NF-κB-dependent MicroRNA-125b Up-regulation Promotes Cell Survival by Targeting p38α upon Ultraviolet Radiation*

Background: UV radiation-induced miRNA expression modulates cellular stress response. Results: UV up-regulates miR-125b expression via ATM-dependent NF-κB activation, which represses p38α expression. Conclusion: UV-induced miR-125b prevents prolonged p38α activation and thereby promotes cell survival. Significance: Up-regulation of miR-125b serves as a negative feedback mechanism to control p38α activity upon UV radiation, which may contribute to tumorigenesis of skin cancer. UV-induced stress response involves expression change of a myriad of genes, which play critical roles in modulating cell cycle arrest, DNA repair, and cell survival. Alteration of microRNAs has been found in cells exposed to UV, yet their function in UV stress response remains elusive. Here, we show that UV radiation induces up-regulation of miR-125b, which negatively regulates p38α expression through targeting its 3′-UTR. Increase of miR-125b depends on UV-induced NF-κB activation, which enhances miR-125b gene transcription upon UV radiation. The DNA damage-responsive kinase ATM (ataxia telangiectasia mutated) is indispensable for UV-induced NF-κB activation, which may regulate p38α activation and IKKβ-dependent IκBα degradation in response to UV. Consequently, repression of p38α by miR-125b prohibits prolonged hyperactivation of p38α by UV radiation, which is required for protecting cells from UV-induced apoptosis. Altogether, our data support a critical role of NF-κB-dependent up-regulation of miR-125b, which forms a negative feedback loop to repress p38α activation and promote cell survival upon UV radiation.

Ultraviolet (UV) radiation is one of the major environmental hazards causing DNA damage in human cells. Exposure to UV may directly induce genomic lesions in nuclear and mitochondrial DNA as well as alternatively through generating reactive oxygen species. To cope with the detrimental effect of UV exposure, eukaryotic cells developed a prompt, inducible, and transient stress response, including cell cycle arrest, DNA repair, and alteration of gene expression. Failure of this response may lead to immune suppression, inflammation, photoaging, and skin carcinogenesis (1). The MAPK p38␣ is one of the primary UV-responsive kinases coordinating the stress response in cells exposed to UV (2). It regulates UV-induced cell cycle arrest, apoptosis, and gene transcription via activating transcription factors, such as p53, AP-1, cAMP-response element-binding protein, STATs, and NF-B.
NF-B (nuclear factor B) is a family of transcription factors that are pivotal in orchestrating cellular response to environmental stresses via modulating responsive gene expression (3). It was shown that NF-B and p53 were two major regulators of gene transcription in response to genotoxic stimuli, such as ionizing radiation (4). Mammalian cells respond to UV radiation by inducing or suppressing the expression of specific genes involved in UV stress response, which is also mediated by transcription factors, such as p53, AP-1 and NF-B (1). Whereas the activation of p53 may involve p38␣-dependent phosphorylation and nuclear signals generated by damaged DNA (2), the molecular mechanisms by which UV light activates NF-B were originally thought to only involve cytoplasmic signals generated independently of DNA damage (5). In contrast to the strong, rapid, and transient NF-B activation induced by inflammatory stimuli, UV-induced NF-B activation appears to be weak, slow, and prolonged. Different mechanisms may be involved in early or later stages of UV-induced NF-B activation (6). Initially, it was proposed that IKK activity is not required for IB␣ degradation because inducible IKK activation cannot be detected after UV treatment (7,8). Instead, UVinduced IB␣ degradation was suggested to be dependent on phosphorylation in its C-terminal PEST domain by casein kinase 2 (CK2) via an interaction with p38␣ (9). However, there was also strong evidence that IKKs are indispensable for UVinduced IB proteolysis, because UV-induced IB␣ degradation and NF-B activation was not detected in IKK knockout cells, particularly in IKK␤-deficient (IKK␤ Ϫ/Ϫ ) cells (7,10,11). Recently, IKK␤ was shown to function as an adaptor protein for IB␣ association with ␤-TrCP in the nucleus, which resulted in IB␣ degradation and subsequent NF-B activation in response to UV (11). Furthermore, a mathematical approach using computational modeling also suggested that constitutive phosphorylation of IB␣ by basal IKK activity, which affects the turnover rate of IB␣, and IB␣ synthesis inhibition are crucial for UV-induced NF-B activation (10,12). Nevertheless, whether UV-induced nuclear DNA damage and the subsequent DNA damage-responsive signaling cascade are involved in UVinduced NF-B activation remains elusive.
