The NAD+ Synthesis Enzyme Nicotinamide Mononucleotide Adenylyltransferase (NMNAT1) Regulates Ribosomal RNA Transcription

Plasmids and Cell Lines

NMNAT1 cDNA was obtained by RT-PCR and cloning from a human cell line. NML plasmid was provided by Dr. Junn Yanagisawa. All constructs used in this study are of human origin. Cells were maintained in Dulbecco's modified Eagle's medium (DMEM) with 10% fetal bovine serum. For glucose starvation treatment, cells were washed twice with PBS before culturing in DMEM with 10% dialyzed FBS and 0 mm or 25 mm glucose. Transfection of H1299 cells was performed using standard calcium phosphate precipitation protocol. U2OS cells with stable expression of NMNAT1 were generated by infection with pLenti-NMNAT1 virus followed by Zeocin selection (ViraPower T-REX lentiviral expression system; Invitrogen). ATP concentration was determined using a commercial kit (BioVision).

RNA Interference (RNAi)

Cells were transfected with 50 nm control siRNA (AAUUCUCCGAACGUGUCACGU), NMNAT1 siRNA1 (GCAGAACUUGCUACCAAGAAUUCUA), and NMNAT1 siRNA2 (GGAAGGAAGAGGAAGUGGACUGAAA) (Invitrogen), NML siRNA pool (Dharmacon), or DBC1 siRNA pool (Dharmacon) using RNAiMAX (Invitrogen) according to instructions from the supplier.

Western Blotting

Cells were washed with cold PBS (pH 7.4) and then scraped into lysis buffer containing 50 mm Tris-HCl (pH 6.8), 1% SDS, 1 mm PMSF, and protease inhibitor mixture. Lysates were heated at 100 °C for 10 min and clarified by centrifugation. Equal amounts of protein were subjected to SDS-PAGE and immunoblotted with specific primary antibodies. Anti-NMNAT1 mouse monoclonal antibody was purchased from Santa Cruz Biotechnology. Rabbit polyclonal serum for NML was immunized using His₆-NML-(1–200). Anti-FLAG polyclonal antibody was purchased from Sigma. Anti-Myc polyclonal antibody was from Cell Signaling. Anti-SirT1 monoclonal antibody 10E4 was purchased from Millipore. Anti-PARP1 antibody was from BD Biosciences.

Immunoprecipitation

Cells were washed in lysis buffer (50 mm Tris-HCl (pH 8.0), 5 mm EDTA, 150 mm NaCl, 0.5% Nonidet P-40, 1 mm PMSF, protease inhibitor mixture) and centrifuged for 10 min at 14,000 × g to remove the insoluble debris. The supernatant was used for immunoprecipitation and Western blotting. Cell lysate (200–1000 μg of protein) was immunoprecipitated with specific antibody and protein A-agarose beads (Sigma) or anti-FLAG M2-agarose beads (Sigma) for 18 h at 4 °C.

GST Pulldown Assay

Bacterial lysates expressing glutathione S-transferase (GST) and GST-NML were applied to glutathione-agarose beads according to the manufacturer's instructions (Pierce). The beads loaded with GST fusion proteins were incubated with recombinant His₆-tagged NMNAT1 protein at 4 °C for 2 h. The beads were washed in lysis buffer (50 mm Tris-HCl (pH 8.0), 5 mm EDTA, 150 mm NaCl, 0.5% Nonidet P-40), boiled in Laemmli sample buffer, and detected by Western blotting.

RNA Isolation and Quantitative PCR

Total RNA was extracted using the Qiagen RNeasy mini kit according to the manufacturer's instructions. cDNAs were prepared by reverse transcription of total RNA using the SuperScript III kit (Invitrogen). The primers used for SYBR Green quantitative PCR of human pre-rRNA, NMNAT1, NML, and GAPDH mRNA were as follows: human pre-rRNA forward, 5′-GAACGGTGGTGTGTCGTTC and reverse, 5′-GCGTCTCGTCTCGTCTCACT; NMNAT1 forward, 5′-TCTCCTTGCTTGTGGTTCATTC and reverse, 5′-TGACAACTGTGTACCTTCCTGT; NML forward, 5′-CCCCAGCCTATGTATAAGTGACT and reverse, 5′-GAGCCTGTTTGTGGCATTTCT; GAPDH forward, 5′-GAGTCAACGGATTTGGTCGT and reverse, 5′-GACAAGCTTCCCGTTCTCAG.

