Critical Role for NAD Glycohydrolase in Regulation of Erythropoiesis by Hematopoietic Stem Cells through Control of Intracellular NAD Content*

Background: The structure and function of NAD glycohydrolases (NADases) with pure hydrolytic activity are unclear. Results: We show that a novel NADase, exclusively expressed in rabbit reticulocytes, affects erythropoiesis. Conclusion: The intracellular NAD level regulated by the NADase is critical for erythropoiesis. Significance: We show that NADase activity regulates erythropoiesis of hematopoietic stem cells. NAD glycohydrolases (NADases) catalyze the hydrolysis of NAD to ADP-ribose and nicotinamide. Although many members of the NADase family, including ADP-ribosyltransferases, have been cloned and characterized, the structure and function of NADases with pure hydrolytic activity remain to be elucidated. Here, we report the structural and functional characterization of a novel NADase from rabbit reticulocytes. The novel NADase is a glycosylated, glycosylphosphatidylinositol-anchored cell surface protein exclusively expressed in reticulocytes. shRNA-mediated knockdown of the NADase in bone marrow cells resulted in a reduction of erythroid colony formation and an increase in NAD level. Furthermore, treatment of bone marrow cells with NAD, nicotinamide, or nicotinamide riboside, which induce an increase in NAD content, resulted in a significant decrease in erythroid progenitors. These results indicate that the novel NADase may play a critical role in regulating erythropoiesis of hematopoietic stem cells by modulating intracellular NAD.

NAD glycohydrolases (NADases) catalyze the hydrolysis of NAD to ADP-ribose and nicotinamide. Although many members of the NADase family, including ADP-ribosyltransferases, have been cloned and characterized, the structure and function of NADases with pure hydrolytic activity remain to be elucidated. Here, we report the structural and functional characterization of a novel NADase from rabbit reticulocytes. The novel NADase is a glycosylated, glycosylphosphatidylinositol-anchored cell surface protein exclusively expressed in reticulocytes. shRNA-mediated knockdown of the NADase in bone marrow cells resulted in a reduction of erythroid colony formation and an increase in NAD level. Furthermore, treatment of bone marrow cells with NAD, nicotinamide, or nicotinamide riboside, which induce an increase in NAD content, resulted in a significant decrease in erythroid progenitors. These results indicate that the novel NADase may play a critical role in regulating erythropoiesis of hematopoietic stem cells by modulating intracellular NAD.
An NADase with relatively pure NAD hydrolytic activity from Neurospora crassa has long been known (1). Another enzyme from the Gram-positive pathogen Streptococcus pyogenes was reported to be an NADase with no apparent ART, ADP-ribosyl cyclase, or cyclic ADPR hydrolase activities (2). We identified and characterized a rabbit erythrocyte enzyme with pure NADase activity (3) that was anchored to the plasma membrane via a glycosylphosphatidylinositol (GPI) linkage and could be solubilized by incubation with Bacillus cereus phosphatidylinositol-specific phospholipase C (PI-PLC) (4,5).
Of the NAD-degrading enzymes, which have the potential to control of NAD(P) levels, CD38, a mammalian ADP-ribosyl cyclase that exhibits significant NADase activity in addition to its intrinsic ADP-ribosyl cyclase activity, has been the most extensively studied (6). CD38 knock-out mice showed significantly higher tissue NAD levels than wild type, suggesting that CD38 may play a role in the control of NAD(P) levels (7). A previous study investigated a correlation between CD38 expression and erythroid differentiation in CD34 ϩ progenitor cells. The CD34 ϩ /CD38 ϩ population included 25-30% clonogenic progenitors with a mature erythroid phenotype, whereas the CD34 ϩ / CD38 Ϫ population was mostly primitive progenitors (8), suggesting that CD38 might affect erythroid differentiation.
In the erythroid lineage, the earliest committed progenitors, the slowly proliferating burst-forming unit-erythroid (BFU-E) cells, divide and further differentiate through the maturation stage into rapidly dividing colony-forming unit-erythroid (CFU-E). CFU-E progenitors divide and differentiate into red blood cells (9). BFU-E cells respond to many hormones and cytokines, including erythropoietin, stem cell factor, insulinlike growth factor 1, glucocorticoids, IL-3, and IL-6, whereas the terminal proliferation and differentiation of CFU-E progenitors are stimulated by erythropoietin, which is induced under hypoxic conditions (10). However, additional regulatory factors for proliferation and differentiation of these progenitor cells are being investigated.