Besides affecting protein-coding gene expression, UV exposure also changes the expression level of microRNAs (13,14). MicroRNAs (miRNAs) 3 are a group of endogenous small noncoding RNAs (ϳ23 nucleotides) that negatively regulate gene expression at the post-transcriptional level, mainly via binding to the 3Ј-untranslated region (3Ј-UTR) of target mRNA (15). The binding of miRNA-loaded RNA-induced silencing complex with target mRNA may lead to blockage of protein translation as well as reduced mRNA stability (16). In mammalian cells, the vast majority of miRNAs are encoded by RNA polymerase II as intergenic miRNAs or as introns of host proteincoding genes (17). In the nucleus, the ϳ70-nucleotide hairpin structures of primary miRNA gene transcripts, termed primary miRNAs, are recognized by RNase III endonuclease Drosha-DGCR8 microprocessor complex and cleaved into precursor miRNAs, which can be exported to the cytoplasm by exportin 5. In the cytoplasm, another RNase III endonuclease, Dicer, will further process precursor miRNAs to yield mature miRNAs, which can be loaded into RNA-induced silencing complex along with Argonaute (Ago) proteins for targeting mRNAs through interactions with sites of imperfect complementarity (18). It was shown that UV induced a cell cycle-dependent relocalization of Ago2 into stress granules and altered miRNA expression in part dependent on ATM/ATR (13). However, whether UV radiation regulates miRNA expression by modulating their gene transcription or altering post-transcriptional maturation has not been determined.
Previous studies have suggested that NF-B may play an important role in regulating transcription of miRNA genes (19 -21). In this report, we found that miR-125b induction upon UV radiation was dependent on activation of NF-B, which enhanced miR-125b gene transcription. ATM activation in response to UV-induced DNA damage was indispensable for NF-B activation in cells exposed to UV radiation. Moreover, the up-regulated miR-125b directly targeted p38␣ and negatively regulated its expression, which prevented prolonged p38␣ activation and protected cells from p38␣-mediated apoptosis upon UV treatment.

EXPERIMENTAL PROCEDURES
Cell Culture, Plasmids, and Reagents-Human embryonic kidney cell line HEK293, mouse embryonic fibroblast cells (wild type, IKK␤ Ϫ/Ϫ , and ATM Ϫ/Ϫ ), human colon cancer cell line HCT116 (wild type and Dicer Ϫ/Ϫ ), and human osteosarcoma cell line U2OS were maintained in DMEM containing 10% inactivated fetal bovine serum. Human keratinocyte cell line HaCaT was grown in DMEM/F-12 medium supplemented with 10% fetal bovine serum. All cell lines were maintained in the presence of penicillin (100 IU/ml) and streptomycin (100 mg/ml) at 37°C with 5% CO 2 . The expression construct of pre-miR-125b was from Origene (Rockville, MD). Control (26164) sponge construct obtained from Addgene and the strategy to generate miR-125b sponge have been described in a previous report (22). The expression construct of p38␣ was generated by cloning the coding region sequence of human p38␣ into pcDNA3 expression vector. To generate p38␣ miRNA-targeting reporter, p38␣ 3Ј-UTR was amplified from human genomic DNA and inserted downstream of the luciferase (Luc2) sequence of pmirGLO Dual-Luciferase reporter construct (Promega, Madison, WI) following the manufacturer's instructions. A similar strategy was used to generate the miR-125b luciferase reporter by inserting synthesized miR-125b-target sequence into the pmirGLO reporter construct. Substitution mutants of p38␣ 3Ј-UTR were generated by QuikChange PCR. Antibodies against p38␣, IKK␤, PARP1, caspase-8, or cleaved caspase-3 were from Cell Signaling Technology (Danvers, MA). Antibodies against IB␣, p65, and tubulin were from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). ATM inhibitor Ku55933, CK2 inhibitor TBB, and p38␣ inhibitor SB203580 were from EMD (Billerica, MA), and IKK␤ inhibitor TPCA-1 was from Sigma. UV radiation was carried out with a GS Gene Linker UV chamber cross-linker (Bio-Rad).
RNA Extraction and Quantitative Real-time PCR-Total RNA was extracted from cells using TRIzol (Invitrogen) and then converted to first-strand cDNA using Superscript III transcriptase (Invitrogen). For microRNA analysis, total RNA was poly(A)-tailed using poly(A) polymerase (Ambion) before reverse transcription as described previously (23). The small noncoding RNA U6 and GAPDH were used as internal control for miR-125b, pri-miR-125b, and p38␣ quantitation, respectively, and quantitative real-time PCR was carried out as described previously (24). The sequences of gene-specific primers used for quantitative real-time PCR are listed in supplemental Table 1.
Immunoblotting-Total cell extracts were prepared and subjected to immunoblotting as described previously (25). For quantitation of the immunoblotting signal, IRdye-labeled secondary antibodies were used, and signals were analyzed by Odyssey scanner (LI-COR, Lincoln, NE).
