Molecular crosstalk between apoptosis and autophagy induced by a novel 2-methoxyestradiol analogue in cervical adenocarcinoma cells

Flow cytometry was used to determine the effect of ESE-16 on the progression of exposed cells through the cell cycle, thus indicating an influence on cell viability. Results were calculated as percentage cells in each stage of the cell cycle. Cell cycle distribution of the cells propagated in growth medium (negative control) revealed an average of 50.82 ± 2.36% in the G₁ phase, 21.41 ± 1.85% in the S phase and 21.73 ± 2.61% in the G₂/M phase (Figure 2ia). Only 0.55 ± 0.19% of the cells were present in sub-G₁ population. There was no statistically significant difference between cells propagated in growth medium (negative control) and the DMSO vehicle control sample, both representing a cell population in logarithmic growth (Figure 2ib). Cell cycle distribution changed significantly with actinomycin D- and ESE-16 exposure. A statistically significant increase of cells in sub-G₁ was demonstrated in the actinomycin D sample (4.93 ± 2.7%, P = 0.04) (Figure 2ic) and the ESE-16-treated cells (27.93 ± 3.86%, P = 0.0002) (Figure 2id). G₁ populations were significantly decreased, to 37.29 ± 5.08% (P = 0.028) and 10.16 ± 3.79% (P = 0.0007) of the cell population in actinomycin D- and ESE-16-treated cells respectively. ESE-16 showed a decrease in viable cells (G₁- and S phases) and a distinct increase of cells in G₂/M phase (51.28 ± 7.09%, P = 0.0034). Figure 2ii summarizes the flow cytometric data of the cell cycle comparing the distribution of cells within the cell cycle in the various samples.

In order to detect and to discriminate between the induction of apoptosis, necrosis and viable cells, flow cytometric analysis using labeled annexin antibodies for the detection of phosphatidylserine flip in ESE-16-exposed cell membranes was done. Analysis of the data generated from dot plots (Figure 3i) via Cyflogic version 1.2.1 software revealed that in a healthy cell population (DMSO as a vehicle control), 92.8 ± 1.84% of the cells were viable, with 7.02 ± 1.83% undergoing apoptosis and an insignificant 0.16 ± 0.02% in necrosis. There were satistically insignificant differences between cells propagated in growth medium and DSMO vehicle controls. Actinomycin D induced apoptosis (21.28 ± 1.76%, P = 0.0006), with a significant decrease of viable cells (77.68 ± 1.89%, P = 0.0005). A decrease in cell viability was demonstrated in the ESE-16-exposed cells (64.22 ± 4.67%, P = 0.0006), with a corresponding rise in apoptotic cell numbers (27.32 ± 2.67%, P = 0.0004). There was no significant difference in the proportion of necrotic cells between the samples (P = 0.08). Results are summarized graphically in Figure 3ii.

A cyclin B1-phycoerythrin conjugated antibody (Milli-Mark™ Anti-Cyclin B1-PE, clone GNS3 (8A5D12)) was used to quantify cyclin B1 protein up-regulation in ESE-16-treated HeLa cells by employing flow cytometric analyses. This was conducted to determine the compound’s ability to induce a metaphase block. Figure 4i displays representative scatter plots, with the fluorescent intensity of the PE-conjugated cyclin B1 antibody plotted against forward scatter as a determinant of cells size (Figure 4ia and ib) and side scatter indicative of cell complexity (Figure 4ic and id). Scatter plots displayed a higher fluorescent signal in more complex and smaller sized cells which are representative of cells in metaphase [34, 35]. Figure 4ii shows the intensity histogram, demonstrating an increased intensity and a right shift of the ESE-16-exposed cells. A 2.47 ± 0.49-fold increase of cyclin B1 expression (P < 0.05) in the ESE-16-exposed cells was revealed when compared to the DMSO vehicle control.

In order to evaluate the induction of autophagy in ESE-16-treated cells, flow cytometric quantification of aggresome formation was conducted. Using the formula provide in the manufacturer’s protocol, the aggresome activity factor (AAF) was calculated. According to the manafacture’s protocol, an AAF of more than 25 is regarded as a statistically significant increase in aggresome formation. There was no statistically significant difference between cells propagated in growth medium and DMSO vehicle control samples. Figure 5 is a graphic representation of the AAF of tamoxifen- and ESE-16-treated cells, calculated at 44.3 ± 6.36 and 40.7 ± 0.53 repectively. Data indicate that ESE-16 induces aggressome formation, indicative of autophagy.

