Compared with early-passage, ‘young' proliferating [Y, at population doubling (pdl) 22] WI-38 cells, the senescent (S, pdl 52) WI-38 cells displayed a flattened morphology and senescence-associated (SA) β-galactosidase (SA-β-gal) activity, a widely used senescence marker [²⁶,²⁷] (Figure 1A). Western blot analysis also revealed that senescent cells expressed lower levels of SIRT1 and HuR, whereas p16 and p53 were upregulated (Figure 1B), in keeping with reported literature [^28-30].

To test how the pattern of expressed microRNAs is affected by replicative senescence, we studied transcriptome-wide changes in microRNAs using miRNome arrays (not shown); we then validated individual microRNAs by reverse transcription (RT) followed by real-time, quantitative (q)PCR amplification (see Materials and Methods). Depicted in Figures 2 and 3 and in Supplementary Table 1 are all of the microRNAs validated using sequence-specific qPCR primers. As shown in Figure 2, several microRNAs were markedly more abundant in senescent cells (e.g., miR-1204, miR-663, miR-548b-3p and miR-431). Other microRNAs were expressed at lower levels in senescent cells [e.g., miR-24, miR-141, and miR-10a (Figure 3, Supplementary Table 1)]. MicroRNAs changing less than twofold with senescence are listed in the Supplementary Table 1.

We were particularly interested in the miR-519 family. miR-519 was recently found to inhibit translation of the RNA-binding protein HuR through its interaction with the HuR coding region [³¹]. In a separate study, miR-519 suppressed the growth of tumor xenografts in an HuR-dependent manner [³²]. Given that HuR promotes cell proliferation and decreases senescence [³³,³⁴], we hypothesized that the elevated miR-519 in senescent cells (Figure 4A) might lower HuR expression in WI-38 HDFs, and hence promote senescence. To test this possibility, we overexpressed miR-519a in young-HDFs (Figure 4B); western blot analysis confirmed that miR-519a overexpression repressed HuR (Figure 4C). In keeping with earlier results [³¹], miR-519a did not influence the levels of HuR mRNA (Figure 4D), in agreement with the view that miR-519a inhibited HuR mRNA translation without affecting HuR mRNA stability. Moreover, sustained miR-519a overexpression for 4 weeks caused a marked reduction in cell number as compared to control transfection groups (Figure 4E). miR-519a-overexpressing cells also showed increased SA-β-gal activity (Figure 4F) and elevated expression of the senescence marker p27 [³⁵,³⁶] (Figure 4G, bottom). Together, these data indicate that miR-519a induced cellular senescence and inhibited cell proliferation, resulting in accelerated senescence. They further suggest that miR-519a-induced senescence may be mediated in part by repression of HuR.

As indicated above, miR-519 was found to suppress tumor growth [³²]. Since cellular senescence is considered to be an anti-tumorigenic process, we examined the effect of miR-519 on the senescent phenotype of cancer cells. Upon miR-519a overexpression (Figure 5A), HeLa cell numbers declined significantly (Figure 5B). Five days after transfection of miR-519a, cells showed a strong increase in SA-β-gal activity compared to the control transfection group (Figure 5C); in addition, miR-519-induced senescence in HeLa cells was accompanied by increased levels of the senescence markers p53 and p27 (Figure 5D). Together, these data indicate that miR-519 reduced HeLa cell proliferation and promoted HeLa cell senescence. Accordingly, we postulate that one of the mechanisms by which miR-519 suppress tumor growth is by inducing senescence, and further propose that miR-519 triggers senescence -at least in part- by reducing HuR levels.

We previously used miRNA microarrays to identify changes in a limited number of microRNAs in senescent cells [¹⁶]. Here, we have expanded this analysis and have verified many individual microRNAs whose abundance changes with replicative senescence. Many of them target key proteins implicated in senescence and cancer. For example, miR-146b is upregulated in senescent cells (Figure 2), in keeping with earlier findings that miR-146a and miR-146b increased with senescence and repressed the senescence-associated inflammatory mediators IL-6 and IL-8 [¹⁵]. miR-34a regulates SIRT1 expression and induced senescence of cancer cells [^39-41]; here, we observed higher miR-34c (not miR-34a) in senescent cells, likely a reflection of the variability and complexity of the senescence process. Several let-7 members were also upregulated in senescent cells (Figure 2 and Supplementary Table 1); this observation supports the view that the ability of let-7 microRNAs can suppress tumor growth [reviewed in 42], which could contribute to the senescence process. Similarly, upregulation of miR-20 in senescence cells correlates with the ability of miR-20 to inhibit proliferation of K562 human erythromyeloblastoid leukemia cells [⁴³]. In conjunction with the finding that miR-519 reduced tumorigenesis in a xenograft model [³²], we propose that the coordinated action of senescence-upregulated microRNAs can suppress tumor growth by reducing the levels of oncogenes or tumor promoters.

Conversely, many microRNAs were downregulated in senescent cells (Figure 3). Among the myriad of senescence-associated proteins that they might regulate, these microRNAs likely repress several tumor suppressors. In this regard, as miR-21 has been shown to lower expression of the tumor suppressor PTEN [⁴⁴], the downregulated of miR-21 in senescent cells (Figure 3) could allow increased PTEN expression, in turn reducing tumor cell proliferation, migration, and invasion [⁴⁵].

We previously reported that miR-519 represses the production of HuR, an RNA-binding protein which is highly abundant in cancer cells and is low in untransformed cells [¹¹,³⁸]. HuR overexpression delays the senescent phenotype while the loss of HuR enhances it [¹¹]. Moreover, while HuR levels are high in tumors and low in normal tissues, miR-519 levels are high in normal tissues and low in cancer tissues [³²]. Since HuR potently enhances the expression of cancer-promoting proteins, and reducing HuR levels promotes HDF senescence [¹¹,³⁸], we propose that miR-519 represses tumor growth at least in part, by lowering HuR and thereby promoting senescence (Figs. 4 and 5). Additionally, miR-519 could further repress tumor growth by lowering the expression of other genes, such as ABCG2 or HIF-1α [⁴⁶,⁴⁷].

In summary, we have identified collections of microRNAs displaying altered abundance with replicative senescence. As shown here for miR-519, we postulate that these changes help to meet the needs of senescent cells in eliciting tumor suppression and growth arrest. Future studies will help to recognize more fully the proteins and processes modulated by senescence-regulated microRNAs.