Anti-NPM monoclonal antibody (mAb) was kindly provided by Dr. P. K. Chan (Balyor College of Medicine, Houston, TX). Mouse monoclonal antibodies against E2F1, β-actin, retinoblastoma tumor suppressor protein (pRB), and rabbit polyclonal antisera against PP1α and PP1γ were obtained from Santa Cruz Biotechnology (Santa Cruz, CA). Anti-PP1β/δ rabbit polyclonal antibodies were purchased from Millipore (Billerica, MA). Rabbit antisera against phospho-NPM (Thr199), phospho-NPM (Thr234/237), and phospho-PP1α (Thr320) were from Cell Signaling Technology (Danvers, MA). All chemicals were purchased from Sigma-Aldrich (St. Louis, MO), except where otherwise indicated.

HeLa, 293T, and H1299 cells were grown as a monolayer in DMEM supplemented with 10% fetal bovine serum and 100 U/ml penicillin and streptomycin, in a 5% CO₂-humidified incubator at 37°C. Cells were transfected using Lipofectamine (Invitrogen, Carlsbad, CA) according to the manufacturer's instructions. Transient transfection was done for 24–54 h before cell harvest, unless otherwise noted.

For the E2F1 (Lin et al., 2006) and p53 (Chan et al., 2005) promoter reporter assay, the cells extracts were obtained with 1× Reporter lysis buffer (Promega, Madison, WI), and the reporter/luciferase activity was measured using Luciferase assay reagent (Promega) following the manufacturer's recommendations. Luciferase activity was measured with Luminometer AutoLunmat LB953 (Berthold, Norwalk, CT) and normalized to the corresponding β-galactosidase activity.

Cells were grown to 80% confluence in 6-cm dishes. DNA damage was achieved by exposing the cells to UV irradiation (50 J/m²). Cells were harvested at indicated times. In some experiments, they were treated with okadaic acid (OA; 50 nM) or calyculin A (5 nM) for 3 h before UV treatment.

Cells were harvested and washed twice in phosphate-buffered saline (PBS) and then lysed in ice-cold Gold lysis buffer (1% Triton X-100, 10% glycerol, 1 mM sodium orthovanadate, 1 mM EGTA, 10 mM NaF, 1 mM sodium pyrophosphate, 10 μM β-glycerophosphate, 20 mM Tris-HCl, 137 mM NaCl, 5 mM EDTA, and cocktail protease inhibitor (Sigma-Aldrich), pH 7.9) for 30 min. Lysates were boiled in 2× urea sample buffer dye (100 mM Tris-HCl, pH 6.8, 4% SDS, 0.2% bromphenol blue, 20% glycerol, 200 mM β-mercaptoethanol, and 8 M urea), and then fractionated by SDS-polyacrylamide gel electrophoresis (PAGE).

Western blot analysis was performed after electrophoretic separation of polypeptides by 8 or 10% SDS-PAGE and transfer to Hybond-polyvinylidene difluoride membrane (GE Healthcare, Piscataway, NJ). Blots were probed with the indicated primary and appropriate secondary antibodies. Immunobands were subsequently detected by the enhanced chemiluminescence reaction (GE Healthcare).

To establish a plasmid-based double-stranded RNAi system targeting endogenous PP1α, PP1β, or PP1γ, annealed oligonucleotides corresponding to partial sequence were designed and ligated to pSuper.neo+GFP (Oligoengine, Seattle, WA) according to manufacturer's instruction. The cDNA sequence of the targeted mRNA region for different genes is as follows: PP1α, 5′-GAGACGCTACAACATCAAA-3′; PP1β, 5′-GTTCGAGGCTTATGTATCA-3′; PP1β-2, 5′-TGTGCAGATGACTGAAGCA-3′, PP1γ, 5′-AGAGGCAGTTGGTCACTCT-3′; E2F1, 5′-CGCTATGAGACCTCACTGA-3′; and E2F1-2, 5′-GTGGACTCTTCGGAGAACT-3′.

To generate phosphorylation-site mutant constructs of NPM for expression in the cells, an appropriate set of oligonucleotide primers were used for site-directed amino acid substitutions (serine or threonine to alanine). We used full-length NPM cDNA in plasmid pCR3.1-FLAG-NPM as a template, and the primer sequences for cloning S125A, T199A, T199D, T234/237A, and T234/237D variants were listed in Supplemental Table S1.