Chromatin Immunoprecipitation

ChIP assay was performed using standard procedures. NMNAT1 complex was immunoprecipitated with Myc antibody (exogenous) or NMNAT1 antibody (endogenous). SirT1 complex was immunoprecipitated with Myc antibody (exogenous) or 10E4 antibody (endogenous). Samples were subjected to SYBR Green real-time PCR analysis using primers for rDNA promoter H0 5′-GGTATATCTTTCGCTCCGAG and 5′-GACGACAGGTCGCCAGAGGA.

NAD⁺/NADH Measurements

The concentrations of NAD⁺/NADH in whole cell extracts were determined using a NAD⁺/NADH Quantification Kit (BioVision) according to instructions from the supplier.

Genomic DNA Isolation

Genomic DNA was prepared from lung tumor cell lines according to the established protocol. The NMNAT1 quantitative PCR data were normalized to interspersed repetitive element LINE1. The primers used for NMNAT1 were 5′-GGCATCATCTCTCCTGTTGGT and 5′-TTTCCCATGTATCAACTTCCACC. The primers for LINE1 were 5′-CAGAATCTCTGGGACGCATT and 5′-ATTGTGATGTTCGGGTGTCA (based on GenBank entry X52235.1).

rRNA Biosynthesis Assay

HeLa cells treated with control siRNA, NMNAT1, or NML siRNA were labeled with 5 μCi of [³H]uridine (38 Ci/mmol; PerkinElmer Life Sciences) for 30 min and washed twice with PBS. RNA was extracted with an RNeasy mini kit. For quantification of rRNA synthesis level, an identical amount of total RNA was analyzed by liquid scintillation counting to determine the incorporation of [³H]uridine.

NMNAT1 Co-purifies with NML

Recent studies identified NML as a novel H3K9me2-binding nucleolar protein that inhibits rRNA transcription through recruitment of SirT1 to the rDNA repeats (8). To determine whether NML interacts with other factors to regulate rRNA transcription, we performed affinity purification of FLAG-NML after transient expression in H1299 cells. Consistent with NML being a nucleolar protein, several ribosomal proteins were co-purified with NML. Also identified in the NML complex was NMNAT1, which is the last enzyme in the NAD⁺ salvage synthesis pathway (Fig. 1a).

To confirm the interaction between NMNAT1 and NML, recombinant His₆-tagged NMNAT1 and GST-NML were purified from Escherichia coli (Fig. 1b, left panel) and tested for binding in vitro. His₆-NMNAT1 was pulled down by beads loaded with GST-NML, but not by GST (Fig. 1b, right panel), suggesting that the two proteins interact directly. Next, we performed co-IP assays using epitope-tagged NMNAT1 and NML. When H1299 cells were co-transfected with Myc-NMNAT1 and FLAG-NML, specific co-precipitation between exogenous NMNAT1 and NML was detected using either FLAG or Myc IP (Fig. 1, c and d). In the same assay, NMNAT1 did not co-precipitate with the SirT1-binding protein DBC1 (Fig. 1c), suggesting that the interaction with NML was specific.

The detection of endogenous NMNAT1 and NML binding by co-IP/Western blotting was inconclusive due to lack of suitable antibody. Therefore, we generated tetracycline-inducible Myc-NMNAT1 expressing U2OS cells. In the absence of tetracycline induction, the cell line expressed Myc-NMNAT1 at a basal level similar to endogenous NMNAT1. IP of the basal Myc-NMNAT1 clearly co-precipitated endogenous NML (Fig. 1e, third lane). As expected, after tetracycline induction of high level Myc-NMNAT1, more endogenous NML was co-precipitated by Myc IP (Fig. 1e, fourth lane). Endogenous NMNAT1 also co-precipitated significantly with Myc-NMNAT1 in the Myc IP (Fig. 1e, third panel, third lane), suggesting that NMNAT1 forms dimer or oligomers in the cell. This is consistent with previous finding that recombinant NMNAT crystallized as a dimer (25). These results suggest that NMNAT1 is a specific binding partner of NML.

