Defects in AU-rich elements (ARE)-mediated posttranscriptional control can lead to several abnormal processes that underlie carcinogenesis. Here, we performed a systematic analysis of ARE-mRNA expression across multiple cancer types. First, the ARE database (ARED) was intersected with The Cancer Genome Atlas databases and others. A large set of ARE-mRNAs was over-represented in cancer and, unlike non-ARE-mRNAs, correlated with the reversed balance in the expression of the RNA-binding proteins tristetraprolin (TTP, ZFP36) and HuR (ELAVL1). Serial statistical and functional enrichment clustering identified a cluster of 11 overexpressed ARE-mRNAs (CDC6, KIF11, PRC1, NEK2, NCAPG, CENPA, NUF2, KIF18A, CENPE, PBK, TOP2A) that negatively correlated with TTP/HuR mRNA ratios and was involved in the mitotic cell cycle. This cluster was upregulated in a number of solid cancers. Experimentally, we demonstrated that the ARE-mRNA cluster is upregulated in a number of tumor breast cell lines when compared with noninvasive and normal-like breast cancer cells. RNA-IP demonstrated the association of the ARE-mRNAs with TTP and HuR. Experimental modulation of TTP or HuR expression led to changes in the mitosis ARE-mRNAs. Posttranscriptional reporter assays confirmed the functionality of AREs. Moreover, TTP augmented mitotic cell-cycle arrest as demonstrated by flow cytometry and histone H3 phosphorylation. We found that poor breast cancer patient survival was significantly associated with low TTP/HuR mRNA ratios and correlated with high levels of the mitotic ARE-mRNA signature. These results significantly broaden the role of AREs and their binding proteins in cancer, and demonstrate that TTP induces an antimitotic pathway that is diminished in cancer. Cancer Res; 76(14); 4068–80. ©2016 AACR.

AU-rich elements (ARE) act as key cis-acting factors in the posttranscriptional control of gene expression; abnormalities involving this pathway can occur in several diseases, including cancer (1). The AREs constitute instability determinant sequences located in the mRNA 3′UTR, and are fundamental to the transient expression of genes that regulate critical cellular functions, such as cell proliferation, apoptosis, cytokine response, and cellular motility (2, 3). A growing body of evidence has linked aberrantly elevated and prolonged expression of ARE-encoding mRNAs to cancer, including those participating in angiogenesis, chemotaxis, and invasion. Examples of such genes include EGF, estrogen receptor (ER), COX-2, VEGF, matrix metalloproteinase (MMP) 1, the chemokine receptor CXCR4, and urokinase-type plasminogen activator (uPA) and its receptor (4–9). The ARE-mRNAs are regulated by trans-acting RNA-binding proteins notably tristetraprolin (TTP/ZFP36) and human antigen R (HuR/ELAVL1). TTP and HuR have antagonistic activities with regard to ARE-mRNA regulation; TTP destabilizes the mRNA and represses protein translation, whereas HuR is an mRNA stability and translation-promoting factor (10–12).

The dysregulation of the expression and function of TTP and HuR has been of particular interest and significance as they affect several cancer processes including cellular growth (13, 14). The expression of TTP and HuR is regulated at the levels of transcription, mRNA stability, and translation; and their activity can be regulated at the posttranslational level by phosphorylation, cellular localization, and protein stability (10, 13, 15, 16). The mRNAs of TTP and HuR contain AU-rich elements, and their proteins are capable of regulating their own expression in an ARE-mediated manner (17–20). In addition, we identified a TTP-HuR axis in which TTP reduces HuR mRNA and translation (18, 21). Furthermore, TTP mRNA levels can be reduced by the high cellular levels of the miR-29a in invasive breast cancer (22). The inhibition of miR-29a using a cell-permeable anti-miR-29a peptide-nucleic acid can normalize the TTP–HuR axis, which leads to a reduction in the levels of ARE-mRNAs in breast cancer cells (21).

Here, we performed a systematic analysis of ARE-mRNA expression in invasive breast cancer and across different types of cancer. We found an ARE-mRNA cluster that is enriched in a mitotic cell cycle, highly coregulated by key RNA-binding proteins, and correlated with the clinical outcome of the patients. We experimentally validated the expression patterns, functional responses to TTP and HuR, and the physical association of these mRNAs with their cognate proteins in cancer cells. Focusing efforts on pathways that significantly impact cancer processes, such as mitosis, may lead to new therapeutic approaches.

Cancer patient data

The Oncomine web-based data mining platform (www.oncomine.com) was used to mine data from The Cancer Genome Atlas (TCGA) and METABRIC databases: http://cancergenome.nih.gov (23). Gene expression levels for 389 and 1,556 invasive ductal breast cancer samples were downloaded for TCGA and METABRIC datasets, respectively, along with the 61 and 144 corresponding matched normal samples. Cancer-upregulated genes with 1.7-fold increase in expression and Q < 0.0001 were retrieved. Intersecting 2,396 upregulated genes with 3,658 genes from the ARE database (ARED; http://brp.kfshrc.edu.sa/ARED/) allowed the identification of the overexpressed AU-rich element-containing mRNAs. Log2 median-centered intensity ratios were used. Other cancer datasets were used and extracted through Oncomine and Nextbio portals (Supplementary Table S1).

Regression and volcano modeling for correlation studies

See Supplementary methods.

Clustering and functional annotation

Normal mixture was used for clustering and obtaining ARE-mRNAs with the most significant negative correlation with the TTP/HuR ratio. The input to the model was the 414 overexpressed ARE-mRNA data from TCGA dataset, specifically: the correlation R-values of TTP mRNA with ARE-mRNAs, R-values of HuR mRNA with ARE-mRNAs, and the fold change of tumor to normal tissues (T/N). The Akaike Information Criterion (AIC) value was used as an indicator of goodness of fit (smallest AIC). The overexpressed ARE-mRNAs were analyzed using the AIC value and Johnson transformation, which resulted in a best-fit normal mixture. Supervised hierarchical clustering and heatmap visualization of gene expression data were performed using Gene-E (Broad Institute). Functional enrichment was performed using several web-based resources, including Geneset analysis http://bioinfo.vanderbilt.edu/ webgestalt/ analysis.php, Database for Annotation, Visualization and Integrated Discovery (DAVID), and STRING protein–protein interaction resource http://string-db.org/.

Cell culture and proliferation

Cell lines were obtained from ATCC (2014 and 2015 batches). The normal-like mammary cell lines MCF10A and MCF12A were cultured in DMEM: F12 medium (Thermo Fisher Scientific) fortified with HuMEC supplement. The tumor breast lines MDA-MB-231, MCF-7, and HEK293 cells were grown in DMEM. The tumor breast cell lines BT-474 and SKBR were cultured in DMEM:F12 and McCoy; respectively. All media were supplemented with 10% FBS and antibiotics. Tet-On Advanced HEK293 cells were obtained from ClonTech (2013) and were cultured in DMEM supplemented with 10% Tet-System Approved FBS (ClonTech), 100 μg/mL G418 (Sigma), and antibiotics. Stocks were made at the original date of obtaining the cells, and were usually passaged for no more than four months. Cellular proliferation was assessed by using real-time label-free electric impedance system (ACEA) as described previously (24).

ARE reporter constructs and transfection

The RPS30 promoter-linked reporter expression vectors (25) containing the AREs (Supplementary Table S4), were constructed by ligating two annealed synthetic complementary oligonucleotides with BamHI and XbaI overhangs. The annealed DNA were cloned in the 3′UTR of the RPS30-SGFP vector. The use of RPS30 promoter along with SGFP, an EGFP generated from UW-reduced coding region (26) allows selective posttranscriptional effects. Cotransfection with RPS23-linked RFP was performed in HEK293 using Lipofectamine 2000 (Invitrogen). The induced expression of HA-tagged TTP and myc-tagged HuR were used as outlined previously (4, 5, 21).

Tet-on TTP induction and effect on ARE-SGFP reporters

Tet-On TTP expressing constructs were prepared by PCR as described previously (27). The HA-TTP construct was originally obtained from Dr. Perry Blackshear (NIH, Bethesda, MD). The HEK-Tet-On advanced cells to study the TTP effect were seeded in six replicates in 96-well clear-bottom black plates and cotransfected overnight using Lipofectamine 2000 with 50 ng of RPS30-SGFP-ARE constructs along with 10 ng of the Tet-On-TTP PCR product. Fluorescence was measured 24 hours after treating the cells with 0.2 μg/mL of doxycycline using the BD Pathway 435 imager (BD Biosciences) as reported previously (27).