In the present study, we report for the first time a novel enzyme from eukaryotes with pure NADase activity (designated as "NADase" here). We characterized this enzyme on a molecular level, especially in comparison with rabbit skeletal muscle ART, which exhibits the most similar primary structure but has different enzymatic activity. The rabbit enzyme showed a restricted pattern of tissue expression limited to erythroid. We also found that the novel NADase plays a critical role in regulating erythropoiesis of hematopoietic stem cells by modulating intracellular NAD content.

EXPERIMENTAL PROCEDURES
Materials-Erythrocytes were obtained from New Zealand White rabbits (3 months old). PI-PLC from Bacillus cereus was purified as described (4 Purification of NADase-NADase was purified from rabbit erythrocytes as described (3). One liter of packed rabbit erythrocytes was washed three times with ice-cold PBS and incu-   were subjected to SDS-PAGE. A summary of the purification is given in Table 1.
In-gel Activity Measurement of NADase-Purified NADase was separated by non-reducing SDS-PAGE (12% gel), and the activity of NADase in the gel was measured with 150 M ⑀-NAD ϩ as described previously (11). The apparent molecular mass of a protein corresponding to a fluorescent band was estimated by marking the band with a needle and, after staining the gel with Coomassie Brilliant Blue, comparing its position with those of marker proteins.
Mass Spectrometry-Quadrupole TOF analysis of trypsin-digested peptide fragments of purified proteins was performed by PROTEINWORKS Inc. (Daejeon, Korea).
5Ј-and 3Ј-Rapid Amplification of cDNA Ends (RACE)-The first cDNA strand was synthesized from 2 g of total RNA using SuperScript II reverse transcriptase by following the manual from the CapFishing cDNA isolation kit. PCR was performed using i-StarTaq DNA polymerase (iNtRON Biotechnology) with 5Ј-or 3Ј-RACE primers and gene-specific RACE primers. Full-length cDNA was generated by two-step PCR following the manufacturer's protocol.
Glycosylation Assay-Purified rabbit erythrocyte NADase was denatured and treated with peptide-N-glycosidase F (PNGase F) (New England Biolabs) according to the manufacturer's instructions before Western blot analysis.
Preparation of the Antibody-Antibodies against purified NADase were raised in mice.
Treatment of NADase-transfected Cells with PI-PLC-For PI-PLC treatment, HEK293 cells transfected with NADase were incubated either in the absence or presence of 1 g/ml PI-PLC at 37°C for 1 h in 7 ml of HEPES buffer (156 mM NaCl, 3 mM KCl, 2 mM MgSO 4 , 1.25 mM KH 2 PO 4 , 10 mM D-glucose, 2 mM CaCl 2 , 10 mM HEPES, pH 7.4). The incubation medium was then collected and concentrated 10-fold using an Amicon Ultra centrifugal filter (3-kDa cutoff).
Western Blot Analysis-Lysed cells and purified samples were reduced and separated by SDS-PAGE (12% gel). The resolved proteins were transferred to a nitrocellulose membrane (Bio-Rad). The blots were blocked with 5% skim milk powder dissolved in TTBS (20 mM Tris-HCl, pH 7.6, 137 mM NaCl with 0.1% Tween 20). The blot was incubated with the primary antibody (mouse anti-rabbit NADase antiserum, 1:2,000 dilution; rabbit anti-FLAG polyclonal antibody (Sigma), 1:2,000 dilution) for 2 h at room temperature. The blot was washed five times with TTBS and then incubated with the secondary antibody (anti-mouse IgG conjugated to horseradish peroxidase, 1:5,000 dilution; anti-rabbit conjugated to horseradish peroxidase (Santa Cruz Biotechnology), 1:2,000 dilution) for 2 h at room temperature. The blots were developed using an enhanced chemiluminescence kit (Amersham Biosciences) and exposed to an LAS 1000 Image Reader Lite (Fujifilm, Tokyo, Japan). Protein concentrations were determined using a Bio- Rad protein assay kit with bovine serum albumin (BSA) as the standard.
Northern Blot Analysis-Total RNA was isolated from rabbit tissues. For Northern blot analysis, 20 g of total RNA was subjected to electrophoresis in a denaturing 1.2% agarose gel containing formaldehyde and ethidium bromide and then transferred to Hybond-N membranes (GE Healthcare). Membranes were prehybridized at 42°C for 1-2 h in hybridization buffer (50% formamide, 5ϫ saline/sodium phosphate/EDTA, 5ϫ Denhardt's solution, 0.1% SDS, 100 g/ml denatured salmon sperm DNA) and then hybridized overnight to the probe labeled with [␣-32 P]dCTP in a hybridization buffer. The membranes were washed twice at room temperature for 10 min in 2ϫ SSC, 0.1% SDS; twice at room temperature for 15 min in 1ϫ SSC, 0.1% SDS; and twice at 65°C for 15 min in 0.1ϫ SSC, 0.1% SDS. After air drying, membranes were evaluated by image analysis using a BAS 2000 system (Fujifilm).
