Tumor suppressor role of cytoplasmic polyadenylation element binding protein 2 (CPEB2) in human mammary epithelial cells
Over-expression of cyclooxygenase (COX)-2 promotes breast cancer progression by multiple mechanisms, including induction of stem-like cells (SLC). Combined gene expression and microRNA microarray analyses of empty vector vs COX-2- transfected COX-2 low MCF7 breast cancer cell line identified two COX-2-upregulated microRNAs, miR-526b and miR-655, both found to be oncogenic and SLC-promoting. Cytoplasmic Polyadenylation Element-Binding Protein 2 (CPEB2) was the single common target of both microRNAs, the functions of which remain controversial. CPEB2 has multiple isoforms (A-F), and paradoxically, a high B/A ratio was reported to impart anoikis-resistance and metastatic phenotype in triple- negative breast cancer cells. We tested whether CPEB2 is a tumor suppressor in mammary epithelial cells.
We knocked-out CPEB2 in the non-tumorigenic mammary epithelial cell line MCF10A by CRISPR/Cas9-double nickase approach, and knocked-down CPEB2 with siRNAs in the poorly malignant MCF7 cell line, both lines being high CPEB2-expressing. The resultant phenotypes for oncogenity were tested in vitro for both lines and in vivo for CPEB2KO cells. Finally, CPEB2 expression was compared between human breast cancer and non-tumor breast tissues.
CPEB2 (isoform A) expression was inversely correlated with COX-2 or the above microRNAs in COX-2-divergent breast cancer cell lines. CPEB2KO MCF10A cells exhibited oncogenic properties including increased proliferation, migration, invasion, EMT (decreased E-Cadherin, increased Vimentin, N-Cadherin, SNAI1, and ZEB1) and SLC phenotype (increased tumorsphere formation and SLC marker-expression). Tumor-suppressor p53 protein was shown to be a novel translationally-regulated target of CPEB2, validated with polysome profiling. CPEB2KO, but not wild-type cells produced lung colonies upon intravenous injection and subcutaneous tumors and spontaneous lung metastases upon implantation at mammary sites in NOD/SCID/IL2Rϒ-null mice, identified with HLA immunostaining. Similarly, siRNA-mediated CPEB2 knockdown in MCF7 cells promoted oncogenic properties in vitro. Human breast cancer tissues (n = 105) revealed a lower mRNA expression for CPEB2 isoform A and also a lower A/B isoform ratio than in non-tumour breast tissues (n = 20), suggesting that CPEB2A accounts for the tumor-suppressor functions of CPEB2.
CPEB2, presumably the isoform A, plays a key role in suppressing tumorigenesis in mammary epithelial cells by repressing EMT, migration, invasion, proliferation and SLC phenotype, via multiple targets, including a newly-identified translational target p53.
KeywordsCPEB2 Tumor suppressor COX-2 EMT EP4 receptor Breast Cancer Stem-like cells MicroRNA-526b MicroRNA-655 p53 Polysome profiling
Analysis of Variance
American Type Culture Collection
Cyclin Dependent Kinase 4/6
Cytoplasmic Polyadenylation Element
Cytoplasmic Polyadenylation Element Binding Protein
Clustered Regularly Interspaced Short Palindromic Repeats
Dulbecco’s Modified Eagle Medium
Eukaryotic Elongation Factor 2
Enzyme-Linked Immunosorbent Assay
Prostaglandin E Receptor
Glyceraldehyde 3-Phosphate Dehydrogenase
Hematoxylin & Eosin
Human Epidermal growth factor Receptor 2
Hypoxia Inducible Factor 1α
Human Leukocyte Antigen
Interleukin 2 Receptor ϒ chain
Michigan Cancer Foundation
Quantitative Reverse Transcription Polymerase Chain Reaction
Standard Error of the Mean
Small Interfering RNA
Triple Negative Breast Cancer
Vascular Endothelial Growth Factor
Upregulation of cyclooxygenase (COX)-2, an inflammation-associated enzyme, noted in half of the breast cancer patients  promotes tumor progression and metastasis through multiple mechanisms, including increased cancer cell proliferation, migration, invasion, epithelial-to-mesenchymal transition (EMT), tumor-associated angiogenesis, lymphangiogenesis, and induction of stem-like cells (SLCs) [2, 3, 4, 5, 6]. SLCs represent a dynamic state regulated by the tumor micro-environment  and resist chemo- and radiation therapies, causing tumor recurrence [8, 9]. COX-2 mediated stimulation of various oncogenic events in breast cancer, as listed above, is largely due to the activation of the PGE receptor EP4 by the endogenous Prostaglandin (PG)E2 .
