High-throughput, Label-Free Quantitative Proteomic Studies of the Anticancer Effects of Electrical Pulses with Turmeric Silver Nanoparticles: an in vitro Model Study

Triple negative breast cancer (TNBC) represents 15–20% of the over one million new breast cancer cases occurring each year. TNBC is an aggressive cancer phenotype, with low 5-year survival rates, high 3-year recurrence rates, and increased risk of metastasis. A lack of three commonly exploited hormone receptors renders TNBC resistant to endocrine therapies and lends to its critical absence of viable therapeutic targets. This necessitates the development of alternate and effective novel therapeutic strategies for TNBC. Towards this, our current work seeks to develop the technique of Electrical pulse (EP)-mediated Turmeric silver nanoparticles (TurNP) therapy, known as Electrochemotherapy (ECT), to effectively target TNBC cells. This technique involves the efficient delivery of natural bioactive molecules with anti-cancer effects via a biophysical means. In these experiments, the bioactive molecules are turmeric, a dried rhizome of Curcuma longa that has been used for centuries, both as a dietary supplement and as a medicine in Ayurveda (science of life) in the Indian subcontinent and in traditional Chinese medicine. Our results reveal the combined effect of TurNP + EP treatment in reducing MDA-MB-231 cell viability to as low as 9% at 12 h. Showing biological selectivity, this combination treatment has a substantially lower effect on non-tumorigenic mammary epithelial MCF10A cells (67% viability). To gain mechanistic insights into the actions of TurNP-based ECT treatment, we performed high-throughput, label-free quantitative proteomics studies. Proteomics results indicate that TurNP + EP treatment significantly influenced expression of a diverse list of proteins, including receptors, transcription factors, structural proteins, kinases, and metabolic enzymes. This include the downregulation of 25 proteins in PI3K-Akt signaling pathway (such as GRB2, EGFR, EPHA2, GNB1, GNB2, 14–3–3 family, and Integrin family proteins), and 12 proteins (AKR1A1, ALDOA, ALDOC, PGK1, PGM1, PGAM1, ENO1, ENO2, GAPDH, TPI1, LDHA, and LDHB) in the glycolytic pathway with concomitant reduction in metabolite levels (glucose uptake, and intracellular- lactate, glutamine, and glutamate). Compared to TurNP alone, TurNP + EP treatment upregulated 66 endoplasmic reticulum and 193 mitochondrial proteins, enhancing several processes and pathways, including Pyruvate Metabolism, Tricarboxylic acid (TCA) cycle, and Oxidative Phosphorylation (OXPHOS), which redirected the TNBC metabolism to mitochondria. This switch in the metabolism caused excessive production of H2O2 reactive oxygen species (ROS) to inflict cell death in MDA-MB-231 cells, demonstrating the potency of this treatment.

shows the viability of MDA-MB-231 upon treatment with EP of various electric field strengths to obtain the optimal EP parameter. No cell death was observed at 600 V/cm, while increased cell deaths were observed at electric field strengths above 600 V/cm. The 12 h viability was 49% at 800 V/cm, and 21% at 1000 V/ cm. It was just 9% at 1200 V/cm. From these, we selected 800 V/cm for the combinational studies. Figure 1c shows the viability of MDA-MB-231 cells at 12 h for TurNP (15 µg/mL), EP (800 V/cm), and their combination (TurNP + EP). The results were normalized with the viability of Control (Ctrl) cells at 12 h (100%). The viability of TurNP treated cells decreased significantly to 64%, and to 39% for EP treatment. It dramatically reduced to 9% for TurNP + EP treatment, indicating the combined effect of TurNP + EP on effective cell death. Figure 1d shows higher viability for the non-tumorigenic mammary epithelial MCF 10 A cells, for all treatments, demonstrating the minimal effect on nearby cells.

outline of proteomics Studies and Lc-MS/MS Reproducibility. The experimental workflow of pro-
teomics studies is shown in Fig. 2a, as explained in detail in Methods. In brief, the samples were run on the Q Exactive Orbitrap HF hybrid MS coupled with the UltiMate 3000 RSLCnano HPLC system to collect LC-MS/MS data. The tandem mass spectra were searched against the UniProt human protein database in MaxQuant. Proteins identified in at least two of the three biological replicates at 1% FDR and with at least 2 MS/MS (spectral) counts were considered for further analysis, resulting in unambiguous identification of 2426 proteins/protein families from 31823 peptides (Tables S1 and S2). The label-free quantification (LFQ) intensity was used to quantify the relative abundance of proteins for each sample.
