Introduction
Current clinical management of breast cancer relies on clinicopathological features as well as expression of biological markers such as estrogen receptor (ER), progesterone receptor (PR) and epidermal growth factor receptor 2 (HER2) [
1,
2]. While tamoxifen has been shown to be highly effective for the treatment of ER/PR-positive breast cancers [
3], there are no specific molecular targets for tumors that don’t express ER, PR or HER2. These triple-negative tumors, which constitute 15–20 % of breast cancers [
4], are more aggressive and less responsive to standard treatment than the more common ER/PR-positive breast cancers, and have a poorer prognosis [
5,
6].
RA and its derivatives, collectively called retinoids, inhibit growth and induce apoptosis in a variety of epithelial cancer cells and hold great promise as chemotherapeutic agents [
7‐
9]. Retinoids inhibit mitogen signalling [
10] and induce downstream signalling pathways implicated in growth arrest, apoptosis and differentiation of precancerous and cancer cells [
11‐
15]. However, with the exception of acute promyelocytic leukemia (APL) [
14,
16,
17], clinical trials designed to test the efficacy of RA and its derivatives in the treatment of cancer have produced disappointing results, primarily because of RA-induced side effects and development of RA resistance [
18,
19].
RA exerts its physiological effects by binding and activating nuclear retinoic acid receptors RARα, β and γ. RAR dimerizes with retinoid-X-receptor (RXR) and binds to retinoic acid response elements (RARE) in the promoters of target genes, regulating their transcription and function [
7‐
9]. As intracellular transporters, the cellular RA binding proteins determine RA subcellular distribution, fate and function [
20‐
22]. Two RA binding proteins, cellular retinoic acid binding protein 2 (CRABP2) and fatty acid-binding protein 5 (FABP5), have been shown to play opposing roles in mediating the RA cellular response by targeting RA to distinct nuclear receptors; i.e., delivery of RA to RARs by CRABP2 leads to inhibition of cell proliferation, whereas delivery of RA to peroxisome proliferator activated receptor beta (PPARβ) by FABP5 increases cell proliferation and causes RA resistance [
23,
24].
We previously reported that FABP5 is preferentially expressed in ER- and triple-negative breast tumors [
25], subtypes that are prone to RA resistance [
26‐
28]. Furthermore, breast cancer cells with an elevated FABP5/CRABP2 ratio show increased resistance to RA [
23,
25]. However, the FABP5/CRABP2 ratio does not always predict breast cancer cell response to RA, and RA resistance in the squamous cell carcinoma cell line COLO 16 cannot be overcome by either restoration of CRABP2 expression or an increased CRABP2/FABP5 ratio [
29], indicating that other factors are involved.
Studies designed to examine the importance of CRABP1 in the clinical outcomes of various cancers have produced conflicting results [
30‐
33]. Prior to this study, the expression and prognostic significance of CRABP1 in breast cancer had not been investigated. In light of CRABP1’s proposed role in attenuating RA activity by enhancing RA metabolism, expression of CRABP1 in breast cancer could have important implications for RA response. Here, we report the expression, clinicopathological association and function of CRABP1 in breast cancer. Our data indicate that CRABP1 is an adverse prognostic factor and a potent inhibitor of RA action in breast cancer which functions by sequestering RA in the cytoplasm rather than by enhancing RA metabolism. We propose that CRABP1 may serve as a biomarker to predict RA response and a target to optimize the efficacy of RA in breast cancer treatment.
Materials and methods
Chemicals, reagents and DNA constructs
All-trans retinoic acid was purchased from Sigma-Aldrich (Oakville, ON, Canada) and dissolved in DMSO (Sigma-Aldrich) at a concentration of 50 mM. Scrambled stealth siRNAs and gene-specific siRNAs targeting different regions of CRABP1 mRNA (nucleotides 381–405 and 484–508 of GenBank mRNA sequence NM_004378) and CRABP2 mRNA (nucleotides 418–442 and 465–489 of GenBank mRNA sequence NM_001878) were purchased from Life Technologies (Burlington, ON, Canada). The Lipofectamine RNAiMAX reagent (Life Technologies) was used for siRNA transfections. The pGL3-RARE-luciferase plasmid DNA was purchased from Addgene (Cambridge, MA, USA) and the luciferase assay system from Promega (Madison, WI, USA). Polyethylenimine (PEI) (Polysciences, Warrington, PA, USA) was used for plasmid DNA transfections. For gain-of-function studies, the entire open reading frame of CRABP1 was PCR-amplified and cloned into pcDNA3 (Life Technologies).