The cells were harvested, washed twice in ice-cold PBS, and the cell pellets were resuspended in hypotonic RSB buffer (10 mM Tris-HCl, pH 7.4, 10 mM NaCl, 3 mM MgCl₂, and 1 mM phenylmethylsulfonyl fluoride) at 4°C for 10 min. An equal volume of an RSB + NP40 (20%) buffer was added next and incubated at 4°C for 10 min. Nuclei were collected by centrifuged at 3000 rpm at 4°C for 5 min and resuspended in ice-cold Gold lysis buffer (20 mM Tris-HCl, pH 7.5, 10% glycerol, 1% Triton X-100, 1 mM sodium pyrophosphate, 100 mM β-glycorophosphate, 1 mM sodium orthovanadate, 137 mM sodium chloride, 5 mM EDTA, 1 mM EGTA, and protease inhibitor cocktail) for 30 min and subsequently centrifuged at 12,000 rpm at 4°C for 30 min. Equal amounts of nuclei extract protein (1 mg) were incubated with the indicated antibodies (2 μg) at 4°C for 2 h with rotation. The immunocomplexes were captured with protein G-Sepharose (30 μl) for 2 h at 4°C with rotation. The protein G–antigen–antibody complexes were washed four times with the Gold lysis buffer and boiled in 2× urea sample buffer dye (100 mM Tris-HCl, pH 6.8, 4% SDS, 0.2% bromphenol blue, 20% glycerol, 200 mM β-mercaptoethanol, and 8 M urea) for subsequent PAGE and immunoblotting analysis as described above.

ChIP experiments were done essentially as described previously (Liu et al., 2007a). Cross-linked, sonicated chromatin was precleared before being incubated with 2 μg of mouse mAb (pRB or E2F1) and rotated at 4°C overnight. Normal mouse immunoglobulin (Ig)G was used for the mock immunoprecipitation. After extensive washes, immunocomplexes were treated with proteinase K and decross-linked. Bound DNA in the pRB ChIP was extracted, purified, and subjected to polymerase chain reaction (PCR) analysis by using primers corresponding to the E2F1 gene promoter sequence (Lin et al., 2006). After 35 cycles of amplification, PCR products were run on a 2% agarose gel and analyzed by ethidium bromide staining. For anti-E2F1 ChIP, bound DNA was subjected to quantitative real-time PCR analysis.

Total cellular RNAs were isolated from cells using the Trizol Reagent (Invitrogen) according to the manufacturer's instructions. Total RNA (2–5 μg) was reverse transcribed with reverse transcriptase (Moloney murine leukemia virus) (Invitrogen) at 37°C for 52 min. cDNA synthesis was primed with random hexamer. The E2F1 and internal control β-actin genes were amplified with the following specific primers: E2F1, 5′-CCGAGGTGCTGAAGGTGCAGA-3′ (sense), 5′-TCTTCCCAGGGCTGATCCCAC-3′ (antisense); and β-actin, 5′-AGAAAATCTGGCACCACACC-3′ (sense), 5′-CCATCTCTTGCTCGAAGTCC-3′ (antisense). The PCR products were resolved on 1.5% agarose gels and stained with ethidium bromide.

