Anti-cancer effect of bee venom on human MDA-MB-231 breast cancer cells using Raman spectroscopy

: We demonstrated the apoptotic effect of bee venom (BV) on human MDA-MB-231 breast cancer cells using Raman spectroscopy and principal component analysis (PCA). Biochemical changes in cancer cells were monitored following BV treatment; the results for different concentrations and treatment durations differed markedly. Significantly decreased Raman vibrations for DNA and proteins were observed for cells treated with 3.0 µg/mL BV for 48 h compared with those of control cells. These results suggest denaturation and degradation of proteins and DNA fragmentation (all cell death-related processes). The Raman spectroscopy results agreed with those of atomic force microscopy and conventional biological tests such as viability, TUNEL, and western blot assays. Therefore, Raman spectroscopy, with PCA, provides a noninvasive, label-free tool for assessment of cellular changes on the anti-cancer effect of BV.


Introduction
Bee venom (BV) extracted from honey bees is commonly used in Korean medicine to treat diseases, including pain, arthritis, tumor, and skin diseases [1][2][3]. It is composed of a complex mixture of biologically active peptides, including melittin (a major component of BV), apamin, adolapin, mast-cell-degranulating (MCD) peptide; enzymes (phospholipase A2, and hyaluronidase), and non-peptide components (histamine, dopamine, and norepinephrine), which have a variety of pharmaceutical properties [2][3][4][5]. Recent studies have shown that BV has anti-cancer effects, including induction of apoptosis and inhibition of proliferation in various cancer cells, such as prostate, breast, lung, liver, ovarian, and bladder [3,[6][7][8][9]. The efficacy of BV appears to be due to the synergistic effect and selective cytotoxicity of melittin, and this anti-cancer peptide might be a better choice than the native form of BV [10,11].

Viability assay
The cells were seeded (2 × 10 5 cells per well) in 96 well plates, subsequently the subconfluent cells were treated with BV (0.7, 1.5, 3.0, 6.0, and 12.5 µg/mL) or vehicle (saline) for 12 h, 24 h, 48 h and 72 h. After treatment, cell viability was measured using the Cell Counting Kit-8 (CCK-8) according to the manufacturer's instructions. Briefly, the CCK-8 solution (10 µL per 100 µL of medium in each well) was added, the plates were incubated at 37°C for 1 h, and the absorbance of each well was read at 450 nm using a microplate reader.

TUNEL assay
The Apoptosis Detection System developed by Promega was used according to the supplier's protocol. Briefly, after treatment with 3.0 μg/mL of BV for 12 h, 24 h and 48 h, the cells were washed twice with PBS, and 5 mL of 1% ice-cold paraformaldehyde was added for 20 min. The cells were incubated for 1 h at 37°C in the dark in 50 μL of equilibration buffer containing fluorescein-12-dUTP in the presence of terminal deoxynucleotidyl transferase to label 3′-OH ends of fragmented DNA. The reaction was stopped by adding 1 mL of 20 mM EDTA, and the cells were washed with PBS containing 0.1% Triton X-100 and 5.0 µg/mL of bovine serum albumin (BSA). The cells were stained with a mixture of 5.0 μg/mL of propidium iodide (PI) and 250 μg of RNase A, incubated at room temperature in the dark for 30 min before analysis by fluorescence microscopy (Axiovert S 100, Zeiss, USA) and captured using an Axiocam MRc5 CCD camera (Carl Zeiss, Germany) at × 200 magnification.

Flowcytometry analysis
Surface exposure of phosphatidylserine (PS) by apoptotic cells was measured by flow cytometer by adding annexin V-V-FITC (Becton Dickinson, USA) according to the manufacturer's specifications (Annexin V Detection Kit), and stained simultaneously with PI. Samples were analyzed on a FACSort flow cytometer using standard Lysis II software (Becton Dickinson, USA). Excitation was set at 488 nm, and the emission filters used were 515-545 nm (green, FITC) and 600 nm (red, PI).