NADase and ART Assays-NADase activity was determined by measuring etheno-ADPR formation fluorometrically using ⑀-NAD ϩ as a substrate (15,16). Samples (20 g) were incubated in the presence of 200 M ⑀-NAD ϩ with or without an appropriate protein in an assay buffer (50 mM potassium phosphate, pH 7.2, 150 mM NaCl, 0.1% Nonidet P-40; final volume of 50 l). The reaction mixture was incubated at 37°C for 20 min. The reaction was stopped by adding 50 l of trichloroacetic acid (10%). The samples were centrifuged at 15,000 ϫ g for 10 min, and the supernatant (80 l) was diluted with 720 l of 100 mM sodium phosphate buffer, pH 7.2. Fluorescence of etheno-ADPR in solution was determined at excitation/emission wavelengths of 297/410 nm (Hitachi F-2500 fluorescence spectrophotometer). Assays were repeated five times.
ART activity was assayed in 300 l of 50 mM potassium phosphate, pH 7.
Purification and Lentiviral Transduction of Bone Marrow-Purification of bone marrow cells was performed according to the method of Lutton et al. (20). Adult New Zealand White rabbits were used as bone marrow donors. Rabbits were sacrificed by anesthesia. Bone marrow cells were harvested from femora and tibiae. Bone marrow was flushed with Iscove's modified Dulbecco's medium. Bone marrow cells were washed three times with ice-cold PBS, and red blood cells (RBCs) were removed by RBC lysis buffer. After removing adherent cells, the cells were prestimulated in Iscove's modified Dulbecco's medium containing 10% FBS and the following growth factors: mouse stem cell factor, human thrombopoietin, and human Flt3 ligand (R&D System) at 10 ng/ml each for 24 h. The prestimulated bone marrow cells were infected with scrambled or NADase shRNA-expressing lentivirus in the presence of Polybrene (8 g/ml) and selected with puromycin at 5.0 g/ml for 3 days.
Quantification of Erythropoiesis-The cells transduced with shRNAs expressing lentivirus were plated in triplicate at a density of 2 ϫ 10 5 cells/plate in MethoCult M3334 (StemCell Technologies) to quantify the formation of CFU-E and mature BFU-E following the manufacturer's instructions. Colonies were scored after 7-10 days.

Measurement of Intracellular NAD Concentration ([NAD] i ) and ADPR Concentration ([ADPR] i )-Cells
were sonicated with 0.2 ml of 0.6 M perchloric acid. Precipitates were removed by centrifugation at 20,000 ϫ g for 10 min at 4°C, and perchloric acid was neutralized with 60 l of 2 M KHCO 3 (21). After centrifugation at 16,000 ϫ g for 10 min at 4°C, the precipitates were removed. The [NAD] i was measured using a cyclic enzyme assay as described previously (22). Briefly, the supernatants (100 l/tube) were further incubated with 100 l of a cycling reagent solution (2% ethanol, 100 g/ml alcohol dehydrogenase, 20 M resazurin, 10 g/ml diaphorase, 10 M riboflavin 5Ј-phosphate, 10 mM nicotinamide, 0.1 mg/ml BSA, 100 mM sodium phosphate, pH 8.0) at room temperature for 2 h. An increase in the resorufin fluorescence was measured at 544-nm excitation and 590-nm emission using a fluorescence plate reader (Spectra-Max GEMINI, Molecular Devices). Various known concentrations of NAD were also included in the cycling reaction to generate a standard curve. The [ADPR] i was measured using LC-MS/MS as described (23). Briefly, to determine the [ADPR] i , the supernatants were loaded onto a Waters ACQUITY UPLC system coupled to a Waters Xevo TQ-S mass spectrometer and separated by a BEH Amide column (Waters ACQUITY UPLC BEH Amide, 130 Å, 1.7 m, 2.1 ϫ 50 mm).