Flow cytometric analysis of autophagy-related protein LC3 B expression was done to quantify the formation of autosomal vacuoles in ESE-16-eposed HeLa cells. The MAP1LC3 B rabbit antibody (Biosensis, Thebarton, Australia) was utilized and the FL1 fluorescence of at least 10 000 cells were analyzed per repeat. There was a statistically significant increase of LC3 B within the tamoxifen (x mean FL1 = 106.5 ± 11.6; P = 0.041) and ESE-16-exposed cells (x mean FL1 = 81.2 ± 4.08; P = 0.040) when compared to the DMSO vehicle control (x mean FL1 = 66.6 ± 0.82) (Figure 6).

The FLICE/Caspase 8 colorimetric kit was used to determine caspase 8 activation in cell lysates of the ESE-16 exposed HeLa cells and of the relevant controls. There was no statistically significant difference noted between readings of cells propagated in growth medium only, DMSO vehicle and actinomycin D controls. A 1.9 ± 0.4-fold increase in caspase 8 induction in the ESE-16-exposed HeLa cells was observed when compared to the baseline (P- value 0.04) (Figure 7). Caspase 3 activity in cell lysates was measured using the BioVision Caspase-3/CPP32 Colorimetric Assay Kit. Cells propagated in growth medium and DMSO vehicle controls showed no statistical difference between these results. A statistically significant up-regulation of caspase 3 activity in the actinomycin D positive control (3.75 ± 0.21-fold increase, P = 0.011) was noted and a pronounced increase in the ESE-16-treated cells (8.03 ± 0.21-fold increase, P = 0.008) (Figure 7).

To substantiate flow cytometric data and to evaluate ESE-16’s anti-proliferative effects, PlasDIC microscopy was used. Cells exposed to DMSO as a vehicle control (v/v%) (Figure 8a) continued to grow in a logarithmic manner resulting in confluent cells in various stages of mitosis (mostly interphase). Cells treated with tamoxifen revealed increased presence of vacuolar structures (most likely to be autophagosomes) and cells in distress (cell protrusions) (Figure 8b). Cells exposed to actinomycin D revealed apoptotic bodies, multiple shrunken cells, ghost cells and scattered cell debris (Figure 8c). Shrunken and round cells in metaphase block, the formation of apoptotic bodies, as well as evidence of cell debris were visible in the ESE-16-treated HeLa cells (Figure 8d). All treated samples displayed a decrease in cell density.

In order to substantiate autophagy induction, a fluorescent microscopic technique using monodansylcadaverine (MDC) was employed as described by Laane et al.[36]. Figure 9a is a representative image of HeLa cells exposed to DMSO as a vehicle control which displayed confluent growth and non-specific MDC staining. Cells exposed to tamoxifen as the positive autophagy control demonstrated the formation of MDC-positive vacuoles (Figure 9b), as did the ESE-16-exposed cells (Figure 9c). A UV filter with excitation 380 nm and emission 420 nm was used for MDC staining detection.

TEM was used to analyze detailed intracellular and membrane properties of HeLa cells exposed to ESE-16 in order to demonstrate morphological evidence of autophagic and apoptotic cell death, both of which were hypothesized to be caused by ESE-16. Figure 10ai demonstrates the smooth cell membrane with normal cell protrusions (Figure 10aii) of cells exposed to the DMSO vehicle control. Comparative cellular morphology to cells propagated in growth medium was observed (not shown). Intracellular organelles including mitochondria, nucleoli and nuclear membranes are clearly identifiable, with no vacuolar formation. In contrast, cells exposed to the positive control agents, tamoxifen and actinomycin D, displayed characteristics of cellular distress and processes of autophagy and apoptosis respectively. Figure 10bi shows the marked increase of vacuolar structures in response to tamoxifen exposure. These vacuolar structures were most likely autophagosomes in the light of increased LC3 B expression observed as determined by flow cytometry. The latter observation was further supported by an increase in aggresomes demonstrated via a significant AAF value. Cells are smaller in size, and the nuclear membrane, nucleolus and cell membrane are still intact. Figure 10bii shows one of the vacuolar structures (most likely an autophagosome), demonstrating the engulfed cellular contents and organelles within a membrane. Actinomycin D treatment had a pronounced effect on cell membrane structure, with an increase of membrane protrusions (Figure 10ci and cii). Additionally, signs of apoptosis are evident and include pyknosis, karyorhexis and the presence of apoptotic bodies. Figure 10d demonstrates the effect that ESE-16 exposure has on HeLa cells. A significantly distressed cell (representative image of cells treated with ESE-16) in the process of undergoing both apoptosis and autophagy is depicted (Figure 10di). Combination of cell death types I and II are represented by the characteristic apoptotic body formation, chromatin hypercondensation and an increased number of cell membrane protrusions of the former and the presence of numerous vacuolar structures (most likely autophagosomes) as part of the latter. The nuclear membrane is absent. Vacuolar structures and organelles such as Golgi bodies were identified (Figure 10dii).

