ABCA1 overexpression worsens colorectal cancer prognosis by facilitating tumour growth and caveolin-1-dependent invasiveness, and these effects can be ameliorated using the BET inhibitor apabetalone

Abstract

At the time of diagnosis, 20% of patients with colorectal cancer present metastasis. Among individuals with primary lesions, 50% of them will develop distant tumours with time. Therefore, early diagnosis and prediction of aggressiveness is crucial for therapy design and disease prognosis. Tumoral cells must undergo significant changes in energy metabolism to meet increased structural and energetic demands for cell proliferation, and metabolic alterations are considered to be a hallmark of cancer. Here, we present the ATP-binding cassette transporter (ABCA1), a regulator of cholesterol transport, as a new marker for invasion and colorectal cancer survival. ABCA1 is significantly overexpressed in patients at advanced stages of colorectal cancer, and its overexpression confers proliferative advantages together with caveolin-1 dependent-increased migratory and invasive capacities. Thus, intracellular cholesterol imbalances mediated by ABCA1 overexpression may contribute to primary tumour growth and dissemination to distant locations. Furthermore, we demonstrate here that increased levels of apolipoprotein A1 (APOA1), a protein involved in cholesterol efflux and high-density lipoprotein constitution, in the extracellular compartment modulates expression of ABCA1 by regulating COX-2, and compensate for ABCA1-dependent excessive export of cholesterol. APOA1 emerges as a new therapeutic option to inhibit the promotion of colorectal cancer to metastasis by modulating intracellular cholesterol metabolism. Furthermore, we propose apabetalone, an orally available small molecule that is currently being evaluated in clinical trials for the treatment of atherosclerosis, as a new putative therapeutic option to prevent colorectal cancer progression by increasing APOA1 expression and regulating reverse transport of cholesterol.

Keywords: ATP-binding cassette transporter; apabetalone; colorectal cancer prognosis; reverse cholesterol transport.

Figure 1

ABCA1, a new putative marker for CRC prognosis. (A) ABCA1 mRNA expression in SII‐CRC and SIII‐CRC patients. (B) Correlation between ABCA1 expression and carcinoembryonic antigen. (C) ABCA1 mRNA levels of expression correlated with disease‐free survival (DFS) in SII‐CRC (left‐hand panel) and SIII‐CRC patients (right‐hand panel).

Figure 2

ABCA1 overexpression in two human CRC‐derived cell lines. (A) mRNA and ABCA1 protein levels in DLD1 and Caco‐2 cells transduced for ABCA1 overexpression. Control cell line was transduced with an empty vector containing a DNA fragment with no open reading frame (NoORF). The values in the histograms correspond to mean ± SEM. DLD1_NoORF: 1.000 ± 3.576e‐007; DLD1_ABCA1: 2.649 ± 0.3553. Caco‐2_NoORF: 1.000 ± 2.146e‐006, Caco‐2_ABCA1: 2.115 ± 0.1029. The significance of the analysis was determined by t‐test. (B) ABCA1 localization in Caco‐2 control and Caco‐2_ABCA1. DNA was stained with 4′,6‐diamino‐2‐phenylindole (DAPI) (blue); ABCA1 in green. Scale bar corresponds to 20 μm. (C) Localization of overexpressed ABCA1 in DLD1 cells. α‐Sodium potassium ATPase was used to stain the plasma membrane (in red). ABCA1 is shown in green. DNA was stained with DAPI (blue). Three sections are shown: the bottom part of the cell corresponds to the first row, then the middle portion of the cells and in the third row, the upper part of the cell is included. (D) Cholesterol efflux in DLD1 and Caco‐2 stable cell lines. n = 3; for each experiment every condition was analyzed in triplicate. The values in the histograms correspond to mean ± SEM. Significance between groups was determined by t‐test. DLD1_NoORF: 1.129 ± 0.1289; DLD1_ABCA1: 1.213 ± 0.2134. Caco‐2_NoORF: 1.000 ± 1.654e‐005; Caco‐2_ABCA1: 1.342 ± 0.08448.

Figure 3

ABCA1 overexpression gives proliferative advantages, promotes EMT and favours invasion. (A) Cell proliferation assay in cells overexpressing ABCA1. n = 3. The values in the histograms correspond to mean ± SEM. The significance of differences between experimental groups was determined by t‐test. (B) Protein localization and mRNA levels of expression of two well established markers for EMT, E‐cad (red) and Vim (green). DNA is shown in blue (DAPI). N = DLD1_NoOF; B = DLD1_ABCA1. Scale bar corresponds to 20 μm. The values in the histograms correspond to mean ± SEM. The significance of the differences between experimental groups was determined by t‐test. (C) Quantification of cell migration (left‐hand panel) and cell invasion (right‐hand panel) in DLD1 and Caco‐2 cells overexpressing ABCA1 and compared with control cells. n = 3. For each experiment, every condition was analyzed in duplicate; six fields per condition were quantified. Mean ± SEM is represented in the histograms. The significance between groups was determine by t‐test analysis.

