NSun2 delays replicative senescence by repressing p27 (KIP1) translation and elevating CDK1 translation

NSun2 represses p27 translation by methylating the 5′UTR of p27 mRNA

The present study was prompted by our earlier findings that high levels of NSun2 lowered p27 abundance. As shown in Fig. 1A, overexpression of NSun2 in HeLa cells by transfection of plasmid pNSun2 reduced p27 protein levels by ~70% while silencing of NSun2 by transfection of small interfering (si) RNA increased p27 protein levels by ~4.2 fold. However, neither overexpression nor knockdown of NSun2 influenced p27 mRNA levels (Fig. 1B). These results suggested that the regulation of p27 by NSun2 may not involve altered p27 mRNA transcription or turnover. Because NSun2 had been shown to regulate the expression of p53, p16, E2F3, Bak1, ErbB2, and CDK1 by methylating the mRNAs encoding these proteins [26–28], we tested if NSun2 was capable of methylating p27 mRNA. To this end, the p27 mRNA fragments described in Fig. 2A were used for in vitro methylation assays (Materials and Methods). The p27 cDNA and p16-CR (coding region of p16 mRNA) were included as negative controls, while bacterial tRNA served as a positive control. As shown in Fig. 2B (left and right), tRNA, 5′UTR, 5′UTRa, and 5′UTRa1 were methylated, while p27 cDNA, p16-CR, p27-CR, p27-3′UTR, p27-5′UTRa2, and p27-5′UTRb were not methylated. Accordingly, the methylation site was located at the p27 5′UTR (positions 1-140).

To determine the formation of m5C or m6A in the methylated fragments, p27-5′UTRa1 methylated in vitro by using nonisotopic S-Adenosyl methionine (SAM) and NSun2, and unmethylated (same reaction without adding NSun2) p27-5′UTRa1 were subjected to MS-HPLC analysis. As shown, m5C was detected in the methylated p27-5′UTRa1 fragment (Supplemental Fig. S1). Identification of m6A from the methylated p27-5′UTRa1 did not show any positive results (not shown). To further identify the methylation site by NSun2, in vitro methylated p27-5′UTRa1 was subjected to bisulfate sequencing analysis. As shown in Fig. 2C, among all of the clones sequenced, C64 was detected in ~92.6% of positive clones. Mutation of C64 (p27 5′UTRΔ) nearly abolished the effect of NSun2 on methylating the p27 5′UTR (Fig. 2D), indicating that C64 is a major methylation site.

To test whether NSun2 was capable of methylating p27 5′UTR in cells, HeLa cells were transfected with a pGL3-derived reporter bearing the p27 5′UTR. Twenty-four hours later, cells were further transfected with a vector expressing NSun2 or with NSun2 siRNA and cultured for an additional 48 h. RNA was isolated and subjected to bisulfate sequencing analysis to assess the rate of C64 methylation. As shown, knockdown of NSun2 reduced the methylation of C64, whereas overexpression of NSun2 increased the rate of C64 methylation (Fig. 2E). These results indicate that NSun2 was able to methylate p27 mRNA in cells.

To further test whether methylation of p27 5′UTR by NSun2 influenced p27 expression levels, pGL3-derived reporters bearing fragments of p27 mRNA were constructed (Fig. 3A, schematic). HeLa cells were transfected with each of these reporters and 24 h later, cells were transfected with control or NSun2-directed siRNAs and cultured for an additional 48 h. Cellular lysates then were collected and subjected to Western blot analysis to assess NSun2, Firefly luciferase, Renilla luciferase, and GAPDH protein levels. As shown in Fig. 3B, knockdown of NSun2 elevated the levels of reporter protein expressed from pGL3-5′UTR, but not that expressed from pGL3, pGL3-CR, pGL3-3′UTR, or pGL3-5′UTRΔ, indicating that methylation by NSun2 at C64 selectively repressed the expression of p27. Because NSun2 regulates the expression of p27 without influencing the levels of p27 mRNA (Fig. 1), we further asked if methylation by NSun2 regulated the translation of p27. To this end, in vitro-transcribed reporter transcripts Luciferase (Luc), Luc-5′UTR, and Luc-5′UTRΔ were methylated in vitro by NSun2 or kept unmethylated. These transcripts then were used for in vitro translation assays and reporter activity was used as readout of the efficiency of translation. As shown in Fig. 3C (left), methylation by NSun2 reduced the reporter activity of Luc-5′UTR (by ~59%), but not that of Luc and Luc-5′UTRΔ. Therefore, methylation of p27 5′UTR by NSun2 represses the expression of p27 at the level of translation.

