The RNA binding protein HuR differentially regulates unique subsets of mRNAs in estrogen receptor negative and estrogen receptor positive breast cancer

Abstract

Background: The discordance between steady-state levels of mRNAs and protein has been attributed to posttranscriptional control mechanisms affecting mRNA stability and translation. Traditional methods of genome wide microarray analysis, profiling steady-state levels of mRNA, may miss important mRNA targets owing to significant posttranscriptional gene regulation by RNA binding proteins (RBPs).

Methods: The ribonomic approach, utilizing RNA immunoprecipitation hybridized to microarray (RIP-Chip), provides global identification of putative endogenous mRNA targets of different RBPs. HuR is an RBP that binds to the AU-rich elements (ARE) of labile mRNAs, such as proto-oncogenes, facilitating their translation into protein. HuR has been shown to play a role in cancer progression and elevated levels of cytoplasmic HuR directly correlate with increased invasiveness and poor prognosis for many cancers, including those of the breast. HuR has been described to control genes in several of the acquired capabilities of cancer and has been hypothesized to be a tumor-maintenance gene, allowing for cancers to proliferate once they are established.

Results: We used HuR RIP-Chip as a comprehensive and systematic method to survey breast cancer target genes in both MCF-7 (estrogen receptor positive, ER+) and MDA-MB-231 (estrogen receptor negative, ER-) breast cancer cell lines. We identified unique subsets of HuR-associated mRNAs found individually or in both cell types. Two novel HuR targets, CD9 and CALM2 mRNAs, were identified and validated by quantitative RT-PCR and biotin pull-down analysis.

Conclusion: This is the first report of a side-by-side genome-wide comparison of HuR-associated targets in wild type ER+ and ER- breast cancer. We found distinct, differentially expressed subsets of cancer related genes in ER+ and ER- breast cancer cell lines, and noted that the differential regulation of two cancer-related genes by HuR was contingent upon the cellular environment.

Figure 1

Immunoprecipitation and RIP in MB-231 and MCF-7 breast cancer cells. Immunoprecipitations were performed from MB-231 or MCF-7 cell lysates using anti-HuR monoclonal antibody (3A2) and IgG1 isotype control. A. IP Western of HuR revealed expected size band as detected by 3A2. Panel on right reveals amounts of HuR in lysates used from both cell lines. B. Verification by quantitative RT-PCR showed fifteen and eleven fold enrichments of B-ACTIN, a known HuR target, in the 3A2 IPs from MB231 and MCF-7, respectively. All ΔΔC_T values were normalized to GAPDH. Experiments were done in duplicate (n = 2).

Figure 2

HuR RIP-CHIP identifies distinct genetic profiles in ER+ and ER- breast cancer cells. HuR immunoprecipitations were performed from MB-231 or MCF-7 cell lysates using HuR antibody and IgG1 isotype control hybridized to Illumina Sentrix arrays (47,000 genes). Control signals were subtracted. Results represent cumulative data from 12 different arrays. Experiments were done in triplicate (n = 3) for each cell line with matching controls. Scales are log₂.

Figure 3

GO Classification of genes found by RIP CHIP of potential HuR targets and their relationship to the Acquired Capabilities of Cancer Model. A. Differentially expressed genes which are more represented in the Biological Processes (BP) GO category than expected. B. Original representation showing subsets of transcripts found to be targets of association with HuR (normal type). New transcripts found in this study with RIP-Chip (bold type). Enhanced expression upon binding to HuR influences several of the acquired capabilities of cancer cells described by Hanahan and Weinberg [23,24].

Figure 4

Validation of target CALM2 and CD9 mRNAs by quantitative RT-PCR. Quantitative RT-PCR using cell lysates, HuR antibody, and IgG1 from RIP-CHIP analysis confirmed results identifying CALM2 mRNA (A) and CD9 mRNA (B) as HuR targets. Change in gene expression is represented as fold increase in HuR immunoprecipitation as compared to IgG1. GAPDH mRNA was used as an endogenous control. Error bars represent SEM. p value is < 0.005. Experiments were done in triplicate (n = 3).

Figure 5

Biotin Pull-down of CD9 and CALM2. A. Scheme of Coding region (CR) and 3'UTR fragments for biotin pull-down assay. The sequences were obtained from Entrez data base. CR and 3'UTR fragments selected for amplification by PCR are as noted. B. 1% agarose gel electrophoresis showing PCR amplified products of the coding regions and 3'UTR's for CD9 (442 bp and 432 bp, respectively) and CALM2 (443 bp and 610 bp, respectively). C. Biotin pull-down assay using lysates prepared from MB-231 cells. The binding of HuR to biotinylated 3'UTR transcripts from both CD9 and CALM2 mRNAs was specific. HuR did not bind a biotinylated control (GAPDH 3'UTR); and did not bind to biotinylated transcripts spanning the CR of CD9 or CALM2. Experiments were done in duplicate (n = 2).

Figure 6

HuR differentially regulates CD9 and CALM2 in MB-231. A. Epitope HA tagged HuR is over-expressed by 142% and 138% respectively, in stably transfected clones 4E1 and 5F1, as compared to empty vector (EV) control clone 2C7. B. HuR knock-down using lentiviral short hairpin (sh) RNA H760 results in a 94% reduction in steady state levels of protein in clone A7 (LL = lentilox control). C. HuR over-expression results in a 40% reduction in CD9 protein levels as assayed by Western analysis; however, HuR knock-down using lentiviral shRNA results in an increase from 100% to 228% of CD9 levels. D. Over-expression of HuR decreases CD9 mRNA levels but not CALM2 expression. Analysis of steady state CD9 and CALM2 mRNA levels by quantitative RT-PCR reveals significant decreases in CD9 mRNA levels, whereas CALM2 levels are unaffected. Although CALM2 expression appears greater, the change is not significant. E. Knocking down HuR levels by shRNA in MB-231 cells shows significant increases in CD9 and CALM2 mRNA levels by quantitative RT-PCR. Decreased levels of HuR mRNA validate HuR shRNA knock-down. F. Graph showing the effects of HuR on the expression of CD9 mRNA. HuR over-expression results in decreases in both mRNA and protein levels, though the decreases are greater in RNA. Whereas, HuR knock-down by shRNA results in significant increases at both the mRNA and protein levels, with greater change at transcript levels. The dashed line represents levels in control cells. Error bars represent SEM. p value is < 0.005; N.S. = not statistically significant; * = statistically significant. All experiments were done in triplicate (n = 3).

Figure 7

Effects of over-expressing or reducing HuR on CD9 and CALM2 expression in MCF-7 cells. A. Western analysis of HuR over-expression in heterogenous population of cells reveals approximately 10% over-expression. B. Lentiviral HuR shRNA efficiently knocks down HuR protein by over 90%. C. HuR over- expression and under-expression results in small changes in CD9 protein levels in MCF-7 cells. D. Levels of both CD9 and CALM2 mRNAs are unchanged in cells which over-express HuR; whereas lentiviral knock-down of HuR in MCF-7 cells results in decreases in steady-state mRNA levels (E). The graph in (F) shows minimal changes in CD9 mRNA and protein levels in HuR over-expressing MCF-7 cells. The CD9 mRNA levels, however, are more affected in HuR knock-down. P value is < 0.005; N.S. = not statistically significant; * = statistically significant.