Luciferase Assay-HEK293 cells were transfected with respective pmirGlo Dual-Luciferase reporter constructs harboring miR-125b target sequence or p38␣ 3Ј-UTR. After 36 h, cells were treated and lysed with passive lysis buffer, and the activity of firefly luciferase and Renilla luciferase in the lysates was measured with the Dual-Luciferase assay system (Promega). For the luciferase transcription reporter assay, NF-B-Fluc2p (pGL4.32) from Promega was used as a positive control. hsa-miR-125b-1/2 gene reporters were generated by inserting the respective promoter regions (WT or B site deletion) into pGL4.11 vector. HEK293 cells were transfected with transcription reporters along with Tk-Rluc reporter. Cells were mocktreated or treated with UV, and luciferase activity was measured as described above.
Electrophoretic Mobility Shift Assay (EMSA)-The Ig-B oligonucleotide probe and conditions for EMSA were described previously (24). The Oct-1 site oligonucleotide used for control was obtained from Promega. Gels were exposed and quantified with a Cyclone phosphor imager (PerkinElmer Life Sciences).
Chromatin Immunoprecipitation (ChIP)-ChIP assays were carried out as described previously (24). Briefly, treated cells were cross-linked with 1% formaldehyde, sheared to an average size of ϳ500 bp, and then immunoprecipitated with antibodies against p65/RelA. The ChIP-PCR primers were designed to amplify the proximal promoter regions containing putative NF-B binding sites within the hsa-miR-125b-1 or hsa-miR-125b-2 promoter as illustrated.
Cell Survival Assay-HEK293 and U2OS cells were mocktransfected or transfected with the indicated plasmids. After 36 h, cells were exposed to UV (20 J/m 2 ) alone or in the presence of SB203580 and harvested at various times after UV treatment. Cells were then stained with trypan blue, and the live cell percentage was obtained with a TC10 automated cell counter (Bio-Rad). Data from three independent experiments were pooled and plotted as shown.
Statistical Analysis-The results were presented as mean Ϯ S.D. and analyzed with Student's t test. p Ͻ 0.05 was denoted as statistically significant.

RESULTS
miRNAs Are Required for UV Radiation-induced Decrease of p38␣ Expression-In our effort to further explore the cellular stress response to UV radiation, we found that the protein level of p38␣, a key UV-responsive kinase, substantially decreased at later time points in human keratinocyte HaCaT cells and HEK293 cells after UV exposure ( Fig. 1A and supplemental Fig.  1A). Proteasome inhibitor failed to block the decrease of p38␣ (supplemental Fig. 1B), suggesting that the mRNA level of p38␣ may decline upon UV radiation. Indeed, we detected a significant reduction of the p38␣ mRNA level in both cell lines in response to UV radiation (Fig. 1B). The decrease of the mRNA level may be caused by reduced gene transcription, increased mRNA decay, or both. Intriguingly, we found that the decrease of p38␣ expression upon UV radiation was remarkably attenuated in Dicer-deficient HCT116 cells compared with wild type cells (Fig. 1C). Dicer is an essential RNase III endonuclease for miRNA biogenesis, and Dicer-disrupted HCT116 cells showed reduced amounts of mature miRNAs and accumulation of Whole cell lysates were immunoblotted with p38␣ and tubulin antibodies. D, similar experiments were carried out as in C. p38␣ and GAPDH mRNA levels at the indicated times after UV treatment were analyzed as in B. **, p Ͻ 0.01; *, p Ͻ 0.05. E, HEK293 cells were transiently transfected with pmirGLO Dual-Luciferase p38␣ 3Ј-UTR reporter (full or short 3Ј-UTR). After 24 h, cells were either left untreated or treated with UV. Luciferase activity was quantified at 12 h after treatment. Data from three independent experiments were pooled and are shown as mean Ϯ S.D. **, p Ͻ 0.01. F, HCT116 cells (wild type or Dicer-deficient) were transiently transfected with pmirGLO Dual-Luciferase p38␣ short 3Ј-UTR reporter and analyzed as in E. Data from three independent experiments were pooled and are shown as mean Ϯ S.D. **, p Ͻ 0.01. miRNA precursors when compared with wild type HCT116 cells (26). Therefore, our data suggested that miRNAs may play an important role in this down-regulation of p38␣ after UV treatment.
Although miRNAs could down-regulate target gene expression via inhibiting protein translation, it was shown that mRNA destabilization usually plays a major role in miRNA-dependent gene repression, especially to those highly repressed targets (16). In accordance with these studies, we found that the mRNA level of p38␣ upon UV radiation was also significantly lower in wild type HCT116 cells than that in Dicer Ϫ/Ϫ cells (Fig. 1D). To further examine the role of miRNA in UV-induced p38␣ downregulation, we generated a luciferase reporter construct in which the luciferase gene was fused with the 3Ј-UTR of p38␣. Interestingly, UV treatment significantly reduced luciferase activity in HEK293 cells expressing this p38␣ 3Ј-UTR reporter (Fig. 1E). Moreover, the UV-induced repression of p38␣ 3Ј-UTR luciferase activity was abrogated in Dicer Ϫ/Ϫ HCT116 cells (Fig. 1F). These results indicate that the p38␣ gene 3Ј-UTR may be subjected to miRNA-dependent repression, resulting in down-regulation of p38␣ expression in response to UV treatment.