Consistency of LFQ intensity is critical for the accurate measurement of protein abundances across multiple samples. Figure 2b shows the boxplot of the log2 LFQ intensity for the triplicate samples for each treatment. Here, Treated cells were incubated in the fresh media with RealTime-Glo MT Cell Viability reagent, and the luminescence was recorded at 12 h to quantify the viable cells relative to Control (Ctrl). The same letter or the same groups of letters indicate that they are not significantly different. The different letters or different group of letters indicate that they are significantly different (P < 0.05).
the median and interquartile range were similar within the triplicates for a treatment, illustrating the consistency of the LC-MS/MS measurements across the replicates.
We also obtained high coefficient of determination (R 2 ≥ 0.95) of LFQ intensities (Fig. 2c), indicating higher correlation between biological replicates from each treatment group than between replicates from different treatments.
Further, we used LFQ expression and MS/MS spectral count to classify the 2426 proteins into four treatment groups. A protein was considered to be present in a treatment, only if the LFQ was non-zero and MS/MS was 2 or higher (LFQ ≠ 0 and MS/MS ≥ 2) in at least two replicates among the triplicates. Figure 3a shows the Venn diagram classification of common and unique proteins present in each treatment. Among 2426 proteins, we identified 1698 proteins (69.9% of 2426) in Ctrl, 2200 (90.7% of 2426) in TurNP, 1485 (61.2% of 2426) in EP, and 1707 (70.4% of 2426) in TurNP + EP, of which 1196 proteins (49.3% of 2426) were common to all treatments (Table S2). The number of unique proteins were 34 in Ctrl, 335 in TurNP, 27 in EP, 118 in TurNP + EP. The highest number of total and unique proteins were found in TurNP. The reduction or loss in both the total and unique number of proteins in the case of TurNP + EP compared to TurNP suggests a substantial impact of TurNP + EP on cellular pathways.
Overview of differentially regulated proteins. The LFQ values were used to calculate the differentially expressed proteins in six pairwise comparison groups: TurNP vs Ctrl, EP vs Ctrl, EP vs TurNP, TurNP + EP vs Ctrl, TurNP + EP vs EP, and TurNP + EP vs TurNP. For a comparison group, all the commonly and uniquely expressed proteins in both the samples were used to perform statistical analysis and to detect significant differentially expressed proteins (e.g. 2275 proteins were used in TurNP vs Ctrl). Figure 3b shows the number of differentially expressed proteins in these comparison groups. Compared to the Ctrl, 269 proteins in the TurNP, 44 proteins in the EP, and 359 proteins in the TurNP + EP were upregulated (P < 0.05), and 39 proteins in the TurNP, and 49 in EP and 380 proteins in TurNP + EP were downregulated (P < 0.05) (Table S3). Compared to TurNP, 408 proteins were upregulated, and 564 were downregulated in the www.nature.com/scientificreports www.nature.com/scientificreports/ TurNP + EP treatment. Compared to EP, 416 proteins were upregulated, and 346 were downregulated in the TurNP + EP treatment, indicating that major changes observed for TurNP + EP treatment.
Heatmap was used to cluster and visualize the expression of these differentially regulated proteins (Fig. 3c). Protein expressions for triplicates within each treatment are clustered together and are consistent. Among treatments, clusters of proteins for Ctrl, TurNP, and EP are very close to each other, and are distinctly separate from the clusters of proteins for TurNP + EP, highlighting the difference in protein expression induced by TurNP + EP treatment. Figure 3(d-i) show the top 10 significantly upregulated and downregulated proteins and their expression (log2 fold change) levels for 6 pairwise comparisons. The cDNA FLJ60818, a protein highly similar to complement C3 (Uniprot: B4DR57), of the immune system which plays a role in the development of inflammation 30 was the most upregulated protein in TurNP vs Ctrl with a fold change of 7.77 (P = 2.8E −07 ), in TurNP + EP vs Ctrl and in TurNP + EP vs EP with a fold change of 9.95 (P = 1.3E −07 ) in each. The antithrombin-III (SERPINC1) was the most upregulated in EP vs Ctrl with a fold change of 6.18 (P = 6.5E −06 ), and in TurNP + EP vs TurNP with a fold change of 6.66 (P = 4.5E −05 ). The ATP synthase subunit delta, mitochondrial (ATP5D) was the most upregulated protein in TurNP vs EP with a fold change of 7.38 (P = 6.1E −07 ).
