Epigenetic suppression of human telomerase (hTERT) is mediated by the metastasis suppressor NME2 in a G-quadruplex–dependent fashion

Transcriptional activation of the human telomerase reverse transcriptase (hTERT) gene, which remains repressed in adult somatic cells, is critical during tumorigenesis. Several transcription factors and the epigenetic state of the hTERT promoter are known to be important for tight control of hTERT in normal tissues, but the molecular mechanisms leading to hTERT reactivation in cancer are not well-understood. Surprisingly, here we found occupancy of the metastasis suppressor non-metastatic 2 (NME2) within the hTERT core promoter in HT1080 fibrosarcoma cells and HCT116 colon cancer cells and NME2-mediated transcriptional repression of hTERT in these cells. We also report that loss of NME2 results in up-regulated hTERT expression. Mechanistically, additional results indicated that the RE1-silencing transcription factor (REST)–lysine-specific histone demethylase 1 (LSD1) co-repressor complex associates with the hTERT promoter in an NME2-dependent way and that this assembly is required for maintaining repressive chromatin at the hTERT promoter. Interestingly, a G-quadruplex motif at the hTERT promoter was essential for occupancy of NME2 and the REST repressor complex on the hTERT promoter. In light of this mechanistic insight, we studied the effects of G-quadruplex–binding ligands on hTERT expression and observed that several of these ligands repressed hTERT expression. Together, our results support a mechanism of hTERT epigenetic control involving a G-quadruplex promoter motif, which potentially can be targeted by tailored small molecules.

hTERT epigenetic control involving a G-quadruplex promoter motif, which potentially can be targeted by tailored small molecules.
Specialized DNA-protein assemblies called telomeres protect chromosome ends from being detected as DNA breaks (1,2). The ribonucleoprotein telomerase adds de novo repeats at the end of telomeres to maintain telomere length (3). Human telomerase comprises the catalytic reverse transcriptase (hTERT) 7 and an RNA component (hTR) that provides the template for addition of telomeric repeats (4,5). Lack of telomerase results in shortening of telomeres because of the end replication problem (6), and cells with critically short telomeres activate the DNA damage response, leading to cell cycle arrest or apoptosis (7,8). This is the case in most normal somatic cells, which lack telomerase. Most cancer cells, however, have high levels of telomerase, and telomere length is maintained for initiation and survival of tumors (9). In normal cells, the limiting factor for telomerase activity is the level of hTERT mRNA, which is under strong transcriptional control (10). In contrast, in about 85% of all cancers, hTERT expression is reactivated (11), leading to malignant transformation and aggressive metastasis in many cases (12). The molecular mechanisms that underlie hTERT reactivation from otherwise tight transcriptional control in normal somatic cells remain poorly understood.
In this context, the metastasis suppressor non-metastatic 2 (NME2; also known as nm23-H2) is of interest (13). Human nm23 has several isoforms; of these, H1 (or NME1) and H2 are the most studied (14 -16). The role of NME2 in metastases suppression is well-described: overexpression of NME2 results in reduced metastasis of human oral squamous carcinoma, breast carcinoma, and murine melanoma cells (17)(18)(19), and the level of NME2 expression negatively correlates with advanced/ metastatic stages in several tumor types (20). Notably, independent studies reported NME2-mediated transcription regulation of c-myc where association of NME2 to a G-rich sequence motif within the nuclease-hypersensitive element of the c-myc promoter was revealed (21). NME2 was also reported to regulate PDGF-A and vinculin transcriptionally, supporting its role as a regulatory factor (22,23).
Herein we show that transcription of hTERT remains repressed in the presence of NME2, and loss of NME2 results in up-regulation of hTERT expression. NME2 binds to the hTERT core promoter, and the REST repressor complex associates with the hTERT promoter in an NME2-dependent manner. Results also revealed that the presence of an intact G-rich DNA secondary structure G-quadruplex (G4) motif in the hTERT core promoter was required for association of NME2 and the REST repressor complex at the hTERT promoter. Notably, in the presence of NME2 and the REST repressor complex, epigenetic alterations restricted permissiveness of the hTERT promoter. Because altered NME2 has been detected in multiple cancer tissues (14,(17)(18)(19), it is of interest to understand the mechanisms underlying low NME2 and enhanced hTERT expression/activation.