For quantitative determination by real-time PCR, target transcripts or ChIP products were analyzed using the LightCycler system and Power SYBR Green PCR Master Mix (Roche Diagnostics, Indianapolis, IN). Transcript levels were normalized to glyceraldehyde-3-phosphate dehydrogenase (GAPDH) values of the respective sample, and represent mean ± SD of at least three independent experiments. Sequences for the primers used in this experiment are as follows: E2F1, 5′-AGATGGTTATGGTGATCAAAGCC-3′ (sense), 5′-ATCTGAAAGTTCTCCGAAGAGTCC-3′ (antisense); GAPDH, 5′-GGTATCGTGGAAGGACTCATGAC-3′ (sense), 5′-ATGCCAGTGAGCTTCCCGT-3′ (antisense); XPC, 5′-CATCATCCCAGCCCGCTTTAC-3′ (sense), 5′-CCACTTCACCAGGTTTGAGAGG-3′ (antisense); DDB2, 5′-GCACCTCACACCTATCAAG-3′ (sense), 5′-GAGCTGACACATCATCTTCC-3′ (antisense); and RPA3, 5′-CGTCACTAAGCAGCCAATC-3′ (sense), 5′-GCACCAATCAGCGAAGAC-3′ (antisense). For quantitative analysis of anti-E2F1 ChIP experiments, primers corresponding to gene promoters were used: XPC, 5′-GAAATAGAGAGAAACCTGTTGT-3′ (sense), 5′-CTAGTCACGCCCCTAAAG-3′ (antisense); and DDB2, 5′-ATGTTTGGCGGGAAGTTG-3′ (sense), 5′-TCTGGGGAGAAACAAGGC-3′ (antisense). The results are given as a percentage of input and represent mean ± SD of at least three independent experiments. Triplicate PCRs were performed for all experiments.

The cells were okadaic acid treated (50 nM) for 8 h before harvest. Nuclei extracts were prepared and subjected to immunoprecipitation by using the anti-NPM antibody. The immunocomplexes were washed with reaction buffer (1 mM MnCl₂ and 5 mM caffeine). One unit of PP1 (New England Biolabs, Ipswich, MA) was added to the immunocomplexes for 1 h at 30°C. For inhibition studies, the immunocomplexes and PP1 were incubated in the presence of 1 μM OA. Degree of dephosphorylation was analyzed by SDS-PAGE and subsequently immunoblotting.

To estimate the DNA repair capacity of cells, the amounts of thymine dimers in UV-irradiated cells were measured by enzyme-linked immunosorbent assays (ELISAs) as described previously (Zhai et al., 2005) using an anti-thymine dimer antibody (Abcam, Cambridge, MA). DNA was extracted from the cells at the indicated times postirradiation. For statistical significance of quantitative comparisons, calculations were done by the SigmaPlot software (Systat Software, San Jose, CA).

This method for measuring nucleotide excision repair (NER) in response to UV irradiation was performed based on a previous report (Athas et al., 1991). The pCAT control vector (Promega) is a nonreplicated plasmid, under control of simian virus 40 promoter and enhancer sequences. The plasmid was treated with or without UV (254 nm) at the dose of 500 J/m² and then cotransfected with unirradiated pGL3luc (as an internal control) as well as plasmids encoding various phosphorylation-site mutants of NPM into HeLa cells. The cells were lysed in 1× Reporter lysis buffer 24 h after transfection, and chloramphenicol acetyltransferase (CAT) activity was detected as described previously (Wu et al., 2002a,b). After normalization, relative CAT activity was expressed as the percent of activity in cells transfected with the UV-irradiated plasmid over those harboring the undamaged plasmid. For statistical significance of quantitative comparisons, calculations were done by the SigmaPlot software.

To characterize the involvement of NPM in cellular response to gentoxic stress, we first tested whether DNA damage affects the phosphorylation of several sites on NPM. HeLa cells were treated with UV radiation, and cell lysates collected at different time points posttreatment were probed with phosphospecific NPM antibodies. Interestingly, UV enhanced dephosphorylation of NPM at Threonine 199, 234, and 237 residues (Figure 1A). Dephosphorylation was detected at 6 h after UV treatment and sustained at 9 h. Similar patterns of modification changes were also evident in the 293T and p53-null H1299 cell lines (Supplemental Figure S1).