The 60S ribosomal protein L35 (RPL35) was the most downregulated protein in TurNP vs Ctrl with a fold change of 7.27 (P = 6.7E −05 ), while, histone H2A (HIST1H2AB) was the most downregulated protein in EP vs  (Table S4). No GO enrichment was found for the regulated proteins in TurNP vs Ctrl, and 3 component, 2 function, and 5 process GO enrichment terms were found for the proteins regulated in EP vs Ctrl, indicating the low count of regulated proteins in these.
The upregulated proteins were primarily localized in mitochondrion, membrane, mitochondrial inner membrane, integral to membrane, endoplasmic reticulum (ER), ER membrane, mitochondrial matrix, and microsome for TurNP + EP compared to TurNP (Fig. 4a). Similar localization profile was also observed for proteins upregulated in TurNP + EP from Ctrl and EP. Majority of downregulated proteins in TurNP + EP vs TurNP were primarily localized in cytoplasm, cytosol, plasma membrane, and ribonucleoprotein (RNP) complex, as also observed for TurNP + EP vs Ctrl and TurNP + EP vs EP. However, only downregulated proteins for TurNP + EP vs TurNP were localized in cytoskeleton and actin cytoskeleton.
Molecular functional analysis showed higher representation of upregulated proteins related to oxidoreductase activity, catalytic activity, NAD binding, electron carrier activity, NADH dehydrogenase (Ubiquinone) activity, transferase activity, and ATPase activity for TurNP + EP (Fig. 4b). Only the calcium ion binding was upregulated for EP vs TurNP. The proteins involved in the protein binding, RNA binding, actin binding, structural molecule activity, actin filament binding, and mRNA binding were downregulated for TurNP + EP. The GO term transferase activity, transferring glycosyl groups, which is involved in catalysis of the transfer of a glycosyl group from one compound (donor) to another (acceptor) was downregulated for EP compared to TurNP.
The biological process analysis showed upregulation of proteins related to transport, transmembrane transport, carbon metabolic process, oxidation-reduction process, and respiratory electron transport chain in TurNP + EP compared to Ctrl, TurNP, and EP (Fig. 4c). This correlates well with the breakdown of cell www.nature.com/scientificreports www.nature.com/scientificreports/ membranes due to electrical pulses. The cellular lipid metabolic process was only upregulated for TurNP + EP compared to TurNP. The processes, such as, mRNA metabolic process, gene expression, RNA metabolic process, and cellular protein metabolic processes were downregulated for TurNP + EP compared to Ctrl and EP. The mRNA metabolic process and blood coagulation were downregulated for TurNP + EP compared to TurNP, while the translation was only downregulated for TurNP + EP from EP.
Collectively these results suggest that EP application with TurNP may upregulate the membrane proteins to facilitate the increased cellular transport process in MDA-MB-231 cells. Additionally, upregulation in organelle (mitochondrion and ER) proteins and activities, such as oxidoreductase, catalytic, NAD binding, NADH dehydrogenase (Ubiquinone), and electron carrier may activate intrinsic pathways to induce apoptotic cell death.
STRING interaction and clustering analysis (Fig. 7) of downregulated proteins including PI3K-Akt pathway proteins (blue color nodes) and glycolytic enzymes (up and down both -red color nodes) indicate that all the integrins family proteins are clustered together and β1 interacts directly with α2, α3, α5, α6, αV, and β4 (cluster #1). All the integrins together with EGFR, and GRB2 also participate in focal adhesion pathway, as indicated by green color nodes, and other downregulated PI3K-Akt pathway proteins, Laminin subunit gamma-1 (LAMC1) and Thrombospondin-1 (THBS1) were also clustered together with integrins. THBS1 is also linked with glycolysis, as highlighted by its interaction with aldolase A (ALDOA). Figure 7 also demonstrates that 26 differentially regulated glycolysis pathway proteins form 6 different clusters (clusters # 2, 4, 5, 6, 7, and 9). Among these proteins, 14 were upregulated, and 12 were downregulated in TurNP + EP for at least one of the three pairwise comparison groups. Table 2 lists these proteins and their expressions.
