Albumin promotes proliferation of G1 arrested serum starved hepatocellular carcinoma cells

Cell culture

The cell line used in this study is C3A [HepG2/C3A, derivative of Hep G2 (ATCC HB-8065)] (ATCC^® CRL-10741™). Cells were routinely cultured in DMEM (Gibco) supplemented with 10% foetal bovine serum (FBS, Sigma) and 2 mM L-glutamine (Biochrom) at a seeding density of 8 × 10⁴ cells/cm² in 25 cm² flasks (Greiner Bio-one) (Fig. S1). Cells were incubated at 37 °C, 98% humidity and 5% CO₂ (BBD 6220 CO₂ incubator, Thermo Scientific). Cells were harvested by detachment using trypsin/EDTA, then neutralised with media containing 10% FBS. They were then pelleted by centrifugation at 300 ×  g, resuspended in media composed of the desired treatments at a total of 2 × 10⁶ cells (8 × 10⁴ cells/cm²). Cell counts were carried out under a light microscope using a Neubauer haemocytometer.

The treatments in this study include serum free media (DMEM with 2 mM L-glutamine, completely without FBS) as the serum starved control, serum free media containing 5 mg/ml albumin (human serum albumin, Octapharma) or serum free media containing 5 mg/ml dextran 70 (Carl Roth). The albumin used in this study is in physiological solution and charcoal treated (Hepalbin, Albutec) to reduce albumin bound stabilisers and other fatty acids prior to applying it to the cells (Chen, 1967). A concentration of 5 mg/ml albumin was applied in our experiments because it is comparable to total protein concentrations in 10% FBS used in routine cell culture.

HepG2 cells typically respond slowly to serum starvation and evidence of cell cycle arrest and apoptosis are usually delayed (Figs. S1 and S2) (Zhuge & Cederbaum, 2006), therefore tests were conducted after 72 h in culture. Cells were routinely viewed using inverted phase contrast microscope (DM IL LED, Leica) and images were captured using a mounted camera (MC120 HD, Leica).

HepG2/C3A cells were authenticated using short tandem repeat analysis and hepatocyte functionality was confirmed by measuring micro albumin synthesis and cytochrome P450 monooxygenase 1A activity. Cells were periodically screened for mycoplasma by extranuclear 4′,6-Diamidine-2′-phenylindole dihydrochloride (DAPI, Carl Roth) staining viewed under ECLIPSE Ti inverted fluorescence microscope (Nikon). Confirmatory testing of mycoplasma enzymatic activity (Mycoalert, Lonza) was carried out according to kit instructions and measured using CLARIOstar plate reader (BMG LABTECH). Both tests have been negative confirming absence of mycoplasma contamination.

Cell cycle analysis

Harvested cells were fixed in 66% Ethanol at a concentration of 10⁶ cells/ml for a minimum of 2 h at 4 °C. Cells were washed with phosphate buffer saline and incubated with 200 µl staining solution consisting of 50 µg/ml propidium iodide and 550 U/ml RNase A (Abcam) in the dark at 37 °C for 20 to 30 min. Measurements were taken at excitation wavelength of 488 nm using the FACSVerse flow cytometer (BD Biosciences). Gating was carried out using FlowJo 10.5.3 software (FlowJo) and calculated using the software’s Watson Pragmatic algorithm to correct for overlaps between the peaks.

TUNEL assay

Suspended cells were fixed with 1% paraformaldehyde (pH 7.4) for 60 min on ice. The cells were washed with PBS followed by 70% ethanol fixation at −20 °C overnight. Cells were treated with terminal deoxynucleotidyl transferase enzyme (TdT) and fluorescein isothiocyanate (FITC) deoxy uridine triphosphate (FITC-dUTP) provided in the kit (Phoenix flow systems). Measurements were taken at excitation wavelength of 488nm using the FACSVerse flow cytometer (BD Biosciences).