NME2 associates with the hTERT core promoter and transcriptionally represses hTERT
We noted a putative NME2-binding site on the hTERT core promoter based on a previously reported motif from NME2 chromatin immunoprecipitation (ChIP)-sequencing experiments (24). Here we performed ChIP-PCR, with primers (spanning from ϩ40 to Ϫ230 bp with respect to the hTERT transcription start site) flanking the putative NME2-binding site, first in HT1080 fibrosarcoma cells and then in HCT116 colon cancer cells to confirm NME2 occupancy at the hTERT promoter ( Fig. 1a and supplemental Fig. S1a). To test the functional significance of the NME2 occupancy, endogenous hTERT expression was checked in NME2-overexpressed or -silenced conditions in HT1080 and HCT116 cells. We found clear repression and an increase in hTERT expression upon NME2 overexpression or silencing, respectively, and similar changes in hTERT protein levels ( Fig. 1, b and c, and supplemental Fig. S1b). The core promoter of hTERT was cloned into a luciferase reporter, and promoter activity was measured under NME2-altered conditions in HT1080 and HCT116 cells. NME2 expression and hTERT promoter activity were found to be inversely correlated (Fig. 1, d and e). Together, these data suggested NME2-mediated transcriptional repression of hTERT in cancer cells.
This was further supported by fluorescence-activated cell sorting (FACS) and immunofluorescence analysis using stable NME2-overexpressing cells, which showed reduced expression of hTERT relative to vector-transformed cells ( Fig. 1f and supplemental Fig. S1, c and d). Telomerase enzymatic activity was measured under NME2-overexpressed or -silenced conditions using a telomere repeat amplification protocol, and a reduction or an increase in telomerase activity, respectively, was observed (supplemental Fig. S1e). Based on this, we asked whether NME2 repressed hTERT in normal primary cells. In primary lung fibroblast MRC5 cells, we found repression or enhanced hTERT expression and promoter activity under NME2-overexpressed or -silenced conditions, respectively ( Fig. 1, g and h).
To validate the occupancy of NME2 on the hTERT promoter, an oligonucleotide pulldown assay was performed. A putative 12-mer (Ϫ115 to Ϫ127 bp from hTERT transcription start site) representing the NME2 DNA-binding motif (23) or its mutated/scrambled negative controls were used. Pulldown using anti-NME2 antibody showed specific binding to the unaltered (wild-type (WT)) relative to mutated (MUT) or scrambled motifs (Fig. 1i). Furthermore, a reporter plasmid in which the mutant NME2 motif was inserted did not suppress promoter activity in the presence of NME2 in HT1080 and HCT116 cells (Fig. 1j). Together, these results suggested that NME2 binding was required for transcriptional repression of hTERT.
Next, we checked the role, if any, of NME1, a close homolog of NME2, in transcriptional repression on hTERT. hTERT expression and protein levels remained unaltered when NME1 was specifically silenced (supplemental Fig. S1, f and g); hTERT promoter activity also was not affected upon NME1 silencing in HT1080 cells (supplemental Fig. S1h).