TTP and HuR overexpression

Transient transfection of TTP in the TTP-deficient MDA-MB-231 cells was performed using Lipofectamine 2000. For HuR overexpression experiments, MCF10A cells were seeded in 6-well plates and transfected with 1 μg of pcDNA3.1-HuR plasmid or vector alone using Lipofectamine 2000. Total RNA and protein was extracted 24 hours after transfection for qPCR of the 11 identified mRNAs and immunoblotting.

RNA interference

Two different small interfering RNA (siRNA) duplexes were synthesized by Metabion: HuR siRNA-1 (sense: 5′UGUGAAAGUGAUCCGCGAC3′ and antisense: 5′GUCGCGGAUCACUUUCACA3′) and HuR siRNA-2 (sense: 5′GCCUGUUCAGCAGCAUUGG3′ and antisense 5′CCAAUGCUGCUGAACAGGC3′) while the scrambled siRNA sequence is as follows (sense: 5′GCCAUGUAUACGCGGUUC3′ and antisense: 5′GAACCGCGUAUACAUGGCC3′). MDA-MB-231 or HEK293 cells were transfected with 50 nmol/L of the siRNA for 48 hours, RNA was extracted for qRT-PCR or fluorescence was quantitated, respectively. HuR silencing efficiency was verified by immunoblotting.

Quantitative real-time PCR

Total RNA was extracted (TRI reagent, Sigma) and reverse transcribed as described previously (5). qPCR was performed in multiplex using FAM-labeled TaqMan primer and probe sets (Applied Biosystems) as outlined in Supplementary Methods, and normalized to VIC-labeled human GAPDH or RPLPO as the endogenous control. Samples were amplified in triplicate in a CFX96 cycler (Bio-Rad), and quantification of relative expression was performed using the ΔΔCt method.

Immunoprecipitation of RNP complexes

HEK-293 cells (10 cc plates) were transfected with Lipofectamine 2000 overnight with 2 μg pcDNA3.1 vectors expressing either HA-tagged-TTP or myc-tagged HuR along with either 100 ng of RPS30-SGFP or RPS30-SGFP-TNF-ARE as controls. TTP and HuR-associated RNA species were obtained using the detailed RNA-IP protocol outlined in Supplementary Methods.

Western blotting

Western blotting was performed as described previously (5). Specific antibodies to TTP, HuR, and serine phosphorylated H3 histone were obtained from Santa Cruz Biotechnology. Anti-HA and anti-β actin was obtained from Roche and Cell Signaling Technology, respectively.

Mitosis assessment: flow cytometry and histone phosphorylation

MDA-MB-231 Tet-Off cells (28) were seeded with or without tetracycline (1 μg/mL). After 24 hours, nocodazole was added to the cells to enrich for G2–M phase for 16 hours. Generally, we used 30 μg/mL of the drug as higher concentrations (e.g., 100 μg/mL) led to overarrested G2–M and toxicity. Cells were trypsinized, treated with DNase (2 U/μL, Ambion), washed with PBS, and fixed with 70% ethanol. Then, cells were washed with PBS, treated with RNase (100 μg/mL), and simultaneously stained with propidium iodide (25 μg/mL) for 30 minutes at 37°C for cell-cycle analysis. Proteins were extracted from the cells for immunoblotting and probed with rabbit anti-phosphorylated Histone H3 (Ser 10, Santa Cruz Biotechnology).

Survival analysis

The Kaplan–Meier survival analyses were performed using the Kaplan–Meier plotter portal (http://kmplot.com/analysis/), a comprehensive dataset for survival analysis that covers cross-normalized expression data of 54,675 genes in 4,142 breast cancer patients (29). The database was built from the gene expression and survival data extracted from the European Genome-Phenome Archive (EGA) and the Gene Expression Omnibus (GEO) repositories. Overall survival, recurrence-free survival, and distant metastasis–free survival were determined using gene cluster stratification. In addition, survival analysis in the METABRIC and TCGA RNA-seq databases was performed separately. Associations between gene expression and patient survival were assessed by the Kaplan–Meier method (log-rank test, GraphPad 6.0; ref. 30). The percentile threshold algorithm (25) was used to determine the optimal cutoff of the ARE-cluster members. The JETSET best probe set was selected in case multiple probe sets measured the same gene to ensure the optimal probe set for each gene. HRs and P values were determined by Cox proportional hazards regression.

ARE gene expression and TTP deficiency are over-represented in cancer

We performed a comprehensive analysis of ARE-mRNA expression across several cancer types (Fig. 1A). We intersected the AU-rich element mRNA database (ARED) with overexpressed genes (tumor/normal: ≥1.7, Q < 0.0001; two-sample t-test) extracted from different large cancer datasets (Supplementary Table S1). We found that the cancer ARE-mRNAs were significantly over-represented (13–23%) when compared with the overall cancer-overexpressed mRNAs (∼7.2%), which makes an enrichment of 2- to 3-fold, (P < 0.0001; one sample one-sided tailed t test; Fig. 1B).

Figure 1.

Systemic analysis of ARE-mRNAs and TTP expression in cancer. A, schematic representation illustrating the approach used for deriving a list of overexpressed ARE-mRNAs in cancer types. Gene expression levels were mined from 20 datasets representing 10 cancer types (Datasets, Supplementary Table S1). The threshold for overexpression was set as 1.7-fold of tumor to normal (T/N) and Q < 0.0001. Overexpressed ARE-mRNAs for each dataset were determined by intersecting total overexpressed genes with the ARED database. B, percentage of ARE-mRNAs that are overexpressed in the different cancer types in proportion to the total expressed mRNA in each cancer. C, tumor to normal level fold of TTP mRNAs was obtained from each cancer dataset. D, density frequency plot of TTP mRNA levels in normal (green) and tumor tissues (brown) using the indicated dataset. E, schematic representation of the three different HuR polyadenylation mRNA variants. The relative probe target locations (arrows) are shown (Affymetrix probe starts with A_ and Illumina with ILMN). pA, poly A signal. Boxes (between vertical lines) denote exons. F, tumor/normal levels of HuR mRNA by probe ID. G, scatter plot using fit model for the correlation between TTP/HuR mRNA ratios and the cancer 414 ARE-mRNAs. H, volcano plots for TTP/HuR mRNA ratio comparative correlation with ARE- and non-ARE cancer mRNAs by significance (−log10P value) as described in Materials and Methods. Pie charts depict the percentages of ARE-mRNAs (top) or non-ARE-mRNAs (bottom) according to the correlation. Vertical dotted lines denote negative correlation (R ≤ −0.3; P < 0.001) or positive correlation (R ≥ 0.3) with TTP expression levels. Pearson correlation with Fisher exact test for significance (P value) was used.

Figure 1.

Systemic analysis of ARE-mRNAs and TTP expression in cancer. A, schematic representation illustrating the approach used for deriving a list of overexpressed ARE-mRNAs in cancer types. Gene expression levels were mined from 20 datasets representing 10 cancer types (Datasets, Supplementary Table S1). The threshold for overexpression was set as 1.7-fold of tumor to normal (T/N) and Q < 0.0001. Overexpressed ARE-mRNAs for each dataset were determined by intersecting total overexpressed genes with the ARED database. B, percentage of ARE-mRNAs that are overexpressed in the different cancer types in proportion to the total expressed mRNA in each cancer. C, tumor to normal level fold of TTP mRNAs was obtained from each cancer dataset. D, density frequency plot of TTP mRNA levels in normal (green) and tumor tissues (brown) using the indicated dataset. E, schematic representation of the three different HuR polyadenylation mRNA variants. The relative probe target locations (arrows) are shown (Affymetrix probe starts with A_ and Illumina with ILMN). pA, poly A signal. Boxes (between vertical lines) denote exons. F, tumor/normal levels of HuR mRNA by probe ID. G, scatter plot using fit model for the correlation between TTP/HuR mRNA ratios and the cancer 414 ARE-mRNAs. H, volcano plots for TTP/HuR mRNA ratio comparative correlation with ARE- and non-ARE cancer mRNAs by significance (−log10P value) as described in Materials and Methods. Pie charts depict the percentages of ARE-mRNAs (top) or non-ARE-mRNAs (bottom) according to the correlation. Vertical dotted lines denote negative correlation (R ≤ −0.3; P < 0.001) or positive correlation (R ≥ 0.3) with TTP expression levels. Pearson correlation with Fisher exact test for significance (P value) was used.