Immunoblotting

Cells collected after 72 h treatments were lysed in RIPA buffer (Millipore) containing protease inhibitor cocktail (cOmplete, Roche). Protein concentrations were measured colorimetrically using Pierce BCA protein assay kit (Thermo Scientific). Total protein concentrations of 15 µg for cyclin D1 and 40 µg for p21 in loading buffer (Roti-Load 1, Carl Roth) were separated by SDS-PAGE in a 12% separating gel and 5% stacking gel (Rotiphorese NF-Acrylamide/Bis-solution 30% (29:1), Carl Roth). Proteins were transferred to PVDF membrane (Immobilon P, Millipore) followed by blocking (Roti-Block, Carl Roth), incubation in primary antibody and then secondary antibody. Rabbit monoclonal to p21 [EPR3993] (1:1000, ab109199, Abcam), Rabbit monoclonal to cyclin D1 [EPR2241] (1:3000, ab134175, Abcam) and Rabbit polyclonal to GAPDH (1:1000, ab9485, Abcam) were used as primary antibodies. The secondary antibody used was HRP conjugated Goat F(ab’)2 Anti-Rabbit IgG (1:7500, ab6013, Abcam). The membrane was washed between incubations with TBST (20 mM Tris–HCl (pH 7.6), 0.137 M NaCl, 0.05% Tween 20). The blots were then developed using Pierce ECL plus (Thermo Scientific) and imaged using Fusion FX (Vilber). The blots were analyzed densitometrically using ImageJ (https://imagej.nih.gov).

Fluorescence imaging

Cells were cultured on a 24 well plate (Greiner) at a seeding density of 8 ×  10⁴ cells/cm² for 72 h with the different treatments. Unfixed adherent cells were directly stained with a mixture of 0.6 µM calcein AM (Invitrogen), 2 µM ethidium bromide (Invitrogen) and 4 µM 4′, 6-Diamidine-2′-phenylindole dihydrochloride (DAPI, Carl Roth) and incubated at 37 °C for 20 min. The cells were viewed and photographed under ECLIPSE Ti inverted fluorescence microscope (Nikon). Cells positive for membrane permeabilization were measured as the proportion of cells stained with ethidium bromide from total cell numbers stained with DAPI. A minimum of 1,000 cells and three fields of view were counted per well.

Statistics

Four biological replicates were used in each treatment unless stated otherwise. F test was used to determine equal variances. Two tailed T-test assuming equal or unequal (Welch test) variances was carried out accordingly to determine statistical significance when comparing two groups. One way-ANOVA followed by Bonferroni’s post hoc analysis was carried out when comparing multiple groups against the control. P value of less than 0.05 is considered significant (α = 0.05).

Albumin alters cell morphology and results in increased cell counts of serum starved HEPG2/C3A cells

To test the immediate effects of albumin, cells were cultured in serum free media containing 5mg/ml albumin and compared with the cells cultured in serum free media without albumin (serum starved control). Cells cultured in serum free media containing 5 mg/ml dextran, with a comparable molecular weight to albumin (70 kDa), was carried out as a control for effects that might relate to oncotic pressure.

Serum starved controls displayed a slight irregular morphology but retained the epithelial polygonal shape and remained as monolayers in colonies. Noticeable morphological differences were observed in albumin treated cells when compared to the serum starved control. Cells remained in colonies but displayed a rounded morphology, formed clusters and grew in more than one layer. Dextran treated cells demonstrated a morphological effect more in resemblance to serum starved cells and did not exhibit the rounded morphology, clusters and layering seen in albumin treated cells (Fig. 1).

The effects of albumin on total cell counts of serum starved cells was evaluated in comparison to Dextran treated cells and serum starved control. Data passed equal variance test and were analysed by one-way ANOVA followed by Bonferroni’s multiple comparison versus control (serum starved cells). Serum starvation for 72 h yielded an average total cell count of 2.8 ×10⁶ cells. Inclusion of albumin in serum starved cells cultured for 72 h significantly increased the average total cell count to 4 ×10⁶ cells (p = 0.019, n = 4, α = 0.05). Dextran treatment of serum starved cells had fewer cells (2 ×10⁶ cells) than the serum starved control but the difference was not significant (p = 0.14, n = 4, α = 0.05) (Fig. 1). These findings suggest that albumin alters cell morphology and results in increased total cell counts in serum starved cells. These effects are not simply due to osmotic pressure and are specific for albumin as these results could not be reproduced in cultures treated with an equal molecular weight of Dextran. Serum starved HepG2/C3a cells demonstrated a similar response of increased total cell counts when higher doses of albumin (25 mg/ml and 50 mg/ml) were included in the cultures (Fig. S3).