NME2 maintains repressive chromatin at the hTERT promoter by engaging the REST-LSD1 repressor complex
To understand the mechanism underlying NME2-mediated hTERT repression, we asked whether other binding partners of NME2 were involved. To find possible nuclear interacting partners of NME2, we used LC-MS/MS after immunoprecipitation with anti-NME2 antibody from nuclear extract of HT1080 cells. Using a stringent analysis cutoff of 1% false discovery rate, 120 interacting partners of NME2 were detected (supplemental Fig. S2a and Table S1) that belonged to several important classes including chromatin remodeling (Gene Ontology analysis; supplemental Fig. S2b). Of these, to specifically understand possible mechanisms underlying NME2-induced hTERT repression, we focused on chromatin modifiers HDAC1 and HDAC2. Interaction of NME2 with both HDAC1 and HDAC2 was further confirmed using co-immunoprecipitation (co-IP) experiments (Fig. 2a).
HDAC1 and HDAC2 are integral members of the REST repressor complex that engages the lysine-specific demethylase LSD-1 (25,26). Therefore, we checked whether NME2 interacts with REST and LSD1. Co-IP with anti-NME2 antibody showed clear enrichment of REST and LSD1 in the anti-NME2-immunoprecipitated fraction compared with its isotype control IgG (Fig. 2a). In addition, reverse co-IP with anti-REST or anti-LSD1 confirmed NME2 interaction with REST-LSD1 (supplemental Fig. S2c).
Next, we checked for REST and LSD1 occupancy on the hTERT promoter and determined whether this was NME2-dependent. Both REST and LSD1 ChIPs showed occupancy on the hTERT promoter where the ChIP signals were significantly reduced under NME2-silenced conditions (Fig. 2, b and c, and supplemental Fig. S2d). We also found HDAC1 and HDAC2 occupancy on the hTERT promoter to be dependent on NME2 Epigenetic regulation of hTERT (Fig. 2d). Moreover, sequential ChIP for NME2-REST and NME2-LSD1 further supported co-occupancy of NME2-REST-LSD1 complex on the hTERT promoter in HT1080 and HCT116 cells (Fig. 2, e and f).
Next, we asked whether NME2-mediated engagement of REST-LSD1 affected hTERT expression. siRNA against REST attenuated NME2-induced repression of hTERT expression, which was further derepressed using the LSD1 inhibitor tranylcypromine (TCP) (Fig. 2g). Silencing of all three, NME2, REST, and LSD1, led to a further increase in hTERT expression (Fig.  2g). An increase in hTERT expression and promoter activity was also evident upon inhibition of REST and LSD1 independently or together (supplemental Fig. S2, e and f). In addition, histone methylation ChIP results showed an increase in histone activation marks (H3K4me2 and H3K4me) following silencing of REST (supplemental Fig. S2g). Taken together these results show role of NME2 in promoting repressive chromatin at the hTERT promoter through the REST-LSD1 complex to suppress hTERT expression.
Conversely, we expected permissive chromatin at the hTERT promoter upon silencing NME2. We checked histone activation marks H3K9ac, H3K4Me2, and H3K4Me and found that all three marks were clearly enhanced in cells treated with siRNA against NME2 relative to control cells (si-control), whereas there was no significant change in total histone H3 occupancy (Fig. 2h). (three biological replicates); * indicates a p value Ͻ0.05. g and h, hTERT expression (g) and promoter activity (h) were measured in primary MRC5 normal lung fibroblasts in NME2-overexpressed and -silenced conditions. Error bars represent S.E. (three biological replicates); * indicates a p value Ͻ0.05. i, oligo pulldown assay for NME2 using biotin-labeled oligos with WT, MUT, and scrambled (Scr) NME2 motifs. The lower panel represents the loading control. j, luciferase reporter assay for hTERT promoter activity after overexpressing NME2 along with wild-type and mutant NME2 motifs present in hTERT promoter construct. Error bars represent S.E. (three biological replicates); * indicates a p value Ͻ0.05.

Epigenetic regulation of hTERT NME2 mutants devoid of DNA-binding function are unable to repress hTERT transcription
Mutants of NME2 that are unable to bind DNA (N69H and R34A) were reported earlier (27,28). Using these mutants, we checked whether DNA binding by NME2 was required for hTERT repression. Both the mutants, NME2(N69H) and NME2(R34A), did not show any repression of hTERT expression or promoter activity in HT1080 cells (Fig. 3, a and b). This was further confirmed using stable NME2 knockdown cells (supplemental Fig. S3a). As expected, on stable knockdown of NME2, telomerase expression and promoter activity increased significantly (supplemental Fig. S3, b and c). hTERT repression and reduced promoter activity were observed on inducing HA-tagged NME2 (HA-NME2) but not when HA-NME2(N69H) or HA-NME2(R34A) were induced (Fig. 3, c and d, and supplemental Fig. S3d). Furthermore, using the mutant NME2(N69H), we checked whether occupancy of the REST-LDS1 complex was altered vis-à-vis wild-type NME2. First, using ChIP with HA antibody, we confirmed relatively reduced occupancy of the NME2(N69H) mutant compared with wild-type NME2 on the hTERT promoter (Fig. 3e). Next, using ChIP, we observed reduced occupancy of REST at the hTERT promoter in the case of NME2(N69H) compared with NME2 ( Fig. 3f) that was not due to any loss of REST interaction with the mutant NME2(N69H) as shown by co-IP experiments in HT1080 cells (Fig. 3g). Consistent with reduced REST occupancy, we further found both LSD1 and HDAC1 occupancy to be reduced at the hTERT promoter in the case of NME2(N69H) relative to wildtype NME2 (Fig. 3f). Together, these studies confirmed DNA binding by NME2 to be a key factor in occupancy of the REST-LSD1 complex at the hTERT promoter and hTERT repression by NME2.

