CRABP1 is associated with a poor prognosis in breast cancer: adding to the complexity of breast cancer cell response to retinoic acid

Abstract

Background: Clinical trials designed to test the efficacy of retinoic acid (RA) as an adjuvant for the treatment of solid cancers have been disappointing, primarily due to RA resistance. Estrogen receptor (ER)-negative breast cancer cells are more resistant to RA than ER-positive cells. The expression and subcellular distribution of two RA-binding proteins, FABP5 and CRABP2, has already been shown to play critical roles in breast cancer cell response to RA. CRABP1, a third member of the RA-binding protein family, has not previously been investigated as a possible mediator of RA action in breast cancer.

Methods: CRABP1 and CRABP2 expression in primary breast tumor tissues was analyzed using gene expression and tissue microarrays. CRABP1 levels were manipulated using siRNAs and by transient overexpression. RA-induced subcellular translocation of CRABPs was examined by immunofluorescence microscopy and immunoblotting. RA-induced transactivation of RAR was analyzed using a RA response element (RARE)-driven luciferase reporter system. Effects of CRABP1 expression and RA treatment on downstream gene expression were investigated by semi-quantitative RT-PCR analysis.

Results: Compared to normal mammary tissues, CRABP1 expression is significantly down-regulated in ER+ breast tumors, but maintained in triple-negative breast cancers. Elevated CRABP1 levels are associated with poor patient prognosis, high Ki67 immunoreactivity and high tumor grade in breast cancer. The prognostic significance of CRABP1 is attributed to its cytoplasmic localization. We demonstrate that CRABP1 expression attenuates RA-induced cell growth arrest and inhibits RA signalling in breast cancer cells by sequestering RA in the cytoplasm. We also show that CRABP1 affects the expression of genes involved in RA biosynthesis, trafficking and metabolism.

Conclusions: CRABP1 is an adverse factor for clinical outcome in triple-negative breast cancer and a potent inhibitor of RA signalling in breast cancer cells. Our data indicate that CRABP1, in conjunction with previously identified CRABP2 and FABP5, plays a key role in breast cancer cell response to RA. We propose that these three RA-binding proteins can serve as biomarkers for predicting triple-negative breast cancer response to RA, with elevated levels of either cytoplasmic CRABP1 or FABP5 associated with RA resistance, and elevated levels of nuclear CRABP2 associated with sensitivity to RA.

Fig. 1

Distinct mRNA expression patterns and prognostic associations of CRABP1 and CRABP2 in breast tumors. a Comparison of CRABP1 and CRABP2 mRNA levels (based on normalized gene microarray signal intensity) in normal mammary tissues (n = 10) and human breast cancer subtypes (n = 176). b, c CRABP1 (b) and CRABP2 (c) mRNA levels in tumor tissues sampled from the 176 breast cancer patient cohort. The MT and GT numbers refer to individual patients. RNA levels were measured by real-time quantitative RT-PCR with GAPDH serving as the internal control. Expression levels from triplicate reactions are shown relative to MT633. d, e CRABP1 (d) and CRABP2 (e) mRNA levels in a panel of 11 breast cancer cell lines. RNA levels were analysed by real-time quantitative RT-PCR and shown as fold change relative to ZR-75-1. f Kaplan-Meier overall patient survival curves generated based on low and high CRABP1 mRNA levels determined by ROC analysis. g Kaplan-Meier overall patient survival curves generated based on low and high CRABP2 mRNA levels. n, denotes sample size; *, p < 0.05; **, p < 0.01

Fig. 2

Immunoreactivity and subcellular distribution of CRABP1 and CRABP2 in a human primary breast tumor TMA. a Western blot of CRABP1 and CRABP2 in MDA-MB-435 cells transfected with a CRABP1 or CRABP2 expression construct, respectively. b Frequency of breast tumors with different subcellular immunoreactivity scores for CRABP1 and CRABP2: 0, negative; 1, weak; 2, intermediate; 3, strong. c Selected tissue sections from a human breast cancer TMA immunostained with anti-CRABP1 and anti-CRABP2 antibodies. Nuclear (Nuc) and cytoplasmic (Cyt) scores are indicated

Fig. 3

Associations of subcellular CRABP1 and CRABP2 levels with breast cancer patient survival and Ki67 immunoreactivity. CRABP1, CRABP2 and Ki67 protein levels were determined by immunohistochemical analysis of a TMA containing triplicate cores for each tumor from a 120 primary breast tumor cohort. a-b Association of cytoplasmic and nuclear CRABP1 levels with patient survival. c-d Association of cytoplasmic and nuclear CRABP2 levels with patient survival. e Positive correlation of cytoplasmic and nuclear CRABP1 protein levels with Ki67 immunoreactivity. f Positive correlation of CRABP1 and Ki67 mRNA levels based on gene microarray analysis. g No significant correlation was observed between subcellular CRABP2 levels and Ki67 immunoreactivity. h No correlation was evident between CRABP2 and Ki67 mRNA levels. HR, hazard ratio; p, statistical significance level; r, correlation coefficient. Scores: 0, negative; 1, weak; 2, intermediate; 3, strong. As the size of samples expressed CRABP1 is small in our TMAs, tumor samples were classified into “positive” and “negative” groups for survival analysis