Western blot analysis
Proteins were extracted from MDA-MB-231 cells treated with BV (0.7, 1.5, 3.0 µg/mL) or vehicle (saline) for 12 h, 24 h, and 48 h. The cells were lysed with protein lysis buffer, and the protein concentrations were determined by Bradford protein assay. Determination of microgram quantities of protein in the Bradford Coomassie brilliant blue assay (Bio-Rad, Laboratories, Inc., CA, USA) is accomplished by measurement of absorbance at 590 nm) [32]. The proteins were separated using 12% SDS-PAGE, and then transferred to Hybond-ECL [enhanced chemiluminescence] membranes (Amersham, Little Chalfont, Buckinghamshire, UK). The membranes were blocked with 6% non-fat milk dissolved in TBST buffer (10 mM Tris-Cl [pH 8.0], 150 mM NaCl, 0.05% Tween 20). The blots were then probed with various rabbit polyclonal antibodies for PARP, Caspase-8, Caspase-9, Caspase-3 (0.5 μg/mL; Cell Signaling Tech., MA, USA), β-actin and or non-immune mouse IgG (0.5 μg/mL; Sigma-Aldrich Co, MO, USA) diluted 1:1,000 in Tris-buffered saline at 4°C overnight; and then incubated with 1:2,000 dilutions of goat anti-rabbit IgG secondary antibody coupled with horseradish peroxidase. The membranes were washed with TBST between each treatment. The blots were developed using the ECL method (GE Healthcare).

AFM measurement
Non-contact mode AFM images were obtained using a NANOstation II (Surface Imaging Systems, Herzogenrath, Germany) equipped with a 42.5 μm XY/4 μm Z scanner and a Zeiss optical microscope (Epiplan 50). AFM was performed on an active vibration isolation table (TS-150, S.I.S., Herzogenrath, Germany) inside a passive vibration isolation table (Pucotech, Seoul, Korea) to eliminate external noise. Non-contact mode AFM imaging was performed at room temperature using a silicon cantilever with aluminum reflex coating (Budget Sensor, Bulgaria). The material properties and dimensions of the probe used in this study was as follows: resonance frequency, 190 kHz; force constant, 48 N/m; cantilever length, 225 μm; cantilever width, 38 μm; cantilever thickness, 7 μm; tip radius, < 10 nm; and tip height, 17 μm. The MDA-MB-231 cells were scanned at a resolution of 512 × 512 pixels with a scan rate of 0.6 lines/s.

Raman spectroscopy
For the Raman spectroscopic measurements, we used gold-coated substrate. A thin 10 nm, Cr layer followed by a 50 nm gold layer were e-beam evaporated on the glass substrate. The Cr layer was included to increase the adhesion between the gold film and the glass substrate. This substrate was minimized spectral contributions from the sample substrate such as glass and enhanced the Raman signal compared to that. Pure metals are known to have no Raman spectral features and very low background signal. For the isolated single cells, the relative position of the laser can potentially affect the spectrum. Thus for all the isolated single cells used, the position of the spot was retained the same in relation to the cell, which the Raman spectra were obtained centrally over the nucleus of the cells when visible. At least ten individual cells were selected from each cell-group for measurement. Raman spectra were acquired using the SENTERRA confocal Raman system (Bruker Optics Inc., Billerica, MA, USA) equipped with a 785-nm diode laser source (100 mW before objective) and a resolution of 3 cm −1 . A 100x air objective (MPLN N.A. 0.9, Olympus), which produces a laser spot size of ~1 µm was used to collect Raman signals and focus the laser on a single cell. The raw spectra were loaded into OPUS software (Bruker Optics Inc., Billerica, MA, USA) to remove cosmic rays manually and subtract the autofluorescence background from every sample spectrum. Raman spectral acquisition and pre-processing of preliminary data, such as baseline subtraction, smoothing, normalization and spectrum analysis were carried out using OPUS software. An automated algorithm for autofluorescence background removal was applied to the measured data to extract pure sample Raman spectra. Baseline correction was performed using the rubber-band method, which was used to stretch between the spectrums endpoints. The baseline-corrected Raman spectra were normalized using the vector normalization method. All Raman measurements were recorded with an accumulation time of 60 s in the 600-1750 cm −1 range. The Raman spectra of the cells were calculated as the average of ten measured samples.

Multivariate analysis
Principal component analysis (PCA) was used for statistical analysis. It is one of the multivariate statistical techniques for reducing high-dimensional data to low-dimensional data. The goal of the PCA is to find the fewest uncorrelated variables, called the principal components (PCs), to account for the maximum variance of the data. As a result, the first principal component has the largest variance possible, and each subsequent principal component has the highest possible variance in orthogonal position with the preceding components. For PCA, MATLAB's statistics and machine learning toolbox was used. The data sets used for PCA, include Raman spectral data sets based on various concentrations and time periods. Raman spectral data based on various concentrations consist of 40 observations; 10 observations for each concentration (i.e., 0, 0.7, 1.5, and 3.0 μg/mL). Data based on various time periods have a total of 40 observations; 10 observations for each time period (i.e., 0, 12, 24, and 48 h) Raman intensities measured at 2301 frequencies (600-1750 cm −1 ) were used as original variables for data analysis. From the PCA-based spectrum data sets, the principal component scores and loadings for each PC were calculated. The three PCs obtained account for a variance in all