TABLE 3 Amino acid sequence identity between NADase and ART family members
The extent of identity (colored boxes) and similarity (white boxes) was calculated with BLASTp and ClustalW. m, mouse; r, rat; h, human; rb, rabbit. The following references were used: Refs. 5, 25, 28, and 32-34. All chromatographic separations were performed using each mobile phase combination at a flow rate of 0.5 ml/min. The column was equilibrated with 100% buffer B (90% acetonitrile, 10% 50 mM ammonium formate) and eluted in a 5-min gradient to 60% buffer A (10 mM ammonium formate in water). The column was then rinsed with 100% buffer A and re-equilibrated with buffer B before the next injection. The following parameters were optimized for ADPR MS analysis: cone gas, 150 liters/h; nebulizer, 7 bars; and desolvation temperature, 350°C. The confirmation ion transitions for quantification were m/z 558. 17
peptides were performed using the activity fractions of Sephacryl S-200HR.
Cloning of NADase-The amino acid sequence of one of the tryptic peptides (amino acids 20 -40) was used to synthesize two sets of degenerate oligonucleotides, which were used as primers in PCR amplification, following rabbit reticulocyte first-strand cDNA synthesis. Significant amounts of PCR products were obtained with primers (rbNA-5Ј, rbAR-3Ј, and rbARr). On the basis of the partial cDNA sequence, oligonucleotide primers for 5Ј-and 3Ј-RACE PCR were designed. The cDNA obtained by RACE-PCR was 1,491 bp in length, encoding a 293 amino acid-long polypeptide with a calculated molecular mass of 32,769 Da (Fig. 2). The deduced amino acid sequence of this protein includes the sequences of all four tryptic peptides from the NADase protein. The NADase gene is located downstream of ART1 and ART1-like gene in chromosome 1 (Fig. 3). To confirm the tissue distribution of NADase, which was red blood cell-specific in the Western blot, we performed Northern blot and RT-PCR analyses with total RNA isolated from various rabbit tissues. The Northern blot analysis showed a 3.6 kb mRNA for the NADase (Fig. 1B), and the NADase-specific band appeared only in reticulocytes in both Northern blot and RT-PCR analyses (Fig. 1, B and C). To further examine whether other blood cells also express NADase, we performed Western blot analysis with various types of rabbit blood cells. NADase was expressed in red blood cells but not expressed in lymphocytes, platelets, and neutrophils (Fig. 1D).
Structural Characterization-To examine whether the NADase is glycosylated, we performed Western blot analysis of purified NADase treated with or without PNGase F. PNGase F-treated NADase showed a molecular weight of 33 kDa compared with 40 kDa for untreated NADase (Fig. 1E), indicating that the NADase is an N-glycosylated protein. Consistent with the findings that the NADase is modified by PNGase F-sensitive N-glycosylation (Fig. 1E) and that GPI-linked proteins are often heavily glycosylated (24,25), three potential sites for N-linked glycosylation were found in the deduced amino acid sequence: Asn 51 , Asn 211 , and Asn 233 (Fig. 2). The hydrophilicity plot showed hydrophobic N and C termini with a hydrophilic center (data not shown). The hydrophobic N-terminal region was reported to serve as a leader sequence, and the hydrophobic sequence near the C terminus was reported to serve as a recognition signal for GPI modification in the endoplasmic reticulum (26,27). To determine localization of NADase on the plasma membrane, we performed immunocytochemical analysis in conjunction with confocal microscopy. GFP-NADase was exclusively localized on the plasma membrane in HEK293 cells (Fig. 4A). To confirm that NADase exists as a GPIanchored form on the plasma membrane, HEK293 cells transfected with FLAG-NADase were incubated with bacterial PI-PLC. Western blot analysis and an NADase activity assay showed release of NADase into the medium by PI-PLC treatment (Fig. 4, B and C), indicating that the NADase is a GPI-anchored protein.
Catalytic Glutamine of NADase Is a Crucial Residue for NADase Activity-A homology search of the deduced amino acid sequence of NADase was performed at the National Center for Biotechnology Information by using BLAST (databases, April 2013) ( Table 3). The highest homology score was obtained for rabbit ART1 protein, which is expressed in skeletal and cardiac muscle (28). In comparison with ART1, the amino acid sequence in the vicinity of the catalytic glutamate residue was notably different in NADase (Fig. 5). Typical ART enzymes, including ART1, have an EEE motif, whereas NADase has 218 QAE 220 , and ART enzymes possessing NADase activity con- tain a QEE motif. This finding suggests that the NADase may function as a NADase but not as an ART as it is enzymatically more similar to rat RT6 rather than to mouse Rt6 (13,24,25,29). To test this hypothesis, we used site-directed mutagenesis with subsequent synthesis of recombinant NADase and ART1 proteins in HEK293 cells (Fig. 6A). A recombinant protein from the full-length NADase cDNA showed NADase activity with very low ART activity, whereas recombinant ART1 exhibited ART activity only (Fig. 6, B and C). Replacement of glutamine with glutamate at position 218 (Q218E) of the recombinant NADase resulted in a loss of NADase activity and an increase of ART activity, similar to the recombinant ART (Fig. 6, B and C). We also examined enzyme activity after replacing glutamine at position 218 with other amino acids. Replacement of glutamine with aspartate at position 218 (Q218D) resulted in a loss of NADase activity, similar to Q218E, but the ART activity was not significantly increased. Replacement of glutamine with alanine (Q218A) resulted in loss of both NADase and ART activities (Fig. 6, B and C). Taken together, these findings indicate that glutamine 218 is a crucial determinant of NADase and ART activities.