It was reported previously that PP1 is a candidate phosphatase for nucleolar proteins such as nucleolin and NPM (Kotani et al., 1998). In line with this observation, we found that the phosphorylation levels of NPM on Thr199 and Thr234/237 were significantly enhanced in cells treated with PP1/PP2A phosphatase inhibitor OA or calyculin A (Supplemental Figure S2, A and B), thus providing further support to the notion that phosphorylation of this protein is counterbalanced by the action of phosphatase under normal growth condition. Based on these results, we next sought to determine whether PP1 is the specific phosphatase that is responsible for UV-induced NPM dephosphorylation. To this end, we first assessed the changes in the PP1 activity upon irradiation by performing immunoblotting with an antibody that specifically recognizes phospho-Thr320 of PP1. This modification was shown previously to be linked to the inactivation of PP1 enzymatic activity (Kwon et al., 1997; Liu et al., 1999). As shown in Figure 1B, our results showed that the level of phospho-Thr320 was reduced after UV treatment, indicating an up-regulation of PP1 phosphatase activity that is responsive to DNA damage. Next, to further investigate whether PP1 can indeed dephosphorylate NPM, we carried out an in vitro phosphatase assay. By monitoring the phosphorylation of endogenous NPM immunoprecipitated from cells, we found that purified recombinant PP1 is capable of removing phosphate groups on the Thr199 and Thr234/237 residues of NPM (Figure 1C and Supplemental Figure S2C). Furthermore, such specific phosphatase activity was lost upon inhibitor treatment (Supplemental Figure S2C). Together, these results showed that UV treatment is capable of inducing dephosphorylation of NPM at several sites and that PP1 is likely to be involved in this removal of phosphorylation after DNA damage.

We next aimed to pinpoint the specific isoform of PP1 responsible for mediating UV-induced dephosphorylation of NPM. Because the mammalian genomes encode three closely related PP1 genes—α, β, and γ—specific ablation of individual PP1 isoform expression was achieved by expressing isoform-specific short hairpin RNA(shRNA) (Figure 2A). HeLa cells depleted of either PP1 isoform as well as control were UV irradiated (Supplemental Figure S3A), and the level of phosphorylation of NPM on Thr199 or Thr234/237 was measured at 0, 3, and 6 h after treatment. Among the three isoforms, depletion of PP1β caused dramatic NPM hyperphosphorylation, whereas PP1α and -γ knockdown had little effect on the basal phosphorylation level (Figure 2B and its shorter exposed version in Supplemental Figure S3B). Furthermore, knockdown of PP1β, but not of PP1α or PP1γ, blocked the UV-induced dephosphorylation of NPM on Thr199 and Thr234/237 (Figure 2B and Supplemental Figure S3B). Similar outcomes of PP1β knockdown also were observed in 293T cells (Supplemental Figure S3C). Consistently, when PP1β was overexpressed in cells, phosphorylation at these threonine residues underwent considerable reduction (Supplemental Figure S3D). An association of transiently overexpressed (Figure 2C) or endogenous (Figure 2D) PP1β with NPM also was found in coimmunoprecipitation experiments. Specificity of such interaction was reinforced by the observations that PP1α and PP1γ did not coprecipitate NPM (Figure 2C). Collectively, these results strongly indicate that PP1β plays an important role in dephosphorylating NPM on Thr199 and Thr234/237 upon UV irradiation.

We also characterized the dephosphorylation of NPM in a more comprehensive manner, by conducting a mass spectrometry-based comparative phosphoproteome analysis on NPM isolated from control and PP1β knockdown cells (see Supplemental Material for detailed methods). Based on phosphopeptide profiling, we identified three additional phosphorylation sites on NPM (Ser4, Ser70, and Ser125; Supplemental Table S2 and Supplemental Figure S3E). However, quantitative comparison of phosphopeptides harboring Ser4 and Ser70 did not indicate significant changes in their abundance between the control and knockdown samples but revealed a slightly lower level of peptides with Ser125 phosphorylation in the PP1β-depleted cells (Supplemental Table S2 and Supplemental Figure S3E). We thus concluded that the phosphorylation levels of these residues are not subject to PP1β-mediated dephosphorylation.

We demonstrated previously an ability of NPM in endowing the cells with a greater NER activity under UV damage (Wu et al., 2002a,b). It is therefore possible that the observed UV-induced dephosphorylation of NPM may be a determinant for this protective function of NPM. To address this issue, we next undertook a reporter plasmid-based assay to assess the effect of NPM phosphorylation on cellular DNA repair capacity. Consistent with the previous studies, overexpression of wild-type NPM rescued the expression of reporter from UV-damaged plasmid, indicating an increased host NER activity (Figure 3A). Interestingly, expression of a variant of NPM in which Thr199/234/237 were changed to Ala (3A) led to an even further elevated reporter activity, whereas the stimulatory effect was abolished in the T199/234/237D (3D) mutant.