UV Treatment Induces miR-125b Expression-To characterize the miRNAs responsible for p38␣ down-regulation by UV, we analyzed 3Ј-UTR of the p38␣ gene by Targetscan and identified two potential sites complementary to the seed region of the miR-22 or miR-125 family, which are highly conserved across species. We then generated a luciferase reporter that harbors a short fragment of the p38␣ gene 3Ј-UTR, including these two miRNA-targeting sites. Interestingly, UV treatment induced a comparable repression of luciferase activity in cells expressing this shorter p38␣ 3Ј-UTR reporter and in cells expressing the full-length p38␣ 3Ј-UTR reporter, suggesting that the miR-22 and/or miR-125 family may play a major role in regulating p38␣ expression upon UV radiation (Fig. 1E). In our previous study, we have found that miR-22 expression was significantly increased in cells exposed to UV (27). However, neither overexpression of miR-22 nor inhibiting miR-22 in UVtreated cells affected p38␣ 3Ј-UTR reporter activity (data not shown), indicating that miR-22 is unlikely to be a major miRNA responsible for p38␣ repression upon UV.
We then examined the expression of members of miR-125 family upon UV radiation and found that miR-125b was significantly induced in both HaCaT and HEK293 cells in response to UV exposure ( Fig. 2A). Consistently, UV treatment significantly inhibited miR-125b reporter activity, which could be rescued by overexpression of a miR-125b sponge inhibitor (Fig.  2B). Because an increased miRNA level may result from transcriptional up-regulation and/or enhanced post-transcriptional maturation upon DNA damage (14), we further determined the expression of primary miR-125b in UV-treated cells. Interestingly, we found that the primary transcripts from both FIGURE 2. miR-125b is up-regulated after UV treatment. A, HaCaT and HEK293 cells were mock-treated or treated with UV (20 J/m 2 ). qRT-PCR was performed to analyze miR-125b expression in HEK293 and HaCaT cells at 3, 6, or 12 h after UV treatment. -Fold change of relative miR-125b expression (miR-125b/U6) was normalized to untreated cells and shown as mean Ϯ S.D. (error bars). B, HEK293 cells were transfected with miR-125b reporter alone or along with pre-miR-125b or miR-125b sponge as indicated. After 36 h, cells were either mock-treated or treated with UV (20 J/m 2 ), and luciferase activity was quantified at 12 h after UV radiation. The histogram represents the normalized relative luciferase activity (firefly luciferase/Renilla luciferase) of three independent experiments, shown as mean Ϯ S.D. C, HaCaT cells were treated with UV (20 J/m 2 ) as indicated, and pri-miR-125b1 and pri-miR-125b2 expression were analyzed by qRT-PCR. -Fold change of relative expression (pri-miR-125b/GAPDH) was normalized to untreated cells and shown as mean Ϯ S.D. D, experiments similar to those in C were performed using HEK293 cells.
human miR-125b gene paralogs, hsa-miR-125b-1 on chromosome 11 and hsa-miR-125b-2 on chromosome 21, were significantly increased in HaCaT and HEK293 cells upon UV treatment (Fig. 2, C and D). Of note, although the expression of primary miR-125b was maintained at an increased level, the mature form of miR-125b expression had dropped at 12 h after UV treatment. It suggests that UV treatment may inhibit miR-125b processing at later times, which is consistent with a previous report showing that UV could modulate miRNA expression by inducing relocalization of miRNA-processing factors Dicer and Ago2 (13). Nevertheless, our data strongly support the notion that UV treatment could induce activation of miR-125b transcription, resulting in increase of the miR-125b level.
Expression of p38␣ Is Negatively Regulated by miR-125b-To determine whether p38␣ is a gene target of miR-125b, we generated a mutated p38␣ 3Ј-UTR luciferase reporter harboring a mutation in the miR-125b target sequence (Fig. 3A). We found that overexpression of miR-125b significantly inhibited luciferase activity in HEK293 cells expressing wild type p38␣ 3Ј-UTR reporter but not the mutated 3Ј-UTR reporter (Fig. 3B). Furthermore, mutation of the miR-125b target sequence also significantly attenuated UV-induced repression of p38␣ 3Ј-UTR reporter luciferase activity (Fig. 3C). It is noteworthy that UVinduced inhibition of p38␣ 3Ј-UTR reporter activity was still observed even in the presence of miR-125b site mutation, suggesting that other UV-induced miRNAs may also contribute to p38␣ repression. However, overexpression of miR-125b was sufficient to down-regulate p38␣ expression at both the protein and mRNA level (Fig. 3D). Accordingly, UV-induced p38␣ repression was rescued by expressing a miR-125b sponge inhibitor (Fig. 3E). Our data also indicated that 3Ј-UTR is required for miR-125b-mediated p38␣ down-regulation because expression of a p38␣ coding region construct was insensitive to miR-125b overexpression as well as UV radiation (Fig. 3, F and G). All of this body of evidence indicates that miR-125b plays a critical role in repressing p38␣ expression through targeting its 3Ј-UTR upon UV radiation.