NMNAT1 Is Recruited into the eNoSC by NML

Previous study showed that NML recruits the NAD⁺-dependent deacetylase SirT1 to form the eNoSC that represses rRNA transcription. The identification of NMNAT1 as an NML-binding protein suggests that it may be recruited by NML into the eNoSC and stimulate SirT1 function by producing NAD⁺ locally. To test whether NML promotes interaction of NMNAT1 and SirT1, a co-transfection and IP-Western blot assay was performed to detect Myc-NMNAT1 with endogenous SirT1. Previous work showed that SirT1 binding to NML was stimulated by glucose deprivation (8). We found that expression of NML resulted in significant SirT1-NMNAT1 co-precipitation after glucose deprivation, correlating with NML-SirT1 binding (Fig. 2a). Therefore, NML can promote complex formation between SirT1 and NMNAT1.

Immunofluorescence staining of Myc-NMNAT1 confirmed the notion that NMNAT1 is localized in the nucleus (data not shown). To test whether a fraction of NMNAT1 was recruited to the nucleolus by NML, we performed ChIP against NMNAT1 and analyzed the binding to rDNA by PCR. The result showed that co-transfection of NML and NMNAT1 stimulated the binding of NMNAT1 to rDNA, which was further enhanced by glucose starvation (Fig. 2b). Furthermore, endogenous NMNAT1 in nontransfected cells also showed moderately increased binding to rDNA after glucose starvation (Fig. 2c). These results suggest that NML is capable of recruiting NMNAT1 to the nucleolus.

NMNAT1 Stimulates SirT1-mediated Deacetylation Reaction

SirT1-mediated deacetylation reaction consumes NAD⁺ and produces nicotinamide. NMNAT1 catalyzes the last reaction in recycling nicotinamide back to NAD⁺. Although NAMPT is thought to be the rate-limiting step in NAD⁺ salvage synthesis reaction based on in vitro enzyme kinetics analysis (10), it is possible that the level of NMNAT1 in the nucleus is important under in vivo conditions. We tested the effect of NMNAT1 level on SirT1 deacetylase function in cells using p53 as a substrate. The level of p53 acetylation on Lys-382 was monitored after co-transfection with p300 in H1299 cells. Expression of NMNAT1 stimulated the ability of SirT1 to deacetylate p53 (Fig. 3a), suggesting that nuclear NMNAT1 level has an impact on SirT1 activity. To further test the effect of endogenous NMNAT1 in p53 acetylation level, U2OS cells were treated with NMNAT1 siRNA in combination with DNA damage. NMNAT1 knockdown stimulated p53 acetylation level before and after DNA damage with doxorubicin (Fig. 3b), suggesting that the level of NMNAT1 was important for regulating SirT1 activity in vivo.

In the co-transfection assay, addition of NML did not further promote p53 deacetylation by SirT1 and NMNAT1 (data not shown). Therefore the assay was not informative for testing the role of NML-mediated SirT1-NMNAT1 complex formation, possibly because the weak binding of SirT1 and NMNAT1 was already sufficient for the p53 deacetylation reaction outside the nucleolus.

NMNAT1 Participates in the Regulation of rRNA Transcription

NML is a suppressor of rRNA transcription. Loss of NML expression dampens the down-regulation of rRNA transcription during glucose starvation (8). The interaction between NML and NMNAT1 suggests that NMNAT1 may be involved in the repression of rRNA transcription. We found that knockdown of NMNAT1 (Fig. 3c) led to increased rRNA synthesis as measured by [³H]UTP incorporation assay (Fig. 3d) or RT-PCR detection of pre-rRNA (Fig. 3e). NMNAT1 knockdown also partially prevented the down-regulation of rRNA synthesis after glucose starvation (Fig. 3d). The effect was similar to knockdown of NML. Ribosomal RNA transcription accounts for >50% of cellular transcription activity and consumes a significant amount of energy. Failure to down-regulate rRNA synthesis during starvation has been shown to cause ATP depletion and cell death (8). NMNAT1 knockdown also accelerated the depletion of cellular ATP level during glucose starvation (Fig. 4c). As expected, NMNAT1 knockdown increased the level of cell death after glucose starvation, similar to NML knockdown (Fig. 4, a and b). Depletion of the SirT1 inhibitor DBC1 did not cause cell death, as expected from the protective effect of SirT1 activation. These results suggest that NMNAT1 participates in regulating rRNA transcription and is necessary for the proper control of ribosome biogenesis during nutrient deprivation.

