Fluvastatin Mediated Breast Cancer Cell Death: A Proteomic Approach to Identify Differentially Regulated Proteins in MDA-MB-231 Cells

Statins are increasingly being recognized as anti-cancer agents against various cancers including breast cancer. To understand the molecular pathways targeted by fluvastatin and its differential sensitivity against metastatic breast cancer cells, we analyzed protein alterations in MDA-MB-231 cells treated with fluvastatin using 2-DE in combination with LC-MS/MS. Results revealed dys-regulation of 39 protein spots corresponding to 35 different proteins. To determine the relevance of altered protein profiles with breast cancer cell death, we mapped these proteins to major pathways involved in the regulation of cell-to-cell signaling and interaction, cell cycle, Rho GDI and proteasomal pathways using IPA analysis. Highly interconnected sub networks showed that vimentin and ERK1/2 proteins play a central role in controlling the expression of altered proteins. Fluvastatin treatment caused proteolysis of vimentin, a marker of epithelial to mesenchymal transition. This effect of fluvastatin was reversed in the presence of mevalonate, a downstream product of HMG-CoA and caspase-3 inhibitor. Interestingly, fluvastatin neither caused an appreciable cell death nor did modulate vimentin expression in normal mammary epithelial cells. In conclusion, fluvastatin alters levels of cytoskeletal proteins, primarily targeting vimentin through increased caspase-3- mediated proteolysis, thereby suggesting a role for vimentin in statin-induced breast cancer cell death.


Introduction
Emerging data suggest that the pleotropic effects of statins (HMG-CoA reductase inhibitors) contribute to their anti neoplastic, anti inflammatory and neuroprotection. In vitro, preclinical and clinical data shows that lipophilic statins are effective against various cancers including breast cancer [1][2][3]. Recently, Nielsen et al., reported that the mortality rate is significantly decreased in prostate cancer patients treated with statins compared to nonstatin users [4]. Lipophilic statins inhibit breast cancer cells (ER positive, HER2 positive and ER negative) growth by decreasing NF-kB, AP-1 activation and phosphorylation of MAPK [5]. Previously, we have shown that both fluvastatin and simvastatin induce MCF-7 breast cancer cell death by inducing nitric oxide via iNOS and arginase dependent pathways [6]. Also, recently we reported that fluvastatin and simvastatin induce triple negative breast cancer (TNBC) cell death by increasing iNOS-dependent nitric oxide levels and dys-regulation of iron homeostasis in MDA-MB-231, MDA-MB-453 and BT-549 cells [7]. Statins are known to deplete mevalonate pathway intermediates including the synthesis of isoprenyl groups necessary for activating the Rho/ Ras/Rac GTPases that play a significant role in cancer cell proliferation and invasion. Though statins are known to inhibit cholesterol biosynthesis through mevalonate pathway, they may target multiple proteins regulating different pro survival pathways thereby inhibiting proliferation of cancer cells. Aka et al., recently compared a functional proteome of two hormone-dependent breast cancer cells lines MCF-7 and T47D and the analyses showed that 164 proteins involved in various proliferative functions are differentially expressed between them [8]. Lovastatin induces breast cancer cell death through modulation of E2F1pathway by altering expression of prohibitin and retinoblastoma (Rb) proteins [9]. Upon exposure to lovastatin in ARO thyroid cancer cells, a set of proteins were altered in their expression which were then mapped to various cellular functions related to protein folding, metabolism, signal transduction, protein expression and protein degradation [10]. Isobaric tags for relative and absolute quantitation (iTRAQ)-based proteome analysis of ZR-75-1 and MDA-MB-231 breast cancer cells treated with chemotherapeutic agent doxorubicin followed by death receptor ligand TRAIL revealed perturbation of various pathways including cellular assembly and organization, molecular transport, oxidative stress, cell motility and cell death. Further, this study also identified three proteins (PPIB, AHNAK, and SLC1A5) that are commonly regulated in both the cell types upon the drug exposure [11]. Recently, stable isotope labeling by/with amino acids in cell culture -based proteomic approach in lovastatin-induced human acute promyelocytic leukemia (HL-60) cells quantified 3200 proteins, among which 120 proteins were significantly altered which were mapped to regulating various cellular pathways including inhibition of cholesterol biosynthesis, estrogen receptor signaling, glutamate metabolism and protein ubiquitination [12].
In the present study, we investigated the comparative proteome of metastatic MDA-MB-231 breast cancer cells exposed to fluvastatin and control treated cells by 2-D gel electrophoresis (2-DE) for protein separation followed by LC-ESI-MS/MS for protein identification. The differentially expressed proteins were analysed by gene ontology and Ingenuity Pathway Analysis (IPA) to understand the molecular functions of proteins and pathways regulated by fluvastatin. The major hubs of significant sub networks and their non canonical pathways were validated by western blot analysis. This systematic analysis revealed the involvement of various signaling networks in determining their key role in mediating fluvastatin-induced MDA-MB-231 cell death. Taken together, this study surmises a newer means of statin induced cancer cell death identifying a set of proteins which may serve as prediction markers up on further validation to monitor fluvastatin treatment in breast cancer patients.