UV-induced NF-B Activation Is Required for miR-125b Induction-To delineate the mechanism involved in miR-125b transactivation in response to UV treatment, we examined the impact of various kinase inhibitors on miR-125b induction in UV-treated HaCaT cells. Intriguingly, UV-induced miR-125b up-regulation was significantly inhibited by either ATM inhibitor Ku55933 or IKK␤ inhibitor TPCA-1 (Fig. 4A). Our previ- ous studies have demonstrated that both ATM and IKK␤ are essential kinases required for genotoxic NF-B signaling (28,29), so this result indicated a critical role of genotoxic NF-B signaling in regulating miR-125b transcription upon UV radiation. Consistently, induction of both pri-miR-125b-1 and pri-miR-125b-2 upon UV treatment was remarkably inhibited in the presence of ATM or IKK␤ inhibitor (Fig. 4B). We also observed that UV-induced p38␣ repression was attenuated by inhibiting either ATM or IKK␤ (Fig. 4, C and D), suggesting that ATM/IKK␤-dependent NF-B activation upon UV treatment is required for p38␣ down-regulation, probably via enhancing miR-125b transcription.
To further investigate the important role of genotoxic NF-B activation in regulating miR-125b up-regulation in response to UV, we examined miR-125b expression in UV-treated MEF cells. We found that UV treatment induced a substantial increase of miR-125b in wild type MEF cells but not in ATMdeficient MEFs, indicating that ATM is indispensable for miR-125b up-regulation upon UV treatment (Fig. 5A). Moreover, the up-regulation of miR-125b by UV was significantly reduced by overexpressing a dominant negative inhibitor of NF-B signaling, IB␣ superrepressor (S32A/S36A) (Fig. 5, B and C), which was shown to inhibit UV-induced NF-B activation (7, 10). Accordingly, UV-induced up-regulation of pri-miR-125b-1/-2 transcription was also abrogated in cells expressing IB␣ superrepressor (Fig. 5D). These data further support an essential role of the ATM-dependent genotoxic NF-B activation in regulating miR-125b transcription in response to UV radiation.
To determine whether NF-B directly regulates miR-125b transcription, we analyzed the upstream sequences of both hsa-miR-125b-1 and -2 and identified two potential NF-B consensus binding sites in the promoter region of each gene (Fig. 5, F  and G, top). Luciferase reporter assay analyses suggested that each potential B-binding site in either hsa-miR-125b-1 or -2 gene promoters are required for transactivation of the respective gene upon UV treatment (Fig. 5E). Remarkably, we detected significant enrichment of RelA/p65 at respective B-binding sites (shaded box in Fig. 5, F and G) within promoter regions of both hsa-miR-125b-1 and -2 genes by Chromatin IP analyses, strongly suggesting that NF-B might activate miR-125b gene transcription via directly binding on its promoter in cells treated by UV (Fig. 5, F and G). Although we did not detect a significant increase of p65 binding on all of the B sites, data from the luciferase reporter assay indicated that all of these sites were critical for miR-125b gene induction upon UV treatment (Fig. 5E), suggesting that additional members of the NF-B family may be recruited to these B sites in response to UV treatment. In addition, inhibition of either ATM or IKK␤ was able to significantly reduce NF-B/p65 recruitment to miR-125b gene promoter regions, which further supports the critical role of ATM and IKK␤ in mediating NF-B signaling upon UV radiation.