Statistica
Values are ex performed for ANOVA. The were consider

Cytotoxic
To evaluate th normal cells, inhibited the p were obtained Fig. 1 To identify the apoptotic effect of BV in MDA-MB-231 cells, we performed the TUNEL assay. We observed many TUNEL-positive cells with small, dense, and fragmented morphology emitting yellow fluorescence, whereas the PI stained nucleus in the control cells exhibited round morphology and emitted red fluorescence as shown in Fig. 2(A). To confirm the mode of BV-induced cell death, we stained the MDA-MB-231 cells with annexin V/PI for 12 h, 24 h and 48 h after exposure to 3.0 μg/mL of BV, and analyzed by flow cytometry. Figure 2(B) shows a time-dependent increase in the number of annexin V-positive cells, such as 12.0%, 13.3%, and 15.2% for 12 h, 24 h and 48 h of exposure, respectively, was observed, suggesting that BV initiated apoptosis in this cell population and it did not increase the number of necrotic cells double-stained with annexin V/PI. Caspases are believed to play a central role in the apoptotic signaling pathway. Therefore, we evaluated the expression of caspase-8, −9 and −3 in MDA-MB-231 cells following treatment with 3.0 μg/mL BV for 12 h, 24 h and 48 h. Interestingly, we observed that BV inhibits the protein level of caspase-8, −9 and −3 of MDA-MB-231 cells in a time-dependent manner ( Fig. 2(C)). Also, BV caused the proteolytic cleavage of PARP, determined by the disappearance of the full-length 116 kDa protein of MDA-MB-231 cells (Fig. 2(C)). Bee venom dose-dependently reduced the expression level of caspase-8, −9 and −3 and PARP of MDA-MB-231 cells (Fig. 2(D)).