Preparation of protein extracts from cells
MDA-MB-231 and MCF-10A cells were treated with fluvaststain (10 mM) in triplicate for a period of 24 h and at the end of the treatment, cells were washed with ice cold DPBS and collected in PBS by cell scraper. Cells were centrifuged at 3000 rpm for 5 minutes at 4uC. Cell pellets were directly lysed in an appropriate volume of 2D-lysis buffer (8 M Urea, 2 M Thiourea, 4% CHAPS, 65 mM DTT, 40 mM Tris-HCl) and homogenized for 60 seconds under ice cold conditions. To remove cell debris, lysates were centrifuged at 12000 rpm for 10 minutes at 4uC. Clear supernatants were collected and protein concentrations were measured using Bradford reagent.

Two-dimensional gel electrophoresis
For two-dimensional gel electrophoresis, isoelectric focussing (IEF) was performed using linear pH 4-7 (11 cm) IPG strips (GE Healthcare) according to the manufacturer's instructions with minor modifications to improve resolution. Briefly, 250 mg protein extracts prepared as described above were diluted to 125 ml with rehydration buffer (8 M urea, 2 M thiourea, 2% CHAPS, 50 mM DTT with 0.5% v/v IPG buffer pH 4-7, (GE Healthcare) and used to passively rehydrate each IPG strip overnight. Proteins were separated by the Ettan IPG Phor 3 (GE) at 20uC. The focusing was started at 250 V for 15 minutes, 500 V for 60 minutes and then increased to 1000 V with 60 minutes and 6000 V to 9000 Vh linear gradient. The samples were maintained at 6000 V until a total run of 12 kVh. After IEF, the IPG strips were equilibrated (10 minutes) in equilibration buffer (0.375 M Tris-HCl, pH 8.8, 6 M urea, 20% glycerol, 2% SDS and 130 mM DTT) and then transferred to equilibration buffer containing 135 mM iodoacetamide (10 minutes) with constant shaking. Equilibrated strips were applied onto the top of 12% SDS-PAGE gels and sealed with 1% agarose prepared in SDS-Tris-glycine buffer with trace amounts of bromophenol blue as a tracking dye to monitor electrophoresis. Electrophoresis was performed at a constant current of 25 mA per gel. At the end of electrophoresis, proteins were visualized by staining gels with colloidal coomassie brilliant blue staining solution (Roth Chemicals, Germany). Briefly, gels were fixed in 40% methanol, 15% acetic acid for 1 h and then soaked with colloidal staining solution overnight. Gels were then de-stained in solution containing 10% methanol, 7% acetic acid to remove background interference.