Cell culture and siRNA transfection
ZR-75-1, MDA-MB-468, MDA-MB-435, BT-20, T47D, BT-474, MDA-MB-231, BT-483, MCF-7, SK-Br-3, BT-549 and Hs578T breast cancer cells were cultured in Dulbecco’s modification of Eagle’s medium (DMEM) supplemented with 10 % fetal calf serum, penicillin (100 units/mL) and streptomycin (100 μg/mL). Cells were grown at 37 °C in a humidified incubator with 5 % CO
2. To knockdown CRABP1 and CRABP2, MCF-7 cells were transfected with 10 nM siRNA. The medium was replaced with fresh medium 18 h after transfection and the cells were cultured for an additional 48 h. Two rounds of siRNA transfections were performed for each experiment. Hs578T, BT-549 and SK-Br-3 cells were transfected with 7 μg of empty (control) or pcDNA3 expression construct (CRABP1 or CRABP2) as previously described [
34]. For cell proliferation assays, 10,000 siRNA-transfected cells were seeded in each well of 12-well plates and cultured overnight in DMEM containing 10 % FBS. The medium was then replaced with FBS-supplemented medium containing the indicated concentrations of RA (or DMSO as a vehicle control). Five days later, cells were counted using a Coulter Particle and Size Analyzer (Coulter Corporation, Mississauga, Canada).
Immunofluorescence analysis
MCF-7 cells were cultured on coverslips for 24 h and treated with 0.5 μM RA (dissolved in DMSO) or vehicle (DMSO) in serum-free DMEM medium for 6 h. Cells were then fixed in 1 % paraformaldehyde in PBS for 10 min and permeabilized in 0.5 % Triton X-100 for 5 min. Cells were immunostained with anti-CRABP1 or anti-CRABP2 antibodies, followed by Alexa 594-conjugated donkey anti-mouse (for CRABP1) or Alexa 555-conjugated donkey anti-rabbit (for CRABP2) secondary antibodies (Life Technologies). Images were acquired using a Zeiss LSM510 confocal microscope (Oberkochen, Germany) with a 40 ×/1.3 oil immersion lens.
Patient population
A total of 176 treatment-naïve primary breast cancer samples and 10 normal breast tissue samples from reduction mammoplasties were obtained from the Canadian Breast Cancer Foundation Tumor Bank and used for gene expression microarray analysis as previously described [
35]. Patient material and clinical information was collected under Research Ethics Board Protocol ETH-02-86-17. Tumor tissues were frozen and histologically analysed as previously described [
35].
Patients received standardized guideline-based chemo- and hormone therapies: i.e., patients with ER-positive tumors received hormone therapy, those with HER2-positive tumors received trastuzumab, high-risk node-negative disease was treated with anthracycline chemotherapy whereas anthracycline plus taxane chemotherapy was used for the treatment of node-positive disease. The 176 patients selected for this study consisted of 88 patients who experienced early relapse (<5 years after the initial treatment) and 88 patients who had not relapsed. ER, PR and HER2 status, stage and time of follow-up were balanced between the two groups. The median follow-up time for surviving patients was 4.5 years. The gene profiling data used in this publication have been deposited in NCBI [GEO Datasets: GSE22820].
RNA preparation, gene expression microarrays and RT-PCR
Total RNA was isolated from frozen human breast tumor biopsies using the TRIzol reagent (Life Technologies) and further purified with Qiagen RNeasy columns (Qiagen, Mississauga, ON, Canada). The average percent of area with tumor cells was 72.6 % and the average percent of cells that were tumor cells was 93.4 %, for all tissues analysed. Microarray hybridization was carried out as previously described [
35]. Reverse transcription-polymerase chain reaction (RT-PCR) conditions were as previously described [
25,
34]. The number of cycles for each primer pair was optimized for quantitative amplification within the exponential PCR product growth phase. PCR primers are listed in Additional file
1: Table S1. PCR amplification of human β-actin mRNA served as positive control as previously described [
25]. For real-time quantitative RT-PCR (qRT-PCR), we used the following TaqMan FAM-labeled Gene Expression Assay primers: human CRABP1 (Hs00171635_m1), human CRABP2 (Hs00275636_m1) and human GAPDH (Hs03929097). cDNA samples were analysed in triplicate, with each cDNA undergoing 40 cycles of amplification in 96-well reaction plates (10 μL volume) using TaqMan Gene Expression Master Mix (Applied Biosystems 7900HT Real-Time PCR System).