ATM Is Indispensable for UV-induced NF-B Activation-It was shown that UV-induced NF-B activation required p38␣dependent CK2 activation (9). We also found that inhibiting either p38␣ or CK2 substantially reduced NF-B activation in UV-treated HaCaT and HEK293 cells (supplemental Fig. 2A). Besides, our previous studies have demonstrated an essential role of ATM in mediating NF-B signaling in response to geno-toxic stimulation (25,28). Interestingly, replication stress inducers, such as hydroxyurea and aphidicolin, also induce NF-B activation in an ATM-dependent manner (24). Similar to hydroxyurea and aphidicolin, UV radiation induces a strong replication stress response featured by the quick activation of ATR along with delayed activation of ATM and DNA-PK (1,30). We found that both ATM and IKK activation appeared to be essential for UV-induced NF-B activation in HaCaT and HEK293 cells (Fig. 6A). Consistent with previous reports (7,10,11), we found that UV-induced NF-B activation was almost abrogated in IKK␤-deficient MEFs compared with wild type MEFs (Fig. 6B). Moreover, reconstitution of IKK␤ dramatically enhanced UV-induced NF-B activation in IKK␤ Ϫ/Ϫ MEFs FIGURE 5. miR-125b induction is regulated by NF-B in response to UV radiation. A, ATM wild type (ϩ/ϩ) and deficient (Ϫ/Ϫ) MEF cells were mock-treated or treated with UV (20 J/m 2 ). After 12 h, miR-125b expression was analyzed by quantitative real-time PCR. -Fold change of relative miR-125b expression (miR-125b/U6) was normalized and shown as mean Ϯ S.D. (error bars). **, p Ͻ 0.01. B, HEK293 and HEK293-IB␣-SR stable cells were treated and analyzed as in A. **, p Ͻ 0.01. C, HEK293 and HEK293-IB␣-SR stable cells were transiently transfected with pmirGLO miR-125b reporter. After 24 h, cells were either left untreated or treated with UV, and luciferase activity was quantified at 12 h after treatment. Data from three independent experiments were pooled and plotted as shown. *, p Ͻ 0.05. D, pri-miR-125b-1 and pri-miR-125b-2 induction upon UV treatment in HEK293 and HEK293-IB␣ SR cells was analyzed by quantitative real-time PCR. -Fold change of relative expression of pri-miR-125b from three independent experiments was normalized and is shown as mean Ϯ S.D. **, p Ͻ 0.01. E, HEK293 cells were transfected with NF-B luciferase reporter or luciferase reporters harboring wild type or mutated promoter regions of hsa-miR-125b-1/-2 genes as shown. Cells were mock-treated or treated with UV (20 J/m 2 ), and luciferase activity was measured at 24 h after treatment. firefly luciferase activity was first normalized to Renilla luciferase, and -fold induction of relative luciferase activity over mock-treated group was plotted as shown. k1 and k2, NF-B binding sites; ϫ, B site deletion. F and G, schematic representation of putative NF-B-binding sites (K1 and K2) within promoter regions of hsa-miR-125b-1 (F) and hsa-miR-125b-2 (G) shown at the top. HaCaT cells were left untreated or treated with UV (20 J/m 2 ), in the presence or absence of Ku55933 (Ku) or TPCA-1. A ChIP assay was performed to analyze p65 binding to B sites within the hsa-miR-125b-1/-2 promoter region. Detected p65 enrichment to NF-B-binding site (shaded box; K2 in the hsa-miR-125b-1 promoter; K1 in the hsa-miR-125b-2 promoter) upon UV treatment is shown below. **, p Ͻ 0.001; *, p Ͻ 0.01. (Fig. 6C). We also found that IKK␤ was essential for UV-induced IB␣ degradation in MEFs (supplemental Fig. 2B), which may involve the recently reported scaffolding role of IKK␤ to facilitate association between IB␣ and ␤-TrCP (11). These data support a critical role of IKK␤ in achieving optimal NF-B activation in cells exposed to UV radiation.
In addition to IKK␤, we found that UV-induced NF-B activation was abolished in ATM-deficient MEF cells (Fig. 6D). Consistently, inhibiting ATM significantly attenuated NF-B activation by UV in wild type MEFs ( Fig. 6D and supplemental Fig. 2C), whereas reconstitution of ATM in ATM Ϫ/Ϫ MEFs dramatically increased NF-B activation in response to UV treatment (Fig. 6E). In addition, ATM knockdown almost abrogated, whereas ATR depletion enhanced, UV-induced NF-B activation in HEK293 cells (supplemental Fig. 3), which is consistent with our previous studies on replication stress-induced NF-B activation (24). All of these results strongly suggest that ATM is indispensable for NF-B activation upon UV radiation. Interestingly, we found that UV-induced p38␣ activation was substantially inhibited by ATM inhibitor, suggesting that ATM may regulate p38␣ activation in response to UV treatment (Fig.  6F). Nevertheless, the mechanisms by which ATM regulates p38␣ activation still remain to be further explored.