Image analysis
Colloidal coomassie brilliant blue stained gels were scanned at 300 dpi resolution using PROPIC II (DIGILAB, Genomic solutions). For the identification of differentially expressed proteins, Image analysis was carried out with PDQUEST 2-D analysis software package (Version 6.0, Bio-Rad). In a match set for comparative image analysis, gels consisting of triplicate treatments were grouped accordingly and analysed for qualitative and quantitative differences. To ensure that the differences in spot volume and density between gels are due to differential expression, all the gels were normalized using total spots density normalization tool. The minimum fold changes between groups was restricted to greater than 1.7 fold and ANOVA test was performed for statistical significance (p,0.001) of differentially altered spots across the gel members in a match set.

Protein identification by LC-MS/MS
Preparation of peptide mixtures for protein identification was performed as described previously [13]. Briefly, selected protein spots were excised manually from Colloidal Coomassie Brilliant Blue stained 2-DE gels and trypsin digestion was performed manually. Gel pieces were destained by washing with 50% (v/v) of 50 mM ammonium bicarbonate/acetonitrile and dehydrated with acetonitrile. After drying, gel pieces were soaked in ammonium bicarbonate solution and then in trypsin solution (20 ng/ml trypsin in 20 mM ammoniumbicarbonate) till the gel pieces were submerged and incubated at 37uC for 15 h. The digestion was stopped by the addition of 5% acetic acid. For peptide extraction, gel pieces were covered with 50% (v/v) ACN/0.1% (w/v) acetic acid and incubated for 30 minutes at 37uC with constant shaking. The supernatant was collected in designated tubes before repeating the extraction procedure. The peptide extracts were pooled into respective tubes and dried in speed vac and reconstituted in buffer with 0.1% acetic acid containing 2% acetonitrile prior to the measurement in the FT-ICR instrument.

Mass spectrometry
Proteins identification from peptide mixtures was performed on LTQ-Fourier transform-ion cyclotron resonance (FT-ICR) mass spectrometer (Thermo Scientific, Germany) equipped with nano electrospray ion source coupled to a nanoACQUITY Ultra Performance LC (UPLC) (Waters Corporation, USA) and controlled by Xcalibur software (Thermo Electron Corporation, Gemany). Briefly, peptides were enriched on a nanoAcquity Symmetry C18 pre-column (2 cm length, 180 mm inner diameter and 5 mm particle size) from Water Corporation and separated using NanoAcquity BEH130 C18 column (10 cm length, 100 mM inner diameter and 1.7 mm particle size from Water Corporation) on nanoAcquity UPLC. The peptides were eluted with a mixture of solvent gradient (1-5% buffer B in 2 min, 5-60% buffer B in 23 min, 60-99% buffer B in 3 min of buffer A (2% acetonitrile containing 0.1% acetic acid) and B (acetonitrile containing 0.1% acetic acid). The eluted peptides were electro sprayed into mass spectrometer. The mass spectrometric data was collected in data dependent mode to switch between FT-ICR MS and LTQ-MS/ MS acquisition automatically. Full scan MS spectra in the m/z range from 300 to 1500 (resolution r = 50000) were acquired in the mass spectrometer. The data acquisition method was set to isolate up to five most intense ions depending upon signal intensity for fragmentation in the linear ion-trap using collision induced dissociation. Target ions already selected for MS/MS were dynamically excluded for 30 seconds. The general conditions of mass spectrometer for data collection were 1.6-1.7 kV electrospray voltage, Ion selection threshold 1000 counts for MS/MS. For identification of proteins from MS data, automated database search was performed using Proteome Discoverer 1.3.0.339 (Thermo Scientific, Germany) with Sequest algorithm. A human uniprot fasta database and a decoy database of 1% FDR were used for identification. Two possible missed cleavages for trypsin enzyme specificity with a mass tolerance of 10 ppm (parent ion) and 0.8 Da (fragment ion), methionine oxidation as dynamic were considered for data base search. The identity of the proteins was confirmed based on high confidence peptide identification containing at least two peptides per protein with rank 1 peptides in proteins and XCorr scores $2.25 or 2.75 for doubly and triply charged peptides respectively.