Figure 4

ABCA1 overexpression favours spheroid formations and promotes invasion in a three‐dimensional model. (A) Representative images of cell growth in a spheroid formation assay in basement membrane matrices (Matrigel™). Scale bar corresponds to 50 μm. Growth of DLD1 spheres is represented in the middle panel. The graph presents the average of 30 different spheres. The percentages of cells that were positive for histone H‐3 phosphorylation are presented in the dot plot graph. n = 3; 10 pictures per condition were taken and quantified. The values in the histograms correspond to mean ± SEM. The significance of differences between experimental groups was determined by t‐test. DLD1_NoORF: 3.311 ± 0.6882; DLD1_ABCA1: 5.736 ± 0.7247. (B) The percentage of spheroids that present protrusions is shown in the left‐hand panel. The values in the histograms correspond to mean ± SEM. ANOVA analysis was performed to determine the significance of differences between groups (F _(4,9) = 6.146; P = 0.0115). In the dot plot graph, quantification of the longest protrusions per spheroid are represented. A t‐test analysis was used to determine the significance of the analysis. DLD1_NoORF: 3.914 ± 3.037; DLD1_ABCA1: 20.92 ± 6.231. The experiment was performed three times; 10–20 spheroids were quantified. Scale bars correspond to 100 μm. (C) Levels of expression of E‐cad and Vim mRNA. The values in the histograms correspond to mean ± SEM. The significance of differences between experimental groups was determined by t‐test. For E‐cad, DLD1_NoORF: 1.003 ± 0.003161; DLD1_ABCA1: 0.6925 ± 0.02835. For Vim, DLD1_NoORF: 0.9942 ± 0.009214; DLD1_ABCA1: 1.515 ± 0.07528.

Figure 5

ABCA1 overexpression stabilizes CAV‐1, affecting the dynamics of cellular focal adhesions. Malignant phenotypes driven by ABCA1 were reversed by APOA1. (A) ABCA1, CAV‐1 and RHOA protein levels in DLD1 stable cell lines. Caveolae staining by CAV‐1 immunoblotting is shown on the left‐hand panel. DNA is shown in blue; CAV‐1 is shown in red. (B) FA areas were measured by vinculin immunostaining. Both DLD1 (upper panel) and Caco‐2 cell lines (lower panel) are shown. Scale bars correspond to 20 μm; r.u., relative units. The experiment was performed three times. For each experiment, 10 pictures were taken and the number of FA was normalized against the number of cells present in the field. The error bars in the histograms correspond to SEM. The significance of the differences between groups was determined by t‐test analysis. DLD1_NoORF: 1.000 ± 0.02838; DLD1_ABCA1: 1.252 ± 0.03057. Caco‐2_NoORF: 1.244 ± 0.01248, Caco‐2_ABCA1: 1.305 ± 0.007392. (C) Correlation between CAV‐1 and ABCA1 mRNA expression in SII‐ and SIII‐CRC patients. (D) Protein levels upon MβCD (4 mm) treatment in DLD1 cells. (E) Migration and invasiveness under MβCD treatment assessed in both DLD1 (left‐hand histograms) and Caco‐2 (right‐hand histograms) cell lines. n = 3. For each experiment, every condition was analyzed in duplicate; six fields per transwell were quantified. The error bars in the histograms correspond to SEM. The significance of differences between groups was determined by t‐test analysis. In the migration analysis with DLD1 cells, the following results were statistically significant: DLD1_ABCA1 control: 1.000 ± 0.07007, DLD1_ABCA1 + MβCD: 0.4931 ± 0.07956. For Caco‐2 cells: Caco‐2_ABCA1 control: 1.000 ± 0.1276; Caco‐2_ABCA1 + MβCD: 0.5360 ± 0.1169. Analysis of invasion in DLD1 cells: DLD1_NoORF control: 1.041 ± 0.08831, DLD1_NoORF+MβCD: 0.7689 ± 0.09205; DLD1_ABCA1 control: 1.063 ± 0.03943; DLD1_ABCA1 + MβCD: 0.9050 ± 0.04589. (F) Analysis of focal adhesion size using vinculin detection and quantification. Cells treated with the vehicle and with MβCD (4 mm) cells are shown. The experiment was performed three times. For each experiment, 10 pictures were taken and the number of FA was normalized against the number of cells present in the field. Mean ± SEM. The significance of differences between groups was determined by t‐test analysis. DLD1_ABCA1 control cells: 1.000 ± 0.02838; DLD1_ABCA1 + MβCD: 1.192 ± 0.03278.