In order to observe the effects of ESE-16 on HeLa cell cytoskeletal microtubule architecture, confocal microscopy was used to visualize α-tubulin. Figure 11a is a representative image of HeLa cells exposed to the DMSO vehicle control. Confluency of cell growth in various stages of mitosis (mostly interphase) was demonstrated with organized spindle formation visible within these cells. A cell undergoing telophase is shown in Figure 11aii with the chromatids being separated by an organised and functional spindle formation. Cells were exposed to 2-ME as a positive control for spindle depolymerization. Figure 11bi and Figure 5bii demonstrate cells in metaphase block and disorganized tubulin structure after 2-ME exposure with a decrease in cell density. Figure 11c depicts the effect that ESE-16 exposure has on the HeLa cell microskeleton. Total disintegration of the tubulin network is visible in Figure 11ci and cii. Additional evidence of apoptosis was identified in these cells and included margination of DNA, a decrease in cell density and apoptotic body formation.

Dulbecco’s Modified Eagle Medium (DMEM) was purchased from Separations (Johannesburg, South Africa). European-grade heat-inactivated fetal calf serum (FCS) and RNase A were obtained from BIOCOM biotech (Pty) Ltd. (Clubview, South Africa). Syringe filters (0.22 μm), sterile cell culture flasks and plates were obtained through Sterilab Services (Kempton Park, Johannesburg, South Africa). Penicillin, streptomycin and fungizone were purchased from Highveld Biological (Pty) Ltd. (Sandringham, South Africa), as was the trypsin/versene. Triton X-100, 100% ethanol, and propidium iodide were purchased from Sigma-Aldrich (St. Louis, Missouri, United States of America (USA). All additional chemicals were of analytical grade and were purchased from Sigma-Aldrich (St Louis, Missouri, USA).

The non-commercially available 2-ethyl-3-O-sulphamoyl-estra-1,3,5(10)16-tetraene (ESE-16) was synthesized by iThemba (PTY) Ltd. Pharmaceuticals (Modderfontein, Gauteng, South Africa). The IC₅₀ of ESE-16 in HeLa cells was established to be 0.5 μM via a crystal violet staining technique in a dose-dependent study (data shown in Additional file 1). As a negative vehicle control, medium of the cell samples were supplemented with an equal concentration of dimethyl sulphoxide (DMSO). The DMSO content of the final dilutions never exceeded 0.1% (v/v). Positive controls constituted actinomycin D at a final concentration of 0.1 μg/ml for apoptosis, 20 μM tamoxifen for autophagy and 1 μM 2-ME for microtubule depolymerization (Sigma-Aldrich, St Louis, Missouri, USA).

The commercially available tumorigenic human epithelial cervical cell line, namely the HeLa cell line, was acquired from Highveld Biological (Pty) Ltd (Sandringham, Johannesburg, South Africa). HeLa cells were propagated and maintained in 25 cm² or 75 cm² tissue culture flasks in a humidified atmosphere at 37°C, 5% CO₂ in a Forma Scientific water-jacketed incubator (Ohio, USA). HeLa cells were cultured in DMEM supplemented with 10% heat-inactivated FCS (56°C, 30 minutes) for optimal cell growth. 100 U/ml penicillin G, 100 μg/ml streptomycin and fungizone (250 μg/L) were added to the medium for infection control. Experiments were performed in either six-well plates or 25 cm² cell culture flasks. For six-well plates, exponentially growing cells were seeded at 350 000 cells per well in 3 ml fresh maintenance medium. For 25 cm² cell culture flasks, exponentially growing HeLa cells were seeded at 1 X 10⁶ cells per flask to a final volume of 5 ml of maintenance medium. After a 24-hour incubation period at 37°C to allow for cell adherence, cells were exposed to 0.5 μM ESE-16 and the relevant controls were included. Cells were incubated for a further 24 hours at 37°C as the exposure time-frame.