The clustering analysis also highlights that the majority of upregulated and downregulated proteins are clustered together in separate clusters, with red arrows indicating the occasional outliers in a cluster from the other (up/downregulated) group. Clusters 4, 5, and 6 represent upregulated proteins, and clusters 2, and 9 represent downregulated proteins, while cluster 7 represents the mixed population, with 2 proteins from each group (PFKL and PFKP: upregulated, ALDOA and ALDOC: downregulated).
The localization studies highlight that while downregulated proteins are cytosolic (yellow color nodes), most of the upregulated proteins are mitochondrial (pink color nodes) correlating well with the global enrichment trends for the regulated proteins as shown previously in Figs. 4 and 5.
The identification and expression of key significantly regulated glycolytic proteins for TurNP + EP compared to TurNP is shown in Fig. 8.
Validation of proteomics results. The proteomics results were validated by verifying the correlation between protein level changes and the transcription level changes for the glycolytic genes LDHB and ENOI. Their mRNA levels were checked using using real time quantitative PCR (qPCR) after 12 h of treatment. Figure 9a,b shows these results. The mRNA/protein levels were normalized with Ctrl mRNA/protein expression levels (level 1). The mRNA expression of LDHB was 0.71 TurNP, while it reduced to 0.43 for EP and 0.04 in TurNP + EP (Fig. 9a). In comparison, the protein expression of LDHB was 0.95 for TurNP, while it reduced to 0.77 for EP and to 0.34 for TurNP + EP treatment. www.nature.com/scientificreports www.nature.com/scientificreports/ For ENO1, the mRNA expression was the lowest at 0.13 for TurNP + EP, which was also the case with the protein level at 0.3 (Fig. 9b). These results demonstrate good correlation between mRNA and protein levels for LDHB and ENO1 and validate that these genes are downregulated at transcription levels upon TurNP + EP treatment. Figure 9c shows the average luminescence (Lum) as relative light units (RLU), which represents the H 2 O 2 reactive oxygen species levels in MDA-MB-231 cells at 12 h. The Lum was 2199 for Ctrl, which increased significantly to 2800 for TurNP treatment and to 3266 for EP treatment. The highest Lum of 3481 was observed for TurNP + EP, a 1.6-fold and significant increase in H 2 O 2 levels compared with Ctrl. We also quantified the uptake and intracellular levels of key metabolites, such as glucose uptake, intracellular lactate, glutamine, and glutamate ( Fig. 9d-g) that suggest a decreasing trend in the metabolites levels upon TurNP + EP treatment, compared to the Ctrl.

Discussion
In this study, we sought to better understand the anticancer effects of TurNP + EP on human, MDA-MB-231 TNBC cells, utilizing cell viability and quantitative proteomic studies. The viability studies showed that this combined treatment substantially promoted tumor cell death (91%) at 800 V/cm and 15 µg/mL TurNP. The String interaction analysis of the significantly regulated proteins in PI3K-Akt signaling and glycolysis pathways. The significantly regulated proteins from these pathways were uploaded to STRING 71 tool to visualize the interaction and functional enrichment with evidence as meaning of network edges, and active interaction sources to be Experiments, Database, Co-expression, Neighborhood, Gene Fusion, and Cooccurrence, with minimum required interaction score as highest confidence (0.9). An MCL clustering analysis with 3 as inflation parameter was run on the network nodes to cluster them in different groups. Here the node color represents the KEGG 66 pathways and the localization of these proteins. The color of the edges is based on the interactions: (known, predicted, or other). Red arrows indicate the occasional outliers in a cluster from the other (up/downregulated) group for regulated proteins in glycolysis.