Fig. 4

CRABP1 and CRABP2 differentially modulate RA-induced growth inhibition and RAR transcriptional activity. a, b Western blot analysis of MCF-7 cells transfected with CRABP1 or CRABP2 siRNAs using antibodies against CRABP1 or CRABP2. c Relative growth rate of MCF-7 cells treated with the indicated concentrations of RA after transfection with non-specific (control) and specific siRNAs targeting CRABP1 or CRABP2. Significance of difference was tested using two-way ANOVA. d RAR transactivation (measured by luciferase activity) of CRABP1a-depleted or CRABP2a-depleted MCF-7 cells transfected with a luciferase reporter construct under the control of a RARE. Cells were treated with DMSO (vehicle control) or RA for 6 h before harvest. Luciferase activity (a measure of RAR activation) is shown as fold change relative to cells that were cultured in the absence of RA. e The luciferase assay was repeated with a second set of siRNAs targeting CRABP1 and CRABP2 (CRABP1b, CRABP2b). f Western blots showing ectopic expression of CRABP1 in three human breast cancer cell lines. g, h, i The effects of ectopic expression of CRABP1 on RAR transactivation (measured by luciferase activity) were examined in BT-549 (g), SK-Br-3 (h) and Hs578T (i) cells co-transfected with a CRABP1 cDNA construct and a RARE-luciferase reporter construct. Cells were treated with RA for 6 h at the indicated concentrations. Luciferase activity (a measure of RAR activation) was adjusted based on protein concentrations of individual lysates and shown as fold change relative to control cells transfected with empty vector and cultured in the absence of RA. KD, denotes knockdown; *, p < 0.05; **, p < 0.01

Fig. 5

a Subcellular localization of CRABP1 and CRABP2 in MCF-7 cells treated with RA. Cells were cultured in medium with serum for 24 h and then treated with 0.5 μM RA in serum-free medium for 6 h. An equivalent amount of DMSO was added to control cells. Cells were immunostained with anti-CRABP1 (red, upper panel) or anti-CRABP2 (red, lower panel) antibodies as described in Materials and Methods. DAP1 (blue) staining was used to visualize the nucleus. b Expression of RA-responsive genes in MCF-7. MCF-7 cells underwent two rounds of transfection with scrambled or CRABP1 siRNAs. Cells were treated with increasing concentrations of RA [lanes 1 to 6 (0, 5 × 10⁻⁵, 5 × 10⁻⁴, 5 × 10⁻³, 5 × 10⁻², 5 × 10⁻¹ μM RA)]. RNA was purified from each culture and semi-quantitative RT-PCR was carried out using gene-specific primers (Additional file 1: Table S1). c Summary of the effect of CRABP1 and RA on downstream genes and pathways. d Western blots showing the subcellular distribution of CRABP2 in SK-Br-3 cells upon CRABP1 overexpression and RA treatment. Densitometric analysis was used to quantitate CRABP2 signal intensity in the cytoplasm and nucleus relative to the cytoplasmic marker (β-tubulin) and nuclear marker (lamin A/C), respectively. Changes in band intensities are shown as fold change in relation to lane 1 (for cytoplasmic CRABP2) and lane 5 (for nuclear CRABP2)

Fig. 6

Schematic representation of the effects of different RA binding proteins on modulation of RA action in breast cancer. a Relative mRNA levels of FABP5, CRABP1 and CRABP2 in ER-negative (n = 64) and ER-positive (n = 112) primary breast cancer tissue samples. The mRNA levels for each gene were determined based on the normalized signal intensity of the gene microarray data and are shown relative to CRABP2 (set as 1) in the case of ER-negative tumors and FABP5 (set as 1) in the case of ER-positive tumors. These data provide insight as to the possible underlying cause of RA resistance in ER-negative tumors. b A schematic model illustrating the distinct roles of CRABP1, CRABP2 and FABP5 in modulating cellular response to RA in breast cancer cells. FABP5 channels RA to PPARδ/β, a nuclear receptor which promotes cell proliferation. CRABP1 sequesters RA in the cytoplasm. CRABP2 delivers RA to RAR, leading to cell growth inhibition. The balance between FABP5/CRABP1 and CRABP2 expression levels determines the cellular response to RA (cell growth promotion or inhibition)