Bioinformatics analysis of the proteomic data
Differential expression data were analysed by unsupervised hierarchical clustering to find unique protein clusters among the altered proteins and sample clusters based on the identified proteins. The clustering was performed using Euclidean distance measure from the log-transformed values of fold change in expression of all significant differentially regulated proteins across samples included in the analysis set. The differentially expressed proteins and their respective biological processes and molecular functions or relationships were determined using Database for Annotation, Visualization, and Integrated Discovery (DAVID) GO (gene ontology) annotation tool. Further, to map protein networks to which differentially regulated proteins are mapped, the lists of all differentially regulated proteins were entered into the IPA software to assess their interaction with other proteins and relevant biological networks. A global master network connecting all differentially expressed proteins was constructed based on annotations published in literature. This network shows most relevant direct and indirect connections of proteins found to be deregulated between fluvastatin treated and untreated MDA-MB-231 cells. From the larger non canonical networks, further sub networks were built to focus activated/inactivated pathways. Based on the inter connectivity of the objects (identified proteins) and edges with in the sub network, major hubs were identified for further validation to understand their role in fluvastatin-induced MDA-MB-231 cell death.

Reverse Transcription-PCR analysis
Following the treatment, total RNA was extracted with 1 mL of TRIzol reagent (Sigma) according to the manufacturer's protocol. First-strand cDNA synthesis was made using five micrograms of RNA according to the manufacturer's instructions (Takara blue print 1 st strand cDNA synthesis kit). One microliter of the cDNA mixture per gene was used to amplify the respective genes. Primer sets for GAPDH (forward-59GCACCACCAACTGCTTAG-CAC39, reverse-59CCCTGTTGCTGTAGCCAAAT39) and vimentin (forward-59 GCCTCTCCAAAGGCTGCAGA39; reverse-59GAGTTTTTCCAAAGATTTAT39).

Western blotting
Cells were washed with ice-cold DPBS and resuspended in 150 mL of radioimmune precipitation assay (RIPA) buffer containing 1 mmol/L of sodium vanadate, 10 mg/mL of aprotinin, 10 mg/mL of leupeptin and 10 mg/mL of pepstatin inhibitor. Cells were homogenized by passing the suspension through a 25gauge needle (20 strokes). The lysate was centrifuged at 750 rpm for 10 min at 4uC to pellet out the nuclei. The remaining supernatant was centrifuged for 30 min at 12000 rpm. Total protein was determined using the Bradford method. Proteins were resolved using 10% SDS-PAGE and blotted onto nitrocellulose membranes. Membranes were washed with TBS (140 mmol/L NaCl, 50 mmol/L Tris-HCl; pH 7.2) containing 0.1% Tween-20 (TBST) and 5% skimmed milk to block nonspecific protein binding. Membranes were incubated with 1 mg/mL of rabbit antivimentin polyclonal antibody, rabbit anti-b-catenin antibody (Cell Signaling Technology) and rabbit anti-GAPDH antibody (Sigma) in TBST for overnight at 4uC, washed five times with TBST, and then incubated with goat anti-rabbit IgG horseradish peroxidaseconjugated secondary antibody (1:5000) for 1.5 h at room temperature. Immunoreactive proteins were detected using the enhanced chemiluminescence method (Amersham Biosciences). Twenty micrograms of 10000 rpm supernatant was diluted with buffer I to a final volume of 900 ml. The fluorogenic proteasome substrate BOC-Leu-Arg-Arg-AMC (trypsin-like) was added at a final concentration of 80 mM. Proteolytic activity was measured by monitoring the release of the fluorescent group 7-amido-4methylcoumarin (excitation 380 nm, emission 460 nm) [14].

Statistical Analysis
Data are expressed as Mean 6 SD and statistically analyzed by the two-tailed, unpaired, Student's t-test and scores were considered significant when p,0.05.