Generation of tissue microarrays and immunohistochemical (IHC) staining
Tissue microarrays (TMA) (TMArrayer, Pathology Devices) were generated using all available formalin-fixed paraffin-embedded breast tumor tissues representing 120 patients out of the 176-patient cohort used for gene expression analysis. The TMA slides contained triplicate core tissue samples (0.6 mm in diameter) from each tumor. TMAs were immunostained with anti-CRABP1 monoclonal antibody (Sigma-Aldrich; 1:200 dilution) and anti-CRABP2 polyclonal antibody (Protein Tech Group; 1:200 dilution). The signal was detected using EnVision + anti-mouse (CRABP1) or anti-rabbit (CRABP2) secondary systems (DakoCytomation, Carpinteria, CA). Tissues were counterstained with hematoxylin. Cytoplasmic and nuclear staining were scored separately based on the average staining signal intensity throughout the tumor tissue on a scale of 0 (negative), 1 (weak), 2 (moderate) and 3 (strong). Of the 120 tumor samples tested, 105 and 106 had sufficient tissue for analysis of CRABP1 and CRABP2 immunoreactivity, respectively.
Western blotting
Cytoplasmic and nuclear extracts were prepared according to Dignam
et al. [
36], and whole cell lysates were prepared using Dignam buffer A plus 0.5 % SDS. Cytoplasmic protein (20 μg), nuclear protein (20 μg) and whole cell protein (40 μg) were separated by SDS-polyacrylamide gel electrophoresis and transferred to nitrocellulose membranes by electroblotting. Membranes were immunostained with primary antibodies in 5 % bovine serum albumin (in 1X Tris-buffered saline) at 4 °C overnight. The signal was detected with horseradish peroxidase-conjugated secondary antibodies using the ECL Western Blotting Detection Reagent (GE Healthcare Life Sciences, USA). The following primary antibodies were used for western blot analysis: anti-CRABP1 (1:1,000), anti-CRABP2 (1:1,000), β −actin (Sigma-Aldrich; 1:100,000), α-tubulin (DSHB; 1:10,000) and Lamin A/C (ThermoFisher; 1:1,000).
Luciferase reporter assay
After siRNA depletion of CRABP1 or CRABP2, MCF-7 cells were seeded in 12-well culture plates at 20,000 cells/well and transfected with the luciferase reporter construct (0.5 μg/well) under the control of a retinoic acid response element (RARE) (pGL3-RARE-luc, Addgene). Alternatively, CRABP1-negative cell lines (BT-549, SK-Br-3, Hs578T) were seeded in 12-well plates, incubated at 37 °C for 24 h and then co-transfected with the CRABP1 expression construct (0.8 μg/well) and pGL3-RARE-Luc (0.5 μg/well) diluted in 250 μL of serum-free medium containing 5 μL PEI. pcDNA3 empty vector served as the negative control for these experiments. Forty-eight h after transfection, cells were treated with RA (in DMSO) at final concentrations of 0, 0.1 and 0.5 μM for 6 h keeping the overall volume of DMSO constant in each well. Cells were then harvested and whole cell lysates prepared using the luciferase cell culture lysis reagent (CCLR, Promega). Luciferase activity was measured with the Luciferase Assay System (Promega) and quantitated using a FLUOstar OPTIMA microplate reader (BMG Labtech) following the manufacturer’s instructions. Triplicate wells were analyzed for each treatment.
