miR-519 suppresses tumor growth by reducing HuR levels

Abstract

The RNA-binding protein HuR is highly abundant in many cancers. HuR expression was recently found to be repressed by microRNA miR-519, which potently lowered HuR translation without influencing HuR mRNA abundance. Here, we examined the levels of HuR and miR-519 in pairs of cancer and adjacent healthy tissues from ovary, lung, and kidney. In the three sample collections, the cancer specimens showed dramatically higher HuR levels, unchanged HuR mRNA concentrations, and markedly reduced miR-519 levels, when compared with healthy tissues. As tested using human cervical carcinoma cells, miR-519 reduced tumorigenesis in athymic mice. Compared with the tumors arising from control cells, cells overexpressing miR-519 formed significantly smaller tumors, while cells expressing reduced miR-519 levels gave rise to substantially larger tumors. Evidence that the miR-519-elicited reduction of HuR was critical for its tumor suppressor influence was obtained by reducing HuR, as HuR-silenced cells formed markedly smaller tumors and were unable to form large tumors even after lowering miR-519 abundance. Together, our data reveal that miR-519 inhibits tumorigenesis in large part by repressing HuR expression.

Figure 1. Inverse relationship between HuR and miR-519 levels in ovarian tissues

(A) Western blot analysis of HuR expression in ovarian tumor (T) and patient-matched normal ovary (N) from four patients (P1-P4). Sample loading was monitored by Ponceau S staining. (B) HuR mRNA abundance was measured by RT-qPCR analysis using HuR-specific primer pairs and normalized to 18S rRNA abundance (also measured by RT-qPCR) in each sample. Data are indicated as percent HuR mRNA in each T sample relative to N sample for each patient. (C) The abundance of miR-519a, miR-519b-3p, and miR-519c-3p in each sample pair was measured by RT-qPCR and normalized to U1 RNA levels. Data are indicated as percent miR-519 abundance in each T sample relative to each N sample for the same patient.

Figure 2. Inverse relationship between HuR and miR-519 expression in kidney tissues

(A) Western blot analysis of HuR expression in kidney tumor (T) and patient-matched normal kidney (N) from four patients (P1-P4). Loading was monitored by Ponceau S staining. (B) HuR mRNA abundance was measured by RT-qPCR analysis as explained in Fig. 1. (C) The abundance of miR-519a and miR-519c-3p (miR-519b-3p was undetectable) in each sample pair was measured as explained in Fig. 1. In panels (B) and (C), data indicate percent RNA abundance in each T sample relative to each N sample for the same patient.

Figure 3. Inverse relationship between HuR and miR-519 expression in lung tissues

(A) Western blot analysis of HuR expression in lung tumor (T) and patient-matched normal lung (N) from three patients (P1-P3). Sample loading was monitored by Ponceau S staining. (B) HuR mRNA abundance was measured by RT-qPCR analysis as explained in Fig. 1. (C) The abundance of miR-519a, miR-519b-3p, and miR-519c-3p in each sample pair was measured as explained in Fig. 1. Data in panels (B) and (C) indicate percent RNA abundance in each T sample relative to each N sample for the same patient.

Figure 4. Tumorigenesis by cells expressing different miR-519 and HuR levels

(A) Forty-eight h after transfecting HeLa cells with the small RNAs indicated, HuR levels were measured by Western blot analysis (top, including loading control β-Actin) and miR-519c-3p levels by RT-qPCR analysis (bottom, normalized using U1 RNA). Western blot analysis is representative of 3 experiments, graph shows the means ±S.E.M. from 3 experiments. (B) HeLa cells transfected as in panel (a) were injected (2 million cells per flank) in athymic mice. Three weeks later, tumors were excised. The individual weights (dots), averages (horizontal lines) and significance (Mann-Whitney U test) from 13 mice per group (two independent injections) are shown.