Decrease in Nucleoside Diphosphate Kinase (NDPK/nm23) Expression during Hematopoietic Maturation*

The nucleoside diphosphate kinase (NDPK/nm23) isoforms H1 and H2 were localized in hematopoietic tissues. Flow cytometric analysis and enzymatic assays were used to quantify the intracellular and extracellular concentrations of NDPK. Bone marrow CD34+ progenitors contained the highest intracellular levels of both nm23-H1 and nm23-H2. Lower levels were measured in more mature bone marrow cells, whereas peripheral blood leukocytes had the lowest expression of nm23. These data suggest a function of NDPK in early hematopoiesis and a down-regulation of NDPK upon differentiation. In addition, an up-regulation of nm23 expression was observed in lymphocytes after induction of proliferation with phytohemagglutinin. Multiparameter flow cytometry demonstrated that this up-regulation occurred during the G0/G1-transition. Flow cytometric analysis also revealed a weak surface expression of nm23 on a number of hematopoietic cell lines, which was not detected on normal hematopoietic cells. Our data also demonstrated the presence of NDPK in human plasma, probably due to a limited in vivo lysis of red blood cells.

Currently, the family of nucleoside diphosphate kinases (NDPK/nm23) 1 consists of four members encoded by the genes nm23-H1, nm23-H2, DR-nm23, and nm23-H4 (1)(2)(3)(4). These enzymes share the ability to transfer the terminal phosphate of nucleoside triphosphates to nucleoside diphosphates. Beside their function to maintain the cellular GTP level (5), several other functions and properties have been assigned to the NDPK family. These include the identity of nm23-H2 to PuF, a transcription factor of the proto-oncogene c-myc (6), the ability of nm23-H1 to suppress the metastatic potential of several tumors and cell lines (1,7,8), the association of NDPK with transcription regulating factors such as the retinoic acid receptor-related orphan receptor ␣ and the retinoic Z receptor ␤, and with the heat shock protein hsc70 (9,10). Recently, it was also reported that NDPK has protein kinase activity (11,12).
The importance of nm23 expression in hematopoiesis has been demonstrated by Okabe-Kado et al. (13), who identified NDPK B as a differentiation inhibiting factor (I-factor) in the conditioned medium of the differentiation-resistant mouse M1 cell line and demonstrated that both nm23-H1 and nm23-H2 can inhibit the in vitro induced differentiation of several hematopoietic cell lines. This inhibition was independent of the phosphotransferase activity as demonstrated with nm23 mutants lacking the enzymatic activity (13)(14)(15).
Moreover, reported data indicated a correlation between proliferation and nm23 expression in hematopoietic cells. In lymphoma and monocytic leukemia, overexpression of nm23-H1 was correlated with a higher tumor aggressiveness and resistance to chemotherapy, respectively (16,17). Up-regulation of nm23-H1 and -H2 was observed in normal lymphocytes induced to proliferate with phytohemagglutinin (PHA) (18,19). In vitro, DR-nm23 mRNA is up-regulated in CD34 ϩ hematopoietic progenitor cells during early stages of myeloid differentiation and is completely down-regulated upon terminal differentiation (3).
Studies on the localization of nm23 mRNA in mouse embryos by "in situ" hybridization with nm23-H1 probes showed an overall staining and an increased expression in differentiating tissue (20). NDPK phosphotransferase activity has also been observed in the cytosol and membrane fraction of neutrophils (21). These studies support the hypothesis that nm23 is ubiquitously expressed in cells. Using enzymatic assays, surfacebound NDPK has been observed on platelets (22). In addition, several hematopoietic tumor cell lines stained positive for NDPK in flow cytometric analysis (23). So far, the presence of NDPK in extracellular fluids has not been demonstrated.
In this article we studied the nm23-H1 and nm23-H2 expression in cells of human hematopoietic origin and demonstrated that both isoforms are highly expressed in hematopoietic progenitors. A decrease in expression levels was observed during myeloid and lymphoid differentiation. Surface expression of nm23 was not detected on normal hematopoietic cells and was only observed on tumor cell lines of hematopoietic origin. We also provide evidence for the presence of NDPK in human plasma. Belgium). The fluorescein isothiocyanate (FITC)-labeled anti-nm23-H1 and anti-nm23-H2-2 antibodies were obtained from Seikagaku Corporation (Tokyo, Japan), the isotype control antibodies, mouse IgG 2a -FITC and mouse IgG 1 -FITC, respectively, and the phycoerythrin-labeled anti-CD14 (Leu TM -M3) and anti-CD34 (anti-HPCA-2) antibodies were obtained from Becton Dickinson (Erembodegem, Belgium). The Hybond-C cellulose membranes and the enhanced chemoluminescence (ECL) kit were from Amersham (Slough, UK). Bovine serum albumin (BSA) was obtained from Merck (Darmstadt, Germany) and Sigma (Bornem, Belgium). Fetal calf serum (FCS, Myoclone Super Plus) was from Life Technologies, Inc. (Paisley, UK) and horse serum from Sera Lab (Crawley, W Sussex, UK). The hematopoietic cell lines K562, HL-60, U937, KG-1, Jurkat, Molt3, Molt4, SupT1, and HSB2 were obtained from ATCC (Rockville, MD) and cultured in IMDM containing 10% (v/v) FCS.