G-quadruplex motif at the hTERT promoter was required for NME2 and REST-LSD1 occupancy at the hTERT promoter
Potential G4 (PG4)-forming sequences were reported in the hTERT promoter (29,51). Interestingly, we observed that the NME2-binding motif (Fig. 4a) was within also within a hTERT promoter PG4-forming sequence, and because NME2 and promoter G4 interactions have been reported before (21,30), we asked whether the G4 structure played any role in hTERT repression by NME2. Oligonucleotide representing the PG4forming sequence adopted mixed parallel/antiparallel G4 structures in solution (Fig. 4b). For further work, we designed nucleotide substitutions such that key bases required for G4 stability were mutated, but the NME2-binding motif remained intact. This resulted in a disrupted G4 structure in solution (Fig.  4b). Next, a luciferase reporter construct was made by substituting these bases (MUT-G4) using site-directed mutagenesis.
Upon expression of NME2, although promoter activity was repressed by ϳ50% in the case of the G4-containing construct, for MUT-G4 the observed repression was relatively less in HT1080 and HCT116 cells (Fig. 4c).
To test NME2 interaction with the G4, we used oligonucleotide pulldown with the G4 or the MUT-G4 sequence. On probing with anti-NME2 antibody, we found relatively reduced affinity of NME2 toward the MUT-G4 compared with G4 ( Fig.  4d and supplemental Fig. S4). In addition, we checked and observed that a significantly reduced amount of REST was pulled down in the case of MUT-G4, supporting NME2-REST interaction (Fig. 4d). Furthermore, we found reduced intracellular occupancy of NME2 at the hTERT promoter in the case of ectopically maintained MUT-G4 relative to G4 in ChIP with anti-NME2 antibody followed by PCR with plasmid-specific primers ( Fig. 4e; also called transient ChIP (31)). This was also confirmed by ChIP using anti-REST or anti-LSD1 where reduced occupancy was observed for both REST and LSD1 on the hTERT promoter in the case of MUT-G4 (Fig. 4f). Together, these results support roles of the G4 structural motif in NME2 binding and recruitment of the REST-LSD1 complex for hTERT repression.
We next studied the effect of G4-binding ligands on telomerase transcription. Because G4 ligands were noted to both activate and repress gene transcription (32,52), we screened 20 previously established intracellular G4-binding ligands (listed in supplemental Table S2). Of these, 11 ligands showed downregulation (ϳ50 -85%) of hTERT in HT1080 cells (Fig. 4g and supplemental Table S2). Five of the 11 ligands also showed more than 50% reduction in hTERT promoter activity in reporter assays (Fig. 4h), and the ligand-induced effect on promoter activity was lost in the case of all five ligands when the G4 motif was mutated in the reporter construct. This further supported a role of G4 -ligand interactions inside cells in hTERT repression (Fig. 4h).

Discussion
Multiple transcription factors including Sp1, c-Myc, p53, SP1, ETS, E2F, and AP1 have been reported to transcriptionally control hTERT expression (10,33,34). However, how these function to control the epigenetic state of the hTERT promoter is not precisely understood. For example, histone deacetylation repressed hTERT transcription, whereas treatment with the HDAC inhibitor trichostatin A resulted in activation of hTERT (35), and a recent report suggested that LSD1, using its demethylation activity, altered activating histone marks on the hTERT promoter to repress hTERT transcription (36). Although these studies indicated the participating regulatory factors, mecha-

Epigenetic regulation of hTERT
nisms of how the repressors/histone modifiers were engaged on the hTERT promoter have remained unclear.
Herein we show that NME2 transcriptionally represses hTERT through recruitment of the REST repressor complex, which includes REST, HDAC1, HDAC2, and LSD1. Using direct and sequential ChIP, we found that association of the REST repressor complex on the hTERT promoter was depen-dent on NME2. Moreover, co-immunoprecipitation experiments confirmed that NME2 interacted with members of the REST complex inside the nucleus. As a result, we observed that the hTERT promoter is maintained in a repressive chromatin state. In other words, these data suggest that the loss of NME2 results in permissive chromatin at the hTERT promoter and up-regulation of hTERT expression.

