MicroRNA-22 Suppresses DNA Repair and Promotes Genomic Instability through Targeting of MDC1

Repeated exposure to both exogenous and endogenous insults challenges the integrity of cellular genomic material. Eukaryotes have evolved a system called the DNA damage response (DDR), which allows cells to sense DNA damage and orchestrate the appropriate cell-cycle checkpoints and DNA repair mechanisms (1). The failure to respond to DNA damage is a characteristic associated with genomic instability and with the onset of diseases, including neurodegenerative diseases, immune deficiency, cancer, and premature aging (2).

DNA double-strand breaks (DSB) activate the DDR by triggering the kinase activity of ataxia telangiectasia mutated (ATM), thereby initiating a signaling cascade in which the histone variant H2AX (γ-H2AX), located at DSB sites, becomes phosphorylated, and other DDR factors, including the adaptor protein mediator of DNA damage checkpoint 1 (MDC1), are recruited. MDC1 amplifies the ATM signaling activity, leading to a higher percentage of phosphorylated H2AX proteins and contributing to the recruitment and retention of additional DDR factors at the sites of DNA damage (3). Thus, MDC1 has been termed a master regulator, modulating the specific chromatin microenvironment required to maintain genomic stability. MDC1 knockout mice show chromosomal instability, defective DNA repair, and radiation sensitivity (4). Furthermore, loss of MDC1 is associated with an increased occurrence of tumors in mice (5), and reduction or lack of MDC1 has been observed in breast and lung carcinoma cells in humans (6). Therefore, cellular levels of MDC1 appear to impact genomic instability and tumorigenicity directly. Although posttranslational modification via small ubiquitin-like modifiers affects the stability of MDC1 and its function in DDR (7–9), little is known about how the expression of MDC1 is regulated and which pathophysiologic conditions are associated with this regulation.

miRNAs are small noncoding RNAs that suppress protein synthesis, usually by interacting with the 3′-untranslated region (3′-UTR) of target mRNAs (10). Several lines of evidence suggest that miRNAs negatively regulate the expression of DDR proteins (11–14). Therefore, miRNAs may play an important role in the regulation of DDR and may contribute to the maintenance of genomic integrity.

To investigate the possibility that some specific miRNA might directly regulate MDC1 expression and its function, we screened for miRNAs that could potentially regulate MDC1 expression and identified miR-22 as an miRNA that could specifically suppress MDC1 expression. We show that miR-22–mediated downregulation of MDC1 induces impaired DDR activation and genomic instability. Further, we demonstrated that this new pathway plays a crucial role in the regulation of DNA repair in sustained activation of Akt1 and senescence and may represent a new therapeutic target.

Polyclonal MDC1 antibody (R2) was raised in rabbit against a glutathione S-transferase fusion protein containing the BRCT domain of MDC1 (residues 1882-2089). Anti-MDC1 polyclonal antibody (Ab11170; Abcam) was used for immunohistochemistry. The following antibodies were used in this study: anti-53BP1 (BD), anti-BRCA1 (Santa Cruz Biotechnology), anti–γ-H2AX (Upstate), anti-Akt (Cell Signaling Technology), anti-phospho Akt1 (Ser473) (Cell Signaling Technology), and anti–α-Tubulin (Santa Cruz Biotechnology).

U2OS-DR-GFP cells were transfected with control or CA-Akt1 vector using lipofectamine 2000, and sequentially transfected with miR-22 inhibitor, and then infected with I-SceI–carrying adenovirus at an estimated multiplicity of infection of 10. After 72 hours, GFP-positive cells were measured by FACS (FACSCalibur; BD Biosciences). The acquired data were analyzed using CellQuest Pro software (BD Biosciences).

After treatment with irradiation, 5 × 10² cells were immediately seeded on 60-mm dish in triplicate and grown for 2 to 3 weeks at 37°C to allow colonies to form. Colonies were stained with 2% methylene blue/50% ethanol and were counted. The fraction of surviving cells was calculated as the ratio for the plating efficiency of treated cells over untreated cells. Cell survival results are reported as the mean value ± SD for three independent experiments.