Combined gene and microRNA expression microarrays with ectopic COX-2 overexpressing and Mock (empty vector-transfected) MCF7 breast cancer cells  revealed 26 genes that are downregulated, along with two microRNAs, miR-526b  and miR-655  that are upregulated by COX-2. MicroRNAs silence their gene targets either by degrading the mRNAs or blocking their translation . We found that both miRNAs - miR-526b and miR-655 were oncogenic and SLC-promoting [11, 12]. The only COX-2 down-regulated gene targeted by both microRNAs was identified as Cytoplasmic Polyadenylation Element-Binding Protein 2 (CPEB2), the functions of which in tumor biology remain controversial.
The CPEB family includes 4 members (CPEB1–4) which regulate translation of their target mRNAs by binding to a Cytoplasmic Polyadenylation Element (CPE) in the 3’untranslated region . Polyadenylation in their target mRNAs depends on both a CPE sequence (UUUUUAAU) and a polyadenylation hexanucleotide signal (AAUAAA) . CPEB proteins can repress or activate translation of target mRNAs by respectively shortening or elongating the poly-A tail . CPEB1 was shown to be a tumor-suppressor, depletion of which in mammary epithelial cells led to Epithelial-Mesenchymal-Transition (EMT) and metastatic phenotype . It restrained proliferation of glioblastoma cells by activating p27 mRNA reanslation . CPEB3 appeared to be a tumor-suppressor, targeted by oncogenic miR-107 in hepatocellular carcinoma . Furthermore, a high CPEB3 protein expression was associated with increased survival in renal cell carcinoma patients . The roles of CPEB4 in tumors remain conflicting. A tumor- suppressor role was demonstrated by its being the target of an oncogenic miR-550A in hepatocellular carcinoma . However, CPEB4 mediated translational activation of oncogenic mRNAs in pancreatic cancer , and EMT induction, growth and metastasis in gastric cancer cells  illustrate pro-oncogenic functions.
The role of CPEB2 in cancer remains paradoxical. A tumor-suppressor role was suggested by CPEB2 binding to HIF1α mRNA and suppressing its translation under normoxic conditions, but releasing it to allow translation under hypoxic conditions . This results from interaction with the elongation factor eEF2 . HIF1α is short-lived under normoxic conditions, but stabilized under hypoxic conditions to stimulate genes promoting angiogenesis, EMT, migration, SLC functions, metastasis and therapeutic resistance . Binding of CPEB2 to the mesenchymal transcription factor TWIST1 down-regulated its translation , suggesting an EMT-suppressor function. A tumor-suppressor role of CPEB2 was further suggested by its down-regulation by microRNA-885-5p, a mediator of EMT, tumorigenesis and metastasis in colorectal cancer .
The roles of CPEB2 in breast cancer appear to be complex, depending on the expression of different CPEB2 isoforms. CPEB2 has six isoforms (A-F). By selecting cells for anoikis-resistance in vitro from triple-negative breast cancer (TNBC) cell lines, Johnson et al.  reported that an alternative splicing of CPEB2, leading to a high isoform B:A ratio mediated anoikis-resistance and metastatic phenotype. They suggested that isoform A which excludes exon 4 is a tumor- suppressor, whereas isoform B that includes exon 4, is a tumor-promoter. This suggestion was validated by next generation sequencing and isoform-specific down-regulation of CPEB2A and B in TNBC lines . They concluded that CPEB2B plays an antagonistic role against CPEB2A by alleviating the translational inhibition of HIF-1α and Twist 1 imparted by CPEB2A.
We adopted a different approach to examine the functions of CPEB2 in breast epithelial cell lines, by depleting the entire CPEB2 gene and observing the resultant phenotypic changes: (a) CPEB2 was knocked out using a double nickase CRISPR plasmid in an immortalized non-tumorigenic human mammary epithelial cell line MCF10A, reported to be a reliable model for normal mammary epithelium ; (b) CPEB2 was knocked down with siRNAs in the MCF7 cell line, a mammary carcinoma of low malignancy . CPEB2KO MCF10A cells exhibited an oncogenic phenotype in vitro, as indicated by increased proliferation, migration, invasion, EMT, stimulation of SLC and a downregulation of p53 tumor suppressor protein owing to a decreased translation of p53 mRNA. They exhibited lung colonization after intravenous injection and subcutaneous tumorigenicity upon inoculation at the mammary sites in NOD/SCID/IL2Rγ null mice. SiRNA-mediated CPEB2 knockdown in MCF7 breast cancer cells also resulted in enhanced oncogenic properties tested in vitro. These results confirm CPEB2 as a tumor-suppressor in breast epithelial and poorly malignant breast cancer cells likely resulting from the CPEB2A isoform prevalence in these cells. This contention was supported by a higher CPEB2A isoform expression and A: B isoform ratio in non-tumorous human breast tissues than in cancerous breast tissues.