Scientific RepoRtS |
(2020) 10:7258 | https://doi.org/10.1038/s41598-020-64128-8 www.nature.com/scientificreports www.nature.com/scientificreports/ Proteomic results revealed that TurNP + EP influenced the expressions of many proteins belonging to multiple cellular components, molecular functions, and biological processes. Among the differentially regulated proteins, we identified receptors, transcription factors, structural proteins, kinases, and metabolic enzymes. The majority of upregulated proteins in TurNP + EP treatment belonged to organelles (ER, and mitochondrion) and membrane structures, while the majority of downregulated proteins resided in cytosol, cytoskeleton, and extracellular regions.
TurNP + EP treatment downregulated 25 key PI3K-Akt pathway proteins (Table 2). This effect may have particular significance given that PI3K is one of the most commonly mutated pathways in TNBC 32,33 . The GRB2 is overexpressed in breast cancers, and is a key molecule in intracellular signal transduction, which can directly interact with RTKs, such as epidermal growth factor receptor (EGFR) to activate several downstream oncogenic signaling pathways, including PI3K-Akt and RAS signaling 34 . The EGFR is frequently overexpressed and it enhances aerobic glycolysis and is linked with poor prognosis in TNBC [38][39][40] . The EPHA2 is uniquely upregulated in TNBC, and its silencing impairs cell growth and stimulates apoptosis 35 . Further, the G family proteins, GNB1 and GNB2 are implicated in cancer proliferation, survival, invasion, metastasis, survival and resistance 41,42 . The 14-3-3 family proteins play important role in regulating multiple signaling pathways, including cell cycle, autophagy, apoptosis, and glycolysis 43 . These proteins are also downregulated in response to EP with curcumin in MDA-MB-231 cells 20 .
Other major changes in protein expression due to TurNP + EP treatment were also observed for proteins involved in glycolysis, TCA cycle and OXPHOS. Majority of key glycolytic enzymes were downregulated in agreement with previous reports [44][45][46][47][48][49][50][51][52][53] . For example, ENO1 is a highly expressed TNBC biomarker and its function is positively related with the distinct TNBC metabolism 44,45 . While LDHA is upregulated in variety of cancers, LDHB is specifically upregulated in basal-like TNBC and is an essential gene in TNBC. Patients with elevated LDHB levels in breast tumors show a poor clinical outcome 46 . LDHB may play a critical role in reverse Warburg effect 47 , an alternate way by which hypoxic and glucose-deprived cells actively utilize lactate secreted from neighboring cells undergoing aerobic glycolysis [46][47][48][49][50] . Downregulation of ALDOA, PGK1, and several other glycolytic enzymes listed in Table 2    www.nature.com/scientificreports www.nature.com/scientificreports/ treatment because these enzymes are known to overexpress in various cancers [51][52][53] . These results were validated by demonstrating a good correlation between mRNA and protein level expressions of LDHB and ENO1 for the different treatments.
The upregulated glycolytic enzymes were either upstream or localized in the mitochondrion. Among these, the upregulated PDC complex enzymes (DLAT, DLD, PDHA1, and PDHB) indicate the restoration in PDC activity, which can revert the Warburg metabolic phenotype by redirecting pyruvate metabolism to the mitochondria and www.nature.com/scientificreports www.nature.com/scientificreports/ enhance apoptosis 54 . The upregulation of TCA cycle and OXPHOS pathway proteins also indicates a shift towards the mitochondrial metabolism upon TurNP + EP treatment, with a larger dependency on oxidative energy substrates for energy production in MDA-MB-231 cells. The upregulation in several other pathways, such as FA degradation, FA metabolism, amino acid (valine, leucine, isoleucine, and lysin) degradation, peroxisome, pyruvate metabolism, and aromatic amino acid (tryptophan) metabolism can produce pyruvate and TCA cycle intermediates, like acetyl COA and oxaloacetate to fuel the TCA cycle. The increased OXPHOS could also be a stress response as cells try to generate energy through alternate sources upon downregulation of glycolysis. Increased OXPHOS activity can also generate ROS to activate cell death pathways, as we showed previously for EP application with cisplatin and curcumin in MDA-MB-231 cells 20,55 . The peroxisome pathway, which involves breaking down the FA for the membrane regeneration can also generate H 2 O 2 to trigger apoptosis 56 . We also measured the H 2 O 2 levels, validating that TurNP + EP treatment causes the oxidative stress in these cells to cause cell death, correlating well with previous results 20, 55 .
