Tris-HCl, sodium chloride, EDTA, NP-40, sodium deoxycholate, urea, thiourea, SDS, 20% glycerol, methanol, acetic acid and iodacetamide were purchased from Sigma-Aldrich (Allentown, PA, USA). Protease and phosphate inhibitors were from Pierce Biotechnology (Rockford, IL, USA); immobilized pH gradient (IPG) buffer pH 3 to 11 was from GE Healthcare (Piscataway, NJ, USA) and dithiothreitol (DTT) was from USB (Cleveland, OH, USA). Lovastatin (in its lactone and hydroxy acid form) was purchased from Toronto Research Chemicals (North York, Ontario, Canada). 3-(4,5-dimethylthiazol-2-Yl)-2,5-diphenyltetrazolium bromide (MTT) cell growth assay kits were from Millipore (Billerica, MA, USA).

MDAMB468 and MDAMB231 cell lines were from American Type Culture Collection (ATCC) and propagated according to the instructions provided. Both cell lines are ER negative, and for this study relevant differences were that MDAMB468 cells lack expression of retinoblastoma (Rb), phosphatase and tensin homolog (PTEN) and steroid and xenobiotic receptor (SXR) proteins.

For proteomics studies, cells were treated for 48 hours with 8 μg/mL lovastatin lactone or hydroxy acid. MTT assays were performed prior to proteomics studies for IC50 determination. To investigate the effects of isoprenoid intermediates of the cholesterol biosynthetic pathway, in particular geranylgeranyl diphosphate (GGPP), farnesyl pyrophosphate (FPP) and mevalonic acid on the proliferation of cells treated with lovastatin, MDAMB231 and MDAMB468 cells were treated with 2 μg/mL, 4 μg/mL and 8 μg/mL lovastatin acid or lactone and were 'rescued' by addition of 10 μM GGPP, 100 μM mevalonic acid or 10 μM FPP.

The cells were cultured in 96-well plates. Treatment occurred with lovastatin in its lactone or acid form or with a combination of lovastatin with GGPP, FPP or mevalonic acid for 48 hours. During the last four hours, 0.02% MTT solution was added and the reaction was stopped with isopropanol and 5% acetic acid. The production of purple formazan in cells treated with an agent was measured relative to the production in control cells and dose-response curves were generated with a Perkin Elmer (Waltham, MA, USA) ELISA plate reader at 525 nm.

IC₅₀ values were estimated using the Prism software (GraphPad Software Inc., Version 4.0, San Diego, CA, USA).

For proteomics studies, cells were washed twice with ice-cold PBS followed by a collection in modified RIPA lysis buffer (50 mM Tris-HCl, pH 7.4; 150 mM NaCl; 1 mM EDTA; 1% NP-40 (v/v); 0.25% sodium deoxycholate (v/v); protease and phosphatase inhibitor cocktail). After complete solubilization, cell extracts were subjected to purification using a 2-D clean-up kit (GE Healthcare, Piscataway, NJ, USA) in accordance with the manufacturer's instructions. The final solubilization was performed in chaotropic lysis buffer containing 7 M urea, 1 M thiourea, 50 mM DTT, 0.4% IPG buffer pH 4 to 7, protease and phosphatase inhibitors. The protein concentrations were determined using a BioRad Bradford protein assay kit (BioRad, Hercules, CA, USA). Samples of 300 μg cell extract (in 200 μL) were loaded onto Immobiline DryStrips (11 cm, pH 3 to 8, GE Healthcare, Piscataway, NJ, USA). Isoelectric focusing was performed on a Protean IEF cell (Biorad, Hercules, CA, USA) with the following voltage program: rehydration: 50 V, 12 hours; 1000 V, 2 hours (gradient); 6000 V, 4 hours (gradient); 8000 V, 6 hours (rapid), maximal current 50 uA per strip. Strips were equilibrated in 20 mL rehydration buffer (6 M urea, 50 mM Tris-HCl pH 8.8, 2% SDS, 20% v/v glycerol) containing 10 mg/mL DTT and 25 mg/mL iodacetamide for 20 minutes each. The second dimension was performed using a Mini-Protean Dodeca chamber (Biorad, Hercules, CA, USA) on 10.5 to 14% Criterion Tris-HCl gels (IPG + 1 well, 11 cm, Biorad, Hercules, CA, USA). Gels were washed with nanopure water and with Biosafe Coomassie-Blue Stain (Biorad, Hercules, CA, USA) or fixed for one hour in 50% methanol, 10% acetic acid and stained with Sypro Ruby protein gels stain (Invitrogen, Carlsbad, CA, USA) overnight. Prior to imaging Coomassie-Blue stained gels were washed in nanopure water for up to 24 hours and imaged on LabScan Image Scanner (GE Healthcare, Piscataway, NJ, USA) with 900 dpi. Sypro Ruby-stained gels were washed twice in 10% methanol and 7% acetic acid for one hour each and imaged on a Typhoon 8600 imager (Amersham Pharmacia Biotech, Piscataway, NJ, USA) with 532 nm laser wavelength.