Induction of miR-125b Promotes Cell Survival in Response to UV Treatment-It was shown that p38␣ activation promoted apoptosis in cells exposed to UV, which may involve direct phosphorylation and activation of p53 (31,32). We also found that inhibiting p38␣ decreased activation of caspase-8 and caspase-3 in cells treated with UV (supplemental Fig. 4A). It is plausible that miR-125b induction may inhibit cell apoptosis via down-regulating p38␣ in UV-treated cells. Indeed, overexpression of miR-125b inhibited caspase-3/8 activation and PARP1 cleavage in cells exposed to UV (Fig. 7, A and B, and supplemental Fig. 4B). Moreover, co-transfection of a construct containing only the p38␣ gene coding region, which is insensitive to miR-125b-dependent repression (Fig. 3F), bypassed the antiapoptotic effect of miR-125b overexpression (Fig. 7A). Accordingly, transfection of miR-125b sponge inhibitor in HaCaT cells significantly enhanced caspase-3 activation and PARP-1 cleavage in response to UV treatment (Fig. 7C). In addition, overexpression of miR-125b sponge inhibitor substantially enhanced and extended p38␣ activation in cells exposed to UV radiation (Fig. 7, D and E). All of this body of evidence indicates that miR-125b induction antagonizes apoptosis in cells exposed to UV, potentially via prohibiting prolonged hyperactivation of p38␣. Consistent with its role in inhibiting caspase activation, up-regulation of miR-125b also substantially enhanced cell survival in response to UV radiation. Overexpression of miR-125b in HEK293 cells significantly improved cell viability after UV radiation, whereas the prosurvival effect was completely blocked by co-transfection of exogenous p38␣ (Fig. 7F). In line with this observation, U2OS cells overexpressing miR-125b were more resistant to UV-induced cell death, whereas overexpression of miR-125b sponge inhibitor sensitized U2OS cells to UV treatment significantly (Fig.  7G). We also found that inhibiting p38␣ showed a protective effect set in cells exposed to UV radiation, similar to that in cells overexpressing miR-125b (supplemental Fig. S4, C and D), suggesting that p38␣ repression may play a major role in miR-125b-dependent protection in cells treated with UV.

DISCUSSION
In response to exposure to genotoxic stimuli, such as UV radiation, eukaryotic cells may induce temporary cell cycle arrest and DNA repair to maintain genomic stability. However, if the level of UV-induced DNA damage is too excessive to be repaired, cells will undergo apoptosis, which might result from prolonged activation of p38␣ and p53 (2). Thus, proper control of p38␣ activation may play a critical role in determining the cell fate between cell survival and apoptotic cell death, which is essential for restoring homeostasis in UV-stressed cells after DNA repair. Therefore, it is not surprising that p53-dependent up-regulation of phosphatase Wip1 was found to dephosphorylate p38␣, and this served as a negative feedback regulation of p38␣ activity in cells treated with UV (33).
In this study, we described a novel miR-125b-mediated negative feedback to control p38␣ activation through post-transcriptional repression of p38␣ in cells exposed to UV radiation. Intriguingly, the UV-induced expression of miR-125b was regulated by NF-B, whose activation has been shown to be dependent on p38␣. Thus, similar to p53-dependent Wip1 induction, p38␣ may also induce miR-125b up-regulation via activating NF-B, which in turn negatively regulates p38␣ expression to avoid prolonged p38␣ activation. Besides miR-125b, miR-141 and miR-200a have been shown to target p38␣ and modulate the oxidative stress response (34). Coincidentally, UV radiation triggers strong oxidative stress, which is one of the major effectors of its genotoxicity. Down-regulation of p38␣ by miR-141 and miR-200a significantly enhanced cell sur-vival and proliferation of human cancer cells exposed to oxidative stress, further supporting a critical role of p38␣ in mediating apoptosis, especially under overwhelming stress conditions. We postulate that there are other UV-induced miRNAs involved in regulating p38␣ expression because overexpression of miR-125b sponge inhibitor was unable to completely block p38␣ repression. Furthermore, we cannot exclude the possibility that transcriptional inhibition may also contribute to p38␣ down-regulation after UV radiation. It was shown that the majority of UV-responsive genes were transcriptionally repressed in cells exposed to UV radiation (35), and we also observed a substantial decrease of p38␣ mRNA in Dicer Ϫ/Ϫ HCT116 cells after UV treatment, suggesting that the decreased p38␣ gene transcription may also play a role in p38␣ down-regulation upon UV treatment. Nevertheless, both protein and mRNA levels of p38␣ were significantly lower in wild type HCT116 cells than that in Dicer Ϫ/Ϫ HCT116 cells when exposed to UV, and overexpressing miR-125b significantly increased cell survival upon UV radiation, which was comparable with the effect of p38␣ inhibition. All of this evidence strongly indicates a pivotal role of miR125b-dependent repression in negatively regulating p38␣ activity in UVtreated cells. p38 is a family of MAPKs consisting of four isoforms: p38␣, p38␤, p38␦, and p38␥ (36). Both p38␣ and p38␤ can be activated by UV exposure; however, their roles in mediating cellular apoptotic response to UV treatment may be distinct. UVinduced p38␣ activation was shown to promote cell death, whereas p38␤ may protect cells from UV-induced apoptosis (37). Using TargetScan, we did not find a miR-125b recognition sequence within the 3Ј-UTR of the p38␤ gene, suggesting that UV-induced miR-125b up-regulation may not directly repress p38␤ expression. However, we cannot rule out the possibility that additional miRNAs may be induced by UV, which could directly target p38␤. Because the p38 inhibitor SB203580 inhibits p38␣ and p38␤ activity, our data suggest p38␣ may play a dominant role in determining apoptotic response in cells exposed to UV (supplemental Fig. 1). Nevertheless, selective inhibition of p38␣ by inhibitors or miRNAs while sparing p38␤ may further improve cell survival in response to UV treatment.