Materials and methods
The immortalized non-tumorigenic mammary epithelial cell line MCF10A (ATCC, at 4–6 passages) was cultured in DMEM (Gibco, Thermofisher, CA) supplemented with 5% horse serum (Invitrogen, Thermofisher, CA), 20 ng/ml EGF, 0.5 mg/ml hydrocortisone (Sigma, Oakville, ON, CA), 100 ng/ml cholera toxin (Sigma), 10 μg/ml insulin (Sigma), and 1% penicillin/streptomycin (Invitrogen) . MCF7 cells (ATCC, Manassas, VA, USA, at 4–6 passages) were grown in EMEM supplemented with 10% FBS, 100 μg/mL of penicillin/streptomycin (Gibco) and 10 μg/mL of insulin (Sigma).
CRISPR knockout of CPEB2 and SiRNA mediated CPEB2 knockdown
Total CPEB2 was knocked out in MCF10A cells with a CRISPR double nickase plasmid (Santa Cruz, Dallas, TX, USA) targeting exon 1 of the gene, that combines a Cas9 nickase mutant with pairs of guide RNAs to introduce targeted double-strand breaks, ensuring a high knockout specificity . Cells transfected with Amaxa Cell Line Nucleofector Kit IV (Lonza, Allendale, NJ, USA) were subjected to 72 h of puromycin selection and expanded. MCF7 cells were transfected with 1 μM of either CPEB2 siRNA (a pool of siRNAs that gave the best results) or Universal Scrambled Control siRNA (OriGene, Rockville, MD, USA). After 24 h media was changed and experiments conducted at 48 h. CPEB2 downregulation in KO and KD cells was validated with qRT-PCR.
Protein extraction and Western blot
Proteins extracted from cell lysates were subjected to western immunoblots , using primary antibodies at the following dilutions: CPEB2 (Origene cat # TA344026, rabbit polyclonal, 1:1000; lacking isoform-specificity), E-Cadherin (Cell Signalling, Danvers, Mass, USA, rabbit monoclonal, 1:1000), Vimentin (Cell Signalling, rabbit monoclonal, 1:1000), β-actin (Santa Cruz, mouse monoclonal, 1:4000 or Cell Signalling, rabbit monoclonal, 1:1000), N-Cadherin (Santa Cruz, Rabbit polyclonal, 1:400), p53 and p21 (both from Novus Biologicals, Centennial, Colorado, USA, mouse monoclonal, 1:200) and β-Catenin (Sigma, rabbit polyclonal, 1:4000). After primary antibody incubation overnight, membranes were incubated for 1 h with appropriate secondary antibody at the following dilutions: Goat Anti-Rabbit (1:10000, Li-COR, Lincoln, Nebraska, USA) and Donkey Anti-Mouse (1:10000, Li-COR). Membranes were scanned using the Odyssey Infrared Imaging System (Li-COR).
Cells grown on glass coverslips were treated with various primary antibodies as reported : E-Cadherin (Cell Signalling, 1:500), Vimentin (Cell Signalling, 1:500), and N-Cadherin (Sigma, 1:500). The cells were then incubated with fluorochrome-conjugated secondary antibodies (Biotum, Cedarlane, Burlington, ON, CA) at the following dilutions: Goat Anti-Rabbit 594 (1:500), Goat Anti-Rabbit 488 (1:500) and Goat Anti-Mouse 488 (1:500). Vectashield anti-fade mounting medium with DAPI (Vector-labs, Burlignton, ON, CA) was used to mount the slides. Fluorescent images were taken with Zeiss LSM 510 Meta Multiphoton Confocal Microscope, and fluorescence intensities calculated with ImageJ software. The raw integrated density was calculated for each cell and normalized to the cell area. Data were presented as an average of all cells.
Fixation, permeabilization, and antibody staining for tumorspheres (minimum 60 μm diameter) were conducted as reported . Incidence of fluorescent cells (for ALDH1, NANOG, and SOX-2) was computed among total number of cells marked by DAPI.