Gel image analysis was carried out using the ImageMaster 2D Platinum II software version 5.0 (GE Healthcare, Piscataway, NJ, USA). The spot auto-detect function was used for all group comparisons using identical parameters. Groups were matched automatically and corrected manually if necessary. Differences in protein expression were identified using the relative volume (%Vol) option of the software. This option allows the data to be independent of experimental variations between gels caused by differences in loading or staining. Relative volume was calculated as follows 1718:

with Vol_s being the volume of spot s in a gel containing n spots.

Raw spot values were normalized using the software's ratio option according to the following equation 171819:

with central tendency being the mean of spot s.

Changes in average volume larger than ± 40% of the average spot volume and the significance level of P < 0.05 (control vs. treated group) was the criterion used for excision. Four replicates were used for each control, lovastatin lactone or acid treatment, respectively.

Proteins from excited gels spots were digested using a modification of the method by Havlis 20. Briefly, spots were destained with acetonitrile and 100 mM ammonium bicarbonate (50/50 v/v), contracted with 100% acetonitrile and then vacuum dried. Spots were rehydrated with 50 μg/ml trypsin (sequencing grade II, Worthington, Lakewood, NJ, USA) and incubated for 10 minutes on ice. Excess liquid was removed and 50 mM ammonium bicarbonate added prior to overnight incubation at 37°C. The supernatants were collected and pooled with two additional extracts using 1% formic acid with 30% acetonitrile. Pooled extracts were vacuum concentrated to approximately 10 μL and stored at -80°C until mass spectrometry analysis.

The analysis of tryptic digests was performed using a 4000 QTRAP liquid chromatography (LC) mass spectrometry (MS)/MS system (Applied Biosystems, Foster City, CA, USA) equipped with a Dionex Ultimate 3000 nano-LC system (Dionex Corporation, Sunnyvale, CA, USA).

Peptides were loaded onto an enrichment column (C18PM, LC packings 0.3 mm ID) with 3% acetonitrile (ACN) and 0.05% trifluoroacetic acid (TFA) at a flow rate of 4 μL/min. After activation of a switching valve, the peptide mixture was back-flushed from the enrichment onto the analytical column (Zorbax 300SB C18, 3.5 um, 150 × 75, Agilent Technologies, Palo Alto, CA, USA) using a gradient. Solvent A was 0.1% formic acid and solvent B was 80% CAN and 0.1% formic acid. The flow rate was 400 nL/min. Buffer B was increased from 5% to 8% in one minute and then from 8% to 45% over 39 minutes. Finally, solvent B was increased to and held at 80% for the next five minutes, after which the settings were returned to initial conditions. Spectra were collected over an m/z range of 350 to 2200 Da. Three MS/MS spectra were collected for the three most abundant m/z values. Then those were excluded from analysis for one minutes and the next three most abundant m/z values were selected for fragmentation.

Proteins were identified by searching the databases of the NCBInr (National Center for Biotechnology Information, non-redundant) and SwissProt (Swiss Institute of Bioinformatics) using ProteinPilot 2.0 with paragorn algorithm (Applied Biosystems, Foster City, CA, USA) software. Parameters used in the database search were as follows: biological modifications; fixed modification: iodacetamide alkylation of Cys; detected protein threshold: more than 1 (90%); thorough ID.

For NMR experiments, the cells were incubated with 5 mmol/L (1-¹³C) glucose (Cambridge Isotope Laboratories, Andover, MA, USA) for the last five hours prior to the perchloric acid (PCA) extraction. All cell extractions were performed using a previously published PCA extraction protocol that allowed for separation of water-soluble and lipid fractions 21. Lyophilized water-soluble cell extracts were re-dissolved in 0.5 mL of deuterium oxide, centrifuged and the supernatants neutralized to pH 7.2 in order to allow for precise chemical shift assignments. Lipid fractions were re-dissolved in a 1 mL CD₃OD/CDCl₃ mixture (1:2).