Epigenetic regulation of hTERT
Low NME2 has been associated with malignant transformation in multiple cancers (14,18). Independently, in oral, breast, and colon cancer, mechanisms of antimetastatic action of NME2 were suggested to be through induction of the epithelial phenotype (mesenchymal to epithelial transition) (37). Conversely, increased telomerase due to telomerase reactivation was noted in progression of various cancers including acute leukemia, breast, prostate, lung, and melanoma (38). Recent studies further reveal that point mutations in the hTERT promoter leading to increased hTERT mRNA synthesis is key to hTERT reactivation in several cancers (10,39). In this context, possible implications of low NME2 in up-regulation of hTERT transcription through epigenetic alteration of the hTERT promoter could be of interest in understanding mechanisms of telomerase activation. Although results suggest engagement of the repressor complex in an NME2-dependent manner, further studies will be required to understand how other transcription factors at the hTERT promoter influence this mechanism. NME2 was also noted to inhibit telomerase catalytic activity in vitro (40). Together with findings reported here, this indicates a possible dual role of NME2 in control of telomerase in cancer cells.
Multiple lines of evidence support a role of G4 motifs in gene expression. Prominent among these are enrichment of potential G4 forms within promoters across species (41)(42)(43) and evidence supporting their role in gene expression (44) including their presence in vivo and identification of proteins that function in association with G4s inside cells (45)(46)(47)(48). Interestingly, NME2 was reported to regulate transcription through interaction with a c-myc promoter G4 motif (21). Together, these data prompted a closer look when we noted that the hTERT promoter harbors G4-forming sequences that were implicated in possible hTERT expression (29). Our results support a role of hTERT promoter G4 motifs in hTERT repression. This encouraged us to test G4 ligand-mediated repression of hTERT expression, which in turn may open potential ways of controlling hTERT activation in cancer cells.

Cells and culture conditions
HT1080, HCT116, and MRC5 primary fibroblast cells were obtained from the American Type Cell Culture (ATCC) and maintained in modified Eagle's medium (MEM), Dulbecco's modified Eagle's medium low glucose (DMEM) with high glucose, and MEM, respectively, and supplemented with 10% fetal bovine serum at 37°C in 5% CO 2 .

ChIP
ChIP assays were performed as described (40) with the following antibodies: rabbit anti-HA antibody (Abcam, ab9110), anti-rabbit IgG (Sigma), anti-REST antibody (Millipore, 17641), anti-LSD1 (Abcam, ab17721), anti-HDAC1 (Abcam, ab7028), anti-HDAC2 (Abcam, ab51832), anti-H3K4me (Abcam, ab8895), anti-H3K4me2 (Abcam, ab7766), anti-H3K4me3 (Abcam, ab8580), anti-histone3 (Abcam, ab1791), and anti-H3k9ac (Abcam, ab10812). For sequential ChIP, samples were treated as described for ChIP (40) before adding 3-4 g of the first antibody and incubating overnight at 4°C followed by 50 l of pro-tein G-Sepharose beads for 4 h at 4°C on a rotator. Samples were then washed with washing buffers (low-salt buffers, highsalt buffers, and lithium chloride) and divided into two fractions. One part was processed further for ChIP-DNA elution, and 3-4 g of the second antibody was added to the second fraction and incubated again overnight at 4°C followed by incubation with 50 l of protein G-Sepharose beads. Beads were washed (washing buffers as above), and the final DNA samples were obtained using a phenol, chloroform, and isoamyl alcohol precipitation method (as for basic ChIP protocol). The ChIP PCR primers used for hTERT promoter were 5Ј-CCAGGCCG-GGCTCCCAGTGGAT-3Ј (forward) and 5Ј-GGCTTCCCAC-GTGCGCAGCAGGA-3Ј (reverse), and the amplicon length was 275 bp. For transient ChIP assays, cells were transfected with wild-type hTERT and mutant hTERT promoter constructs (2 g of plasmid/1 ϫ 10 6 cells). Transfected cells were harvested after 48 h, and ChIP was performed against REST, NME2, and LSD1 as described above. The eluted ChIP-DNA was checked for hTERT promoter amplification by using primers derived partially from the vector construct and from the hTERT promoter sequence.