Proteomic analysis of Fluvastatin treated MDA-MB-231 breast cancer cells
In this study we analyzed the differentially regulated protein profiles in MDA-MB-231 metastatic breast cancer cells treated with fluvastatin. For this, cells were treated with fluvastatin (10 mM) for 24 h and protein extracts were prepared according to the procedure mentioned in the methods section. Proteins were resolved using 2D SDS-PAGE in a pH range of 4-7 with a molecular weight range between 10 and 150 kDa. With this approach, about 800 protein spots per gel were detected with more sensitive colloidal comassie staining (Fig. 2). The protein expression patterns were compared using the PDQUEST software for quantitative and qualitative differences. The 2-DE image analysis employed for the identification of proteins from protein spots of altered expression revealed 39 spots with differential expression in MDA-MB-231 cells treated with fluvasatin. Among the identified proteins, 25 proteins displayed increased levels and 14 proteins displayed reduced levels in fluvastatin treated cells compared to untreated conditions. After in-gel digestion, these proteins were identified by LC-MS/MS. The list of altered proteins and the degree of differential expression is presented in Table 1. From these identified proteins, we found more than one spot containing the same protein identity with different molecular weight and pI, which may be due to protein fragmentation induced by fluvastatin treatment. A selection of differentially expressed proteins both up regulated and down regulated are presented in Fig. 3 A and B respectively as enlarged gel images. To further confirm the reproducibility of the observed differences between the triplicate gels within each group, we have performed the correlation coefficient analysis using PDQUEST software. The results showed that the correlation coefficient between the untreated cell lysate gels was 0.94 ( Fig. 3C and 3D) and between the fluvastatin treated cell lysate gels was 0.97 ( Fig. 3E and 3F).

Hierarchial clustering, Gene Ontology (GO) and Network analysis classification of differentially regulated protein
To understand the patterns of differentially regulated proteins, unsupervised hierarchical clustering has been performed. From the presented heat map (Fig. 4A), we identified one major (C1) and two (C2&C4) minor clusters as up regulated and two major clusters (C3&C5) comprising of down regulated proteins. However, this analysis showed that none of these protein clusters represented any of the particular biological pathway or molecular functions related to cancer but nonetheless, protein clusters represented multiple pathways and cellular functions. Therefore to further understand the role and relevance of different proteins that are precisely involved in statin mediated breast cancer cell death, all the differentially regulated proteins were mapped to various biological networks using IPA analysis. DAVID GO annotation analysis identified that most of the proteins that were differentially regulated between fluvastatin and untreated cells clustered in biological processes and molecular functions regulating cell adhesion, cell communication and cell cycle ( Fig. 4B and  4C). In addition to the GO analysis, protein network analysis was computed using the Ingenuity Pathways Analysis (IPA) tool and the list of most significant networks is presented in Table 2. This analysis showed that the networks presented in Fig. 5A were significantly altered in MDA-MB-231 cells up on exposure to 10 mM fluvastatin for a period of 24 h. Global proteome network of fluvastatin-mediated differentially regulated proteins directly or indirectly are linked with many protein hubs that are highly interconnected to RhoA, Akt, ERK1/2, PI3K, NF-kB complex, vimentin, ubiquitin C and 26S proteasome networks, which are known to be involved in controlling cell proliferation. From sub network analysis as shown in Fig. 5B, it was observed that the altered cytoskeletal proteins are highly interconnected through vimentin as a central node.