High-resolution ¹H and ¹³C-NMR experiments were performed using a Varian INOVA NMR 500 MHz spectrometer equipped with a 5 mm HCN PFG probe (Varian, Palo Alto, CA, USA). For ¹H-NMR analysis of water-soluble extracts we have used fully relaxed spectra with a standard water presaturation pulse program, whereas for analysis of lipids no presaturation pulse was used. Spectra were obtained at 12 ppm spectral width (10 ppm for lipids), 32 K data arrays, and 64 scans with 90-degree pulses applied every 14.8 seconds. The pool size of metabolites was determined based on fully relaxed ¹H-NMR spectra of extracts using trimethylsilyl propionic-2,2,3,3,-d₄ acid (TSP) as an external standard and chemical shift reference (0 ppm). The absolute concentrations of each metabolite [metabolite] were determined and normalized according to cell wet weight, as previously described 222324 and calculated using the following equation:

where integral_met is integral of respective metabolite signal divided by the number of protons; integral_TSP is integral of TSP signal divided by the number of protons; [TSP] is TSP nominal concentration; V_S is sample volume; wet weight is sample weight.

¹³C-NMR spectra with proton decoupling (composite pulse decoupling) were recorded using the C3-lactate peak at 21 ppm as chemical shift reference (spectral width was 150 ppm, 16 K data arrays, with 20 K scans applied every three seconds). For quantification of absolute concentrations of ¹³C metabolites, calculations were made according to 2526. The ¹³C-enrichments in C3-lactate were determined by the heteronuclear spin-coupling pattern in ¹H-NMR spectra as follows:

where the sum (area [¹H-¹²C] + area [¹H-¹³C]) is equivalent to the pool size of lactate. The values were corrected for 1.1% natural abundance ¹³C. ¹³C-enrichments in individual carbons of amino acids were derived from ¹³C-NMR spectra using the known ¹³C-enrichment in lactate:

where A_Met represents ¹³C carbon peak area of the metabolite, A_n.a. is its natural abundance signal intensity, and 1.1 is the percentage factor of the ¹³C-isotope. The natural abundance of ¹³C, contributing to the total intensity A_n.a. (Met), was determined using the known ¹³C-enrichment and natural abundance of lactate and correction for the pool size:

A_Lac represents the carbon peak area of lactate, [Lac] or [Met] are the pool sizes of lactate or metabolite of interest, respectively, and E_Lac is the percentage ¹³C-enrichment in lactate. The ¹³C signal intensities were corrected for nuclear Overhauser enhancement effects by comparison with the standard mixture of amino acids.

The absolute amount of ¹³C in specified carbon positions is the product of the pool size multiplied by the fractional ¹³C-enrichment.

Western blot analysis was carried out to validate proteomics 'hits'. Aliquots of frozen extracts were loaded onto Biorad 4 to 12% Bis-Tris Criterion gels and proteins separated using a Biorad Criterion cell electrophoresis system (BioRad, Hercules, CA, USA) operating for approximately two hours at 120 V and then transferred (200 mA, 5 hours) from the gel to an Immobilon-P membrane (Millipore Corporation, Billerica, MA, USA). Membranes were incubated overnight at 4°C with the primary antibody following blocking with 5% milk/BSA in PBS-Tween buffer. Antibodies used in this study included: proliferating cell nuclear antigen (PCNA), prohibitin, E2F-1, RhoGDI, Ras homolog gene family member A (RhoA), pRb, cell division cycle 42 (CDC42), PTEN (Cell Signaling Technology, Inc., Danvers, MA, USA); high-mobility group protein B1 (HMGB1), N-myc downstream regulated gene 1 (NDRG1), DJ-1, pAkt (Abcam, Cambridge, MA, USA); MutS homolog 2 (MSH2), phospho-GTPase activating protein binding protein 1 (G3BP1), pG3BP1 (GenScript, Piscataway, NJ, USA); minichromosome maintenance protein 7 (MCM7) (Biolegend, San Diego, CA, USA). After the membranes were washed three times, the secondary antibody (horseradish peroxidase (various hosts), Pierce, Rockford, IL, USA) was applied for three hours at room temperature. Membranes were subsequently treated with Pierce SuperSignal^® West Pico Solution (Pierce, Rockford, IL, USA) in accordance with the method described by the manufacturer's protocol. A UVP BioImaging Systems UV detector (BioImaging Systems, Upland, CA, USA) was used to detect the horseradish peroxidase reaction on the membrane. Densitometry data were normalized by the amount of β-actin.