Analysis was performed using the BioVision Annexin V-FITC reagent kit acquired from BioVision Research Products (Mountain view, California, USA. The 1X binding buffer (supplied in the kit) was prepared as directed. According to manufacturer’s protocol, 500 000 exposed cells from each flask were washed in PBS, centrifuged and resuspended in 1X binding buffer. Double staining was done by adding 1 μl Annexin V-FITC and 1.2 μl 40 μg/ml propidium iodide to the cells, which were incubated for five minutes at room temperature in the dark. Annexin V (FL1) and propidium iodide (FL3) fluorescence were measured in 10 000 cells per sample with a FACS FC500 System flow cytometer (Beckman Coulter South Africa (Pty) Ltd.) equipped with an air-cooled argon laser excited at 488 nm. Results were expressed as a percentage of cells in three categories namely; viable cells, apoptotic cells and necrotic cells.

Exposed cells were trypsinized, collected, centrifuged and washed with 1 ml ice-cold PBS. Cells were resuspended in 200 μl ice-cold PBS containing 0.1% FCS. Cells were fixed with 10 ml ice-cold 70% ethanol and stored at 4°C for 24 hours. Cells were centrifuged and washed twice with PBS to remove the ethanol. A working antibody solution (Milli-Mark™ Anti-Cyclin B1-PE, clone GNS3 (8A5D12), Millipore Corporation, Temecula, California, USA) was prepared by diluting the primary antibody in PBS in a 1:5 ratio. A 1:10 antibody solution to 0.1% Triton X-100 PBS was added per 1 X 10⁶ cell sample. Cells were incubated in the conjugated cyclin B1 antibody solution for 40 minutes at 37°C. Cells were washed with PBS. All procedures after the incubation with the antibody were conducted in a light protected environment. PE (FL3) fluorescence was measured with a FC500 System flow cytometer (Beckman Coulter South Africa (Pty) Ltd.) equipped with an air-cooled argon laser excited at 488 nm. Aggregated and aneuploid cells were excluded from the analysis. 10 000 cells were analyzed per sample. Measurement of PE-conjugated cyclin B1 fluorescence of control and exposed HeLa cells was done using the normalized area of the dot-plot.

The Enzo Life Sciences’ ProteoStat® Aggresome Detection Kit was purchased from Enzo Life Sciences Inc. (New York, USA). The kit provided the ProteoStat® Aggresome Detection Reagent and the 10X assay buffer. Subsequent to the 24 hour exposure, cells were harvested via trypsinization and resuspended in 200 μl PBS prior to adding 4% formaldehyde in a drop-wise manner while vortexing the tube. Cells were left at room temperature for 30 minutes. Samples were centrifuged and a PBS washing step followed. While vortexing, permeabilizing solution was added in a drop-wise fashion, after which the sample was incubated for 30 minutes on ice. Cells were collected by centrifugation, after which they were washed in PBS. Cells were resuspended diluted ProteoStat® Aggresome detection reagent. Cells were incubated at room temperature for 30 minutes in the dark. Fluorescence within 10 000 cells of each sample was analyzed in the FL3 channel of the FACS FC500 System flow cytometer (Beckman Coulter, South Africa (Pty) Ltd.) equipped with an air-cooled argon laser excited at 488 nm. After obtaining the mean fluorescence intensity (MFI) for the samples (cells propagated in medium only as the negative control, ESE-16-treated cells and a positive autophagy control), the aggresome activity factor (AAF), as determined according to the supplier’s manual, was calculated as follows:

where AAF is the aggresome activity factor; MFI_Rx is the mean fluorescence intensity of the treated sample; and MFI_CONT is the mean fluorescence intensity of the control sample. An AAF of more than 25 was indicative of a positive result and a significant increase in aggresome formation.