Fluvastatin cause dysregulation of vimentin, a major cytoskeletal protein in MDA-MB-231 cells: Effect of mevalonate
Notably, from the network analysis and GO results it was observed that the cytoskelatal protein vimentin formed a major hub connecting other differentially regulated proteins directly or indirectly. Intriguingly, our proteomic data revealed that vimentin was identifed in 2D gel spots at three different pI's and molecular weight regions. To gain more insights on this discrepancy, we then measured mRNA levels of vimentin using full length primers to see if there are any alterative spliced isoforms of vimentin in untreated and fluvastatin treated MDA-MB-231 cells. However, we could not detect multiple gene products of vimentin which indicates that vimentin has no alternatively spliced isoforms in MDA-MB-231 cells (Fig. 6A). Next, to see the possibility of the involvement of proteolysis playing a role in the degradation of vimentin and thereby leading to the appearance of the observed 3 spots with low molecular weight (,48, ,28 and ,20 kDa) compared to full length protein (,57 kDa), we have analyzed the tryptic peptides identifiedby mass spectrometry. Based on the MS-MS analysis, it appears that vimentin is truncated at either N-terminal or Cterminal regions leading to the generation of two spots spanning from amino acids 101-236 and 207-466 respectively with different molecular weights as seen in the 2D gel separation (Fig. 6B). The other spot (with observed molecular weight of ,48 kDa) was identified from peptides spanning from N-terminal to C-terminal region. These observations suggest that vimentin undergoes controlled proteolysis after fluvastatin treatment in MDA-MB-231 cells. In the vimentin sub network, we have also observed that many of the cytoskeletal and their regulatory proteins such as vinculin, actin c1, cofilin and tubulin a 1C are directly or indirectly connected to the regulation of vimentin. This suggests that vimentin may play a pivotal role in fluvastatin mediated cytotoxicity in MDA-MB-231 breast cancer cells. Furthermore, to validate the altered regulation of vimentin with fluvastatin treatment that was identified with 2DE coupled MS-MS analysis; we have checked its levels by Western blot analysis. For this, cells were treated with various concentrations of fluvastatin (0-20 mM) as a function of time. The results showed that fluvastatin dose-and -time dependently decreased vimentin levels ( Fig. 6C and 6D). Similar results were observed in BT-549 cells (Fig. 6E). These results are in agreement with the altered vimentin levels seen in proteomic studies upon fluvastatin treatment. On the other hand, interestingly, fluvastatin did not affect vimentin levels in normal mammary epithelial (MCF-10A) cells (Fig. 6F)wherein, fluvastatin did not induce any appreciable cell death (Fig. 1C). This result indicates that fluvastatin exerts distinct effects in normal and cancer cells. To determine whether the statin-induced decrease in vimentin is mediated through HMG-CoAR inhibition, MDA-MB-231 cells were co-treated with fluvastain (10 mM) in the presence or absence of mevalonate (25 mM) for 24 h. It was found that mevalonate significantly restored fluvastatin-mediated down regulation of vimentin levels (Fig. 6G). This result implicate that mevalonate down stream products are responsible for regulating vimentin and a reduction in vimentin levels may in part contribute to fluvastatin-mediated cell death in MDA-MB-231 cells.

Caspase-3 but not proteasomal activation down regulates vimentin levels in MDA-MB-231 cells
To gain more insights into the statin-mediated dysregulation of vimentin levels through increased proteolysis, we initially measured 26S proteasomal activities (both trypsin-like and chymotrypsin-like) in cells treated with various concentrations (0-20 mM) of fluvastatin for 24 h. I Fluvastatin dose-dependently increased the trypsin-like activity of the 26S proteasome from 0-5 mM of fluvastatin and tapered off from there (Fig. 7A). However, the chymotrpsin-like activity was unchanged under the same experimental conditions (Fig. 7A). Fluvastatin started to show a significant activation of trypsin-like activity by 12 h and reached peak by 24 h (Fig. 7B). Next, to see if the increased proteasomal activity was responsible for the regulation of vimentin in fluvastatin treated cells, we measured both b-catenin, an upstream regulator of vimentin in the presence or absence of MG-132 (proteasomal inhibitor). Results showed that fluvastatin-mediated down-regulation of b-catenin was significantly reversed in the presence of MG-132 (Fig. 7C). Treatment of cells with MG-132 alone greatly accumulated b-catenin even when compared to untreated conditions (Fig. 7C). However, under the same experimental conditions, to our surprise, vimentin levels were unchanged with MG-132 treatment (Fig. 7C). This result suggests that an increased level of b-catenin due to proteasomal inhibition per se was not responsible for the regulation of vimentin in fluvastatin treated MDA-MB-231 cells. To resolve this discrepancy, we then went on to check the effect of caspase-3 inhibitor in regulating vimentin levels because we recently reported that fluvastatin treatment doseand time-dependently increased caspase-3 like protease activity in TNBC cells like MDA-MB-231, MDA-MB-453 and BT-549 cells [7]. Interestingly, caspase-3 inhibitor, Boc-D-FMK (50 mM) treated for 12 h in the presence of fluvastatin (5 mM) greatly rescued the fluvastatin-mediated down regulation of vimentin (Fig. 7D). Thereby suggesting that caspase-3 activity but not 26S proteasomal activity regulates vimentin levels in these cells under fluvastatin treated conditions.