Preparation of Platelet-poor Plasma-Blood of healthy volunteers was collected in CTAD-tubes (Becton Dickinson, Erembodegem, Belgium), without using a tourniquet and placed on ice for at least 10 min in order to avoid platelet activation. Umbilical cord blood was collected from the placenta of healthy full-term infants. Bone marrow samples were aspirated by sternal puncture from hematologically normal patients undergoing cardiac surgery after informed consent. Umbilical cord blood and bone marrow samples were collected in tubes containing 2 ml of IMDM with 10% (v/v) FCS and 5 units/ml preservative-free heparin (Novo Industries, Bagsvaerd, Denmark). Subsequently, the samples were centrifuged at 400 ϫ g for 15 min at 4°C. The middle part of the plasma column was collected and centrifuged at 700 ϫ g for 10 min at 4°C. The supernatant was collected and filtered through a 0.22-m microfilter (Millipore, Molsheim, France).
Isolation and Lysis of Blood Cells-White blood cells were isolated from normal blood or bone marrow by a 1/100 dilution of the sample in hemolysis buffer (155 mM NH 4 Cl, 0.1 mM EDTA, 10 mM KHCO 3 , pH 7.3). After an incubation for 5-10 min at room temperature, cells were washed three times with PBS and labeled for flow cytometric analysis.
Peripheral blood lymphocytes were isolated from normal blood by density gradient centrifugation on Ficoll-Paque (Pharmacia Biotech, Uppsala, Sweden) and washed three times with IMDM. Purified lymphocytes were cultured in IMDM supplemented with 20% (v/v) FCS and 5 g/ml PHA.
Platelets were isolated from normal blood samples, collected in CTAD-tubes. The samples were centrifuged at 400 ϫ g for 15 min and the intermediate, platelet-containing fraction of the plasma collected. Platelets were washed three times with PBS containing 3.7 mM adenosine, 15 mM theophylline, and 0.198 mM dipyridamole and pelleted by a 10-min centrifugation at 700 ϫ g (24). Erythrocytes were isolated by repeated (3 times) centrifugation of normal blood samples at 400 ϫ g for 15 min and removal of the buffy coat and supernatant. Cell lysates were obtained by four repetitive cycles of freeze-thawing and sonication for 5 min in the presence of 0.02% (w/v) CHAPS.
Enzymatic NDPK Assay-Based on an exploratory screening, plasma samples and cell lysates were diluted with PBS containing 0.1% (w/v) casein to approximately equal NDPK phosphotransferase activities. The diluted samples were incubated at 37°C in a buffer containing 50 mM Tris-HCl (pH 7.6), 100 mM NaCl, 10 mM MgSO 4 , 1 mM ATP, 2 mM CDP, and 0.1 mCi/ml [␥-32 P]ATP (ICN Biochemicals Inc., Costa Mesa, CA). At relevant time intervals, samples were taken and the reaction was stopped by addition of an equal volume of 0.5 M EDTA. Radiolabeled ATP and CTP were separated by polyethyleneimine-cellulose thin layer chromatography in 750 mM KH 2 PO 4 (pH 3.65). The spots were visualized by autoradiography, excised, and counted in a liquid scintillation counter.
Determination of Protein and Hemoglobin Content of Plasma Samples-The protein content was measured by the biuret method in a Vitros 750XRC analyzer (Johnson & Johnson Clinical Diagnostics Inc., Rochester, NY). Free plasma hemoglobin was determined according to Crosby and Furth (25). In brief, 20 l of plasma was added to a solution of 1% (w/v) benzidin in 90% (v/v) acetic acid. The reaction was initiated by addition of 1 ml of 1% (w/v) H 2 O 2 and stopped after an incubation of 20 min at room temperature by addition of 10 ml of 10% (v/v) acetic acid. After 10 min, the optical density (OD) of the samples was measured at 595 nm. In this assay, 0.1% (w/v) hemoglobin has an OD of 1.5.