Previous study implicated that ATM may be involved in regulating NF-B signaling in UV-treated melanoma cells (38). Using pharmacological and genetic approaches, we demonstrated that ATM is indispensable for NF-B activation by UV treatment, which is conserved among different cell types and between species. Our data indicated that UV-induced nuclear DNA damage response is involved in mediating NF-B signaling, even at a relatively early stage (3 h after UV treatment), yet how ATM regulates NF-B activation upon UV radiation is not entirely clear. Our data indicated that ATM was required for p38␣ activation by UV. A recent study showed that a cytosolic ATM-NEMO-RIP1 complex mediated p38␣ activation through recruiting TAK1 in response to genotoxic treatment (39). Our previous studies and others (40,41) also demonstrated that genotoxic stress may promote ATM nuclear export in association with NEMO, which regulates TAK1 activation in cytoplasm. It is possible that, in response to UV radiation, NEMO may facilitate ATM nuclear export into the cytoplasm, where it mediates TAK1 activation, which in turn activates p38␣ through phosphorylating MKK3/6. This hypothesis is consistent with the observation that NEMO is required for UVinduced NF-B activation (7). In parallel, UV radiation was recently shown to induce the nuclear translocation of IB␣ and association with IKK␤, which constitutively interacts with ␤-TrCP through heterogeneous ribonucleoprotein-U, leading to IB␣ ubiquitination and degradation (11). Thus, nuclear IKK␤ can act as an adaptor protein for IB␣ degradation in UV-induced NF-B activation. We also found that IKK␤ was essential for optimal NF-B activation in response to UV treatment. Because ATM may interact with IKK␤ via NEMO association (27), nuclear ATM may also modulate association between IKK␤, IB␣, and heterogeneous ribonucleoprotein-U upon UV radiation. Nonetheless, further investigation is required to examine these hypotheses and explore the mechanistic role of ATM in UV-induced NF-B signaling.
NF-B activation is generally considered to be prosurvival. However, UV-induced NF-B activation has also been shown to actively repress transcription of antiapoptotic genes, such as XIAP and BCL2L1, which may promote cell death (42). We found that UV-induced NF-B activation was required for upregulation of miR-125b, which promoted cell survival via repressing p38␣. NF-B was also considered as one of the major contributors in the oncogenesis of UV-induced skin cancers, probably through up-regulating its prosurvival target genes, such as COX-2 and BCL2 (43). Therefore, the impact of NF-B activation on cell survival and proliferation in response to UV probably needs to be weighed in the context of cell types and the dose and wavelength of UV as well as its extent and duration. Similarly, the roles of miR-125b in regulating cell growth and tumor progression are being actively studied but remain inconclusive. Increased expression of miR-125b has been observed in a variety of hematopoietic malignancies, including acute myeloid leukemia, myelodysplastic syndrome, and acute lymphoblastic leukemia (44). The oncogenic function of miR-125b was associated with reduction of proapoptotic genes, such as BAK1, TP53, and PUMA, or skewing hematopoietic lineage differentiation via repressing Lin28A (45). The roles of miR-125b in solid tumors are relatively controversial. Overexpression of miR-125b was found in prostate cancer and glioma patient samples (46,47). Up-regulation of the miR-125 family was also associated with increased cancer metastasis (48). However, it was found that miR-125b could down-regulate certain oncogenes, such as MUC1, ERBB2, and ERBB3, resulting in growth inhibition and enhanced sensitivity to genotoxic anti-cancer agents in breast cancer cells (49,50). In contrast, miR-125b was found to be up-regulated in Taxol-resistant cancer cells, which may inhibit Taxol-induced apoptosis and confer therapeutic resistance in breast cancer cells (51). All of these findings further underlined the notion that miRNAs can function as either oncogenes or tumor suppressors, depending on the genes they target in the context of different cancers (52).
Contradictory data regarding the NF-B-dependent regulation of miR-125b transcription have been reported. The NF-B activation has been shown to both transactivate and repress miR-125b expression in immune response elicited by LPS (21,53). Here we found that UV-induced NF-B activation was required for miR-125b transactivation, suggesting that NF-B may exert a distinct role in regulating miR-125b expression, depending on cell types and stimuli. Nevertheless, in skin cells exposed to UV radiation, NF-B probably plays an oncogenic role through promoting miR-125b-dependent p38␣ repression.