, grown in the presence of 260 µg/mL of the particles for 24 h, were centrifuged at 2000 rpm (25 °C, 5 min). Protein extraction was carried out using the lysis buffer (20 mM Tris-HCl, 150 mM NaCl, 1% Triton X-100, 0.5% sodium deoxycholate, 1 mM DTT, 1 X protease inhibitor cocktail, 1 X phosphatase inhibitor cocktail).
Precipitation, reduction, alkylation, and digestion of the proteins. The extracted protein samples (control and treated) were precipitated by acetone precipitation 40 . The protein pellets thus obtained were dissolved by adding 8 M urea. The samples were then reduced by the addition of 100 mM DTT and heated in a dry bath at 90 °C for 15 min. After cooling the samples, they were alkylated by adding 200 mM IAA and incubated in the dark (25 °C, 15 min). ABC (100 mM) was then added and proteins were digested with 1 mg/mL trypsin protease (37 °C, 16 h). The reaction was stopped by addition of concentrated TFA. The peptides were then dried using vacuum centrifugation for 24 h, and dissolved in 0.1% TFA for the MS-analysis.
Peptide separation and identification. The samples (5 µL, each) were analyzed using a High-Performance Chip (Chip ID: G4240-62030) connected to Agilent 1260 infinity HPLC-Chip/MS system (Agilent Technologies, Santa Clara, CA). Charged peptides from the HPLC-Chip system were directly infused into mass-spectrometer for detection, as reported 41 .
Spectra analysis, protein database searches and relative quantification. Agilent Mass Hunter software (Mass Hunter Qualitative Analysis B.08.00 Service Pack 1 (SP1)), was used for data acquisition and analysis of total ion chromatograms. Protein searches were carried out using Morpheus software (Howell, MI) with reference to human proteome database 22 . Summed mass spectra from each chromatogram were analyzed manually for accurate identification. Protein data analysis: The differentially-expressed proteins were subjected to pathway enrichment, protein-protein interaction (PPI) analysis and gene ontology (GO) analysis. The analyses were performed using DAVID (for the database for annotation, visualization and integrated discovery) bioinformatics resources 6.8 42,43 and involved molecular function enrichment (MF), cell component enrichment (CC), biological process enrichment (BP), and Kyoto Encyclopedia of genes and genomes (KEGG) pathway enrichment. Significantly modulated network nodes were reported using the protein dataset 44 . The protein-protein interactions (PPI) were analyzed using the Search Tool for the Retrieval of Interacting Genes/Proteins (STRING/P) database (version 10.5).
Purification of tubulin. PEM buffer (50 mM Pipes, 3 mM MgSO4, 1 mM EGTA, pH 6.8) was used for the isolation and purification of tubulin from goat brain through temperature and GTP-dependent multiple polymerization and depolymerization cycles, as reported 38 and kept at −80 °C.
Tryptophan-quenching assay. Tubulin (2 μM) was incubated in the absence or presence of the T-GNPs (500 µg/mL) in a water-circulating bath (35 °C; 45 mins). After the incubation, the samples excited at 295 nm, and the emission spectra in the range 310 nm-400 nm were obtained. A FlouroMax ® 4 spectrofluorometer (Horiba Scientific, Edison, NJ) supported by FluorEssence 3.5 software was used for the spectrofluorimetric titrations 45 .
Measurement of reactive oxygen species. The cells (5 × 10 4 cells/mL), grown in 12-well plates and treated with the T-GNPs (130 µg/mL or 260 µg/mL), H 2 O 2 (400 µM), or N-acetyl cysteine (NAC, 16.3 µg/mL) were examined for evidence of ROS generation after incubating them with DCFH-DA (2.4 mg/mL), as reported earlier 46 . The intensity of the DCF fluorescence was measured using Nikon Eclipse 90i fluorescence microscope and the images were analyzed using Image J software. The fluorescence intensity corresponding to ROS level were quantitated using Image J software. At least 200 cells were counted and their fluorescent intensity was recorded and calculated.
Corrected total cell fluorescence (CTCF). Analysis of mitochondrial membrane potential. The cells (5 × 10 4 cells/mL) grown on the poly-L-lysine-coated coverslips in 12-well plates were treated with the T-GNPs (130 µg/mL or 260 µg/mL) and visualized for evidence for loss of mitochondrial membrane potential using Rhodamine 123, as described earlier 47 .

Statistical analysis.
Values are expressed as mean ± standard deviation (SD). Statistical analysis was performed using GraphPad Instat software (San Diego, CA) using one-way analysis of variance (ANOVA). The values were considered statistically significant, if the p-value was <0.05.

Data availability
The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.