After a 24 hour exposure to ESE-16 and the relevant controls, cells were trypsinized and washed with ice-cold PBS. Cells were fixed with 0.01% paraformaldehyde/PBS for 10 min, then pelleted and resuspended in ice-cold PBS. Cells were permeablized with ice-cold methanol (−20°C), added in a drop-wise manner. Cells were washed twice with cold PBS. Pelleted cells were resuspended in the primary antibody solution (0.5:1000 MAP1LC3 B rabbit antibody (Biosensis, Thebarton, Australia) to PBS, 1% BSA) and incubated for 2 hours at 4°C protected from light. Cells were washed thrice with washing buffer (PBS, 0.05% Triton, 1% BSA). Blue carrier protein (FL1) fluorescence was measured with a FC500 System flow cytometer (Beckman Coulter SA (Pty) Ltd.) equipped with an air-cooled argon laser excited at 488 nm. Data from at least 10 000 cells were analyzed. Measurement of Blue Carrier Protein-conjugated MAP1LC3 B antibody fluorescence of control and exposed HeLa cells was done, using the normalized area of the dot-plot. An overlay histogram was constructed from the dot plots using Cyflogic version 1.2.1 software (Pertu Therho, Turko, Finland).

The FLICE/Caspase 8 colorimetric kit was used to determine caspase 8 activation and the BioVision Caspase-3/CPP32 Colorimetric Assay Kit was used to quantify caspase 3 activity. Both kits were purchased from BioVision Research Products (Mountain View, California, USA). Included were the cell lysis buffer, 2X reaction buffer, 4 mM IETD-p NA (in the caspase 8 kit), 4 mM DEVD-p NA (in the caspase 3 kit), 1 M dithiothreitol (DTT) and the dilution buffer. 1–5 × 10⁶ treated cells were pelleted by centrifugation after trypsinization, and resuspended in ice-cold cell lysis buffer and incubated on ice for 10 minutes. Following centrifugation, the supernatant was transferred to eppendorf tubes and placed on ice. The protein concentration was determined using the Pierce® BCA protein assay kit (Thermo Fisher Scientific Inc. Rockford, Illinois, USA) as per manufacturers protocol. Protein (50 μg) of each assay was diluted in 50 μl cell lysis buffer and placed in a 96-well plate, to which an equal volume of 2× reaction buffer (containing 10 mM DTT) was added. The reaction was initiated by adding 5 μl DEVD-p NA for caspase 3 or IETA-p NA for caspase 8. Plates were incubated for 120 minutes at 37°C, protected from light and the absorbance was read at 405 mn on the BioTek Epoch multi-volume spectrophotometer system (BioTek Instruments Inc., Analytical Diagnostic Products, Weltevreden Park, South Africa). Background readings from the buffers were first subtracted from the total readings, after which the fold increase of caspase activity was determined by using the DMSO vehicle control as the baseline reading.

Viable cells (3.5 × 10⁵) determined by trypan blue exclusion, were seeded in six-well plates and allowed to attach for 24 hours. PlasDIC images were obtained using the Zeiss Axiovert-40 microscope (Göttingen, Germany) and Zeiss Axiovert MRm monochrome camera (Carl Zeiss MicroImaging GmbH, Göttingen, Germany) before and after a 24 hour exposure to vehicle controls, positive controls or ESE-16 respectively.

Fluorescent microscopy using monodansylcadaverine (MDC) was employed to determine the effect that ESE-16 has on acidic vesicular organelle formation [36]. After discarding the growth medium, ESE-16 exposed cells (and relevant controls) were incubated in 0.05 mM MDC for 10 minutes at 37°C. Cells were washed four times with PBS. Cells were examined with a Zeiss inverted Axiovert CFL40 microscope and Zeiss Axiovert MR monochrome camera (Carl Zeiss (Pty) Ltd., Johannesburg, South Africa). A UV filter with excitation 380 nm and emission 420 nm was used for detection of MDC staining. In order to curtail fluorescent dye quenching, all procedures were performed in a dark room.