Real-time PCR
Total RNA was isolated using TRIzol reagent (Invitrogen, Life Technologies) according to the manufacturer's instructions. RNA was quantified, and 2 g of RNA was used for cDNA preparation using an Applied Biosciences kit. The relative transcript expression level for genes was measured by quantitative real-time PCR using a SYBR Green-based method. Average -fold change was calculated by the difference in threshold cycles (Ct) between test and control samples. GAPDH gene was used as an internal control for normalizing the cDNA concentration of each sample. For TCP (concentration, 1 M)-treated cells, treatment was done after 24 h of transfection, and RNA was isolated 24 h post-treatment. G-quadruplex ligand (concentration, 2 M) treatment was done for 48 h.

siRNA transfection
SMARTpool ON-TARGET siRNAs against NME2, NME1, and REST were procured from GE Dharmacon and used according to the manufacturer's protocol.

Luciferase assay
The minimal promoter region of hTERT harboring the NME2 motif and G-quadruplex sequences was cloned into pGL3-basic vector and transfected into HT1080 and HCT116 cells using Lipofectamine 2000 (Invitrogen). Plasmid (pGL4.73) containing a CMV promoter driving Renilla luciferase was cotransfected as a transfection control for normalization. After 48 h, cells were harvested, and luciferase activities of cell lysate were measured using a Dual-Luciferase reporter assay kit (Promega). For TCP (concentration, 1 M)-treated cells, treatment was done after 24 h of transfection, and luciferase activity was measured 24 h post-treatment. G-quadruplex ligand (concentration, 2 M) treatment was done for 48 h.

Preparation of nuclear extracts
HT1080 cells grown in MEM supplemented with 10% FBS (Sigma) were collected and washed in cold 1ϫ PBS, and nuclear Epigenetic regulation of hTERT extract was isolated using a nuclear extraction kit (CelLytic, Sigma) according to the manufacturer's protocol.

Co-IP
For immunoprecipitation experiments, 500 g of nuclear extract was incubated for 4 h at 4°C with 4 g of anti-NME2 antibody (Kamiya, KM1121 MC412). Immunoprecipitation was performed using a Catch and Release co-immunoprecipitation kit (Millipore) according to the manufacturer's protocol.

Immunofluorescence microscopy
Cells were grown on coverslips and at 100% confluence were fixed with 4% paraformaldehyde by incubating for 10 min at room temperature. Cells were permeabilized with 0.5% Triton TM X-100 and treated with blocking solution (3% BSA in PBS) for 30 min at room temperature. After one PBS (1ϫ) wash, cells were treated with anti-hTERT antibody (1:100) overnight at 4°C. The next day cells were washed alternately with 1ϫ PBS and PBS with Tween 20 three times and probed with Alexa Fluor 594 for 1 h at room temperature. Cells were washed again alternately with PBS and PBS with Tween 20 three times and mounted with Prolong Gold antifade reagent with DAPI. Images were taken as maximum intensity projections on a Leica TCS-SP8 confocal microscope.

Flow cytometry (FACS)
After trypsinization, both stable GFP-NME2-expressing and corresponding GFP vector-transformed cells were fixed in 4% paraformaldehyde for 10 min. After washing in PBS, cells were permeabilized in 0.5% Triton X-100 in PBS for 10 min. After blocking in 3% BSA in PBS for 1 h at 4°C followed by washing in PBS, cells were incubated with primary anti-hTERT antibody (Abcam, ab94523) (1:100) at 4°C overnight. After washing in PBS, cells were incubated with Alexa Fluor 594-conjugated secondary antibody for 2 h at 4°C. As a negative control, cells were fixed and incubated with secondary antibody only without incubation with primary antibody. Finally, after washing in PBS, cells were resuspended in PBS for acquisition by flow cytometry (BD Biosciences FACSAria III) using the Cy3 channel (for Alexa Fluor 594) to detect hTERT expression.

Oligonucleotide pulldown assay
For pulldown assays, 2 g of biotinylated NME2 motif, motif mutant M1, wild-type G4, and MUT-G4 and 300 g of nuclear extract (prepared from HT1080 cells) were incubated for 1 h at room temperature in binding buffer. Then the whole complex was incubated with 60 l of streptavidin-agarose (Invitrogen) for 4 h at 4°C. In all experiments, the beads were washed three times with washing buffer (50 mM Tris-HCl (pH 7.5) and 150 mM NaCl), and the bound proteins were eluted by boiling in 5ϫ SDS sample buffer (20 mM Tris-HCl (pH 6.8), 10% glycerol, 4% SDS, 100 mM dithiothreitol, 4 mM EDTA, and 0.025% Coomassie Brilliant Blue R-250) and subjected to Western blot analysis using anti-REST antibody (Abcam) and anti-NME2 antibody (Kamiya).