Close modal

The TTP mRNA is generally found deficient in different cancers, with a 3- to 6-fold reduction compared with normal tissues (Fig. 1C). It was most deficient in breast cancer; density plots clearly illustrate the dramatic differences in TTP levels in normal versus cancer tissues in two large datasets (Fig. 1D). Because of this particular reduction in TTP gene expression and the large size of data available, further systemic analyses of ARE-mRNA expression patterns and their relation to TTP and HuR levels were conducted initially in breast cancer. All overexpressed genes with at least 1.7-fold increase (P < 0.0001, two-sample t test) were retrieved from TCGA-IDBC data and crossed with the ARED (2). A total of 414 overexpressed ARE-mRNAs were derived (Supplementary Table S2).

Correlation modeling of TTP/HuR axis and alternative polyadenylation of HuR

We looked at the relationship between HuR and the expressed levels of the cancer ARE-mRNAs. HuR has three polyadenylation variants as previously reported by us and others (18–20). Therefore, we examined the probe sets (Affymetrix) in the TCGA breast cancer dataset, and found there were two different sets for HuR (ELAVL1) mRNA, one that targets the most abundant variant (2.7 kB) and another that targets the long 6 kb variant (Fig. 1E). The probe set that detects abundant HuR variants exhibits overexpression of HuR in breast cancer. In contrast, the probe set that specifically targets the distal (6 kb mRNA) 3′UTR appears to be underexpressed compared with normal tissues (Fig. 1F). In METABRIC dataset (23), there is only one probe (Illumina), which is directed to the distal 3′UTR and thus the 6 kb variant, and, indeed, gave underexpressed levels of HuR in tumor tissues compared with normal tissues (Fig. 1F). This observation is also found with other types of cancer; particularly, brain and lung cancer (Supplementary Fig. S1). Thus, we focused on the probe set that targets the most abundant form of HuR transcript variants, which also is overexpressed in cancers.

Next, we performed expression/correlation analysis by integrating both the levels of TTP and HuR mRNAs. The TTP/HuR mRNA ratios were computed since the balance of these RNA-binding proteins can influence the expression of certain ARE-mRNAs (21). We tested the correlation of the TTP/HuR mRNA ratio with the expression patterns of the 414 ARE-mRNAs using the fit model; the result was significant, R2 = 0.98 (P < 0.001; Pearson correlation with Fisher test of significance; Fig. 1G). Moreover, the volcano plot shows that a large proportion (47%) of ARE-mRNAs have a negative correlation with TTP/HuR mRNA level ratio (198 ARE-mRNAs with R-values < −0.3 and P values < 0.0001-(−Log10 ≥ 4) compared with non-ARE mRNAs (only 2.8% of the 1979 non-ARE-mRNAs, Fig. 1H). Supplementary Table S3 shows the identities of the ARE genes, R-values, and P-values that correlate with TTP, HuR, or TTP/HuR mRNAs. In general, TTP levels correlated negatively with ARE-mRNAs, while HuR positively correlated with the ARE-mRNAs when compared with non-ARE-mRNAs (Supplementary Fig. S2). Overall, these data clearly show that it is much more likely for overexpressed ARE genes to negatively correlate with low TTP/HuR ratio than overexpressed non-ARE genes.

Step-wise functional clustering analyses for cancer ARE-mRNAs

To derive further correlation and functional clustering of the cancer ARE-mRNAs that depend on TTP/HuR mRNA ratio, the Normal Mixture model was used to find the most significant cluster of genes (Fig. 2A). Although we have used stringent criteria on the expense of sensitivity, several known TTP/HuR ARE-mRNAs such as uPA (PLAU), PLAUR, and MMP13 were found (Supplementary Table S2). Five clusters were generated, and the total squared correlation among the three parameters was highest (R = 0.81) in cluster five, which contained a 23-ARE-mRNA cluster (Fig. 2B). The expression levels of this cluster in the breast tumors and the normal tissues from the TCGA dataset are illustrated on a heatmap demonstrating significant cancer upregulation for each cluster member (Fig. 2C).

Figure 2.

Clustering analysis of cancer ARE-mRNAs with TTP and HuR expression. A, normal mixture clustering analysis based on the correlation (R) of the tumor to normal expression ratios (fold) of the 414 overexpressed breast cancer ARE-mRNAs with that of TTP and HuR. B, heatmap visualizing the most significant cluster (Cluster Five) containing 23 ARE-mRNA showing fold (tumor to normal mRNA ratio) and ARE-mRNA correlations (R-values) with HuR and TTP mRNA levels. Legend shows the scale for each parameter. C, heatmap visualizing the overexpression of the resultant 23 ARE-mRNA cluster in the invasive ductal breast cancer. D, functional annotation of the 23 genes using Geneset analysis; enriched GO categories are in red (multiple tests adjusted P value) while those in black are their nonenriched parents. Pie charts show the most enriched GO categories using DAVID. E, modules of enriched protein interactions (red lines) on directed acyclic graphs (DAG). Green nodes, mitotic proteins. The large inset shows the STRING analysis of known and predicted protein–protein interactions among the 11 ARE-mRNA gene products. The nodes indicate proteins, and the edges indicate interaction between these genes.

Figure 2.

Clustering analysis of cancer ARE-mRNAs with TTP and HuR expression. A, normal mixture clustering analysis based on the correlation (R) of the tumor to normal expression ratios (fold) of the 414 overexpressed breast cancer ARE-mRNAs with that of TTP and HuR. B, heatmap visualizing the most significant cluster (Cluster Five) containing 23 ARE-mRNA showing fold (tumor to normal mRNA ratio) and ARE-mRNA correlations (R-values) with HuR and TTP mRNA levels. Legend shows the scale for each parameter. C, heatmap visualizing the overexpression of the resultant 23 ARE-mRNA cluster in the invasive ductal breast cancer. D, functional annotation of the 23 genes using Geneset analysis; enriched GO categories are in red (multiple tests adjusted P value) while those in black are their nonenriched parents. Pie charts show the most enriched GO categories using DAVID. E, modules of enriched protein interactions (red lines) on directed acyclic graphs (DAG). Green nodes, mitotic proteins. The large inset shows the STRING analysis of known and predicted protein–protein interactions among the 11 ARE-mRNA gene products. The nodes indicate proteins, and the edges indicate interaction between these genes.

Close modal

Directed acyclic graphs (DAG) enriched in Gene Ontology (GO) categories, each as a node in a network, were built through Geneset analysis (Fig. 2D). The mitotic (M) cell cycle was the most significant node and comprises 11 genes, which are CDC6, CENPA, CENPE, KIF11, KIF18A, NCAPG, NEK2, NUF2, PBK, PRC1, and TOP2A. An enrichment (7-fold) of this category was observed when this ARE-mRNA cluster is compared with all TCGA overexpressed genes (P < 0.0001; χ2 test). The protein interaction module, which is based on DAGs, reveals that the highest enrichment (P = 6.70e−11) is in DNA replication-independent nucleosome assembly and microtubule motor activity; both are within the M-phase of cell-cycle category (Fig. 2E). The use of STRING analysis confirmed the significant protein interactions among the gene nodes (Fig. 2E, large inset).

Bioinformatics and gene expression of M cell cycle ARE-mRNAs

The weighted means of the different ARE-clusters between tumor and normal tissue ratios were calculated (Fig. 3A). As a whole, the 414 overexpressed ARE-mRNAs were slightly, although statistically significantly (P < 0.001) higher than non-ARE overexpressed control set genes (n = 1,979). However, the TTP/HuR–correlative 23-ARE-mRNA cluster weighted mean has significantly higher expression ratios (∼5-fold) compared with both the total overexpressed ARE- and non-ARE-mRNAs (Fig. 3A). The 11 M cell cycle ARE-mRNAs were significantly higher than other gene sets (6-fold of the weighted mean, P < 0.001, two-sample t test; Fig. 3A). The data of individual expression levels and correlations of the 11 ARE-mRNA cluster demonstrate significant upregulation and negative correlation with the TTP/HuR mRNA ratio (Supplementary Figs. S3 and S4).