To further examine the role of NPM dephosphorylation in cellular response to UV damage in vivo, we measured by ELISA the cellular capacity of removing UV-induced thymine dimers. HeLa cells were transiently transfected with various phosphorylation-site mutants of NPM and subsequently irradiated (Figure 3C). In cells that expressed control vector and the phosphomimetic mutant (3D) of NPM, ∼30% of thymine dimers detected immediately after UV irradiation (at time 0) were removed within 3 h after irradiation, and the remaining damage was slightly decreased up to 9 h (Figure 3B). In contrast, phosphorylation-defective version of NPM (3A) substantially enhanced the ability of the cells to repair thymine dimers, especially within the first 6 h of UV treatment. Together, these data clearly show that the dephosphorylated species of NPM promotes repair of DNA damage induced by UV.

Next, we sought to further delineate the mechanism underlying NPM's involvement in DNA repair. Recent evidence has supported a role of NPM in the transcriptional regulation of E2F1 (Lin et al., 2006; Liu et al., 2007b). Coincidentally, E2F1 protein was reported to undergo accumulation after DNA damage (Lin et al., 2001; Ianari et al., 2004). As we uncovered the UV-induced NPM dephosphorylation, we also discovered a concomitant rise in E2F1 mRNA and protein levels in response to DNA damage (Supplemental Figure S4, A and B). Berton et al. (2005) further demonstrated in an in vivo context that such up-regulation promotes DNA repair. Based on these observations, we therefore set out to examine whether the role of dephosphorylated NPM in DNA repair is linked to E2F1 regulation.

First, quantitative RT-PCR was carried out to examine the expression profile of E2F1 gene in cells that harbor the wild-type, 3A, or 3D variant of NPM (Figure 4A). As shown previously (Lin et al., 2006), overexpression of the wild-type construct led to an increase in the E2F1 transcript level compared with the control. Interestingly, the hypophosphorylated form of NPM (3A) augmented this expression even further. Consistently, this NPM mutant could highly activate the E2F1 promoter activity, as shown by a reporter assay (Figure 4B). In contrast, expression of the phosphomimetic variant (3D) did not result in any significant alteration of the E2F1 mRNA level (Figure 4A). Another phosphodefective mutant that was altered on one of the MS-identified phosphorylation sites (S125A) also did not further activate E2F1 promoter activity compared with wild-type NPM (Supplemental Figure S4C), suggesting a lack of involvement of this residue in this functional aspect.

Our above-mentioned finding of PP1β's role in NPM dephosphorylation (Figure 2) also suggests its potential involvement in modulating E2F1 expression under DNA damage condition. To resolve this issue, we next assessed the effect of PP1β knockdown in a promoter reporter assay. Indeed, we found that reduction of PP1β by shRNA but not that of PP1α or PP1γ, abolished the UV-induced activation of E2F1 promoter (Figure 4C and Supplemental Figure S4D). Moreover, the increase in E2F1 protein levels in response to UV irradiation was lost upon depletion of PP1β (Figure 2B). In line with these observations, inactivation of phosphatase activity by calyculin A treatment also down-regulated the UV-induced E2F1 protein expression (Supplemental Figure S4E). Collectively, our data imply that the UV-responsive transcriptional activation of E2F1 may be mediated through the action of PP1β on NPM.

pRB is known to, via complex formation with E2F1, counterbalance the activation potential of E2F1 at the transcriptional level, and its occupancy of the E2F1 gene promoter has been linked to its gene regulation (Polager and Ginsberg, 2008). Recent studies have pointed to an involvement of NPM in the regulation of such transcriptional network (Lin et al., 2006; Liu et al., 2007b). Therefore, to elucidate how dephosphorylated NPM could activate E2F1 gene expression, we next assessed the binding of pRB to the E2F1 promoter in the context of DNA damage by using ChIP. After UV irradiation, we found a loss of pRB binding to the E2F1 binding site within the E2F1 gene promoter (Figure 5Aa, lanes 1 and 2). Interestingly, compared with the wild-type protein, expression of the nonphosphorylatable variant of NPM (3A) also led to a decrease of pRB promoter binding (Figure 5Ab, lanes 3 and 4). On the contrary, knockdown of PP1β expression augmented promoter recruitment (Figure 5Ac, lanes 5 and 6). Collectively, these findings are in close agreement with the above observations that linked PP1β-mediated dephosphorylation of NPM to E2F1 gene activation (Figure 4).