Migration and invasion assays using Transwells
Semi-confluent MCF10A and MCF7 cells were serum-starved overnight and seeded on microporous membranes (8 μm pore diameter) coated without or with Matrigel respectively for migration and invasion assays using transwell inserts . Bottom chambers included 5% horse serum to stimulate migration or invasion respectively for 24 and 48 h. The cells at the bottom of the membranes were fixed with cold methanol, stained with Eosin (cytoplasm) and Thiazine (nucleus), and mounted onto glass chamber slides to count all cells. Assays were done in triplicate.
Migration (wound-healing) assays using scratch method
Cells grown to semi-confluency were serum-starved over-night. Proliferation inhibitor mitomycin C (1 ng/μl) was added 2 h before the plate was scratched with a microtip. Cells were then incubated in medium including 1% horse serum and mitomycin C for 72 h, replacing media every 24 h. Migration, un-affected by proliferation was measured as the distance covered per unit time.
Cell-free conditioned media were collected at 24 h and frozen at − 20 °C. 10% zymogram gels were made with Gelatin A to measure gelatinolytic activity. Conditioned medium mixed 1:1 with zymogram sample buffer, was loaded and run at 110 V for 90 min. The gel was incubated in zymogram renaturation buffer for 1 h, then in zymogram developing buffer for 1 h, and finally overnight at 37 °C in fresh zymogram developing buffer (all from BioRad, Mississauga, ON). Gels stained with Coomassie Blue were imaged using the BioRad XR+ Gel documentation system.
Tumorsphere (spheroid) formation assay
MCF7 and MCF10A cells were grown as spheroids on ultra-low attachment plates as reported . In brief, cells grown to 70–80% confluency were trypsinized and spun down. They were then suspended in basal HuMEC media (Gibco) with added B-27 supplement (Gibco), EGF (20 ng/mL, Invitrogen) and FGF (20 ng/mL, Invitrogen), taken up by a 1 mL syringe and put through a 40 μm cell strainer (Falcon) to collect a single cell suspension. The cells were then counted and seeded at a density of 10 cells/well in an ultra-low attachment 96-well plate (Thermo Fisher) to measure spheroid forming efficiency (SFE), as well as 1 × 103 cells/well in an ultra-low attachment 6-well plate (Corning, NY, USA) for for RNA analysis and immunofluorescence. SFE was calculated at 4 days as the total number of spheroids (minimum 60 μm in diameter) divided by total number of cells plated.
Polysome profiling was performed as reported . WT or CBEB2KO MCF10A cells were pre-treated with cycloheximide (CHX, 100 μg/mL) for 5 min at 37 °C. Cells were washed twice with ice-cold PBS containing 100 μg/mL CHX and lysed in hypotonic buffer (100 mM KCl, 50 mM Tris–HCl pH 7.4, 1.5 mM MgCl2, 1 mM DTT, 1 mg/mL heparin, 1.5% NP40, 100 μg/mL CHX, supplemented with mini cOmplete™ Protease Inhibitor Cocktail tablet and 100 Unit RiboLock RNase inhibitor). The lysates were incubated on ice for 5 min and cleared by centrifugation at 12000 rpm for 5 min at 4 °C, prior to loading on 10–50% sucrose gradients to isolate the sub-polysomal and polysomal fractions. Total RNA was isolated from each fraction using Trizol reagent and converted to cDNA using SuperScript III First-Strand Synthesis System (Life Technologies, Burlington, ON). RT-PCR was performed for each gene using a PCR thermocycler (Bio-Rad) and visualized by agarose gel electrophoresis.
Cell proliferation assays
WT or CPEB2KO MCF10A cells were incubated for 24 h in complete media with or without EDU (negative control for autofluorescence) using Click-iT EdU Alexa Fluor 488 Imaging kit (Invitrogen). EdU incorporation was measured in an analytical flow cytometer. BrdU incorporation was measured in WT, MOCK and CPEB2KD MCF7 cells using the Cell Proliferation ELISA, BrdU (colorimetric) Kit (Roche, Sigma) and an Epoch Microplate Spectrophotometer at a wavelength of 370 nm with a reference wavelength of 492 nm.