Figure 3.

Expression of the ARE-mRNA mitotic cluster in cancer. A, the weighted means of tumor to normal ratio for the indicated ARE-mRNA clusters that correlated with TTP/HuR. Level 2 raw data was obtained from TCGA invasive ductal breast cancer dataset. B, a heatmap was generated to visualize the overexpression of the mitotic cell cycle ARE-mRNAs in 10 different cancer types. C, the mean expression (nonweighted mean) of the combined fold increase of the 11 M ARE-mRNAs was derived from the Nextbio portal using several cancer types. No SEM data are given from the Nextbio output, but P values are reported (numbers are shown). D, the normal MCF-10A and MCF-12A, as well as the breast tumor cell lines triple-negative MDA-MB-231 and ER-positive/Her-negative, MCF-7, were grown in 6-well culture plates. Total RNA was extracted, and the mRNAs were quantitated by qRT-PCR using TaqMan-specific primers and were normalized to the housekeeping gene, RPLPO. E, the mRNA levels of the 11 ARE-mRNA cluster in MCF10A and the HER2-positive cell lines BT474 and SK-BR3 using qRT-PCR as above. Data are from one experiment are representative of at least two independent experiments. D and E, statistical significance was assessed by one-way ANOVA, with underlined values of *, P < 0.05; **, P < 0.01; ***, P < 0.0001. Post hoc tests, Dunnett test for comparison with MCF10A or Tukey (with each normal–tumor pair) were used; *, P < 0.01; **, P < 0.0001.

Figure 3.

Expression of the ARE-mRNA mitotic cluster in cancer. A, the weighted means of tumor to normal ratio for the indicated ARE-mRNA clusters that correlated with TTP/HuR. Level 2 raw data was obtained from TCGA invasive ductal breast cancer dataset. B, a heatmap was generated to visualize the overexpression of the mitotic cell cycle ARE-mRNAs in 10 different cancer types. C, the mean expression (nonweighted mean) of the combined fold increase of the 11 M ARE-mRNAs was derived from the Nextbio portal using several cancer types. No SEM data are given from the Nextbio output, but P values are reported (numbers are shown). D, the normal MCF-10A and MCF-12A, as well as the breast tumor cell lines triple-negative MDA-MB-231 and ER-positive/Her-negative, MCF-7, were grown in 6-well culture plates. Total RNA was extracted, and the mRNAs were quantitated by qRT-PCR using TaqMan-specific primers and were normalized to the housekeeping gene, RPLPO. E, the mRNA levels of the 11 ARE-mRNA cluster in MCF10A and the HER2-positive cell lines BT474 and SK-BR3 using qRT-PCR as above. Data are from one experiment are representative of at least two independent experiments. D and E, statistical significance was assessed by one-way ANOVA, with underlined values of *, P < 0.05; **, P < 0.01; ***, P < 0.0001. Post hoc tests, Dunnett test for comparison with MCF10A or Tukey (with each normal–tumor pair) were used; *, P < 0.01; **, P < 0.0001.

Close modal

Furthermore, we looked at the expression patterns of the ARE-mRNA cluster across 10 different types of cancers. The expression of 11 ARE-mRNAs was upregulated in a number of different cancers as shown in the heatmap (Fig. 3B). The upregulation of the mean expression (nonweighted mean) of the M ARE-mRNA cluster is also obvious across the different cancers (Fig. 3C).

By qRT-PCR, we evaluated the expression levels of the M cell-cycle ARE-mRNAs in a number of cell lines (Fig. 3D and E). Overall, the mRNA levels were higher in the triple-negative MDA-MB-231 cells compared with moderately-tumorigenic ER-positive, MCF-7, and normal-like breast cancer cells MCF10A and MCF-12A, respectively (Fig. 3D). Using different panel of cells, HER-positive/ER-negative status, we also found that all the ARE-mRNAs (except CENPA, which had very low levels) were highly expressed when compared with the normal-like cells (Fig. 3E). In particular, the mRNA levels in HER-positive ER−/PR-negative SKBR3 was the highest when compared with other lines in the panel (Fig. 3E). In general, these results are in agreement with the observations made in the patient tissues' datasets in which these ARE-mRNAs are overexpressed in invasive ductal breast cancer.

Antimitotic effect of TTP

On the basis of the computational approach, TTP could be functionally related to M-phase of the cell cycle. Thus, we have utilized TetO/TTP–regulated MDA-MB-231 cell line (28); with removal of tetracycline, the cells express TTP and exhibit reduced proliferation (Fig. 4A). When cells were treated with nocodazole, TTP behaved as an antimitotic protein and blocked cell cycle at G2–M phase as determined by flow cytometry (Fig. 4B). Furthermore, we showed that histone H3 phosphorylation, a marker of mitotic arrest, is higher when TTP is expressed in the nocodazole-treated cells (Fig. 4C). A confirmatory experiment was performed using transient expression of TTP in MDA-MB-231 cells (Fig. 4D).

Figure 4.

The antimitotic effect of TTP. A, continuous monitoring of cell proliferation of MDA-MB-231 Tet-off cells in the presence or absence of tetracycline (1 μg/mL). B, after 24 hours of tetracycline treatment, nocodazole (30 μg/mL) was added to the cells for 16 hours to enrich for G2–M–positive cells. Data are representative of three flow cytometry experiments. C, MDA-MB-231 Tet-Off cells were treated with tetracycline and nocodazole as indicated in B and protein was extracted for immunoblotting with anti-p-Histone H3 or β-actin. The experiment is representative of two independent experiments. D, transient expression of TTP and histone H3 phosphorylation. MDA cells were transfected with TTP or vector, followed by nocadzole treatment (16 hours). Western blotting showed anti-phosphorylated H3 and actin control.

Figure 4.

The antimitotic effect of TTP. A, continuous monitoring of cell proliferation of MDA-MB-231 Tet-off cells in the presence or absence of tetracycline (1 μg/mL). B, after 24 hours of tetracycline treatment, nocodazole (30 μg/mL) was added to the cells for 16 hours to enrich for G2–M–positive cells. Data are representative of three flow cytometry experiments. C, MDA-MB-231 Tet-Off cells were treated with tetracycline and nocodazole as indicated in B and protein was extracted for immunoblotting with anti-p-Histone H3 or β-actin. The experiment is representative of two independent experiments. D, transient expression of TTP and histone H3 phosphorylation. MDA cells were transfected with TTP or vector, followed by nocadzole treatment (16 hours). Western blotting showed anti-phosphorylated H3 and actin control.

Close modal

Reporter assessment of ARE activity and response to TTP induction

The putative AREs of the 11-gene cluster and their context sequences were cloned in the 3′ UTR of fluorescent reporter assays (Fig. 5A). A heatmap was constructed to reflect the strength of the TTP target sequence UUAUUUAUU/UAUUUAU (Fig. 5A), which shows that NEK2, CENPE, and NCAPG AREs have Class-II–like ARE while the rest are largely Class-I AREs; that is, AUUUA in a U-rich context (Fig. 5A; Supplementary Table S4). We used HEK293 cell line, which is suitable for weak fluorescence-based reporter assays. Seven of the AREs caused a reduction in the reporter activity demonstrating functional and active AREs (Fig. 5B). Statistically significant reductions ranged from 25% (CDC6) to 50% (CENPE, NEK2), which share the strongest TTP target sequence score (Fig. 5B). The IL8 ARE caused 52% reduction in reporter expression. The mRNAs for CENPA, PRC1, and TOP2A, although not statistically significant, they demonstrated a modest 25% reduction in reporter fluorescence compared with the non-ARE control (Fig. 5B). Also, most responded to TTP induction or HuR knockdown to varying degrees of reduction (Fig. 5C and D).

Figure 5.

Reporter assessment of ARE activity and response to TTP induction. A, a scheme representing the AREs of the M cycle ARE-mRNAs along with the cluster membership according to the ARED. The color score represents strength of TTP sequence in which 1 = perfect match with UUAUUUAUU. B, HEK293 cells were transfected with the RPS30-SGFP-ARE reporters along with control vector (non-ARE) and positive control (IL8 ARE) as described in Materials and Methods. Fluorescence was measured after 24 hours and presented as percent mean of control ± SEM. C, TTP effect on ARE-reporter expression. A Tet-O–inducible TTP expression cassette was cotransfected with each of the reporter constructs in HEK-Tet-ON cells. TTP was induced by 0.2 μg/mL doxycycline treatment, and the effect of TTP was assessed 24 hours later by quantifying the level of the reporter fluorescence. D, HuR knockdown and ARE reporter activity. HEK293 cells were treated with 50 nmol/L control siRNA or HuR siRNA for 24 hours and transfected as above. Fluorescence was quantified after another 24 hours. *, P < 0.05; **, P < 0.005; ***, P < 0.001; ****, P < 0.0001 using Student t test.