In summary, our results provide preliminary novel evidences and insights into the anticancer effects of ECT with TurNP against MDA-MB-231 cells. The combined TurNP + EP suppressed key proteins implicated in cancer cell proliferation, differentiation, migration, survival, and evasion of cell death and apoptosis. These proteins were involved in multiple pathways, such as PI3K-Akt signaling and glycolysis and their downregulation altered the metabolic profile of the MDA-MB-231 cells. These results suggest that the suppression of glycolytic metabolism in TNBC could be a potential therapeutic avenue against TNBC.  www.nature.com/scientificreports www.nature.com/scientificreports/ (Gibco) supplemented with 5% horse serum (Atlanta Biologicals), 20 ng/ml human EGF (Sigma-Aldrich, USA), 0.5 mg/mL hydrocortisone (Sigma-Aldrich), 100 ng/mL cholera toxin (Sigma-Aldrich), 10 mg/mL bovine insulin (Sigma-Aldrich), 100 IU/mL penicillin and 100 mg/mL streptomycin was used for MCF10A. Using trypsin, cells were detached and were centrifuged at 1000 rpm for 5 min at 4 °C and were resuspended in fresh media at 1×10 6 cells/mL for treatment. turnp preparation. TurNP synthesis was carried out with some modification 57,58 . First, an aqueous extract from the dried Turmeric tuber (Curcuma longa) was prepared. The 2 g of the manually ground fine powder from Turmeric tuber was added to 20 mL of sterile distilled water (10% w/v), and was incubated overnight at 40 °C and 100 rpm (Scigenics biotech pvt. Ltd, India). The solution was boiled for 1 minute in the microwave and incubated for 10 minutes at 1000 rpm at 25 °C. The supernatant was filtered through Whatmann No:1 filter paper, and was concentrated to a final volume of 5 mL 59,60 .

Methods
For TurNP synthesis, 5mL of filtered Turmeric extract was added to 45 mL of 1 mM AgNO 3 solution, and incubated for 3 days in dark at 100 RPM at 40 °C. 1 mM AgNo 3 solution was used as control. The TurNP synthesis was observed by UV-vis spectroscopy between 300 nm to 650 nm. The TurNP synthesis was confirmed by a peak at 440 nm in UV-vis spectrum, and visually by the change in color from yellow to brown (not in AgNO 3 control). After confirmation, the extract was centrifuged at 15,000 rpm for 20 min at 4 °C to obtain pellet. The pellet was washed once in sterile water and was dried to obtain TurNP powder used in the study. turnp treatment. A stock solution of 1 mg/mL was prepared by suspending the TurNP into sterile double-distilled water with 10% DMSO, and sonicated for 2 minutes at 5% intensity for homogenization. The required volume from the stock solution was added into the cell suspension to achieve the desired final treatment concentration (5,10,15, and 25 µg/mL) of TurNP. the electrical pulse application. Eight, 600-1200 V/cm, 100 μs unipolar, square wave pulses at 1 Hz repetition rate were applied using a BTX-ECM830 electroporator (Genetronics Inc., USA). For this, 600 μL cell suspension (1 × 10 6 cells/mL) with or without TurNP in BTX electroporation cuvettes (4 mm gap) was used, as previously 20,55 . No electrical pulses were applied to Ctrl and TurNP. After treatment, cells were transferred and cultured with fresh-media for 12 h for various assays.