Labeling of Cells for Flow Cytometric Analysis-Cells were washed three times with PBS. For the detection of membrane-bound NDPK, 5 ϫ 10 5 nucleated cells or 10 7 thrombocytes or erythrocytes, were suspended in a volume of 50 -200 l, blocked for aspecific staining with 10 g of unlabeled mouse gamma globulins (Jackson ImmunoResearch Laboratories, West Grove, PA) for 10 min and incubated for an additional 15 min with 0.5 g of antibody, washed with PBS, and analyzed on a FACScan flow cytometer (Becton Dickinson, Erembodegem, Belgium).
For detection of intracellular NDPK, 5 ϫ 10 5 cells were suspended in 50 -200 l, fixed, and permeabilized with Fix&Perm (An-der-Grub, Kaumberg, Austria) according to the manufacturer's instructions. The cells were labeled with 0.5 g of antibody and washed once before analysis. FITC-labeled mouse IgG 1 and IgG 2a were used as isotypematched controls for the anti-nm23-H2 and anti-nm23-H1 antibodies, respectively. All data were obtained with the same batches of antisera.
Multiparameter Flow Cytometry-Nm23 expression and DNA content were simultaneously measured according to the method of Clevenger et al. (26). In brief, 4 ϫ 10 6 cells were fixed in 1 ml of 0.5% (w/v) paraformaldehyde for 15 min at 4°C. After centrifugation at 400 ϫ g for 5 min, the cells were resuspended in 1 ml of 0.1% (v/v) Triton X-100 (BDH Chemicals, Poole, UK) for 5 min at 4°C. After centrifugation, the pellet was divided in four equal parts: one part was incubated with 0.5 g of anti-nm23-H1-FITC antibody, one part with 0.5 g anti-nm23-H2-FITC antibody, and the two remaining parts with 0.5 g of the isotype-matched control antibodies, mouse-IgG 2a -FITC, and mouse-IgG 1 -FITC, respectively. After an incubation for 1 h at 4°C, all samples were washed with PBS and resuspended in 400 l of PBS containing 10 g/ml ethidium bromide and 0.1% (w/v) RNase A. After an incubation at room temperature for at least 20 min, cells were analyzed on a FACScan flow cytometer with standard optics except for the 515 Ϯ 4.5 nm bandpass filter in front of the Fl-1 photomultiplier.
Statistical Analysis-Comparisons of nm23 expression in different cell types was performed with a paired Student's t test using Statview 4.5 (Abacus Concepts, Berkely, CA). p Values smaller than 0.05 were considered as statistically significant. Regression analysis was performed in Excel 7.0 (Microsoft Corp., Redmond, WA).
FIG. 1. NDPK activity in plasma. NDPK phosphotransferase activity is plotted against the hemoglobin content of plasma from peripheral blood (A), umbilical cord blood (B), and bone marrow (C). NDPK activity and free plasma hemoglobin were quantified as described under "Experimental Procedures." The data were evaluated by linear regression. All regression coefficients (R) were significantly different from zero (p Ͻ 0.01).

RESULTS
NDPK/nm23 in Plasma-NDPK was measured in the physiological medium of hematopoietic cells. Platelet-free plasma was prepared from human peripheral blood, bone marrow, and umbilical cord blood and was assayed for NDPK phosphotransferase activity, protein content, and free plasma hemoglobin. NDPK assays were performed in conditions of linear kinetics. The activity in bone marrow plasma exceeded by far that in peripheral blood and umbilical cord blood plasma. All plasma samples contained approximately equal amounts of protein and a statistically significant correlation (p Ͻ 0.01) between NDPK activity and free hemoglobin was observed (Fig. 1). However, peripheral blood samples containing very low levels of hemo-globin still contained NDPK activity. Extrapolation to zero hemoglobin revealed a basal NDPK phosphotransferase activity of 33 Ϯ 5 milliunits/ml in peripheral blood plasma (p ϭ 0.05). The NDPK activity was also quantified in several protein solutions routinely used in cell cultivation, i.e. FCS, horse serum, BSA, and casein. With the exception of casein, all assayed solutions contained NDPK activity (Table I).