Statistical analysis
Statistical analysis was performed using MedCalc Statistical Software version 12.7.2 (MedCalc Software, Ostend, Belgium) as previously described [
25]. Briefly, gene profiling data for CRABP1 and CRABP2 were classified as “low” or “high” by receiver operating characteristic (ROC) curve analysis. Student
t-test or chi-square test was used to examine the significance of associations between CRABP mRNA levels or immunoreactivity and clinical outcome parameters. Two-way ANOVA was used to test the significance of the effects of siRNA knockdown and RA treatment on cell proliferation. The prognostic significance of CRABP1 and CRABP2 was analyzed by logrank test on Kaplan-Meier survival curves using both gene profiling and TMA immunoassay data.
Discussion
Cellular response to RA is believed to depend on two different classes of nuclear receptors, RARs and PPARs [
23,
24]. RAR activation by RA results in cell growth inhibition whereas PPARδ/β activation stimulates cell proliferation. These two RA signaling pathways are in turn modulated by two intracellular RA binding proteins: CRABP2 which channels RA to the nucleus to target RAR, and FABP5 which delivers RA to the nucleus thereby activating PPARδ/β [
23,
24]. We previously showed that FABP5 is preferentially expressed in ER- and triple-negative breast cancers, molecular subtypes believed to be resistant to RA treatment [
25]. High levels of FABP5, as well as a low ratio of CRABP2 to FABP5, are associated with poor prognosis. In this study, we identify a third RA-binding protein, CRABP1, as an inhibitor of RA action and an adverse factor for clinical outcome in breast cancer. Like FABP5, CRABP1 is preferentially expressed in ER- and triple-negative breast cancer. We propose a model whereby CRABP1 can compensate or synergize with FABP5 to compete with CRABP2 for RA, by sequestering RA in the cytoplasm, thereby reducing RA access to RAR (Fig.
6b).
The role of CRABP1 in carcinogenesis and tumor progression is poorly understood and contradictory. For example, CRABP1 is down-regulated in some human cancers and cell lines [
31‐
33], with DNA methylation proposed to contribute to CRABP1 silencing [
33,
58‐
60]. DNA methylation-mediated silencing of CRABP1 has been observed in a subset of breast carcinoma tissues [
60]. CRABP1 has been proposed to be a tumor suppressor in esophageal squamous cell carcinoma, with reduced CRABP1 levels associated with increased cell growth and distant lymph node metastasis [
33]. Reduced levels of CRABP1 are also associated with poorer prognosis in serous (
n = 40) and clear cell ovarian adenocarcinoma (
n = 59) [
31]. On the other hand, high levels of CRABP1 have been linked to lymph node metastasis and poor differentiation/high grade in pancreatic neuroendocrine tumors [
30].
In agreement with a pro-carcinogenic role for CRABP1, we found an association between CRABP1 expression and worse clinical outcomes in breast cancer using both gene profiling and TMA analysis. Similar to a previous report indicating that CRABP1 is differentially expressed in different subtypes of pituitary adenomas [
37], we found that CRABP1 is downregulated in ER+ breast tumors, but expressed in ER- and triple-negative tumors. These results suggest that downregulation of CRABP1 expression in cancer cells may be modulated by specific signaling pathways in different cancer subtypes (e.g. estrogen signalling), and that its role in tumor progression may differ between cancer types. It is noteworthy that estrogen signaling has been linked to the regulation of DNA methylation, perhaps explaining to some extent the reduced expression of CRABP1 observed in ER+ compared to ER- breast cancers [
61].
CRABP1 has the highest RA binding affinity of all RA binding proteins [
42]. It is generally believed that CRABP1 represses cellular response to RA by sequestering RA and/or promoting RA catabolism, reducing its availability in the nucleus for activation of RARs [
51,
62‐
64]. Several studies have demonstrated that CRABP1 promotes RA metabolism [
49‐
51]. However, the fact that RA metabolites can activate RAR
in vitro [
65], combined with the observation that a positive relationship exists between RA metabolism and cell growth inhibition in several cancer cell lines [
66], suggest that CRABP1-mediated RA metabolism may not account for RA resistance. In light of our observations that: (i) cytoplasmic CRABP1 in breast tumors is an adverse factor in clinical outcomes and (ii) CRABP1 accumulates in the cytoplasm of cells treated with RA, we propose that the primary role of CRABP1 in breast cancer is to sequester RA in the cytoplasm, thereby preventing RAR activation in the nucleus (Fig.
6b). The up-regulation of the RA-metabolizing gene
CYP26A1 observed upon CRABP1 depletion at high RA concentration further suggests a negative association between CRABP1 and RA metabolism.