Additional materials and methods can be found in Supplementary Information.

To search for miRNAs that regulate MDC1 expression, we carried out a comprehensive bioinformatics analysis to generate a selective miRNA library that could then be used for screening. From this analysis, a total of 8 miRNAs were identified as candidates (Supplementary Table S1), and each was reversely screened for the effect on MDC1 expression by using a luciferase assay. The results of the luciferase assays revealed that overexpression of miR-22 led to remarkably lower luciferase activity compared with scrambled control miRNA (Fig. 1A). Consistent with these results, when overexpressed, only miR-22 significantly reduced endogenous MDC1 protein expression in HEK293T cells (Fig. 1B). The MDC1 expression level decreased as the concentrate of transfected miR-22 or its premature hairpin (pre–miR-22) was increased (Supplementary Fig. S1). miR-22–mediated repression of MDC1 was not restricted to HEK293T cells, as we also observed specific repression in HeLa and U2OS cells (Fig. 1C). Using real-time quantitative PCR (qPCR), we found that MDC1 mRNA was reduced more than 50% when miR-22 was overexpressed (Fig. 1D), indicating that miR-22 posttranscriptionally downregulates MDC1, probably by promoting both mRNA decay and inhibiting translation.

We next analyzed putative miR-22 target site using TargetScan algorithm (MIT; release 6.2) and found that bases 317 to 339 in the MDC1 3′-UTR are complementary to the target sites of miR-22 (Fig. 1E). To determine whether miR-22 binds to this site to repress MDC1 expression, we constructed a mutant MDC1 3′-UTR luciferase reporter that was lacking predicted seed region for the miR-22. As shown in Fig. 1F, deletion of the miR-22 binding site in the seed region of the MDC1 3′-UTR abrogated the suppressive ability of miR-22. These results together demonstrate that miR-22 represses MDC1 expression by directly targeting its 3′-UTR.

To determine whether the miR-22 could affect the recruitment of MDC1 and its downstream targets to sites of DNA damage, we introduced miR-22 into U2OS cells and subjected them to ionizing radiation (IR), and examined MDC1, 53BP1, and BRCA2 foci formation by immunofluorescence. We observed that miR-22 greatly reduced the percentage of cells expressing IR-induced MDC1 foci (Supplementary Fig. S2A). In addition, formation of both 53BP1 and BRCA1 foci was also decreased in miR-22–overexpressing U2OS cells as compared with control cells (Supplementary Fig. S2B and S2C). Because miR-22 did not affect 53BP1 and BRCA1 protein levels (Fig. 1C), these results suggest that miR-22 inhibits IR-induced MDC1 foci formation, which in turn suppresses the recruitment of 53BP1 and BRCA1 to sites of DNA damage.

MDC1 plays an important role in DSB repair (15,16) and checkpoint activation (17–19). Therefore, we examined the effect of miR-22 on intra-S phase cell-cycle checkpoint. We found that the radiation-resistant DNA synthesis was induced in U2OS cells by overexpressing miR-22 (Fig. 2A, empty squares). Further, overexpression of miR-22 in U2OS cells had significantly more residual DSBs than control cells, as evidenced by the increase in signal intensity of γ-H2AX staining and by the increase in comet tail moments (Fig. 2B and C, middle columns; Supplementary Fig. S3A and S3B). We then analyzed metaphase spreads of control and miR-22–expressing U2OS cells after IR exposure. The results showed that overexpression of miR-22 into U2OS cells could significantly increase the number of chromosome breaks as compared with control cells (Fig. 2D, middle column).

To determine whether the effect of miR-22 on radiation-resistant DNA synthesis and DSB repair was mediated via its effect on MDC1, we cotransfected U2OS cells with miR-22 and an miR-22–insensitive MDC1 expression plasmid. Our results showed that the nontargetable MDC1 fully rescued the radiation-resistant DNA synthesis (Fig. 2A, black squares), DSB repair defect (Fig. 2B and C, third columns; Supplementary Fig. S3), and chromosome breaks (Fig. 2D, third column). This evidence shows that miR-22 inhibits DDR through the downregulation of MDC1.