To further investigate the mechanisms underlying the translational regulation of p27 by NSun2, the polysomal presence of p27 mRNA in cells with overexpressed or silenced NSun2 was examined. As shown in Fig. 3D, overexpression of NSun2 decreased the presence of p27 mRNA in the polysomal fraction (by ~66%, p=0.0005) but increased it in the non-polysomal fraction (by ~2.0 fold, p=0.0254). On the other hand, knockdown of NSun2 increased the presence of p27 mRNA in the polysomal fraction (by ~60%, p=0.0005) but decreased it in the non-polysomal fraction (by ~2.0 fold, p=0.0024). Taken together, these findings indicate that the assembly of p27 mRNA in the polysome was repressed by NSun2-mediated methylation, in turn repressing the translation of p27.

By binding to the p27 5′UTR, RNA-binding proteins HuR and CUGBP1 repress p27 translation [15–16]. These findings raised the question that HuR or CUGBP1 might influence the function of NSun2 in repressing the translation of p27. As shown in Supplemental Fig. S2A by RNA pull-down assays, HuR associated with the p27 5′UTR, 5′UTRb, and 3′UTR, while CUGBP1 only associated with the p27 5′UTR and 5′UTRb, in agreement with previous findings [15, 16]. NSun2 interacted with the p27 5′UTR and 5′UTRa, consistent with the results that NSun2 methylated p27 5′UTR and 5′UTRa (Fig. 2B). Although NSun2 was capable of interacting with the p27 3′UTR, this interaction did not lead to the methylation of p27 3′UTR (Fig. 2B). Furthermore, methylation by NSun2 did not influence the association of HuR or CUGBP1 with the p27 mRNA (Supplemental Fig. S2B), and the association of NSun2 with p27 mRNA did not influence the association of HuR and CUGBP1 with p27 mRNA or vice versa (Supplemental Fig. S2C-E). Moreover, knockdown of NSun2 increased the activities of reporters pGL3-5′UTR and pGL3-5′UTRa, but not those of pGL3-5′UTRb or pGL3-5′UTRΔ (Supplemental Fig. S2F). In contrast, knockdown of HuR or CUGBP1 increased the activities of reporters pGL3-5′UTR, pGL3-5′UTRb, and pGL3-5′UTRΔ, but not that of pGL3-5′UTRa (Supplemental Fig. S2F). These results suggest that the translational repression of p27 by NSun2-mediated mRNA methylation is independent of the effects elicited by HuR or CUGBP1.

NSun2 regulation of p27 and CDK1 impacts upon replicative senescence

Apart from p27, NSun2 was also found to regulate the expression of senescence-associated proteins p53, p16, CDK1, and CDC25C [26–28]. To further investigate whether NSun2 regulated the process of replicative senescence, we assessed the levels of proteins NSun2, p53, p16, CDK1, p27, CDC25C, and GAPDH in early-passage [proliferating, ‘Young’ (Y), ~PDL 28], middle-passage [middle (M), ~PDL 40], and late-passage [senescent (S), ~PDL 55] human diploid fibroblasts (2BS) by Western blot analysis. As shown in Fig. 4A (left), the levels of p16 and p27 increased with replicative replicative senescence, while the levels of p53 remained unchanged. However, the levels of proteins NSun2, CDK1, and CDC25C were reduced with replicative senescence. The inverse correlation between NSun2 and p27 levels was also observed in another model of senescence, proliferating and senescent IDH4 cells (Fig. 4A, right). In addition, the levels of CDK1 mRNA decreased (by ~45% in 2BS cells, p=0.0000; by ~63% in IDH4 cells, p=0.0000) but p27 mRNA levels remained unchanged during senescence in both cell types (Fig. 4B). Both endogenous p53 and p16 are constitutively expressed in IDH4 cells, regardless of whether they are maintained in a proliferative state or they are induced to undergo senescence (Fig. S3). Methylated RNA-specific PCR analysis revealed that the levels of methylated CDK1 and p27 mRNAs decreased in the process of replicative senescence (by ~76% for p27 mRNA; by ~46% for CDK1 mRNA) (Fig. 4C), while the levels of unmethylated CDK1 and p27 mRNAs increased in replicative senescence (by ~2.7 fold for p27 mRNA, p=0.0023; by ~5.9 fold for CDK1 mRNA). These results indicate that NSun2-mediated RNA methylation may be able to regulate the expression levels of CDK1, p27, and CDC25C, but not those of p53 and p16, as cells progress to senescence.