Figure 5.

Reporter assessment of ARE activity and response to TTP induction. A, a scheme representing the AREs of the M cycle ARE-mRNAs along with the cluster membership according to the ARED. The color score represents strength of TTP sequence in which 1 = perfect match with UUAUUUAUU. B, HEK293 cells were transfected with the RPS30-SGFP-ARE reporters along with control vector (non-ARE) and positive control (IL8 ARE) as described in Materials and Methods. Fluorescence was measured after 24 hours and presented as percent mean of control ± SEM. C, TTP effect on ARE-reporter expression. A Tet-O–inducible TTP expression cassette was cotransfected with each of the reporter constructs in HEK-Tet-ON cells. TTP was induced by 0.2 μg/mL doxycycline treatment, and the effect of TTP was assessed 24 hours later by quantifying the level of the reporter fluorescence. D, HuR knockdown and ARE reporter activity. HEK293 cells were treated with 50 nmol/L control siRNA or HuR siRNA for 24 hours and transfected as above. Fluorescence was quantified after another 24 hours. *, P < 0.05; **, P < 0.005; ***, P < 0.001; ****, P < 0.0001 using Student t test.

Close modal

Physical association and response of the M ARE-mRNAs to TTP and HuR

We investigated the possible physical association of TTP with the 11 M cell-cycle ARE-mRNAs by overexpressing an HA-tagged TTP (HA-TTP) construct in HEK-293 cells. The immunoprecipitation of TTP was confirmed by Western blotting (Fig. 6A). The RNA-IP clearly shows that the SGFP-TNF-ARE reporter mRNA is highly enriched on the HA-coupled beads compared with the myc-coupled beads (>30-fold enrichment). The ARE-containing known TTP target, IL8 mRNA, is highly enriched in the TTP-HA IP sample (∼23-fold) but not mRNAs of non-ARE housekeeping genes (Fig. 6A). Next, the association of 10 of the ARE-mRNAs was investigated (CENPA expression was below detection levels) by qRT-PCR and all were found to be enriched with TTP (Fig. 6A). The enrichment degree varied between 6-fold (CENPE) and 20-fold (PBK). As independent experiments, TTP-associated ARE-mRNAs were assessed in MDA-MB-231 using an alternative RNA-IP protocol (21) and further confirmed that TTP bound the ARE-mRNAs except CENPA, in which its levels were too low (Fig. 6B). Exogenous expression of TTP led to reduction of the majority of the 11 ARE-mRNAs (except NCAPG; Fig. 6C).

Figure 6.

RNA/protein immunoprecipitation and modulation of TTP and HuR. A, RNA/protein coimmunoprecipitation of TTP with the putative M ARE-mRNA cluster. HA-TTP HEK293-transfected cells were lysed and TTP protein was immunoprecipitated with anti-HA coupled beads. The negative control (NC) immunoprecipitation was performed with anti-myc. TTP-IP was confirmed by Western blotting (insets). TTP-associated mRNA levels were assessed by qRT-PCR and normalized to RPLPO as background control. The non-ARE mRNAs of GAPDH, GUS, and RPLP1 were used as internal negative controls while IL8 ARE-mRNA was used as control. B, MDA-MB-231 cells were transfected with TTP expression plasmid for 24 hours. Cells were lysed and TTP protein was immunoprecipitated using anti-TTP antibody or normal IgG control antibody. Associated mRNA was quantified by qRT-PCR using gene-specific probes normalized to GAPDH. C, TTP-mediated downregulation of M ARE-mRNA in MDA-MB-231 cells. Cells were transfected with 10 ng TTP-pcDNA3.1 plasmid and 500 ng DNA with pcDNA3.1 empty vector or transfected with 500 ng empty vector alone. Total RNA was extracted after 24 hours and qRT-PCR of 11 genes plus positive and negative controls were performed as in A. D, the binding of myc-HuR to the 11 M ARE-mRNAs was assessed in a similar manner to A. The myc-HuR IP was performed on anti-myc coupled beads and the negative control immunoprecipitated with anti-HA coupled beads. The same positive and negative control mRNAs were used as in A. E and F, HuR silencing and the M ARE-mRNA cluster in MDA-MB-231. The cells (0.5 × 106) were transfected with 50 nmol/L HuR siRNA-1, HuR siRNA-2, or scrambled siRNA control for 48 hours. HuR silencing efficiency was determined by Western blotting. The mRNA levels were determined by qRT-PCR using HuR siRNA-2 samples. Data are from one representative experiment of at least two independent experiments. G, effect of HuR overexpression on the M ARE-mRNA levels in MCF10A. Cells (0.5 × 106 cells) were transfected with 1 μg of pcDNA3.1-HuR plasmid or vector alone and RNA was extracted after 24 hours for qRT-PCR. The results are from one experiment representative of two independent experiments. For all data above: *, P < 0.05; **, P < 0.005; ***, P < 0.0001 (Student t test) shown for each ARE mRNA in comparison with the control.

Figure 6.

RNA/protein immunoprecipitation and modulation of TTP and HuR. A, RNA/protein coimmunoprecipitation of TTP with the putative M ARE-mRNA cluster. HA-TTP HEK293-transfected cells were lysed and TTP protein was immunoprecipitated with anti-HA coupled beads. The negative control (NC) immunoprecipitation was performed with anti-myc. TTP-IP was confirmed by Western blotting (insets). TTP-associated mRNA levels were assessed by qRT-PCR and normalized to RPLPO as background control. The non-ARE mRNAs of GAPDH, GUS, and RPLP1 were used as internal negative controls while IL8 ARE-mRNA was used as control. B, MDA-MB-231 cells were transfected with TTP expression plasmid for 24 hours. Cells were lysed and TTP protein was immunoprecipitated using anti-TTP antibody or normal IgG control antibody. Associated mRNA was quantified by qRT-PCR using gene-specific probes normalized to GAPDH. C, TTP-mediated downregulation of M ARE-mRNA in MDA-MB-231 cells. Cells were transfected with 10 ng TTP-pcDNA3.1 plasmid and 500 ng DNA with pcDNA3.1 empty vector or transfected with 500 ng empty vector alone. Total RNA was extracted after 24 hours and qRT-PCR of 11 genes plus positive and negative controls were performed as in A. D, the binding of myc-HuR to the 11 M ARE-mRNAs was assessed in a similar manner to A. The myc-HuR IP was performed on anti-myc coupled beads and the negative control immunoprecipitated with anti-HA coupled beads. The same positive and negative control mRNAs were used as in A. E and F, HuR silencing and the M ARE-mRNA cluster in MDA-MB-231. The cells (0.5 × 106) were transfected with 50 nmol/L HuR siRNA-1, HuR siRNA-2, or scrambled siRNA control for 48 hours. HuR silencing efficiency was determined by Western blotting. The mRNA levels were determined by qRT-PCR using HuR siRNA-2 samples. Data are from one representative experiment of at least two independent experiments. G, effect of HuR overexpression on the M ARE-mRNA levels in MCF10A. Cells (0.5 × 106 cells) were transfected with 1 μg of pcDNA3.1-HuR plasmid or vector alone and RNA was extracted after 24 hours for qRT-PCR. The results are from one experiment representative of two independent experiments. For all data above: *, P < 0.05; **, P < 0.005; ***, P < 0.0001 (Student t test) shown for each ARE mRNA in comparison with the control.

Close modal

The binding of HuR to the 11 ARE-mRNAs were assessed in a similar manner using transfected anti-myc tagged HuR. (Fig. 6D). The immunoblot shows the specificity of the antibody (Fig. 6D inset). Although the bindings to the different 11 ARE-mRNAs were modest compared with TTP, it is within the same degree as with the HuR mRNA target of IL-8 and is statistically significant (Fig. 6D). NEK2 mRNA has the strongest binding (3-fold) while TOP2A and PRC1 mRNAs were the weakest. These results demonstrate the physical association of the RNA-binding proteins TTP and HuR with the ARE-mRNA mitotic cluster.