NMNAT1 Expression Is Induced by DNA Damage

Ribosomal RNA transcription is repressed after DNA damage. We found that DNA damage by doxorubicin treatment (Fig. 5a) or γ-irradiation (Fig. 5b) caused an increase in NMNAT1 protein level. RT-PCR analysis showed that doxorubicin induced NMNAT1 mRNA expression by 5-fold (Fig. 5c). Several other stress treatments that inhibit rRNA transcription (actinomycin D), inhibit mTOR (rapamycin), or activate p53 (Nutlin) did not significantly affect NMNAT1 expression. NMNAT1 induction by doxorubicin was observed in multiple tumor cell lines, independent of p53 status (Fig. 5d). To test whether NMNAT1 induction is a protective response to DNA damage, U2OS cells were treated with NMNAT1 siRNA in combination with doxorubicin. The combination treatment caused more cell death than doxorubicin alone (Fig. 5e). Furthermore, the levels of DNA damage markers γH2AX and phosphorylated p53 (pSer15) were elevated after γ-irradiation and down-regulated less efficiently at late time points, suggesting a delayed repair of DNA damage (Fig. 5f). This effect may be mediated by NMNAT1 stimulation of PARP, which is a major NAD⁺-consuming enzyme after DNA damage. Our results showed that NMNAT1 is also a DNA damage-responsive gene, suggesting that NMNAT1 plays a role in cellular response to multiple types of stress signals.

NMNAT1 Is Frequently Deleted in Tumors

Ribosomal rRNA transcription is significantly increased in tumors. Therefore, alteration of NMNAT1 expression during tumor development may facilitate rRNA transcription and cell growth and confer a selective advantage under certain conditions. A query of the Broad Institute database (26) for somatic gene copy number alteration showed that NMNAT1 is located in a chromosomal region that undergoes heterozygous deletion in ∼20% of several human tumor types (Fig. 6a). In contrast, NMNAT2/3 loci were not deleted (data not shown). We examined a panel of lung tumor cell lines and found that NMNAT1 level was significantly reduced in a subset (14 of 36) of cell lines (Fig. 6b). In contrast, SirT1 expression level did not show such a wide range of variation. RT-PCR analysis showed that the level of NMNAT1 protein was generally correlated with mRNA level (Fig. 7, a and b).

The copy number of the NMNAT1 gene was analyzed for 28 cell lines by quantitative PCR and normalized to the signal of interspersed repetitive element LINE1. Of 9 cell lines with unambiguous heterozygous reduction of NMNAT1 copy number, 6 (67%) showed low level NMNAT1 expression (Table 1, upper half). Of 11 cell lines with unambiguous diploid NMNAT1 copy number, 9 (80%) showed high or medium level NMNAT1 expression (Table 1, lower half). Therefore, the results indicate a correlation between NMNAT1 expression and gene copy number, suggesting that heterozygous deletion of NMNAT1 locus was responsible for reduced expression in a significant fraction of the cell lines.

NMNAT1 Expression Level Correlates with Sensitivity to Doxorubicin

Because NMNAT1 knockdown increased sensitivity to doxorubicin, we tested whether endogenous NMNAT1 level correlates with drug sensitivity. Thirteen lung tumor cell lines with high or low NMNAT1 levels were treated with doxorubicin and analyzed for cell survival after 48 h. The result suggested a correlation between low NMNAT1 expression level and higher sensitivity to doxorubicin (Fig. 7c), which was consistent with the effect of transient NMNAT1 knockdown. These results suggest that tumors with low level NMNAT1 expression may be more sensitive to treatment with DNA-damaging drugs.

To test whether the NMNAT1 level in cell lines correlates with the NAD⁺ level or NAD⁺/NADH ratio, we determined the levels of total cellular NAD⁺ and NADH in several cell lines with high and low levels of NMNAT1 expression. The results showed that there was no clear correlation between NAD⁺ level (Fig. 7d) or NAD⁺/NADH ratio (data not shown) in these cell lines. Furthermore, transient overexpression or knockdown of NMNAT1 only caused minor changes (10%) in total NAD⁺ level (data not shown). Therefore, the NMNAT1 expression level has negligible influence on total cellular NAD⁺ level, which may be largely determined by cytoplasmic NMNAT2/3 homologs. Because the assay did not distinguish cytoplasmic/mitochondria and nuclear compartments, the impact of NMNAT1 on nuclear NAD⁺ level is unknown.