To test this idea using a different approach, 2BS cells were stably infected with lentiviruses expressing either NSun2 or NSun2 shRNA. As shown in Fig. 5A, infection with the NSun2-expressing lentiviruses reduced the levels of p27 by ~70%, while infection with NSun2 shRNA-expressing lentiviruses increased the levels of p27 by ~5.1 fold. In contrast, the levels of proteins CDK1 and CDC25C increased in cells with overexpressed NSun2 (by ~2.8 fold for CDK1; by ~1.5 fold for CDC25C) but decreased in cells with silenced NSun2 (by ~70% for both CDK1 and CDC25C). However, the levels of proteins p53 and p16 increased moderately in cells with overexpressed NSun2 and decreased only moderately in cells with silenced NSun2 (Fig. 5A). These modest changes may be an indirect reflection of the fact that 2BS cells with overexpressed NSun2 or silenced NSun2 have a constitutively altered senescence phenotype, perhaps as a result of compensatory changes in response to the permanent alterations in NSun2 levels. In agreement with the findings in Fig. 1B and in previous studies [28], neither overexpression of NSun2 nor knockdown of NSun2 altered the levels of p27 and CDK1 mRNAs in a measurable manner (Fig. 5B). In addition, the levels of cellular methylated p27 mRNA increased in cells with overexpressed NSun2 but decreased in cells with silenced NSun2 (Supplemental Fig. S4).

Finally, knockdown of NSun2 inhibited cell growth and increased the proportions of cells displaying the protein marker senescence-associated (SA)-β-galactosidase (~37.5% vs. ~91.0%), while overexpression of NSun2 promoted cell growth and reduced SA-β-galactosidase activity (~33.5% vs. ~10.6%). We next tested if overexpression of the p27 5′UTR fragment could function as a decoy fragment and rescue the effect of NSun2 overexpression in modulating p27, CDK1, and CDC25C levels, in promoting cell growth, and in delaying replicative senescence. As shown in Figure 6, overexpression of the p27 5′UTR fragment in 2BS cells by infecting 2BS cells with a lentivirus expressing luc-5′UTR (which increased p27 5′UTR levels by ~944 fold) rescued the effect of NSun2 overexpression in influencing the expression of p27, CDK1, and CDC25C (Figure 6A), in promoting DNA replication (progression through the S phase; Figure 6B), and in delaying replicative senescence (Control vs. pNSun2, ~46.4% vs. ~20.0%; control+pHBLV-5′UTR vs. pNSun2+pHBLV-5′UTR, ~51.5% vs. ~46.8%) (Figure 6C). Overexpression of the p27 5′UTR modestly increased the SA-β-galactosidase activity in control vector-transfected cells (~46.4% vs. ~51.5%), although the alteration was not statistically significant. In sum, the influence of NSun2 on the expression of p27, CDK1, and CDC25C has a robust impact upon replicative senescence.

Cell culture, transfection, and SA-β-galactosidase activity

Human IDH4 fibroblasts were generously provided by J. W. Shay and cultured in Dulbecco's modified Eagle's medium (DMEM) (Invitrogen) supplemented with 10% fetal bovine serum, 100 units/ml penicillin, 100 μg/ml streptomycin, and dexamethasone (Dex, 1 μg/ml) for constitutive expression of SV40 large T antigen to suppress senescence and stimulate proliferation [29]. Proliferating (‘young’) and senescent IDH4 cells were obtained by incubating cells in charcoal-stripped serum (10%) in the presence (Young) or absence (Senescent) of Dexamethasone (Dex, 1 μg/ml) for 5 days. Early- [Proliferating (P), Young, ~28 population doublings (pdl)], middle- [(M), ~40 pdl], and late-passage [(S) Senescent, ~55 pdl] human diploid 2BS fibroblasts (National Institute of Biological Products, Beijing, China), and HeLa cells were cultured in DMEM supplemented with 10% fetal bovine serum, 100 units/ml penicillin, and 100 μg/ml streptomycin, at 37°C in 5% CO₂. All plasmids were transfected using lipofectamine 2000 (Invitrogen) and collected 48 to 72 h after transfection for further analysis. SA-β-galactosidase activity (β-gal staining) was performed as described previously [42].