We then determined whether the suppressed DSB repair by miR-22 leads to increased cellular sensitivity to IR. We found that the survival fractions of colonies in miR-22–expressing cells following IR were significantly reduced relative to those of control cells (Fig. 2E, empty squares). Of notice, overexpressing miR-22–insensitive MDC1 significantly reduced the extent of miR-22–mediated sensitization to IR in miR-22–expressing cells (Fig. 2E, black squares), further supporting working model that miR-22 modulates the DSB repair through MDC1.

To assess the subsequent genomic effects resulting from the numerous chromosomal breaks in miR-22–expressing cells, we performed array comparative genomic hybridization (array CGH) using human fibroblast GM00637 cells. From this analysis, we conclude that there was a high frequency of chromosomal abnormalities in miR-22–expressing cells, including clonal amplifications and deletions in discrete regions (Fig. 2F; Supplementary Fig. S4). Taken all together, these results provide evidence that miR-22–mediated downregulation of MDC1 results in defects in DSB repair and allows cells to bypass an intra–S-phase checkpoint causing a decrease in chromosome integrity.

Endogenous miR-22 expression appears to be dependent on particular abnormal pathologic contexts, including high Akt1 activity (20) and replicative-induced senescence (21). Akt1 is frequently activated in many tumor types (22), and its overstimulation leads to a strong repression of homologous recombination (HR), causing genomic instability (23–25). Thus, miR-22–mediated downregulation of MDC1 is hypothesized to abrogate HR in Akt1-activated cancer cells. To determine whether activated Akt1 negatively regulates MDC1 through the induction of miR-22, U2OS and HCT116 cells were transfected with constitutively active Akt1 (CA-Akt1). As expected, overexpression of CA-Akt1 in either U2OS or HCT116 cells led to increased miR-22 expression (Fig. 3A, top panel). Intriguingly, in comparison with control cells, CA-Akt1–transfected cells exhibited diminished levels of MDC1 protein (Fig. 3A, bottom panel) and mRNA (Fig. 3B). To demonstrate whether activated Akt1 controls MDC1 expression through miR-22, we cotransfected U2OS and HCT116 cells with the CA-Akt1 expression vector and a miR-22 antisense oligonucleotide (anti–miR-22). We found that inhibition of endogenous miR-22 led to an increase in MDC1 protein and mRNA in CA-Akt1–overexpressing cells (Fig. 3C and D). Moreover, CA-Akt1 significantly inhibited MDC1 3′-UTR luciferase activity relative to control (Fig. 3E). Under these experimental conditions, CA-Akt1–induced decrease of luciferase activity was significantly attenuated when anti–miR-22 was introduced, suggesting that CA-Akt1 negatively regulates MDC1 via miR-22.

It has been reported that miR-22 is frequently overexpressed in prostate tumor tissues, and its expression is positively correlated with the level of phosphorylated Akt1 (pAkt1; ref. 26). Using this system, we asked whether Akt1 activity directly affects MDC1 levels in vivo. To this end, we examined levels of the pAkt1 and MDC1 in prostate tumor tissue array by immunohistochemistry. We observed an increase in the level of Ser473 phosphorylation of Akt1 in the prostate tumor specimens, which was the region with the low MDC1 expression (Fig. 3F, left panel). On the other hand, specimens with weak pAkt signals had strong MDC1 staining (Fig. 3F, right panel). The Pearson correlation analysis demonstrated a statistically significant inverse relationship between pAkt1 and MDC1 levels (Fig. 3G). To further investigate the role of miR-22 on pAkt-mediated suppression of MDC1 levels in prostate cancer, we measured the miR-22, p-Akt, and MDC1 levels of fresh human prostate cancer tissues. We observed that miR-22 levels were significantly higher in prostate cancer samples with high pAkt and low MDC1 expression than in those with low pAkt and high MDC1 expression (Supplementary Fig. S5A and S5B). These results further strengthen the notion that activated Akt1 downregulates MDC1 protein through upregulation of miR-22.