Animal experiments for tumorigenicity assays
NOD/SCID/IL2Rγ-null (NSG) mice (deficient in T, B and NK cells) were bred locally by Dr. David Hess (co-author) and maintained according to the Canadian Council of Animal Care guidelines with food and water ad libitum at the Robarts Research Institute mouse barrier. The protocol was approved by the Animal Care and Veterinary Services (ACVS) committee on animal use. Six-week-old females were used as hosts for tumorigenicity assays by intravenous or subcutaneous routes using 5 × 105 CPEB2KO and WT MCF10A cells per mouse. They were euthanized with CO2 at the appropriated time. Tail vein injected mice (n = 6) were sacrificed at 8 weeks to isolate lungs, spleen and liver to assess metastases. Cells mixed 1:1 with Matrigel were injected S.C. into both right and left inguinal mammary regions of 5 mice each for CPEB2KO or parental cells (total = 10 sites each). After 12 weeks at sacrifice, one mammary fat pad was stored in O.C.T. for frozen sectioning and the other in Bouin’s solution for paraffin embedding. Lungs, liver and spleen were also harvested to assess spontaneous metastasis from mammary sites. All lungs were inflated with PBS prior to isolation. All mice were weighed once per week.
Tissue processing from mice
Lungs, liver and spleen were stored either in O.C.T for cryo-sectioning or Bouin’s solution/Neutral-Buffered Formalin for fixation and paraffin embedding. Paraffin blocks were sectioned at 5 μm and stained with H&E. Frozen organs were sectioned at 8 μm for lungs, liver and spleen, while mammary fat pads were sectioned at 10 μm. Sections were then fixed in 10% formalin, permeablized in 0.1% Triton-X-100, and blocked with M.O.M (mouse-on-mouse Ig, Vector labs, Burlington, ON) prior to immuno-staining. Mouse anti-HLA antibody (1:100, BD, Mississauga, ON), followed by horse anti-mouse FITC (1:200, Vector Labs) was applied to detect human cells. Sections were mounted with Vectashield (Vector Labs), stained with DAPI and viewed under a fluorescent microscope. Micrometastases were arbitrarily scored as single cells, clusters (2–8 cells) and colonies (> 8 cells) as reported earlier  and averaged within 3 sections, 5 images of 1600μm2 each per section.
Measurements of CPEB2 expression in human breast cancer cell lines, breast cancer and non-tumor breast tissues
Using the blast search, we found that primers used by Johnson et al.  for CPEB2A covered all 6 isoforms, whereas primers used for CPEB2B covered both isoforms B and D. In our study we used probes with increased isoform specificity: a Taqman probe for isoform A/E and another probe for isoform B/D (Applied Biosystems, ThermoFisher) to conduct qPCR in a panel of human breast cancer tissues (n = 105) and non-tumor breast tissues (n = 20) obtained from the Ontario Tumor bank (OTB) maintained by the Ontario Institute of Cancer Research (OICR) following ethics approval by the OICR committee. This bank receives tumor tissues from donors in Ontario hospitals following donor consent. Taqman probes (Applied Biosystems) for GAPDH and β-Actin were used as the internal loading control. Delta Ct values were calculated by subtracting the average Ct values (triplicate) from the control and analyzed as previously described .
The probe for isoform A/E was also utilized to compare CPEB2A expression in a panel of COX-2 disparate human breast cancer cell lines MCF7, MCF7-COX2, MDA-MB-231 and SKBR3. To conduct qPCR for CPEB2 in MCF7 cells we used a Taqman probe from Life Sciences, which is not isoform-specific. To conduct RT-PCR for CPEB2 in MCF10A cells, we designed oligos to cover all isoforms: Forward: ACACTCTTACCCTTACAGGTG; Reverse: CGCCCATAACTCCTTGCATT.
Statistical analyses were performed using Graphpad Prism Software 5.0 (Graphpad Software Inc. 2007). All parametric data were analyzed with one-way ANOVA followed by Tukey-Kramer or Dunnett’s post hoc comparisons. Spheroid and lung colony numbers were analyzed both by parametric and nonparametric (Mann-Whitney test) methods, giving similar results. Student’s t test was used to compare two datasets. Statistically relevant differences between means were accepted at p < 0.05.
CPEB2 is the single COX-2 down-regulated gene targeted by miR-526b and miR-655
We found that miR-526b and miR-655 collectively targeted a total of 13 COX-2 -down-regulated genes, 12 of which are tumor-suppressor-like. The remaining single gene targeted by both microRNAs was identified as CPEB2 (Additional file 4: Table S1).