Over five hundred genes are known to be regulated by RA, including
CRABP2 which has a functional RARE in its promoter region [
67,
68]. CRABP2 serves as a positive regulator of RA signalling in breast cancer cells [
23,
24,
42] and its expression can be induced by RA in various cell types [
67‐
69]. In this study, we found that CRABP1 and CRABP2 have inverse expression patterns in breast tumors and play an opposing role in the mediation of RA action in breast cancer cells. We further report that CRABP1 negatively regulates
CRABP2 expression. While increased CRABP2 expression has been observed in AB1 embryonic stem cells with homozygous deletion of
CRABP1 [
70], this is the first report demonstrating that CRABP1 has an inhibitory effect on CRABP2 expression in cancer cells. Intriguingly, CRABP1 not only inversely regulates
CRABP2 expression, but also affects nuclear translocation of CRABP2 in breast cancer cells. We postulate that CRABP1 plays a key role in attenuating RA activity in breast cancer cells, with high levels of CRABP1 reducing availability of RA in the nucleus. In turn, RA sequestration to the cytoplasm represses RA-mediated nuclear translocation of CRABP2 and induction of CRABP2 expression.
In addition to
CRABP2, our data indicate that CRABP1 modulates the expression of various genes implicated in RA biosynthesis, metabolism and action. With the exception of
RBP7, all these genes have previously been shown to be RA-regulated [
68]. We observed that the genes encoding CYP26A1 (catalyzes RA metabolism) and ALDH1B1 (catalyzes RA biosynthesis) are up-regulated and down-regulated, respectively, in the presence of RA upon CRABP1 depletion. We speculate that this is due to a feedback process as CRABP1 depletion may increase levels of free RA (especially when cells are exposed to high doses of RA), in turn resulting in accelerated RA metabolism and reduced RA synthesis. Therefore, CRABP1 may play a role in regulating cellular levels of free RA. RBP1 and RBP7 are retinol binding proteins which facilitate retinol storage and/or retinol-RA conversion [
53‐
56]. The different expression patterns observed for these two genes in control and CRABP1-depleted MCF-7 cells suggests opposite roles. We propose that
RBP1 (upregulated by RA but not affected by CRABP1) may be involved in retinol storage whereas
RBP7 (down-regulated by RA and CRABP1 depletion) may be involved in retinol metabolism producing RA.
TFAP2A is a RA-inducible gene whose expression increased upon CRABP1 knockdown in MCF-7 cells.
TFAP2A encodes AP-2α, recently shown to be essential for RA action [
71], has previously been reported to stimulate CRABP2 expression in mammary epithelial cells and breast cancer cells [
72]. AP-2α significantly enhances RA-induced RAR activation in breast cancer cells (our unpublished data). These combined data suggest the presence of a CRABP1-AP-2α-CRABP2 axis which modulates RA action in breast cancer cells.
In summary, we show that CRABP1 expression is maintained in ER- and triple-negative breast tumors, and that elevated levels of CRABP1 is a significant indicator of high tumor grade, Ki67 immunoreactivity, and poor prognosis. Our data indicate that cytoplasmic CRABP1, like FABP5, is a potent inhibitor of RA signalling. Elevated levels of CRABP1 may lead to RA resistance in breast cancer cells through sequestration of RA in the cytoplasm thereby preventing RA-mediated induction of RAR. We further demonstrate that CRABP1 attenuates RA activity by modulating the expression of important RA-regulated genes implicated in cellular RA availability, traffic and action. Thus, both CRABP1 and FABP5 represent potential therapeutic targets to overcome RA resistance in breast cancer. The discovery that there are at least three proteins involved in RA transport in breast cancer cells (CRABP1, CRABP2 and FABP5 [
23‐
25]), helps to address the molecular mechanism governing RA resistance in ER-negative or triple-negative breast cancer, and provides molecular tools to predict and eventually overcome RA resistance in breast cancer prevention and therapy.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made.
The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
RZL conceived, carried out experiments, analysed the data and wrote the manuscript. EG, DDG and HYP carried out experiments. JRM was involved in the study design and data collection. RG was involved in the study design, data interpretation and manuscript editing. All authors reviewed and approved the manuscript.