All numerical data is presented as mean ± standard deviation from replicate experiments. Student's t test, or when applicable one-way analysis of variance (ANOVA) were used to determine differences between groups. Tukey's test was used as a post-hoc test in combination with ANOVA to test for significances among groups. The significance level was set at p < 0.05 for all tests (SigmaPlot-version 11.0, Systat Software, Point Richmond, CA, USA) and PASW version 18.0 (SPSS Inc., Chicago, IL, USA).

Lovastatin induces inhibition of cell proliferation in MDAMB468 and MDAMB231 cells (Figure 1). The lovastatin hydroxy acid form was slightly more effective in both cell lines with a half maximum inhibition concentration (IC₅₀) of 8 μg/mL in MDAMB468 and 5 μg/mL in MDAMB231 cells, whereas the IC₅₀ values for lovastatin lactone were 9 μg/mL and 7 μg/mL, respectively. All subsequent experiments were carried out using 8 μg/mL lovastatin in its lactone or acid form.

In rescue experiments, when cells were co-incubated with lovastatin and GGPP, FPP or mevalonate, only mevalonate and GGPP were able to fully rescue cells from the anti-proliferative effect of lovastatin, whereas FPP could only achieve a partial rescue. Upon GGPP and mevalonate co-exposure with 8 μg/mL lovastatin (acid or lactone), cells regained 92 to 98% of the proliferation rate of control cells, whereas only 67% was regained with the co-administration of lovastatin and FPP.

In order to obtain a comprehensive view of changes in the protein synthesis in response to lovastatin treatment, proteome analyses using two-dimensional gel electrophoresis were performed on MDAMB468 and MDAMB231 breast cancer cell lines (Figure 2). Both forms of lovastatin (lactone and hydroxy acid form with 8 μg/mL for 48 hours) were used for cell treatment.

Each identified protein was assigned a functional classification based on the gene ontology annotation in the Database for Annotation, Visualization, and Integrated Discovery (DAVID). The DAVID annotation tool was used for functional clustering and pathway mapping of identified protein hits. A comparison between the expressional changes of spots in the lactone or hydroxy acid group revealed that both chemical forms of lovastatin followed the same directional change through an increase or decrease in the relative protein abundance. For this reason, we combined the treatment groups and these combined protein hits were then subjected to DAVID annotation tool analysis.

Seventy-four proteins were identified as significantly changed upon treatment with 8 μg/mL lovastatin (lactone or acid form) in MDAMB231 cells, and 42 such proteins were identified in MDAM468 cells (Table 1). Despite the stronger response of MDAMB231 cells, impact by lovastatin on the biological processes was similar in both cell lines. For example, the addition of lovastatin not only influenced the major metabolic cellular pathways, such as glycolysis or pentose-phosphate shunt, it also changed expression of proteins involved in the regulation of apoptosis, stress response, cell differentiation and actin-filament morphogenesis. Furthermore, lovastatin lactone and acid exposure-induced changes in cell cycle regulatory proteins and small GTPases mediated signal transduction members.

Small GTPase family members, some of which are known to modulate Ras protein signal transduction, have been described in the literature as major targets of statins other than HMG-CoA reductase 2728.

Our proteomics data revealed a decrease in total expression of RhoA (Table 1). In addition to the total expression, a western blot analysis on membrane-bound, geranylgeranylated RhoA in MDAMB231 cells was performed and it was found that lovastatin acid caused a significant decrease in the expression of this activated RhoA form, and that only a slight decrease was caused by the lactone (Figure 3). Our data also showed that the expression of GDP dissociation inhibitor 2 (GDI-2), a protein stabilizing the inactive RhoA form, experienced a significant increase and was more pronounced in MDAMB231 than in MDAMB468 cells (Figure 3). Lovastatin also induced downregulation of unmodified and G3BP1 (Table 1, Figures 2a and 3; down-regulation of phospho-G3BP1 only with lovastatin lactone) and cofilin 1/2 proteins (Figure 2a), and an overexpression of CDC42 protein (Figure 3).