Albumin promotes proliferation but does not prevent cell death in serum starved HEPG2/C3A cells

Albumin triggered increased cell counts in serum starved HEPG2/C3A cells requires further explanation into whether these effects are a result of promoting proliferation or preventing cell loss. HEPG2/C3A cells cultured in media containing 10% FBS were maintained for 120 h until confluent (Fig. S4) and possibly contact inhibited (Kelly et al., 1992; Davis, Ho & Dowdy, 2001; Cho et al., 2005). Cell cycle analysis demonstrated that the cells cultured in 10% FBS for 120 h were 86.6% in G1 phase, 6.2% in S phase and 7.2% in G2/M phase (Fig. S4). The confluent cells were harvested then seeded at the regular seeding density (2 ×10⁶ cells) in serum starved media and serum starved media containing 5mg/ml albumin. Cell cycle analysis after 72 h in culture demonstrated a significant reduction in G1 phase in albumin treated cells (78.3 ± 3.2%) compared to serum starved cells (86.4 ± 2.3%) (equal variance t.test, p = 0.006, n = 4, α = 0.05). Whereas S phase was significantly higher in albumin treated cells (14.3 ± 3.6%) compared to serum starved cells (6.5 ± 1.5%) (unequal variance (Welch) t.test, p = 0.016, n = 4, α = 0.05). There was however no significant difference in G2/M phase between albumin treated cells (7.4 ± 1.9%) and serum starved cells (7 ± 2.2%) (equal variance t.test, p = 0.8, n = 4, α = 0.05) (Fig. 2 and Fig. S1). These results suggest that serum starvation resulted in a cell cycle arrest in G1 phase, while inclusion of albumin promoted cell cycle transition into S phase.

Western blot analysis demonstrates that levels of p21 (average relative density) expressed in serum starved cells (1 ± 0.64) is significantly reduced (equal variance t.test, p = 0.034, n = 4, α = 0.05) by the inclusion of 5 mg/ml albumin (0.12 ± 0.08). Levels of cyclin D1 between serum starved cells (1 ± 0.72) and albumin treated cells (1.07 ±0.87) were not significantly different (equal variance t.test, p = 0.0, n = 3, α = 0.05) (Fig. 2). This suggests that albumin promotes cell cycle progression in serum starved cells by suppressing p21.

Serum starvation causes cell apoptosis and necrosis in cultured cells. The effects of serum starvation are however slow in HepG2 cells (Zhuge & Cederbaum, 2006; Liang et al., 2013). This slow response is recapitulated in HEPG2/C3A cells (Figs. S2 and S5); hence, the 72-hour time point was selected for this study. Apoptosis and necrosis were measured in HEPG2/C3A cells by TUNEL assay and ethidium bromide staining respectively. Serum starved cells were 16.6 ± 7.2% apoptotic and albumin treated cells were 11.6 ± 10.2% apoptotic (Fig. 3). Although TUNEL positive cells were perceptibly lower in some of the samples when treated with albumin (Fig. S6), the effect was not statistically significant (unequal variance (Welch) t.test, n = 4, p = 0.46, α = 0.05) (Fig. 3).

Cells in culture for 72 h were stained with calcein AM (green) and ethidium bromide (red) to demonstrate presence of live and necrotic cells respectively (Decherchi, Cochard & Gauthier, 1997). DAPI (blue) nuclear stain was carried out to quantify total number of cells in the field of view. Live cells were characterised by esterase activity that causes calcein AM to fluoresce in the cytoplasm and is clearly present in all treatments. Necrotic cells were characterised by damaged cell membrane that is permeable to ethidium bromide allowing it to stain the nucleus (Fig. 4). Serum starvation resulted in 13.8 ± 4.8% necrosis. Inclusion of 5mg/ml albumin resulted in an increase in necrosis to 16.9  ± 8.9% but is not significantly higher than serum starved cultures (equal variance t.test, n = 4, p = 0.5, α = 0.05) (Fig. 4). Cells grown in 10% FBS were used as negative controls for TUNEL assay (0.8 ±0.2% apoptotic) and ethidium bromide staining (1.7 ± 0.8% necrotic) (Fig. S7). While inclusion of albumin significantly accounts for an increase in cell counts and proportion of cells in S phase during serum starvation it does not significantly protect against cell death by either apoptosis or necrosis.