We have also assessed the outcome of HuR knockdown in MDA-MB-231 (Fig. 6E), which has high levels of HuR (21), on the expression of the 11 ARE-mRNAs, and found that siRNA-mediated silencing caused significant reduction in the mRNA levels (Fig. 6F). Using the nontransformed MCF10A, which generally has low levels of HuR (21), we found that overexpressed HuR was able to augment the M cell cycle ARE-mRNA levels (Fig. 6G).

Association of M cell cycle ARE cluster and the survival of cancer patients

Patient classification to low and high mean expression of the M cell cycle ARE-mRNA cluster (Fig. 7A) inversely corresponds as expected with the TTP/HuR mRNA ratio (Fig. 7B). Poor relapse-free survival (n = 4,073 patient data) was observed with the high M cluster cohort (P < 10e−10, log-rank test; and HR = 2.4; Fig. 7C). Similarly, poor outcome was observed in distant metastasis-free survival (DMFS, HR = 2.4; Fig. 7D) and in overall survival (OS, HR = 2.13, Supplementary Fig. S5A). We repeated the survival analysis in the non-Affymetrix METABRIC database (n = 1959) and the RNA-seq data (n = 781) and data demonstrate poor survival with higher mean expression of the mitotic ARE-mRNA cluster (Supplementary Fig. S5B and S5C).

Figure 7.

Clinical outcome of the M cell cycle TTP/HuR correlated ARE-mRNA expression. A, a heatmap for M ARE-mRNA expression cohorts. Data in Kaplan–Meier portal was used to classify the data into low- and high-expressing M ARE-mRNA cohorts in relation to the clinical outcome. The dotted line represents the best cutoff. B, TTP/HuR mRNA ratio levels of the two expression cohorts. The best cutoff was based on the expression levels of the ARE-mRNA cluster shown in A. C and D, Kaplan–Meier survival curves generated in accordance with expression classifiers in A using relapse-free survival (RFS; C) or distant metastasis–free survival (DMFS; D). E, Kaplan–Meier survival curve using TTP/HuR mRNA classifiers. F and G, grade, lymph node involvement, and receptor status in patients with TTP/HuR mRNA classification (F) and M cell cycle ARE-mRNA cluster (G). P values (numbers) and HR are indicated. Q2 test was used to evaluate the statistical significance between each group member, *, P < 0.01; **, < 0.001; ***, P < 0.0001. NS, nonsignificant.

Figure 7.

Clinical outcome of the M cell cycle TTP/HuR correlated ARE-mRNA expression. A, a heatmap for M ARE-mRNA expression cohorts. Data in Kaplan–Meier portal was used to classify the data into low- and high-expressing M ARE-mRNA cohorts in relation to the clinical outcome. The dotted line represents the best cutoff. B, TTP/HuR mRNA ratio levels of the two expression cohorts. The best cutoff was based on the expression levels of the ARE-mRNA cluster shown in A. C and D, Kaplan–Meier survival curves generated in accordance with expression classifiers in A using relapse-free survival (RFS; C) or distant metastasis–free survival (DMFS; D). E, Kaplan–Meier survival curve using TTP/HuR mRNA classifiers. F and G, grade, lymph node involvement, and receptor status in patients with TTP/HuR mRNA classification (F) and M cell cycle ARE-mRNA cluster (G). P values (numbers) and HR are indicated. Q2 test was used to evaluate the statistical significance between each group member, *, P < 0.01; **, < 0.001; ***, P < 0.0001. NS, nonsignificant.

Close modal

Testing direct correlation of patients' survival with TTP/HuR mRNA ratio stratification, we found that the cohort with the low TTP/HuR mRNA ratio had poor survival outcome (Fig. 7E). Poor survival (higher HR) in relation to the low TTP/HuR mRNA ratio is more prominent with ER-positive cases and negative lymph node involvement (Fig. 7F). The 11 M ARE-mRNA cluster was also more significantly associated with poor survival in case of ER-positive and HER-negative tissues (Fig. 7G). It appears that the TTP/HuR mRNA ratio and the ARE-mRNA cluster is more associated with survival in patients with lower grade tumors (Fig. 7F). Overall, high expression of the mitotic ARE cluster and their association with a low TTP/HuR mRNA ratio is indicative of poor survival in breast cancer patients.

Alterations in ARE-mediated pathways are prominent posttranscriptional events that are increasingly studied in cancer. Only a small set of ARE-mRNAs have been previously shown to be overexpressed in cancer and regulated by the key RNA-binding proteins, TTP and HuR [for an updated list, see our review (1)]. In this study, we expanded cancer ARE-mRNAs to nearly 200 ARE-mRNAs that correlate negatively with TTP and positively with HuR expression. Further statistical and functional enrichment, and experimental validation led to the identification of 11 ARE-mRNAs, which code for proteins involved in the mitotic (M)-phase of the cell cycle. The 11 mitosis members perform critical roles in M phase of the cell cycle (31). Thus, TTP may promote G2–M arrest by the fine-tuning the temporal expression of these ARE-mRNAs.

The systemic study conducted here shows that ARE-mRNA expression is over-represented in multiple cancer types adding to the significance of the ARE-mediated posttranscriptional control in cancer. In these same types of tumors, TTP deficiency is observed and is linked to nearly half (198 genes) of the overexpressed ARE-mRNAs in breast cancer as assessed here. A deficiency of TTP can lead to the increased growth of cancer cells, because TTP targets those mRNAs involved in cellular proliferation, such as the previously identified cyclin D1 myc, E2F1, and IL6 (32, 33) and also the M cell cycle ARE-mRNAs newly identified here. Modulation of TTP in breast and other cancer cells by ectopic expression impacts cellular growth and attenuates tumor xenografts in mice (28, 34, 35), which is in agreement with the patients' data here.

While TTP is deficient in cancer, HuR, which stabilizes ARE-mRNAs is often overexpressed in tumors or its cytoplasmic localization is increased (36–41). HuR is more of a general RNA-stability promoting protein with preference towards AU/U-rich sequences (42). Here, we also expanded (more than 100 mRNAs) HuR-correlated ARE-mRNAs that are linked to cancer. HuR mRNA has three alternative polyadenylation forms of 1.5 kb, 6 kb long variant, and the most abundant form is 2.7 kb transcript (18). HuR autoregulates its own mRNA at the posttranscriptional level (18–20). The lower expression of the long variant in contrast to the overexpression of the shorter (and abundant) mRNA variant in cancer can be explained by defects in the polyadenylation process, which can favor the abundant variant. This may be consistent with the cell line work of two groups (19, 20). In general, HuR has been shown to block polyadenylation and cleavage at polyadenylation sites that are flanked by U-rich sequences (43). Thus, our data notably extend these laboratory HuR transcript expression patterns to patients' tissues. Furthermore, this brings attention to the fact that different probe sets directed towards different regions of the same transcript, including polyadenylation variants, can have an impact on the interpretation of microarray expression results, including the expression direction.

We focused analysis on TTP/HuR mRNA aberrations and correlation with the cancer genes because a TTP–HuR axis exists in cells: TTP competes for HuR binding and can reduce HuR mRNA expression (18, 21). Further mechanistic details showed that phosphorylation of TTP by MK2 alters that balance towards ARE-mRNA stabilization and translation (11, 44), and more recent work confirmed that TTP targets the HuR mRNA (45). Another evidence for the TTP–HuR axis is the ability of the miR-29a that targets TTP 3′UTR in breast cancer cells (22) to alter the normal TTP–HuR balance, while miR-29a inhibitor restores the reversed balance. This leads to reduction of the cellular levels of ARE-mRNAs and their proteins; for example, uPA and CXCR4 in cancer cells (5, 21).

The M cell cycle ARE-mRNAs either as individuals or as a group (weighted mean) were not only overexpressed in the invasive breast tumor tissues but were overexpressed across a number of solid tumors, including liver, lung, bladder, brain, stomach, and kidney. This implies that there is a wide-spread effect of dysregulated ARE-mediated posttranscriptional processes in controlling M cell cycle in cancer. Other findings do not rule out other regulatory mechanisms that are abnormal in cancer such as transcription, as these genes are tightly coexpressed.