Several proteins present in breast cancer cells that are involved in regulation of cell proliferation and cell-cycle activity were significantly altered when exposed to lovastatin. Changes in the expression of the two E2F activity related cell-cycle regulatory proteins prohibitin and MCM7, were also detected. Although the expression of prohibitin increased nearly two-fold (Table 1 and Figure 4), the expression of MCM7, an essential component of the replication helicase complex 29, decreased to 28% of control (Table 1 and Figure 3). Lovastatin-induced DNA damage also had an impact on damage repair regulating pathways. We observed a downregulation of a representative member of DNA-mismatch repair (MMR) systems, MSH2 (Figure 4 and Table 1). Expression of PCNA is downregulated by both forms of lovastatin in MDAMB231 cells, with a stronger reduction in presence of the lactone form (Table 1 and Figure 5).

In both cell lines, lovastatin treatment was accompanied by the loss of cell viability. Functional clustering facilitated the identification and subsequent inclusion of a large group of proteins related to the apoptosis signaling. These included: TNF type 1 receptor-associated protein (TRAP-1), 70 kDa subunit of Ku antigen (Ku70), disulfide isomerase ER-60, DJ-1 (PARK-7; Figure 5), cofilin 1/2, heat shock 27 kDa, HMGB1, glutathione S-transferase Pi, annexins A1 and A4, and nucleophosmin (Table 1).

Lovastatin treatment altered the expression of proteins involved in the regulation of metabolic processes such as pentose-phosphate pathway (NADP metabolic process): 3-hydroxyisobutyrate dehydrogenase, 6-phosphogluconolactonase, triosephosphate isomerase 1; glycolysis: triosephosphate isomerase 1, alpha enolase, dihydrolipoamide S-acetyltransferase; and tricarboxylic acid cycle activity as indicated by decreased expression of succinate dehydrogenase (ubiquinone) flavoprotein subunit (SDHA) and dihydrolipoamide S-acetyltransferase. ATP citrate lyase, an enzyme involved in synthesis of acetyl-CoA, was downregulated as well (Table 1 and Figure 2b).

The expression of ROS scavengers peroxiredoxin 2 and peroxiredoxin 3 was upregulated, while the expression of a protein related to the family of thioredoxins, the thioredoxin domain-containing protein 12, was down-regulated (Table 1). An increase in expression levels of two isoforms of glutathione S-transferase, GST-Pi and GST omega-1, was observed (Table 1). Both of these isoforms are active in the detoxification of ROS-induced damage (Figure 2b).

In order to confirm the two-dimensional gel electrophoresis proteomics mass spectrometry data, western blot analysis was performed on selected proteins, the results of which are presented in Figures 3 to 5. When not performed on both cell lines, the analysis was performed only on the more sensitive of the two, the MDAMB231. The results obtained from the western blot analysis corresponded well with the results from the proteomics database search. The expression of the small GTPases, GDI-2 and CDC42, showed an increase in MDAMB231 cells. Analysis of the expression of the membrane-bound, active RhoA surprisingly indicated no change after exposure to lovastatin lactone, in contrast to a significant decrease during treatment with lovastatin acid. In the protein group associated with the E2F1 pathway, the expression of E2F1, as well as MSH2, MCM7 and HMGB1 was more pronounced in the lovastatin acid group than in the lovastatin lactone treatment group. Time-dependent changes were, again, more prominent in MDAMB231 than in MDAMB468 cells. The same specific trend towards higher sensitivity of MDAMB231 cells to lovastatin acid continued in the expression of proteins related to Akt signaling. Although the expression of PTEN increased, its associated regulator protein DJ-1 was down-regulated, as was pAkt itself. Conversely, NDRG1, an Akt downstream target, was upregulated by lovastatin lactone and acid.

Dihydrolipoamide S-acetyltransferase and ATP citrate lyase are enzymes that are involved in the production of acetyl-CoA. A reduction in their expression decreases production of acetyl-CoA. This has a negative effect on fatty acid and cholesterol synthesis. Our NMR data revealed a significant reduction of choline-containing phospholipids, fatty acids and cholesterol concentrations as a result of lovastatin treatment. Additionally, we identified a transporter, the sterol carrier protein-X/2, which is not only involved in cholesterol, fatty acids and phospholipids trafficking 66, but also has a high affinity for isoprenyl pyrophosphates (GGPP, FPP, GPP) 67. Its downregulation suggests that both, the production of isoprenylated intermediates and their transport are influenced by lovastatin.