CPEB2 expression levels in multiple breast cancer cell lines
Validation of CPEB2 down-regulation in CPEB2KO MCF10A cells and CPEB2KD MCF7 cells
CPEBE2 mRNA and protein levels were very high in MCF10A cells, in which we knocked-out CPEB2 (inclusive of all isoforms) using a double nickase CRISPR plasmid to ensure high specificity. A comparison of mRNA and protein levels in WT and KO cells (Fig. 1c) demonstrated approximately 80% knock-out efficiency (Fig. 1d). PCR data revealed a single band corresponding to isoform A, both in WT and CPEB2KO MCF10A cells confirming that MCF10A cells lack in the B isoform . We also knocked down CPEB2 in the poorly malignant, CPEB2A dominant breast cancer cell line MCF7, using a pool of siRNAs. Scrambled siRNAs served as controls (Mock cells). An efficient knock-down (approximately 75% relative to WT or Mock cells) was noted at the mRNA level (agarose gel picture in Fig. 1e; quantification of Fig. 1e presented in Fig. 1f).
CPEB2 downregulation in breast epithelial cells induces EMT and promotes migration and invasion
Migration was also measured in transwells (chemokinesis assay), in which serum-starved cells were allowed to migrate through microporous membranes for 24 h into a medium containing 5% Horse Serum. CPEB2KO cells migrated faster than WT cells (Fig. 3c). Similarly, CPEB2KD MCF7 cells also migrated faster than the Mock (scrambled siRNA-transfected) or WT MCF7 cells (Additional file 2: Figure S2A).
Ability to invade basement membrane components is an important prerequisite for metastasis. Invasion was measured as above for 48 h, in which the microporous membranes were coated with a basement membrane analog Matrigel. CPEB2KO MCF10A cells exhibited a higher invasive ability than WT cells (Fig. 3d). CPEB2KD MCF7 cells also showed significantly higher invasiveness than Mock or WT MCF7 cells (Additional file 2: Figure S2B). Increased matrix-degrading ability of CPEB2KO MCF10A cells was corroborated by gelatin zymography, showing a 2.9 fold increase in gelatinase A (MMP-9) activity (Fig. 3h).
CPEB2KO promotes proliferation
Sustained proliferative ability is a hallmark of cancer cells . Flow cytometry for 5′- ethynyl-2′-deoxyuridine (EdU) incorporation (measuring DNA synthesis) for 24 h revealed a four-fold increase in CPEB2KO MCF10A cells compared to WT cells (Fig. 3e and f) in 3 replicate preparations. CPEB2KD MCF7cells, however, exhibited only a minor increase (p = 0.06) in proliferative ability compared to mock cells (Additional file 2: Figure S2C).
P53 translation is repressed in CPEB2-KO cells
To interrogate whether p53 protein was translationally regulated by CPEB2, we compared polysome profiling between WT and CPEB2KO MCF10A cells. The profile appeared almost identical indicating no significant difference in global translation between the two cell types (Fig. 4e). RT-PCR was performed to evaluate the distribution of β-actin and p53 mRNAs across a sucrose density gradient. While β-actin distributions were similar in both cells, p53 mRNA was significantly shifted toward light polysomes in CPEB2KO cells (Fig. 4f). These results, combined with the findings of the absence of any change in p53 mRNA in CPEB2KO cells, clearly reveal that CPEB2 knock-out decreased the translation of p53 by shortening the Poly A tail leading to decreased p53 protein.
CPEB2KO/KD stimulates SLC phenotype
CPEB2KO MCF10A cells reveal an upregulation of VEGF-D, COX-2 and EP4
We reported that ectopic COX-2 over-expressing MCF7 cells displayed an upregulation of VEGF-A, VEGF-D and EP4 receptor . Furthermore two COX-2-upregulated miRNAs miR-526b and miR-655, which target CPEB2, when ectopically over-expressed in miRNA-low MCF7 cells led to an upregulation of EP4 and COX-2, indicating a positive feed-back loop for pathways in SLC sustenance [11, 12]. These findings prompted us measure VEGF, EP4 and COX-2 mRNAs in CPEB2KO MCF10A cells. We found a significant upregulation of VEGF-D, COX-2 and EP4 mRNAs (Additional file 3: Figure S3). Conversely, inhibition of COX-2 or EP4 activity in MCF7-COX2 cells which was found to suppress SLC activity  also upregulated CPEB2 (data not presented).
CPEB2KO MCF10A cells form tumors in immune-compromised mice
CPEB2 isoforms a and B expression in human breast cancer tissues
The roles of CPEB2 in human breast tumorigenesis have so far remained a paradox, until isoform-specific roles were identified [29, 30]. Selecting cells from TNBC cell lines for anioikosis-resistance, these authors observed that alternative splicing resulting in the loss of CPEB2A with concomitant increase in CPEB2B was responsible for their metastatic phenotype. CPEB2 was shown to suppress the translation of two oncogenic transcription factors, TWIST1  - an EMT inducer, and HIF1α  associated with many oncogenic functions [24, 25]. Deligio et al.  reported that anoikosis-resistance and metastasis phenotype of TNBC cell lines resulted from CPEB2B isoform mediated translational activation of HIF1α and TWIST1. They suggested that CPEB2B plays an antagonistic role against CPEB2A by alleviating the translational inhibition of HIF1α and TWIST1 imparted by CPEB2A.