Another potential and important ramification of TTP deficiency and HuR overexpression is the impact on angiogenesis, tumor invasion, and metastasis as it targets ARE-mRNAs that mediate such processes, including IL8, VEGF, IL6, uPA, uPAR, CXCR4, and MMP13 (4–6, 46, 47). TTP has been shown to significantly reduce tumor cell migration and invasion while HuR increases these activities (4, 5, 21, 48, 49). The combined impact on mitosis and tumor invasion due to lack of TTP in breast cancer cells and patients may confirm the putative tumor suppressor role of TTP, including its anti-invasive role. The strikingly low number of patient samples with normal TTP and TTP/HuR levels allows the suggestion that the repression of TTP might be an "essential factor" at least in invasive breast cancer. Those who have levels that are not dramatically reduced had better prognostic outcomes. There was significant stratification of survival based on TTP/HuR mRNA ratios and the M-phase cell cycle ARE-mRNA clusters in breast cancer patients, particularly with low-grade, lymph-node negative and ER-positive status. Thus, targeting the TTP-HuR axis, and thereby reducing the M cell cycle and invasiveness of the cancer cells, either by increasing TTP or directly by reducing HuR, may have beneficial effects in patients. Our results can give more insights to the role of the posttranscriptional network in cancer maintenance with new therapeutic approaches in mind.

No potential conflicts of interest were disclosed.

Conception and design: K.S.A. Khabar

Development of methodology: E. Hitti, N. Al-Souhibani, W. Moghrabi

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): R. Grepin, G. Pagès

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): E. Hitti, T. Bakheet, N. Al-Souhibani, W. Moghrabi, M.M. Shoukri, B. Győrffy, K.S.A. Khabar

Writing, review, and/or revision of the manuscript: E. Hitti, N. Al-Souhibani, B. Győrffy, K.S.A. Khabar

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): W. Moghrabi, S. Al-Yahya, M. Al-Ghamdi, M. Al-Saif, A. Lánczky