Separation of immunoprecipitate and preparation of peptide mixtures
Protein identification using mass spectrometry was performed as follows. First, the immunoprecipitate was resolved by 12.5% SDS-PAGE followed by visualization with colloidal Coomassie Brilliant Blue. The bands that appeared specifically in the immunoprecipitate were excised manually using sterile blades. Further procedures including distaining and digestion with trypsin followed by extraction of peptides were performed manually. In brief, gel pieces were distained in 50 mM ammonium bicarbonate and acetonitrile (50%, v/v) and dehydrated with 95% (v/v) acetonitrile. The dehydrated gel pieces were soaked with a sufficient volume of trypsin solution (20 ng/l trypsin in 25 mM ammonium bicarbonate) and incubated at 37°C overnight. Upon digestion, the trypsinized solution was pooled with a further peptide extraction from gel pieces using 50% (v/v) acetonitrile containing 0.1% (v/v) acetic acid and incubating for 30 min at 37°C with constant shaking. The peptide extract collected as supernatant was pooled, vacuum-dried to about 10 l, and purified using a micro ZipTip (Millipore). The purified mixture was vacuum-dried, and peptides were reconstituted in 2% acetonitrile containing 0.1% (v/v) acetic acid and further analyzed by LC-MS/MS.

Identification of proteins using LC-MS/MS
Protein identification from peptide mixtures was performed on an LTQ Orbitrap Velos mass spectrometer (Thermo Scientific, Germany) equipped with a nanoelectrospray ion source coupled with an on-line Proxeon Easy nLC (Thermo Scientific). Briefly, the peptides were separated using an Acclaim PepMap 100 analytical column (C18; particle size, 3 m; 75-m inner diameter; 15 cm long; 100 Å; LC Packings, Germany) at a flow rate of 300 nl/min. Before separation, the peptide mixtures were enriched using an NS-MP-10 Biosphere precolumn (C18; particle size, 5 m; 100-m inner diameter; 20 mm long; pore size, 120 Å; Nano Separations, Netherlands). The peptides were eluted with a mixture of solvent gradient (2-5% buffer B in 1 min, 5-25% buffer B in 59 min, 25-40% buffer B in 10 min, and 100% buffer B in 8 min) of buffer A (2% acetonitrile containing 0.1% acetic acid) and buffer B (acetonitrile containing 0.1% acetic acid). The eluted peptides were electrosprayed into the mass spectrometer. The mass spectrometric data were collected in data-dependent mode to switch between Orbitrap MS and LTQ MS/MS acquisition automatically. Full-scan MS spectra in the m/z range from 300 to 1700 (resolution r ϭ 30,000) were acquired in the Orbitrap mass spectrometer. The data acquisi-Epigenetic regulation of hTERT tion method was set to isolate up to 20 of the most intense ions depending upon signal intensity for fragmentation in the linear ion trap using collision-induced dissociation. Target ions already selected for MS/MS were dynamically excluded for 60 s. The general conditions of the mass spectrometer for data collection were 1.6 -1.7-kV electrospray voltage and ion selection threshold of 2000 counts for MS/MS. For identification of proteins from MS data, an automated database search was performed using Proteome Discoverer 1.3.0.339 (Thermo Scientific) with the Sequest algorithm. A human UniProt FASTA database and a decoy database of 1% false discovery rate were used for identification. Two possible missed cleavages for trypsin enzyme specificity with a mass tolerance of 10 ppm (parent ion) and 0.8 Da (fragment ion) and methionine oxidation dynamics were considered for the database search. The identity of the proteins was confirmed based on high-confidence peptide identification containing at least two peptides per protein (with rank 1 peptides in proteins; XCorr score, Ն2.45 or 2.85 for doubly and triply charged peptides, respectively).

Site-directed mutagenesis
A QuikChange site-directed mutagenesis kit (Agilent technologies) was used to generate various mutants of NME2 gene. Mutagenesis reactions were performed according to the manufacturer's instructions with the wild-type c-DNA cloned in pRSET-A used as a template. The positive clones were screened for the mutation by sequencing. Note added in proof-An incorrect gel section was inadvertently used in the REST ChIP panel in Fig. 3f of the version of this article that was published as a Paper in Press on July 17, 2017. This error has now been corrected and does not affect the results or conclusions of this work.