Transcript preparation

cDNA was used as a template for PCR amplification of the p27 fragments. All 5′ primers contained the T7 promoter sequence (CCAAGCTTCTAATACGACTCACTATAGGGAGA). To prepare templates for the p27 5′UTR (positions 1-472), p27 5′UTR-A (positions 1-210), p27 5′UTR-B (positions 211-472), p27 5′UTRa1 (positions 1-140), p27 5′UTR a2 (positions 141-210), p27 CR (positions 473-1069), and p27 3′UTR (positions 1070-2290), we used the following primer pairs: (T7) CTTCTTCGTCAGCCTCCC and CTTTCTCCCGGGTCTGCA for 5′UTR, (T7) CTTCTTCGTCAGCCTCCC and AGCGGAGAGGGTGGCAAAGCCC for 5′UTR-A, (T7) TGCCTGGTCCCCTCTCCTCT and CTTTCTCCCGGGTCTGCA for 5′UTR-B, (T7) CTTCTTCGTCAGCCTCCC and CTCTCCAAACCTTGCCGGCGTC for 5′UTRa1, (T7) CGGCTGGGTTCGCGGGAC and AGCGGAGAGGGTGGCAAAGCCC for 5′UTRa2, (T7) ATGTCAAACGTGCGAGTGTC and TTACGTTTGACGTCTTCTGAG for CR, (T7) ACAGCTCGAATTAAGAATATG and TAACAAAAGAGGGGAAAACCTATTC for 3′UTR. These PCR products were transcribed in vitro following the manufacturer's instructions (Thermo).

The p27 5′UTR mutants were prepared by overlapping PCR. The p16 3′UTR fragment was described previously [27].

Western blot analysis

Monoclonal anti-CUGBP1, monoclonal anti-p53, monoclonal anti-p16, monoclonal anti-CDC25C, and monoclonal anti-HuR were from Santa Cruz Biotechnology. Polyclonal anti-NSun2 and monoclonal anti-GAPDH were from Abcam. After secondary antibody incubations, signals were detected by Odyssey Imaging System (Gene Company Limited) following the manufacturer's instruction and quantitated by densitometric analysis with Image J software.

Knockdown of NSun2, HuR, and CUGBP1

To silence NSun2, HuR, or CUGBP1, cells were transfected with siRNAs (10 nmol/L each) targeting NSun2 (AGAUGUUAAGAUACUGUUGACCC), HuR (AAGAGGCAAUUACCAGUUUCAUU) or CUGBP1 (GAGCCAACCUGUUCAUCUAUU), or with a control siRNA (UUGUUCGAACGUG UCACGUTT) using RNAiMAX (Invitrogen). Unless otherwise indicated, cells were collected for analysis 48 h after transfection. All knockdown interventions caused less than 1% cell death (by FACS analysis, data not shown).

RNA isolation and real-time qPCR analysis

Total cellular RNA was prepared using the RNeasy Mini Kit (Qiagen). For reverse-transcription (RT) followed by real-time, quantitative (q)PCR analysis of CDK1 and p27, we used following primer pairs: AAATGTGTGTAGGTCTCAC and ATGATTTAAG CCAACTCAAA for CDK1 mRNA, and GGCTCCGGC TAACTCTGA and TCTTCTGTTCTGTTGGCTCTTT for p27 mRNA.