Discussion
A very recent study showed that the basal type breast cancers are highly sensitive to fluvastatin and it is negatively associated with expression of estrogen receptor alpha (ERa) [15]. Though different mechanisms have been proposed for the anti cancer potential of statins, information is still limited on the global proteome or gene expression changes associated with statin treatment to identify possible molecular signatures responsible for their anti cancer properties. Hence, in the present study we focussed on a proteome analysis of MDA-MB-231, a triple negative metastatic breast cancer cell type challenged with fluvastatin to determine the protein alterations associated with this treatment employing 2DE coupled mass spectrometry. Our proteomic analysis data showed differential expression of 39 spots representing 35 different proteins in MDA-MB-231 cells treated with fluvastatin compared to untreated conditions. The identified proteins represent various biological functions known in different cancers. Lipophilic statins are effective against triple negative breast cancer and by and large; it appears that different lipophilic statins induce a similar response in mediating anti-cancer effects.
In an earlier study with lovastatin, it was shown that GTPase mediated cell signaling proteins like CDC42, pG3PI1 were up    lovastatin for a period of 48 h and in the current study, cells were treated with 10 mM fluvastatin for a period of 24 h. Nevertheless, we have observed ANXA1, CLIC, CFL1, EEF1D, CAPG, RhoA and VCL proteins are being regulated commonly by either lovastatin or fluvastatin [9]. In search of the molecular functions and biological processes involved with the list of identified proteins by 2DE coupled mass spectrometry, we adapted Gene Ontology (GO) approach [16]. Using GO, these proteins were classified into several groups related to the regulators of cell proliferation, migration and apoptosis. Many of these altered proteins are involved in binding and within this class, majority of them are enriched with calcium ion binding and GTPase activity protein binding. Small G-proteins regulate many cellular functions including cytoskeletal rearrangement, cell motility, intracellular trafficking, transcriptional regulation, cell growth, and development [17]. It is possible that statins by depleting isoprenoids may alter many of the cytoskeletal protein functions and disrupts cell morphology. Statins are known to involve in inhibiting cancer cell proliferation by arresting the cell cycle at the G0/G1 phase and induce apoptosis [6,18]. To corroborate this, GO analysis showed that cell cycle regulatory proteins such as SKP1 and S-100A6 were up regulated with fluvastatin treatment which in turn have a role in inducing apoptosis in breast cancer cells. Furthermore, mapping of the differentially expressed proteins to biological networks curated in database Ingenuity pathway Analysis (IPA) has identified several significant subnet works. Among the networks built from the current analysis, most significant sub network formed with vimentin as a central node connecting to several sub nodes. This interesting observation suggests a possible role for vimentin in statin mediated cell death. Vimentin, an epithelial-mesenchymal transition marker, is elevated in various epithelial cancers including breast cancer and recently it was also highlighted as a potential therapeutic target for cancer because of its role in proliferation, migration and invasion [19,20]. Further, it was reported that loss of cytokeratin and gain of vimentin expression are indicators of biologically aggressive breast carcinoma [21]. Vimentin, Zeb1 and Sip1 are high in TNBC compared to non-TNBC cells and it was also shown that over expression of vimentin is associated with poor prognosis of breast cancer [22].  [23]. Along these lines, we have also recently reported that fluvastatin induces caspase activity in breast cancer cells [7]. In the present study, we found that statin mediated down regulation of vimentin was rescued in presence of caspase-3 inhibitor (BOC-D-FMK). Therefore, it is likely that fluvastatin induce proteolysis of vimentin at Asp85, which in turn facilitates apoptosis in MDA-MB-231 cells by activating caspase-mediated   proteolysis. Vimentin, apart from its cytoskeletal regulation, has been identified as a downstream target of major cell signaling cascades including ERK, AKT, STAT3 and PI3K [24][25][26]. Hence, statins by altering these signaling pathways may accelerate vimentin proteolysis which in turn disrupts the cellular organization, reduces metastasis and induce apoptosis.