Morphological changes in MDA-MB231 cells by BV treatment
We performed AFM measurement to observe the morphological changes in MDA-MB-231 cells following bee venom treatment. Figure 3 shows representative AFM topography images (40 × 40 μm 2 ) taken from fixed MDA-MB-231 cells that were treated with 0, 0.7, 1.5, and 3.0 μg/mL BV for 48 h (Fig. 3(A)), and 3.0 μg/mL BV for 0, 12, 24, and 48 h (Fig. 3(B)). The control cells had a regular spindle-like shape and the nuclei were elliptical and plump [33], and the cell surface seemed relatively smooth. Even after the cells were treated with 0.7 μg/mL of BV for 48 h, the morphology of the MDA-MB-231 cells did not change. However, when the cells were treated with 1.5 μg/mL of BV for 48 h, the surface around nuclei became a little rougher than that of the control cells. As the concentration of BV increased to 3.0 μg/mL, the cells changed greatly in size and shape; the cells shrunk and became round in shape. In addition, the cell membrane seemed to be rough due to protrusions and holes. And the change in surface roughness might be attributed to the damaged cell membranes by BV. Then, the time-dependent changes in MDA-MB-231 cells at 3.0 μg/mL of BV were examined ( Fig. 3(B)). A little change in cell morphology was observed after 12 h of treatment. The cell seemed deformed after 24 h treatment, for example, the cell tails shrank a little. Over the course of 48 h, the cell morphology changed greatly in size and shape, and the tail also disappeared. Our previous study reported that the toxic effects of sodium lauryl sulfate (SLS) not only changed the spindle-like shape of L929 cells into a round shape, but also made the cell surface rough [34]. Therefore, the morphological changes in MDA-MB-231 cells might be attributed to the apoptotic cell death caused by BV.  of the average wn in Fig. 4  However, the most dramatic cellular changes in the MDA-MB-231 cells, associated with DNA, proteins, and lipids, were observed upon treatment with 3.0 µg/mL of BV (Fig. 4(D)). The major spectral differences were in the peaks associated with proteins (phenylalanine at 1003 cm −1 , 1032 cm −1 , amide III at 1230-1320 cm −1 and amide I at 1658 cm −1 ), nucleic acids (U, C, and T at 779 cm −1 , and PO 2 stretching at 1098 cm −1 ), and lipid (C-H deformation at 1449 cm −1 and C = C stretching 1658 cm −1 ). The magnitude of Raman intensity at 1003 cm −1 , 1032 cm −1 , corresponding to proteins, showed a decrease, which is due to the degradation of proteins as a related of cell death. Amide III (1230-1320 cm −1 ) and amide I (1600-1700 cm −1 ) are the basic components of the protein structure, and also extremely sensitive to changes in the structure of the protein. As shown in Fig. 4(D), the amide III and amide I band widths in the BV-treated cells were narrower, compared to that of the control (gray color), which indicates the presence of a mixture of conformations, reflecting a difference in their protein secondary structure. The amide III band originates from the N-H in-plane deformation at 1276 cm −1 , coupled to the C-N stretching mode at 1243 cm −1 [43][44][45]. The amide I mode is a property of the peptide group, including the C = O stretching vibration with minor contributions from C-N and N-H motions [46][47][48]. Since the C = O stretching is differently involved in secondary structure elements via hydrogen bonds to the peptide N-H group, the band position correlates with different protein conformations. Thus, amide III and amide I bands are superimposed to get a multitude of single bands at different wavenumbers, which can be resolved in multiple components and attributed to different secondary structures [49][50][51]. Moreover, the 3.0 µg/mL BV-treated MDA-MB-231 cells showed a shift from 1449 cm -1 to 1443 cm -1 , and 1658 cm -1 to 1654 cm -1 . These Raman bands are frequently used to assign secondary structure to the proteins, and the shifts observed in these bands can be used as Raman markers for the conformational changes in the associated proteins as a result of the action of the BV. The band at 1098 cm −1 is considered as an internal standard for DNA content. As can be seen in Fig. 4(D), the intensity of the peak at 1098 cm −1 decreased in 3.0 μg/mL BV-treated cells in comparison to the control, suggesting changes in the DNA content due to drug intercalation. This observation is also supported by a similar variation observed in the guanine band at 725, and 779 cm −1 . These changes indicated nuclear DNA fragmentation as a result of apoptosis. Therefore, these bands can be considered as Raman signature for DNA intervention. It was previously reported that decrease in the Raman intensity was related to the degradation and conformational changes of proteins and DNA fragmentation, which correlated with cell death [19,21]. The intensity of the 1449 cm −1 (C-H deformation) and 1658 cm −1 (C = C stretching) bands increased the lipid-related Raman signal, indicating the presence of intracellular vesicles. This behavior is consistent with the formation of the lipid vesicles at the cell surface as well as blebbing of the cell membrane during apoptosis [52,53]. A similar increase in these bands was reported previously after the MDA-MB-231 cells were exposed to etoposide, an anti-cancer agent, resulting in apoptotic cell death [28].

BV induc
To investigate the correlation between the time dynamics of the identified apoptotic molecular events, Raman spectra of the MDA-MB-231 cells treated with 3.0 µg/mL BV were determined 0, 12, 24, and 48 h post-treatment as shown in Fig. 5(A). For a clear distinction between the spectral profiles, the BV-treated spectra were subtracted from the control (0 h) spectra ( Fig. 5(B-D)). The changes in Raman spectra were insignificant at 12 and 24 h, whereas they demonstrated dramatic cellular and molecular changes at 48 h. After 48 h, a decrease in the magnitude of Raman intensity was observed at 1003 and 1032 cm −1 (belonging to protein), and for the bands at 725, 779, and 1098 cm −1 (assigned to DNA). The intensity of the bands at 1449 and 1658 cm −1 reflected an increase in lipid-related Raman signal. As mentioned above, these results might indicate protein degradation, nuclear DNA fragmentation, and presence of intracellular vesicles leading to apoptotic cell death.

Conclusion
In this study, we demonstrated the effects of concentration-and time-dependent BV treatment using Raman spectroscopy along with multivariate analysis. The Raman spectrum for the cells treated with 3.0 µg/mL of BV for 48 h corresponded to the Raman bands assigned to DNA and protein, and demonstrated a decrease in signal intensity, which was attributed to nuclear fragmentation and protein degradation. Differences in Raman spectra between the control and BV-treated cells correlated with the cellular events during apoptosis. The results of Raman spectroscopy showed good agreement with AFM and the conventional biological assays, such as viability, TUNEL, and western blot assays performed on the same types of cells. This study provides a new method to monitor the concentration-and time-dependent multi-molecular events via measurement of the vibrations of various biomolecules in living cells. Therefore, Raman spectroscopy along with multivariate analysis might be a useful tool for a label-free and noninvasive investigation of the anti-cancer effect of BV on human breast cancer cells.

Disclosures
All authors declare that they have no conflict of interest.