Constructs and reporter gene assays

For construction of the pGL3-derived reporter vectors bearing the p27 5′UTR and 5′UTR mutating C64 (C-G, 5′UTRΔ), the 5′UTR or 5′UTRΔ fragment was amplified by PCR by using primer pairs CCCAAGCTTGGGCTTCTTCGT CAGCCTCCC and CATGCCATGGCATGCTTTCT CCCGGGTCTGCA and inserted between the Hind III and Nco I sites of pGL3-promoter vector (Promega). For constructing the pGL3-derived reporter vectors bearing the p27 5′UTRa, 5′UTRb, CR, and 3′UTR, 5′UTRa, 5′UTRb, CR, and 3′UTR were amplified by PCR using primer pairs CCCAAG CTTCTTCGTCAGCCTCCC and CATGCCAGCGGA GAGGGTGGCAAAGCCC for 5′UTRa, CCCAAGTGCCTGGTCCCCTCTCCTCT and CATGCCCTTTCTCCCGGGTCTGCA for 5′UTRb, CCCAAGCTGTCAAACGTGCGAGTGTC and CATGCCTTACGTTTGACGTCTTCTGAG for CR, and CCCAAGACAGCTCGAATTAAGAATATG and CATGCCTAACAAAAGAGGGGAAAACCTATTC for 3′UTR. Each of these fragments was inserted between HindIII and NcoI sites of the pGL3-promoter vector (Promega). The pcDNA 3.1 vector expressing NSun2 was described previously [27].

For reporter gene assays, each of the pGL3-derived vectors was co-transfected with pRL-CMV vector by Lipofectamine 2000 (Invitrogen). Forty eight hours after transfection, cell lysates were collected and the firefly and renilla luciferase activities were measured with a double luciferase assay system (Promega) following the manufacturer's instructions. All firefly luciferase measurements were normalized to renilla luciferase measurements from the same sample.

Lentiviruses and stable expression of ectopic genes

For the construction of lentivirus vectors expressing NSun2 (pHBLV-NSun2), the coding region of NSun2 was amplified by PCR by using primer pairs CGGAATTCATGGGCGGCGGTCGCGGGGT and CGGGATCCGGTCACCGGGGTGGATGGACC and inserted between EcoRI and BamHI sites of pHBLV-CMVIE-IRES-Puro vector (HANBIO Biotechnology Co., Ltd, Shanghai, China). For construction of the lentivirus vector expressing Luc-5′UTR (pHBLV-luc-5′UTR), the Luc-5′UTR fragment was amplified from the pGL3-5′UTR reporter by PCR by using primers CCGGAATTCTTCTTCGTCAGCCTCCC and CGGGATCCCGTTACACGGCGATCTTTCCGCCCT and inserted between EcoRI and BamHI sites of pHBLV-CMVIE-IRES-Puro vector. For construction of the lentivirus vector expressing NSun2 shRNA (pHBLV-shNSun2), complementary fragments GATCCGCGGCCTCATCATAAGATCTTAGATATTCAAGAGATATCTAAGATCTTATGATGAGGCCGTTTTTTC and AATTGAAAAAACGGCCTCATCATA AGATCTTAGATATCTCTTGAATATCTAAGATCTTATGATGAGGCCGCG were annealed and inserted between the BamHI and EcoRI sites of pHBLV-U6-Zsgreen-puro vector (HANBIO Biotechnology Co., Ltd, Shanghai, China). Viruses were packaged in 293T cells following the instructions from the manufacturer (HANBIO Biotechnology Co., Ltd, Shanghai, China). Stably transfected cells (2BS) were selected in the presence of puromycin (0.5 μg/ml) for 5 days.

Preparation of polysomal fractions

A total of 20 million cells were incubated for 15 min with 100 mg/ml cycloheximide, and total lysates (500 μl) were layered onto a cushion of 30% sucrose in ice-cold buffer containing 20 mM HEPES (pH 7.4), 50 mM potassium acetate, 5 mM magnesium acetate, 1 mM dithiothreitol, 1 unit of RNasin per μl, 1 μg of leupeptin per ml, 1 μg of aprotinin per ml, and 0.5 mM phenylmethylsulfonyl fluoride. After centrifugation (Beckman SW40; 100,000 × g for 2 h, 4°C), RNA from the supernatant (nonpolysomal fraction) and the pellet (polysomal fraction) was prepared and used for RT-qPCR analysis.