Consistent with an activation of E2F1, there was a disruption of the E2F1–pRB complex after UV treatment (Figure 5B). Intriguingly, coimmunoprecipitation experiments also showed that there was a coprecipitation of pRB with NPM upon UV irradiation, which was not detectable under mock treatment (Figure 5C and Supplemental Figure S5). Such association of NPM with pRB was further evident when Thr199/234/237 were substituted with Ala, compared with wild-type and the 3D mutant of NPM (Figure 5D), strongly suggesting that the NPM–pRB interaction depends on dephosphorylation of these residues. Consistently, upon siRNA-mediated reduction of intracellular PP1β, which was shown to increase the phosphorylation of NPM (Figure 2B), a loss of interaction between NPM and pRB was observed (Figure 5E). Conversely, lack of PP1β led to an up-regulated complex formation between E2F1 and pRB (Figure 5F). Together, these results demonstrate that the interaction of pRB with hypophosphorylated NPM, in response to UV damage and mediated by PP1β, relieves its transcriptional repression activity on E2F1.

Finally, to clarify whether E2F1 underlies the protective effect of NPM dephosphorylation in DNA damage response, we again performed the in vivo DNA repair assay in the context of E2F1 knockdown. Compared with control, cells lacking E2F1 exhibited diminished ability to repair UV-induced DNA damage (Figure 6A and Supplemental Figure S6). Moreover, although overexpression of NPM-3A enhanced the DNA repair activity under UV treatment, down-regulation of E2F1 expression effectively abolished such augmentation of DNA repair. Consistent with a repressive role on E2F1, overexpression of pRB also decreased the cellular capacity of removing thymine dimers (Figure 6B), mimicking the effect of E2F1 knockdown as well as the phospho-mimetic mutant 3D (Figure 3B). Furthermore, when expressed along with the NPM-3A mutant, pRB antagonized the repair-promoting activity of the hypophosphorylated NPM. Collectively, these results demonstrate the connection between NPM-mediated regulation of pRB-E2F1 pathway and the consequent activation of DNA repair.

To provide further mechanistic basis for the promotion of DNA repair in the context of NPM-mediated E2F1 up-regulation, we next evaluated possible expression of E2F1 target genes that encode DNA repair proteins. To this end, quantitative RT-PCR experiments demonstrated that the dephosphorylated form of NPM promotes the expression of several NER genes, such as RPA3, XPC, and DDB2, that were previously linked to E2F1 (Polager et al., 2002; Nichols et al., 2003; Prost et al., 2007; Lin et al., 2009; Figure 6C). Concomitantly, there was an increase in E2F1 occupancy of the XPC and DDB2 promoters in the presence of NPM-3A (Figure 6D), in line with the notion that these genes are regulated by E2F1 in a UV-dependent manner (Prost et al., 2007; Lin et al., 2009). Conversely, no significant changes in these E2F-dependent attributes were evident in the context of NPM–3D overexpression. Finally, we performed additional experiments (Supplemental Figure S7) to determine whether dephosphorylated NPM also could functionally interact with p53, another critical regulator of DNA damage response. As shown by the reporter assay, whereas UV treatment caused a dramatic up-regulation of p53 activity, overexpression of wild-type NPM exerted a negative regulation, consistent with previous observations (Maiguel et al., 2004). However, overexpression of NPM-3A, compared with the wild-type form, did not lead to discernible alteration in the p53 transactivation activity (Supplemental Figure S7A). Results of the flow cytometry-based cell cycle profiling also revealed that overexpression of NPM-3A failed to affect the progression of cell cycle and extent of apoptosis (Supplemental Figure S7B). These data therefore suggest that up-regulation of E2F1 and its downstream DNA repair genes may be a major mechanism through which dephosphorylated NPM favors repair of UV-induced DNA damage. Considered together, our findings provide strong evidence that, rather than acting as a repair protein per se, NPM serves as an integral component of the DNA damage response pathway that regulates expression of DNA repair genes.