In the present study, by depleting the entire CPEB2 gene, we demonstrated a robust role of CPEB2 in suppressing a variety of oncogenic functions in both MCF10A and MCF7 cell lines, evidently due to the CPEB2A isoform, prevalent in both cell lines. Reportedly, no alternative splicing of CPEB2 or presence of isoform B was detectable MCF10A cell line . In both cell lines, we found that CPEB2 downregulation promoted EMT, proliferation, migratory and invasive functions, as well as SLC phenotype measured with spheroid-forming ability. The spheroids formed by CPEB2KO MCF10A cells exhibited a higher growth rate, indicating a faster self-renewal of the SLC population. Increased β-Catenin protein noted in CPEB2KO cells may be responsible for a higher self-renewal capacity of stem-like cells or increased proliferative ability of non-stem cells. For example, increased β-Catenin signaling can lead to increased expression of CCND1, a nuclear protein, that forms a complex with CDK4 and CDK6 leading to progression of cells from G1 into S-phase .
During EMT, polar epithelial cells assume an elongated nonpolar mesenchymal morphology, associated with an acquisition of mesenchymal cell markers, increased migratory ability, cell survival, and invasiveness . E-Cadherin, an epithelial cell junction-associated protein mediates cell-cell adhesions, contact inhibition and control of cell proliferation . It is lost during EMT , as shown with CPEB2KO or KD cells. However, EMT also requires “Cadherin switching” in which expression of E-Cadherin is exchanged for other cadherins, such as N-Cadherin [44, 45]. Ectopic over-expression of N-Cadherin in MCF7 cells stimulated migration and invasiveness through upregulation of MMP-9 and metastasis . Here, in CPEB2KO MCF10A cells, a switch from E-Cadherin to N-Cadherin was clearly evident. Increased MMP-9 activity shown with gelatin zymography explained their increased invasive capacity. While transcriptional regulation of N-Cadherin during cadherin switching is unknown, Twist1 can modulate N-Cadherin expression by binding to the E-box on CDH2 (N-Cadherin) [45, 47].
Numerous transcription factors may have contributed to the EMT phenotype in CPEB2KO cells. TWIST1 can mediate EMT , and also inhibit apoptosis through evasion of p53-induced cell death . CPEB2 mediated translational repression of TWIST1, a function ascribed to the A isoform  could be a mechanism by which EMT is suppressed. In support, in CPEB2KO cells exhibited no change in TWIST1 mRNA expression, suggesting post- transcriptional regulation by CPEB2. An increase in SNAI1 and ZEB1, observed in CPEB2KO cells, could additionally contribute to EMT. Both molecules are known to suppress transcription of E-Cadherin and play important roles in invasion and metastasis [50, 51].
P53 is a master tumour-suppressor, responsible for suppressing EMT, migration, and invasion through transcriptional regulation of many other molecules . Here, we show for the first time that CPEB2 is a novel translational regulator of p53, as was reported for CPEB1 . Polysome profiling confirmed that a reduction in p53 protein in CBEB2KO cells is due to reduced p53 mRNA translation resulting from shortened Poly A tail, rather than an indirect mechanism by binding to another target molecule. Additionally, another tumor suppressor protein p21, a downstream partner of p53 was also reduced in CPEB2 KO cells.
We have shown that ectopic COX-2 overexpression in MCF7 cells promoted SLC phenotype associated with increased expression of EP4 receptor, Notch and Wnt . Wnt pathway protein β-Catenin and the downstream Wnt pathway genes CCND1, AXIN1 and AXIN2 were all significantly upregulated in MCF7-COX2 spheroids . In the present study CPEB2KO cells exhibited an upregulation of β-Catenin protein, AXIN1 gene, and interestingly also COX-2 and EP4 genes linking with SLC phenotype . It is likely that an upregulation of COX-2/EP4 provided a positive feed-back loop for SLC sustenance in CPEB2KO cells, as shown for miR-526-B and miR-655 overexpressing cells [11, 12]. Conversely, treating MCF7-COX2 cells with either a COX-2 inhibitor celecoxib or with an EP4 antagonist ONO-AE3–208 significantly up-regulated CPEB2 in these cells (data not presented), concomitant with a marked drop in their spheroid-forming capacity . The molecular mechanisms underlying this CPEB2-COX2/EP4 regulatory loop remains to be investigated. One possible pathway is binding of CPEB2 to HIF-1α , a known upregulator of COX-2. Furthermore, increased β-Catenin protein noted in CPEB2KO cells can stabilize COX2 mRNA by interacting with AU-rich elements of 3′-UTR . Collectively, the SLC stimulation in CPEB2KO MCF10A cells could be due to an upregulation of COX2/EP4, and Wnt pathway genes.