In vitro methylation assays. For in vitro methylation assay, His-tagged NSun2 was prepared as described previously [27]. Briefly, reaction mixtures (50 μl) containing 0.2 nM His-NSun2, 0.01 nM RNA, and 1 μCi of ³H-labeled S-adenosyl-L-methionine (SAM, Amersham Bioscience) in reaction buffer (5 mM Tris HCL[pH 7.5], 5 mM EDTA, 10% glycerol, 1.5 mM dithiothreitol, 5 mM MgCl₂) supplemented with inhibitors (leupeptin [1 μg/ml], aprotinin [1 μg/ml], 0.5 mM phenyl-methylsulfonyl fluoride, and RNasin [5 U/μl]), incubated for 45 min at 37°C, as described [27]. E. coli tRNA (0.01 nM, Sigma) and p16 CR (0.01 nM) served as a positive and negative controls, respectively. Unincorporated ³H SAM was removed by using Qiaquick Spin Columns (Qiagen) and incorporated radioactivity was measured by liquid scintillation counting. The non-isotopic methylated RNA fragments were prepared using cold SAM (Sigma) and in vitro-transcribed RNA fragments under similar conditions.

Bisulfate RNA sequencing

This method could identify the methylation site (m5C) within an RNA fragment shorter than 150 nt. Briefly, fragment 5′UTRa1 (positions 1-140), which was methylated by NSun2, was amplified by using primers (T7) CTTCTTCGTCAGCCTCCCTT and CCTCCATTCC CAACAAACCTTGCCGGCGTCGGAGTCGCAGAGCCGTGA.

This fragment (1 μg) was transcribed in vitro and methylated by using non-isotopic SAM. Samples then were dissolved in 10 μl of RNase-free water and mixed with 42.5 μl 5 M sodium bisulfate (Epitect) and 17.5 μl DNA protection buffer (Epitect), incubated in 70°C for 5 minutes and 60°C for 60 min, repeating for 3-5 cycles. After desalting using Micro Bio-spin6 columns, samples were de-sulfonated by 1 M Tris (pH 9.0, 1/1, V/V) at 37°C for 1 h, followed by ethanol precipitation. The bisulfate-converted fragments (0.2 μg) were reverse-transcribed by RevertAid First Strand cDNA Synthesis Kit (Thermo) using primer GTCGTATCCAGTGCAGGGTCCGAGGTATTCGCACTGGATACGACCCTCCATTCCCAACAAAC. The reverse-transcribed products (cDNA) were amplified by PCR by using primers GGGAGATTTTTTTGTTAGT and GCAGGGTCCGAGGTATTC. The PCR products were inserted into the pGEM-T Easy Vector System (Promega). The plasmids purified from single clones were sequenced. The sequences were aligned with the corresponding p27 mRNA sequence and the cytosines retained were considered to be methylated.

Methylation-specific RT-PCR

Two micrograms (2 μg) of cellular RNA or 1 μg of in vitro-methylated RNA fragment was bisulfite-converted as described in ‘bisulfate RNA sequencing’. The converted p27 5′UTR was reverse-transcribed by using primer GTCGTATCCAGTGCAGGGTCCGAGGTATTCGCACTGGATACGACCCAACCACTCTCCAAACCTT. For methylation-specific p27 mRNA RT-PCR analysis, primer pairs TAATAAGGTTAAAAAAGGGGGTTTG TTTGGGC and GCAGGGTCCGAGGTATTC (for methylated fragment) as well as AAAAAAGGGGG TTTGTTTTAAT and GCAGGGTCCGAGGTATTC (for unmethylated fragment) were used.

In vitro translation assays

For in vitro translation assays, a cell-free translation system (Promega) in rabbit reticulocyte lysate (RL) was used. Luciferase (Luc) transcript was in vitro transcribed from pGL3 by using primer pairs (T7) ATGGAAGACGCCAAAAA CAT and TTTTTTTTTTTTTTTTTTTTTTTTTTTTT TACACGGCGATCTTTCCGCCCT. Luc-5′UTR and Luc-5′UTRΔ transcripts were transcribed in vitro from pGL3-5′UTR and pGL3-5′UTRΔ reporters by using primer pairs (T7) CTTCTTCGTCAGCCTCCCTT and TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTACACG CGATCTTTCCGCCCT. Transcripts (0.01 nM) were either methylated by NSun2 or kept untreated. The methylated and non-methylated transcripts were used for in vitro translation assays. The translation efficiency was determined by measuring the activity of firefly luciferase.