Figure 6

The presence of endogenous and exogenous APOA1 indicated a reduction in cell proliferation and decreased invasiveness by ABCA1 downregulation and CAV‐1 reduction. (A) Cell proliferation of DLD1 and Caco‐2 cells overexpressing ABCA1 and APOA1 alone or in the same population (DLD1_Double). The values in the histograms correspond to mean ± SEM. The significance of differences between experimental groups was determined by t‐test. (B) Matrigel™‐coated transwell analysis for invasiveness relative to control DLD1_NoORF cells. The values in the histograms correspond to mean ± SEM. The significance of differences between experimental groups was determined by t‐test. (C) Proliferation rate plots under hAPOA1 treatment. (D) Migration and invasion assays in cells treated with human recombinant APOA1. n = 3. For each experiment, every condition was analyzed in duplicate; six fields per transwell were quantified. The values in the histograms correspond to mean ± SEM. The significance of differences between experimental groups was determined by t‐test. DLD1_NoORF+DMSO: 1.026 ± 0.1288; DLD1_NoORF+hAPOA1: 0.4850 ± 0.05796. DLD1_ABCA1 + DMSO: 1.002 ± 0.08675; DLD1_ABCA1 + hAPOA: 0.6076 ± 0.1068. (E) Western blot for ABCA1 and CAV‐1 detection in cells treated and not treated with hAPOA1 (80 μg·mL⁻¹). (F) Quantitative real‐time PCR for the detection of COX‐2 mRNA in the presence of hAPOA1. In the histogram on the right, the levels of ABCA1 mRNA expression are shown for control and DLD1_ABCA1 cells. The PCR was performed in triplicate with different cell stocks. Each sample was analyzed in triplicate. The values in the histograms correspond to mean ± SEM. The significance of differences between experimental groups was determined by t‐test. For migration, DLD1_NoORF + hAPOA1: 1.123 ± 0.07083. DLD1_ABCA1 + hAPOA1: 0.5527 ± 0.06850. For invasion, DLD1_NoORF + hAPOA1: 0.8321 ± 0.1226; DLD1_ABCA1 + hAPOA1: 0.5001 ± 0.09493.

Figure 7

Treatment with apabetalone reverses the malignancy driven by ABCA1. (A) Levels of expression of EMT markers, Vim (left‐hand panels) and E‐cad (right‐hand panels). Cells were treated with RVX‐208 (40 μm) or DMSO as control. The values in the histograms correspond to mean ± SEM. The significance of differences between experimental groups was determined by t‐test. Each analysis was performed by triplicate. (B) Growth curve analysis and quantification. Both vehicle‐treated and RVX‐208‐treated cells were plotted. n = 3. D_NoORF = DLD1_NoORF; D_ABCA1 = DLD1_ABCA1; C_NoORF = Caco‐2_NoORF; C_ABCA1 = Caco2_ABCA1 cells. The values in the histograms correspond to mean ± SEM. The significance of differences between experimental groups was determined by t‐test. (C) ABCA1, APOA1 and CAV‐1 protein levels upon RVX‐208 treatment; α‐tubulin was used as loading control. (D) Quantification of migration and invasion assays, both in DLD1 (black) or Caco‐2 (grey) control cells and cells overexpressing ABCA1. n = 3. For each experiment, every condition was analyzed in duplicate; six fields per transwell were quantified. (E) Cholesterol efflux analysis in DLD1 control (black) cells and cells overexpressing ABCA1 (grey). The experiment was performed in triplicate, and for each repetition, every condition was analyzed in triplicate. (F) mRNA levels of COX‐2 together with immunodetection of ABCA1 and CAV‐1 by Western blot. Vinculin was used as loading control.

Figure 8

APOA1 silencing reversed the phenotype promoted by RVX‐208. (A) Cell proliferation assay in control cells treated with DMSO and RVX‐208 (80 μm). Scr, scrambled shRNA; shApoA1, short hairpin RNA against APOA1 expression. The histogram represents the slopes of the curves. DLD1_NoORF transfected with shApoA1 and DMSO; in red, DLD1_NoORF transfected with shAPOA1 and treated with RVX‐208. The values in the histograms correspond to mean ± SEM. Significance between experimental groups were determined by t‐test. (B) Cell proliferation assay in ABCA1 overexpressing cells treated with DMSO and RVX‐208. Scr: scrambled short hairpin RNA; shAPOA1: short hairpin RNA against APOA1 expression. The histogram represents the slopes of the curves. (C) Invasion assay performed in DLD1 cells. Left panel represents DLD1_NoORF data; right‐hand panel corresponds to DLD1_ABCA1. n = 3. For each experiment, every condition was analyzed in duplicate; six fields per transwell were quantified.