Finally an examination of breast cancer and non-tumor breast tissues revealed a lower expression CPEB2 isoform A and higher expression of isoform B in cancerous tissues. This difference was more evident by measuring the ratios of A: B isoforms, presented in Fig. 8a. These results are in concordance with reported tumor-suppressor vs tumor-promoter functions of A and B isoforms . However our probes could not demonstrate a preferential expression of A (A/E) or B (B/D) isoform in any of the tumor subsets based on the information on their ER, PR or HER2 status. Since these samples contained variable proportion of stromal and immune cells, it is possible that they may have masked the differences.
In summary, we demonstrate here that CPEB2, presumably the isoform A, plays a significant role in suppressing tumorigenesis in mammary epithelial cells by repressing EMT, migration, invasion, proliferation and SLC phenotype, possibly through multiple targets, one of them identified here as p53. Figure 8c presents a schema for the proposed mechanisms in tumor suppressor functions of CPEBE2.
Present study, utilizing in vitro and in vivo functional assays in CPEB2-depleted mammary epithelial cells as well as human breast-derived non-tumor and cancerous tissues, demonstrates that CPEB2, presumably the isoform A, plays a key role in suppressing tumorigenesis in mammary epithelial cells. The underlying mechanisms involved suppression of EMT, migration, invasion, proliferation and SLC phenotype, via a newly-identified translational target p53.
We thank Dr. Trevor Shephard (Departments of Anatomy and Cell Biology, and Obstetrics & Gynecology, University of Western Ontario) for his expert advice on the CRISPR/Cas9-double nickase method.
We have adhered to the ARRIVE guidelines/methodology.
All authors made substantial contributions, as follows. JT conducted the experiments related to MCF10A cells (presented in his Master’s thesis), and wrote the first draft; MM performed the experiments with human tissues, analyzed the data and revised manuscript draft; MA performed and analyzed the polysomal profiling; AH conducted experiments related to MCF7 cells (presented in her Master’s thesis); DH bred and provided the NSG mice, supervised tumor transplantation and staining for HLA; checked the results on tumorigenicity; PKL directed and supervised the overall study and wrote the final draft of the manuscript. All authors approved the final submitted version of the manuscript and ensured the accuracy.
The study was supported by funds of the Ontario Institute of Cancer Research (OICR) and Natural Sciences and Engineering Research Council of Canada (NSERC) to PKL, and NSERC to MM. AH was a scholar of the Translational Breast Cancer Research Unit, with funds derived from the Breast Cancer Society of Canada. The funders had no role in the studies.
Ethics approval and consent to participate
Animal were used according the guidelines of the Canadian Council of Animal care, and experimental protocols approved by the ACVS committee. Human tissues were obtained from the Ontario Tumor bank (OTB) of the Ontario Cancer Research Institute (OICR), (a repository of tumor tissues created on the basis donor consent), following approval of human ethics by the Ethics Review Board of the OICR.
Consent for publication
Not applicable. No personal data is included in this manuscript.
The authors declare that they have no competing interests.
- 20.Foundation KAW. The human protein atlas project 2005 [updated 2018-11-15. Available from: https://www.proteinatlas.org/ENSG00000107864-CPEB3/pathology.
- 33.Gandin V, Sikstrom K, Alain T, Morita M, McLaughlan S, Larsson O, et al. Polysome fractionation and analysis of mammalian translatomes on a genome-wide scale. J Vis Exp. 2014;(87). https://doi.org/10.3791/51455.
- 44.Chen A, Beetham H, Black MA, Priya R, Telford BJ, Guest J, et al. E-cadherin loss alters cytoskeletal organization and adhesion in non-malignant breast cells but is insufficient to induce an epithelial-mesenchymal transition. BMC Cancer. 2014;14:552. https://doi.org/10.1186/1471-2407-14-552.CrossRefPubMedPubMedCentralGoogle Scholar
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.