NADase Is Involved in Erythropoiesis in Bone Marrow Cells-Because NADase is exclusively expressed in erythrocytes, we speculated that NADase may play a role in erythroid differentiation of bone marrow cells. We quantified colony formation (CFU-E and BFU-E). For lentivirus transduction, we first prestimulated bone marrow cells with stem cell factor, thrombopoietin, and Flt3 ligand and then infected them with scrambled or NADase shRNA-expressing lentivirus. NADase was significantly reduced in bone marrow cells treated with NADase shRNAs compared with those treated with scrambled shRNA (Fig. 7A). CFU-E formation was significantly reduced in NADase knockdown cells compared with those with scrambled shRNA lentivirus (Fig. 7, B and C), suggesting that NADase is involved in erythropoiesis of bone marrow cells. Mature BFU-E formation was not observed in bone marrow cells treated with NADase shRNAs probably due to the effect of puromycin in the process of selection (Fig. 7C).
To investigate further the mechanism by which NADase affects erythropoiesis and a possible role of the metabolites generated by the NADase, we examined concentrations of substrate, NAD, and a product, ADP-ribose, in control cells or cells in which the NADase had been suppressed. Bone marrow cells with NADase reduced by shRNA treatment showed a significant increase in NAD content and a decrease in ADPR content compared with control cells (Fig. 7, D and E). We reasoned that an increased NAD level might affect erythropoiesis of bone marrow cells. Therefore, we examined whether erythropoiesis is affected in the bone marrow cells with increased NAD level by supplementation with an NAD precursor, nicotinamide riboside (NR) (30,31). NR-treated cells showed a significant decrease in CFU-E formation and BFU-E formation (Fig. 8A). The NR-treated cells showed an increase in intracellular NAD but not in ADPR level compared with control cells (Fig. 8, B and C). These results indicate that the NAD level affects erythropoiesis of bone marrow cells. To further corroborate these findings, we examined the effects of extracellular NAD, ADPR, and nicotinamide on erythropoiesis of bone marrow cells. Treatment of bone marrow cells with NAD or nicotinamide, which resulted in an increase in intracellular NAD level, significantly decreased CFU-E and BFU-E formation, whereas ADPRtreated cells showed no difference from control vehicle-treated cells (Fig. 9). These results confirm that the NAD level regulates erythropoiesis of bone marrow cells. A previous report on hematopoietic progenitor cell content of mouse bone marrow cells showed that CFU-E formation and mature BFU-E formation of CD34 ϩ /CD38 Ϫ cells were significantly decreased compared with those of CD34 ϩ /CD38 ϩ cells (8), suggesting that another NADase family, CD38, also plays a role in erythropoiesis. Given that CD38 is also an NAD-degrading enzyme, the expression of the enzyme in bone marrow cells may regulate erythropoiesis in hematopoietic stem cells by modulating the NAD level, similar to the rabbit NADase. Therefore, to confirm whether a decrease in erythropoiesis in the bone marrow cells from CD38-deficient mice was due to an increase in NAD level,  we examined erythropoiesis and NAD levels of bone marrow cells prepared from CD38 wild-type (WT) and knock-out (KO) mice. As expected, bone marrow cells from CD38 KO mice showed a significant increase in NAD level and a decrease in erythropoiesis compared with those of CD38 WT mice (Fig.  10). These findings further support the hypothesis that the NAD level affects erythropoiesis in hematopoietic stem cells.
In conclusion, we have cloned and characterized a novel NADase specifically expressed in rabbit reticulocytes. This novel NADase showed only NADase activity but not ART activity. Although homologs of this enzyme were not found in other species, this enzyme shows unique characteristics in enzymatic function as well as in exclusive expression in erythrocytes, consistent with its role in regulating erythropoiesis.