Study supervision: K.S.A. Khabar

The research was supported by intramural funding (RAC-2150004 to K.S.A Khabar) from the King Faisal Specialist Hospital & Research Centre.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Khabar
KS
. 
Post-transcriptional control of cytokine gene expression in health and disease
.
J Interferon Cytokine Res
2014
;
34
:
215
9
.
2.
Halees
AS
,
El-Badrawi
R
,
Khabar
KS
. 
ARED Organism: expansion of ARED reveals AU-rich element cluster variations between human and mouse
.
Nucleic Acids Res
2008
;
36
:
D137
40
.
3.
Bakheet
T
,
Frevel
M
,
Williams
BRG
,
Greer
W
,
Khabar
KSA
. 
ARED: human AU-rich element-containing mRNA database reveals an unexpectedly diverse functional reportiore of encoded proteins
.
Nucleic Acids Res
2001
;
29
:
246
54
.
4.
Al-Souhibani
N
,
Al-Ahmadi
W
,
Hesketh
JE
,
Blackshear
PJ
,
Khabar
KS
. 
The RNA-binding zinc-finger protein tristetraprolin regulates AU-rich mRNAs involved in breast cancer-related processes
.
Oncogene
2010
;
29
:
4205
15
.
5.
Al-Souhibani
N
,
Al-Ghamdi
M
,
Al-Ahmadi
W
,
Khabar
KSA
. 
Posttranscriptional control of the chemokine receptor CXCR4 expression in cancer cells
.
Carcinogenesis
2014
;
35
:
1983
92
.
6.
Essafi-Benkhadir
K
,
Onesto
C
,
Stebe
E
,
Moroni
C
,
Pages
G
. 
Tristetraprolin inhibits Ras-dependent tumor vascularization by inducing vascular endothelial growth factor mRNA degradation
.
Mol Biol Cell
2007
;
18
:
4648
58
.
7.
Devaney
JM
,
Tosi
LL
,
Fritz
DT
,
Gordish-Dressman
HA
,
Jiang
S
,
Orkunoglu-Suer
FE
, et al
Differences in fat and muscle mass associated with a functional human polymorphism in a post-transcriptional BMP2 gene regulatory element
.
J Cell Biochem
2009
;
107
:
1073
82
.
8.
Balmer
LA
,
Beveridge
DJ
,
Jazayeri
JA
,
Thomson
AM
,
Walker
CE
,
Leedman
PJ
. 
Identification of a novel AU-Rich element in the 3′ untranslated region of epidermal growth factor receptor mRNA that is the target for regulated RNA-binding proteins
.
Mol Cell Biol
2001
;
21
:
2070
84
.
9.
Young
LE
,
Dixon
DA
. 
Posttranscriptional regulation of cyclooxygenase 2 expression in colorectal cancer
.
Curr Colorectal Cancer Rep
2010
;
6
:
60
7
.
10.
Hitti
E
,
Iakovleva
T
,
Brook
M
,
Deppenmeier
S
,
Gruber
AD
,
Radzioch
D
, et al
Mitogen-activated protein kinase-activated protein kinase 2 regulates tumor necrosis factor mRNA stability and translation mainly by altering tristetraprolin expression, stability, and binding to adenine/uridine-rich element
.
Mol Cell Biol
2006
;
26
:
2399
407
.
11.
Tiedje
C
,
Ronkina
N
,
Tehrani
M
,
Dhamija
S
,
Laass
K
,
Holtmann
H
, et al
The p38/MK2-driven exchange between tristetraprolin and HuR regulates AU-rich element-dependent translation
.
PLoS Genet
2012
;
8
:
e1002977
.
12.
Kratochvill
F
,
Machacek
C
,
Vogl
C
,
Ebner
F
,
Sedlyarov
V
,
Gruber
AR
, et al
Tristetraprolin-driven regulatory circuit controls quality and timing of mRNA decay in inflammation
.
Mol Syst Biol
2011
;
7
:
560
.
13.
Sanduja
S
,
Blanco
FF
,
Young
LE
,
Kaza
V
,
Dixon
DA
. 
The role of tristetraprolin in cancer and inflammation
.
Front Biosci
2012
;
17
:
174
88
.
14.
Abdelmohsen
K
,
Gorospe
M
. 
Posttranscriptional regulation of cancer traits by HuR
.
Wiley Interdiscip Rev RNA
2010
;
1
:
214
29
.
15.
Eberhardt
W
,
Doller
A
,
Pfeilschifter
J
. 
Regulation of the mRNA-binding protein HuR by posttranslational modification: spotlight on phosphorylation
.
Curr Protein Pept Sci
2012
;
13
:
380
90
.
16.
Brook
M
,
Tchen
CR
,
Santalucia
T
,
McIlrath
J
,
Arthur
JS
,
Saklatvala
J
, et al
Posttranslational regulation of tristetraprolin subcellular localization and protein stability by p38 mitogen-activated protein kinase and extracellular signal-regulated kinase pathways
.
Mol Cell Biol
2006
;
26
:
2408
18
.
17.
Tchen
CR
,
Brook
M
,
Saklatvala
J
,
Clark
AR
. 
The stability of tristetraprolin mRNA is regulated by mitogen-activated protein kinase p38 and by tristetraprolin itself
.
J Biol Chem
2004
;
279
:
32393
400
.
18.
Al-Ahmadi
W
,
Al-Ghamdi
M
,
Al-Haj
L
,
Al-Saif
M
,
Khabar
KS
. 
Alternative polyadenylation variants of the RNA binding protein, HuR: abundance, role of AU-rich elements and auto-Regulation
.
Nucleic Acids Res
2009
;
37
:
3612
24
.
19.
Dai
W
,
Zhang
G
,
Makeyev
EV
. 
RNA-binding protein HuR autoregulates its expression by promoting alternative polyadenylation site usage
.
Nucleic Acids Res
2012
;
40
:
787
800
.
20.
Mansfield
KD
,
Keene
JD
. 
Neuron-specific ELAV/Hu proteins suppress HuR mRNA during neuronal differentiation by alternative polyadenylation
.
Nucleic Acids Res
2012
;
40
:
2734
46
.
21.
Al-Ahmadi
W
,
Al-Ghamdi
M
,
Al-Souhibani
N
,
Khabar
KS
. 
miR-29a inhibition normalizes HuR over-expression and aberrant AU-rich mRNA stability in invasive cancer
.
J Pathol
2013
;
230
:
28
38
.
22.
Gebeshuber
CA
,
Zatloukal
K
,
Martinez
J
. 
miR-29a suppresses tristetraprolin, which is a regulator of epithelial polarity and metastasis
.
EMBO Rep
2009
;
10
:
400
5
.
23.
Curtis
C
,
Shah
SP
,
Chin
SF
,
Turashvili
G
,
Rueda
OM
,
Dunning
MJ
, et al
The genomic and transcriptomic architecture of 2,000 breast tumours reveals novel subgroups
.
Nature
2012
;
486
:
346
52
.
24.
Al-Ahmadi
W
,
Al-Haj
L
,
Al-Mohanna
FA
,
Silverman
RH
,
Khabar
KS
. 
RNase L downmodulation of the RNA-binding protein, HuR, and cellular growth
.
Oncogene
2009
;
28
:
1782
91
.
25.
Hitti
E
,
Al-Yahya
S
,
Al-Saif
M
,
Mohideen
P
,
Mahmoud
L
,
Polyak
SJ
, et al
A versatile ribosomal protein promoter-based reporter system for selective assessment of RNA stability and post-transcriptional control
.
RNA
2010
;
16
:
1245
55
.
26.
Al-Saif
M
,
Khabar
KS
. 
UU/UA dinucleotide frequency reduction in coding regions results in increased mRNA stability and protein expression
.
Mol Ther
2012
;
20
:
954
9
.
27.
al-Haj
L
,
Al-Ahmadi
W
,
Al-Saif
M
,
Demirkaya
O
,
Khabar
KS
. 
Cloning-free regulated monitoring of reporter and gene expression
.
BMC Mol Biol
2009
;
10
:
20
.
28.
Griseri
P
,
Bourcier
C
,
Hieblot
C
,
Essafi-Benkhadir
K
,
Chamorey
E
,
Touriol
C
, et al
A synonymous polymorphism of the Tristetraprolin (TTP) gene, an AU-rich mRNA-binding protein, affects translation efficiency and response to Herceptin treatment in breast cancer patients
.
Hum Mol Genet
2011
;
20
:
4556
68
.
29.
Gyorffy
B
,
Lanczky
A
,
Eklund
AC
,
Denkert
C
,
Budczies
J
,
Li
Q
, et al
An online survival analysis tool to rapidly assess the effect of 22,277 genes on breast cancer prognosis using microarray data of 1,809 patients
.
Breast Cancer Res Treat
2010
;
123
:
725
31
.
30.
Mihaly
Z
,
Kormos
M
,
Lanczky
A
,
Dank
M
,
Budczies
J
,
Szasz
MA
, et al
A meta-analysis of gene expression-based biomarkers predicting outcome after tamoxifen treatment in breast cancer
.
Breast Cancer Res Treat
2013
;
140
:
219
32
.
31.
Nigg
EA
. 
Mitotic kinases as regulators of cell division and its checkpoints
.
Nat Rev Mol Cell Biol
2001
;
2
:
21
32
.
32.
Lee
HH
,
Lee
SR
,
Leem
SH
. 
Tristetraprolin regulates prostate cancer cell growth through suppression of E2F1
.
J Microbiol Biotechnol
2014
;
24
:
287
94
.
33.
Marderosian
M
,
Sharma
A
,
Funk
AP
,
Vartanian
R
,
Masri
J
,
Jo
OD
, et al
Tristetraprolin regulates Cyclin D1 and c-Myc mRNA stability in response to rapamycin in an Akt-dependent manner via p38 MAPK signaling
.
Oncogene
2006
;
25
:
6277
90
.
34.
Milke
L
,
Schulz
K
,
Weigert
A
,
Sha
W
,
Schmid
T
,
Brüne
B
. 
Depletion of tristetraprolin in breast cancer cells increases interleukin-16 expression and promotes tumor infiltration with monocytes/macrophages
.
Carcinogenesis
2013
;
34
:
850
7
.
35.
Kim
CW
,
Vo
MT
,
Kim
HK
,
Lee
HH
,
Yoon
NA
,
Lee
BJ
, et al
Ectopic over-expression of tristetraprolin in human cancer cells promotes biogenesis of let-7 by down-regulation of Lin28
.
Nucleic Acids Res
2012
;
40
:
3856
69
.
36.
Nabors
LB
,
Gillespie
GY
,
Harkins
L
,
King
PH
. 
HuR, a RNA stability factor, is expressed in malignant brain tumors and binds to adenine- and uridine-rich elements within the 3′ untranslated regions of cytokine and angiogenic factor mRNAs
.
Cancer Res
2001
;
61
:
2154
61
.
37.
Niesporek
S
,
Kristiansen
G
,
Thoma
A
,
Weichert
W
,
Noske
A
,
Buckendahl
AC
, et al
Expression of the ELAV-like protein HuR in human prostate carcinoma is an indicator of disease relapse and linked to COX-2 expression
.
Int J Oncol
2008
;
32
:
341
7
.
38.
Wang
J
,
Zhao
W
,
Guo
Y
,
Zhang
B
,
Xie
Q
,
Xiang
D
, et al
The expression of RNA-binding protein HuR in non-small cell lung cancer correlates with vascular endothelial growth factor-C expression and lymph node metastasis
.
Oncology
2009
;
76
:
420
9
.
39.
Yi
X
,
Zhou
Y
,
Zheng
W
,
Chambers
SK
. 
HuR expression in the nucleus correlates with high histological grade and poor disease-free survival in ovarian cancer
.
Aust N Z J Obstet Gynaecol
2009
;
49
:
93
8
.
40.
Yuan
Z
,
Sanders
AJ
,
Ye
L
,
Jiang
WG
. 
HuR, a key post-transcriptional regulator, and its implication in progression of breast cancer
.
Histol Histopathol
2010
;
25
:
1331
40
.
41.
Mazan-Mamczarz
K
,
Hagner
PR
,
Corl
S
,
Srikantan
S
,
Wood
WH
,
Becker
KG
, et al
Post-transcriptional gene regulation by HuR promotes a more tumorigenic phenotype
.
Oncogene
2008
;
27
:
6151
63
.
42.
Mukherjee
N
,
Corcoran
DL
,
Nusbaum
JD
,
Reid
DW
,
Georgiev
S
,
Hafner
M
, et al
Integrative regulatory mapping indicates that the RNA-binding protein HuR couples pre-mRNA processing and mRNA stability
.
Mol Cell
2011
;
43
:
327
39
.
43.
Zhu
H
,
Zhou
H-L
,
Hasman
RA
,
Lou
H
. 
Hu proteins regulate polyadenylation by blocking sites containing U-rich sequences
.
J Biol Chem
2007
;
282
:
2203
10
.
44.
Tiedje
C
,
Holtmann
H
,
Gaestel
M
. 
The role of mammalian MAPK signaling in regulation of cytokine mRNA stability and translation
.
J Interferon Cytokine Res
2014
;
34
:
220
32
.
45.
Dai
W
,
Li
W
,
Hoque
M
,
Li
Z
,
Tian
B
,
Makeyev
EV
. 
A post-transcriptional mechanism pacing expression of neural genes with precursor cell differentiation status
.
Nat Commun
2015
;
6
:
7576
.
46.
Suswam
E
,
Li
Y
,
Zhang
X
,
Gillespie
GY
,
Li
X
,
Shacka
JJ
, et al
Tristetraprolin down-regulates interleukin-8 and vascular endothelial growth factor in malignant glioma cells
.
Cancer Res
2008
;
68
:
674
82
.
47.
Griseri
P
,
Pages
G
. 
Control of pro-angiogenic cytokine mRNA half-life in cancer: the role of AU-rich elements and associated proteins
.
J Interferon Cytokine Res
2014
;
34
:
242
54
.
48.
Van Tubergen
EA
,
Banerjee
R
,
Liu
M
,
Vander Broek
R
,
Light
E
,
Kuo
S
, et al
Inactivation or loss of TTP promotes invasion in head and neck cancer via transcript stabilization and secretion of MMP9, MMP2, and IL-6
.
Clin Cancer Res
2013
;
19
:
1169
79
.
49.
Kakuguchi
W
,
Kitamura
T
,
Kuroshima
T
,
Ishikawa
M
,
Kitagawa
Y
,
Totsuka
Y
, et al
HuR knockdown changes the oncogenic potential of oral cancer cells
.
Mol Cancer Res
2010
;
8
:
520
8
.