After a 24 hour exposure to ESE-16 and the relative controls, cells were collected via trypsinization and fixed in diluted Karnovsky’s fixative (2.5% glutaraldehyde and 2.5% formaldehyde in 0.075 M phosphate buffer at pH 7.4–7.6) for 45 minutes at room temperature. Cells were washed with 0.075 M phosphate buffer. Cells were fixed in 0.5% aqueous osmium for two hours, followed by a rinsing step with distilled water. Dehydration steps followed and entailed increasing concentrations of ethanol (30%, 50%, 70%, 90%, 100%, 100%, 100%). Cells were infiltrated with a series of increasing percentage of quetol epoxy resin (30% for 30 minutes, 60% for 30 minutes and 100% quetol for four hours). Embedding of samples followed and polymerization of the specimens was allowed for 36 hours at 60°C. Ultra-thin sections were then prepared with a microtome and mounted on a copper grid. Samples were contrasted with 4% aqueous uranyl acetate and Reynolds’ lead citrate. All chemical were obtained from Merck (Darmstadt, Germany). Samples were viewed with a JEM-210 °F field emission transmission electron microscope (JEOL, Tokyo, Japan) at the Laboratory for Microscopy and Microanalysis, University of Pretoria (Pretoria, South Africa).

Exponentially growing HeLa cells were seeded at 350 000 cells per well in 6-well plates on flame-sterilized cover slips. After exposure, cells on cover slips were washed with pre-warmed cytoskeletal buffer (CB) (60 mM PIPES, 27 mM, 10 mM EGTA and 4 mM magnesium sulphate with the pH adjusted to 7.0 with NaOH). Fixation of cells was done with in pre-warmed 0.3% gluteraldehyde for 10 minutes at 37°C. This was followed by a wash step with warmed cytoskeletal buffer while agitating. Cell membranes were permeablized using the permeabilization buffer (1% Triton X-100 in CB) and agitated for 15 minutes. Cells were washed once with CB buffer and then washed twice with phosphate-buffered saline (PBS). Non-reacting aldehydes were removed by treating samples three times with the reducing agent (5 mg sodium borohydrite in 5 ml PBS). Cover slips were incubated in a 10% normal host serum-blocking buffer (10% FCS in PBS containing 0.05% Triton X-100) for 60 minutes. Cover slips were incubated in a primary mouse anti-tubulin alpha antibody (monoclonal Antibody to Tubulin-alpha (clone DM1A), Imgenex, San Diego, California, USA) cocktail (1:100 ratio antibody to PBS-Triton wash buffer diluted with 50% blocking buffer) for 90 minutes at 37°C in a humidity chamber. Cover slips were washed at room temperature with PBS-Triton wash buffer (PBS with 0.05% Triton X-100) with 1% FCS. Samples were incubated while protected from light for 90 minutes at 37°C after adding secondary fluorescent labelled (alexa Fluor® 488 dye) anti-mouse antibodies (raised in donkey) (Invitrogen, Paisley, United Kingdom) cocktail (1:125 as above) with a final concentration 2 mg/ml. Samples were washed with PBS-Triton wash buffer with FCS. Nuclear staining was achieved by incubating the cover slips in 50% DAPI (Thermo Fisher Scientific Inc. Rockford, Illinois, USA) to equal volumes of PBS for five to ten minutes while protected from light. Counterstained cover slips were then washed with distilled water, mounted with a glycerol-based mounting fluid and cells were examined with a Zeiss LSM 510 Meta Confocal Microscope (Zeiss, Jena, Germany) at the Laboratory for Microscopy and Microanalysis, University of Pretoria, Pretoria, South Africa. Cells were visualized using immunofluorescent techniques in which green emissions from the Alexa-488 probe were collected at 519 nm (excited at 495 nm) and the blue DAPI at 461 nm (excited at 358 nm).

Quantitative data were obtained from three independent experiments. In addition, each of the independent experiments was conducted in three replicates and is shown as the mean ± standard deviation (SD). Quantitative data were statistically analyzed for significance using the analysis of variance (ANOVA)-single factor model followed by a two-tailed Student’s t-test. Means are presented in bar charts, with T-bars referring to standard deviations. P-values of less than 0.05 were regarded as statistically significant and are indicated by an asterisk (*). For flow cytometry, three independent experiments were done in triplicate and data from at least 10 000 cells were analyzed using Cyflogic version 1.2.1 software (Pertu Therho, Turko, Finland). Flow cytometric measurement of cell cycle progression and protocols involving FITC-labels and the aggresome kit are expressed as relative fluorescence (a percentage of the value measured for vehicle-treated negative control cells). Morphological microscopy studies (qualitative data) were repeated three times.