Viability assay. Following treatment, cells were transferred to 96-well plates (MDA-MB-231: 20,000 cells and MCF 10 A: 10,000 cells (as suggested by manufacturer's protocol)) with fresh-media, as previously 20,55 . The cells were incubated for 12 h to assess the metabolic activity using RealTime-Glo MT Cell Viability Assay (Promega, USA), as per manufacturer's protocol. Synergy LX Multi-Mode Reader (BioTek Instruments, USA) was used to record luminescence (Lum) at 1 s integration time. The sample Lum values were normalized with Ctrl to quantify viability using equation (1  proteomics Studies. Following treatment, MDA-MB-231 cells were incubated in 6 well plates (600,000 cells/well) containing 2 mL of fresh-media and were cultured for proteomics experiments. The following methodology was adopted for sample preparation, mass spectroscopy run, and data analysis, as previously: 20 www.nature.com/scientificreports www.nature.com/scientificreports/ Sample preparation for mass spectrometry analysis. After treatments, MDA-MB-231 cells were incubated in 6 well plates (600,000 cells/well) containing 2 mL of fresh-media. After 12 h, cells were collected by scraping, washed thrice in ice-cold 1×PBS and re-suspended in 4 M urea. Protein extraction and proteomics sample preparation was done, as described previously 20,55 . In brief, cells were homogenized in Precellys 24 Bead Mill Homogenizer (Bertin Corp., USA) and extracted proteins were precipitated overnight at −20 °C with pre-chilled (−20 °C) acetone. Precipitated proteins were dissolved in 8 M urea and protein concentration was estimated using BCA assay. Protein (50 µg) from each sample was digested overnight with Trypsin/Lys-C Mix (Promega) following reduction of disulfide bonds with dithiothreitol and cysteine alkylation with iodoacetamide. Peptides were desalted using C18 micro spin columns (The Nest Group Inc., USA) prior to LC-MS/MS analysis 20,55 .
LC-MS/MS data collection. LC-MS/MS data were collected on Q-Exactive (QE) High Field (HF) Hybrid Quadrupole Orbitrap MS (Thermo Fisher Scientific) coupled with an UltiMate 3000 RSLCnano HPLC and a Nano-spray Flex ion source (Thermo Fisher Scientific) using a standard data-dependent acquisition. Peptides (1 µg) were loaded onto a trap column (300 µm ID × 5 mm, 5 µm 100 Å PepMap C18 medium) and then separated on a 15-cm long Acclaim ™ PepMap ™ (75 µm, 3μm 100 Å PepMap C18 medium, Thermo Fisher Scientific) analytical column. All the MS measurements were performed in the positive ion mode using 120 min LC gradient method as described elsewhere 61,62 , MS data were collected using Top20 data dependent MS/MS scan method. All default settings were used except the following settings: precursor mass tolerance was 10 ppm, enzyme specificity was Trypsin/P and Lys-C with up to 2 missed cleavages, variable modification was oxidation of methionine (M); fixed modification was carbamidomethylation of cysteine (C); false discovery rate (FDR) of peptides and proteins was 0.01. Unique plus razor peptides were used for protein quantitation. Proteins were quantified using LFQ intensity. Post search data analysis was performed as described previously 20,55 . Zero LFQ values were imputed with 983600, half of the lowest LFQ value (1967200) observed across three treatments. The protein fold-change was calculated by subtracting the average log2 values [Δlog2 (LFQ intensity)] between proteins from each comparison group. Proteins with fold-change of | Δlog2 | >0.5, and P < 0.05 (Student's unpaired, two-tailed, t test) were considered significant.
Enrichment and string interaction analysis. Significant proteins were compared among treatments using KEGG database 66 in DAVID 6.8 67,68 . The proteins from KEGG pathway analysis were uploaded to the Cytoscape 3.6.1 software 69 and matched using the WikiPathway app (beta), with the degree of shading representing the fold change. GO enrichment analysis was performed using Genecodis 70 , using total 2426 proteins as background. STRING 71 was used to visualize the interaction and functional enrichment with minimum required interaction score as highest confidence (0.9). An MCL clustering analysis with 3 as inflation parameter was run on the network nodes to cluster them in different groups.
Statistical analysis. One-way ANOVA was used to calculate statistical significance for cell viability assays, coupled with Tukey's multiple comparison test, as previously 55 . Prior to ANOVA analysis, the data were checked against normality and homoscedasticity assumptions. Tukey's test tags each treatment with a letter or a group of letters to indicate their significance. The same letter or the same groups of letters indicate that they are not significantly different. The different letters or different group of letters indicate that they are significantly different (P < 0.05).
Statistical significance for proteomics data (log2 transformed), and all other experimental data was calculated using Student's unpaired, two-tailed, t-test.
All experiments were performed in triplicates or more.

Data availability
RAW proteomics data files, parameters used, and LC-MS/MS methodology and statistics, and the other datasets generated and/or analyzed during the current study are available from the corresponding author on reasonable request.