We next examined the effect of activated Akt1 on the DDR function of MDC1 by monitoring the recruitment of MDC1 to sites of DNA damage after IR. As shown in Fig. 4A, CA-Akt1 greatly reduced the percentage of cells with IR-induced MDC1 foci. However, transfection of the anti–miR-22 could completely rescue the inhibitory effect of CA-Akt1 on MDC1 foci formation (Fig. 4B), suggesting that inhibition of MDC1 foci by CA-Akt1 is due to upregulation of miR-22.

Because it has been demonstrated that MDC1 plays an important role in the HR (15), we tested whether activated Akt1 inhibits HR via suppressing MDC1 expression. To this end, we assayed for HR-mediated repair of an I-SceI–induced DSBs, in U2OS cells, using a recombination substrate DR-GFP (27). When DSBs are repaired by HR, GFP is expressed and levels can be quantitated using flow cytometry (Fig. 4C). Consistent with the role of activated Akt1 in HR (23–25), cells overexpressing CA-Akt1 had significantly reduced HR efficiency (Fig. 4D, middle bar). Intriguingly, anti–miR-22 treatment enhanced the recovery of HR activity in CA-Akt1–overexpressing U2OS DR-GFP cells (Fig. 4D, third bar). Moreover, the efficiency of HR significantly improved with miR-22–insensitive MDC1 transfection (Fig. 4E and F). Together, these results suggest that induction of miR-22 and subsequent downregulation of MDC1 are responsible for the repression of HR in high Akt1-activated cells.

Given that our data support a role for miR-22 in abrogation of MDC1 expression and its DDR function and recent studies have shown that miR-22 family is upregulated during replicative senescence of human fibroblasts (21), we next sought to assess whether upregulation of miR-22 could downregulate MDC1 expression during both replicative senescence and premature stress-induced senescence. Senescence was induced in both human embryonic lung fibroblasts MRC-5 and IMR-90 cells by either serial passaging or by treatment with either hydrogen peroxide (H₂O₂) or busulfan (BU). As expected, the level of miR-22 was markedly higher in replicatively senescent cells than in young cells. Like replicative senescent cells, H₂O₂- and BU-induced prematurely senescent cells also showed much higher miR-22 expression compared with control young cells (Fig. 5A; Supplementary Fig. S6A). Importantly, all of these senescent cells showed significantly reduced MDC1 protein and mRNA (Fig. 5B and C; Supplementary Fig. S6B and S6C). Thus, downregulation of MDC1 is considered as a general phenomenon in cells undergoing replicative senescence or stress-induced premature senescence.

The DDR function of MDC1 during senescence was then examined by immunofluorescence for IR-induced MDC1 foci formation. When young cells were exposed to IR, clearly visible MDC1 foci formed (Supplementary Fig. S7A and S7B). In contrast, induction of MDC1 foci formation was significantly reduced in replicatively senescent and H₂O₂-induced prematurely senescent cells. To confirm the role of miR-22 in this process, we transfected anti–miR-22 into H₂O₂-induced prematurely senescent cells and showed that MDC1 levels (Fig. 5D) and IR-induced MDC1 foci formation (Fig. 5E) were restored.

We then determined whether miR-22 inhibits MDC1 expression contributing to repressing DSB repair in senescent cells. To test this, we introduced either anti–miR-22 or miR-22–insensitive MDC1 into senescent cells and examined the effect on DSB repair. Young cells that were exposed to IR repaired the majority of DSBs within 6 hours (Supplementary Fig. S8). In contrast, senescent cells still had numerous unrepaired DSBs 6 hours after IR exposure, but this effect was reversed if anti–miR-22 was present (Fig. 6A). The DSB repair defect in senescent cells was also fully rescued by overexpressing miR-22–insensitive MDC1 (Fig. 6B; Supplementary Fig. S9), suggesting that miR-22 acts to regulate DSB repair in senescent cells by modulation expression of MDC1.