Recently, it was reported that simvastatin in combination with irinotecan induces apoptosis and cell cycle arrest by inhibiting the proteasome activity in A-549, NCI-H460 human non-small cell lung cancer cells [27]. Furthermore, farnesyl and geranyl geranyl transferase inhibitors and other agents containing active lactone moiety arrest cells at G1stage by inhibiting proteasome and upregulation of p21 in HS578-T, MDA-MB-231 and MDA-MB-436 breast cancer cells [28]. In contrast to this, it was also reported that statins (both lipophilic and hydrophilic) did not alter the 20s proteasome activity in mammalian endothelial and vascular smooth muscle cells [29]. This difference may be due to the different cell types employed in those studies i.e., transformed vs normal cells. Along these lines, in this study we were interested in understanding the involvement of proteasomal enzyme activities in fluvastatin-mediated MDA-MB-231 cell death and also their relevance in the regulation of vimentin proteolysis. We observed that fluvastatin increases trypsin-like but not chymotrypsin-like proteosomal activity of the 26S proteasome. But to our surprise, fluvastatin-mediated vimentin down regulation was not rescued in presence of MG-132, a proteasomal inhibitor. This suggests fluvastatin-induced vimentin degradation is not mediated by proteasomal enzymes. It is known that the vimentin promoter is a direct target of b-catenin/TCF transactivation in breast cancer cells and the cytoplasmic and nuclear localization of b-catenin in mammary human epithelial cell lines is associated with vimentin expression leading to high invasive/migratory potential [19]. Interestingly, in the present study it was found that b-catenin, an upstream regulator of vimentin was significantly down regulated with fluvastatin treatment which was reversed in the presence of MG-132. However, this reversal was not seen with vimentin levels. Treatment of MG-132 alone to MDA-MB-231 cells greatly accumulated b-catenin indicating that b-catenin but not vimentin is regulated by proteasomal machinery. This clearly indicates that b-catenin is not the only upstream regulator of vimentin in MDA-MB-231 cells. It is known that Smad interacting protein-1(SIP1) directly regulates vimentin expression in human mammary epithelial (MCF-10A) cells and also the activation of the vimentin promoter by SIP1 was not associated with the activation of the bcatenin/TCF/LEF signaling pathway [30]. Therefore, it is tempting to speculate that SIP1 regulates vimentin levels in MDA-MB-231 cells. Apart from vimentin, translationally controlled tumor protein (TPT1) and elongation factor 1-d (EEF1D) are among the statin mediated down regulated proteins that were identified as multiple protein spots in 2D-gel. Interestingly, TPT1/ TCTP protein is known to act as anti apoptotic protein in cancer cells and is highly expressed in various cancers including breast cancer. Recently it was also reported that high-TPT1 status associates with poorly differentiated, aggressive grade breast tumors, predicting poor prognosis [31]. Similarly, Elongation factor-1 (EF-1) delta is one of the four subunits of elongation factor-1 complex involved in elongation step during the protein synthesis. Previously it was shown that EF-1 delta mRNA expression was significantly higher in cancerous compared to noncancerous oesophageal tissues, and a higher expression of EF-1 delta mRNA was correlated with lymph node metastases, advanced disease stages and poorer prognosis for patients with oesophageal carcinoma [32].

Conclusions
The present study reports that fluvastatin induces breast cancer cell death by altering various cellular proteins. Among the differentially regulated proteins, three proteins were observed at more than one spot identified as vimentin, translationally controlled tumor protein and eukaryotic elongation factor 1 delta. Perhaps for the first time, we show that statins target vimentin by inducing caspase mediated proteolysis. Vimentin proteolysis was rescued in the presence of mevalonate and may play a role in protecting from statin-induced breast cancer cell death as vimentin expression was not altered in normal mammary epithelial cells. Importantly, statin mediated vimentin down regulation and other differentially regulated proteins may be used as biomarkers upon further validation in TNBC patients.