Several studies have proposed that miR-22 is a tumor suppressor because it induces senescence-like phenotypes in cancer cells and it triggers both growth suppression and apoptosis (21, 28–31). However, our data highlight an oncogenic function of miR-22, in which it inhibits DSB repair and consequently increases the risk of genomic instability, a hallmark of cancer cells. Thus, we examined whether DNA damage accumulation accompanied miR-22–induced cellular senescence in cancer cells. To this end, human breast cancer cells (MCF7 and MDA-MB-231) and human cervical cancer cells (si-Ha) were transiently transfected with the miR-22, exposed to IR, and then stained cells for foci of γ-H2AX, a marker of unrepaired DSBs. Consistent with a previous report (21), introduction of miR-22 into cancer cells caused a senescence-like phenotype, as observed by the increased senescence-associated β-galactosidase (SA-β-gal) activity (Supplementary Fig. S10). Notably, miR-22–induced senescent cells showed a dramatic reduction in MDC1 expression (Fig. 6C, middle lanes in each panel) and a significant increase in the number of unrepaired DSBs as compared with control cells (Fig. 6D, middle row in each panel). However, when miR-22–induced senescent cells were transfected with miR-22–insensitive MDC1 (Fig. 6C, third lanes in each panel), the suppression of the DSBs repair was almost abolished (Fig. 6D, bottom rows in each panel). These results suggest that the miR-22–mediated decrease in MDC1 plays a critical role in the accumulation of DSBs in senescent cells.

We next sought to assess whether miR-22 expression inversely correlates with the MDC1 expression in aged human and mouse tissues. To test this, we measured miR-22 and MDC1 expression levels in lung and colon tissues from young and old mice using real-time qPCR and Western blotting. For both lung and colon tissues, the older mice had higher levels of miR-22 (Fig. 7A) and lower levels of both MDC1 mRNA and protein expression (Fig. 7B) than younger mice. Based on our previous observations, it is very likely that the decreased MDC1 levels in lung and colon tissues in older mice were a direct result of the high levels of miR-22. We were able to show the same effect in human cells using peripheral blood mononuclear cells from young (n = 7) and old (n = 11) human donors (Fig. 7C). Furthermore, the same was true for human tissue samples. When colon biopsies from young (17 and 24 years old) and old (64 and 68 years old) donors were subjected to immunohistochemical analysis, staining for MDC1 was much darker in younger colon tissue than the older tissue, again showing that MDC1 abundance decreases with aging (Fig. 7D). Thus, miR-22–mediated downregulation of MDC1 expression is a common mechanism in aging cells and is utilized in diverse and widespread tissue types in mammals.

Akt1 is a serine/threonine kinase, which is a key downstream target of the signaling pathway mediated by PI3K, and plays a pivotal role in the regulation of diverse cellular process, including cell growth, proliferation, and survival (32). Aberrant activation of the PI3K/Akt1 pathway is a common event in a wide range of human cancers (22). Activated Akt1 stimulates NHEJ repair, which contributes to chemo- or radioresistance in some tumor cells with constitutive Akt1 activation (33–35). In contrast, activation of Akt1 inhibits HR due to suppression of the DDR under pathologic circumstances (23–25, 36). Because defective HR can lead to genome instability and predisposition to cancer (37), the inhibition of HR by the activation of Akt1 may contribute to tumorigenesis. However, the precise mechanism by which activated Akt1 exerts its influence on HR needs to be elucidated. In the present study, we detected a decline in HR in cells expressing constitutively active Akt1. This decline in HR was associated with the upregulation of miR-22, which caused the loss of MDC1 function. These data show that CA-Akt1 can reduce the efficiency of HR following DSBs, which correlates with the decreased recruitment of MDC1 to sites of DNA breaks. These results also showed that both the inhibition of miR-22 and overexpression of MDC1 completely restored the function of MDC1 in the DDR and HR in cancer cells with high Akt1 activity. This scenario clarifies how elevated Akt1 activity might increase genomic instability and foster an environment for cancer development. Remarkably, this inverse correlation between pAkt1 and MDC1 expression occurs in vivo in human prostate tumor tissues. Thus, miR-22–mediated MDC1 downregulation could be an underlying mechanism behind Akt1-mediated oncogenesis and anti–miR-22 may be a new potential therapeutic agent for Akt1-induced tumorigenesis.

During cellular senescence or organismal aging, mammalian cells accumulate mutations in their genome and often end up with abnormal chromosomal rearrangements (38). These mutations and genomic rearrangements arise from aberrant DSB repair (39–42). However, the molecular mechanism for the diminished capacity to repair DSBs during replicative senescence and aging is poorly understood. Our experiments reveal an important role of miR-22 in the control of MDC1 function and DSB repair in different types of senescence. Thus, the miR-22–mediated decrease in MDC1 expression that occurs during replicative senescence and stress-induced premature senescence explains how senescent cells become defective in DSB repair capacity, and this may provide a representative molecular mechanism of increased genomic instability in senescent cells.

The incidence of carcinoma, the most common cancer in humans, increases exponentially with age (43). This is most likely due to the mutations and genomic rearrangements that accumulate during normal aging, and that could contribute to the transformation of functionality in aged tissues (44, 45). Hence, cellular senescence is a potent tumor-suppressing mechanism, but at the same time, it also contributes to cancer promotion at an advanced age (46). miR-22 has been proposed to be a tumor suppressor because it could inhibit cell proliferation and induce a senescence-like phenotype in human breast, cervical, and colon cancer cells (21, 28, 31). However, it was also proposed that miR-22 had oncogenic functions because it targets ten eleven translocation (TET) tumor suppressors in breast cancer cells and hematopoietic stem cells (47, 48). We observed the senescence-like phenotype when we overexpressed miR-22 in human cancer cells; however, we also found that miR-22 induced extensive DSB accumulation in the same cells. Furthermore, our in vivo data using both mouse and human tissues strongly suggest a positive correlation between miR-22 upregulation and MDC1 deficiency. Thus, even though miR-22–induced senescence may act as a barrier to cancer development, the defects in DDR and DSB repair that result from the downstream effects of miR-22 regulation cause detrimental chromosomal abnormalities, leading to the accumulation of secondary insults that might establish a cellular environment fostering tumorigenesis and cancer progression independent of proliferation-related phenotypes, and this may provide an underlying molecular mechanism for the increased incidence of cancer with advanced age.

In summary, we have shown that miR-22 is a key player in DSB repair and genomic stability through modulation of MDC1 expression under particular pathophysiologic contexts, including high Akt1 activity and replicative/stress-induced senescence. Our findings reveal the miR-22/MDC1 interaction as a previously unsuspected link between DNA damage signaling pathways under aberrant physiologic conditions, and this regulatory network is a key process in compromising the maintenance of genomic integrity and thus provides new insight into the role of MDC1–miR-22 network in Akt1- and aging-related tumorigenesis (Fig. 7E). Antagonists of endogenous miR-22 in these cells may thus be useful therapeutic strategies for enhancing MDC1 expression, given that it could block dysregulated DDR, accumulated DNA damage, and chromosomal abnormalities.

No potential conflicts of interest were disclosed.

Conception and design: J.-H. Lee, J. Yong, H.J. You

Development of methodology: S.-J. Park, S.-Y. Jeong, M.-J. Kim, S. Jun

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): S. Jun, H.-S. Lee, I.-Y. Chang, S.-C. Lim, S.P. Yoon

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): J.-H. Lee, S.-J. Park, S. Jun, H.-S. Lee, S.-C. Lim, S.P. Yoon, J. Yong

Writing, review, and/or revision of the manuscript: J.-H. Lee, J. Yong, H.J. You

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): J.-H. Lee, S. Jun

Study supervision: J.-H. Lee, H.J. You

The authors thank the members of the DNA Damage Response Network Center for technical assistance and helpful comments on the article.

This work is supported by the National Research Foundation of Korea (NRF), funded by the Ministry of Science, ICT, and Future Planning (NRF-2011-0018686, NRF-2011-0029629, and NRF-2013M2B2A9A03051397).

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.