Intracellular Levels of NDPK/nm23 during Hematopoietic Maturation-Intracellular nm23 in blood and bone marrow cells was monitored by flow cytometry after intracellular staining with FITC-conjugated anti-nm23-H1 and anti-nm23-H2 antibodies. In peripheral blood leukocytes, distinction between lymphocytes and granulocytes was made by the scatter profile of the cells. Monocytes were identified by labeling with a phycoerythrin-conjugated anti-CD14 antibody ( Fig. 2A). In bone marrow, a distinction was made between mononuclear and myeloid precursors by their scatter profile. CD34 ϩ progenitors were identified by labeling with a phycoerythrin-conjugated anti-CD34 antibody (Fig. 2B). All analyzed cells showed a positive intracellular immunofluorescence for nm23-H1 and nm23-H2. The results presented in Fig. 3 demonstrated that the expression of nm23 is high in bone marrow CD34 ϩ cells and lower in more differentiated cell types. In addition, peripheral blood polymorphonuclear cells have a higher expression of nm23-H1 than lymphocytes, which in turn express more nm23-H2 than polymorphonuclear cells. Among the peripheral

FIG. 3. Decrease in nm23 expression upon maturation of hematopoietic cells. Nm23 expression is determined from the mean fluorescence intensity (MFI) measured by flow cytometric analyis of cells, labeled with FITC-conjugated antibodies for intracellular nm23.
The plotted MFI is the measured value substracted with the MFI of cells labeled with isotype-matched control antibodies. Nm23-H1 expression is indicated in the upper bar chart, nm23-H2 expression in the lower bar chart. Error bars represent the standard deviation. Note that nm23-H1 and nm23-H2 cannot be compared due to different affinities and fluorescence intensities of the used antibodies. Statistical analysis using the Student's t test revealed a significantly different expression of nm23-H1 and nm23-H2 in all cells, except when mononuclear precursors are compared with monocytes and granulocytes. p Values smaller than 0.05 were considered as statistically significant.  blood leukocytes, monocytes have the highest nm23 expression. Due to cell lysis of platelets and red blood cells during fixation and permeabilization, the intracellular nm23 expression of these cells could not be measured by flow cytometry. Therefore, the presence of NDPK was measured by enzymatic assays and Western blotting. In lysates of red blood cells and platelets, a NDPK phosphotransferase activity of 1.3 Ϯ 0.6 nanounits/cell and 0.40 Ϯ 0.08 nanounits/cell (mean Ϯ S.E., n ϭ 3) was measured, respectively. Both nm23-H1 and nm23-H2 were identified in these cells by Western blotting (data not shown).
NDPK/nm23 Expression in PHA-stimulated Lymphocytes-Nm23 was rapidly up-regulated in lymphocytes after induction of proliferation by PHA. Using single parameter flow cytometry on fixed and permeabilized lymphocytes, we observed a more than 10-fold increase in immunoreactivity of nm23-H1 and nm23-H2 72-96 h after PHA stimulation (Fig. 4). Cells were also simultaneously examined for nm23 and DNA content using a multiparameter flow cytometric approach. From the obtained data, we were able to correlate the nm23 content with the cell cycle phase. The results presented in Fig. 5, are representative of five independent experiments. The most important conclusions which can be drawn from these data are: (i) unstimulated lymphocytes in G 0 phase have a basal level of nm23 expression; (ii) the onset of proliferation takes place between 24 and 48 h after stimulation; (iii) there is an up-regulation of nm23 expression after 24 h; (iv) cells that enter the S-phase have a nm23 content which exceeds that of unstimulated cells; (v) there is no further up-regulation of nm23 during the Sphase. The obtained results were similar for nm23-H1 and nm23-H2.
Membrane-bound NDPK/nm23-The surface expression of nm23-H1 and nm23-H2 on normal hematopoietic cells was investigated by flow cytometric analysis. In five independent experiments, we could not demonstrate nm23 on the surface of any normal hematopoietic cell, i.e. bone marrow CD34 ϩ progenitors, mononuclear precursors (lymphoid and erythroid), myeloid precursors, lymphocytes, monocytes, granulocytes, and purified thrombocytes and erythrocytes. Apart from the latter two, cells were identified by flow cytometric analysis of the scatter profiles and the CD34 membrane marker (Fig. 2, C  and D). PHA stimulation of lymphocyte proliferation did not induce surface expression of nm23. In contrast, of the hematopoietic cell lines examined for surface-bound nm23, the myeloid cell lines (K562, KG1, HL-60, and U937) showed a weak but reproducible positive staining for nm23-H1, whereas the lymphoid ones (HSB2, SupT1, Jurkat, Molt3, and Molt4) were negative for both isoforms. K562 also showed a comparable positive staining for nm23-H2. Four representative examples are shown in Fig. 6.

DISCUSSION
Flow cytometric and enzymatic analysis were used to study the NDPK/nm23 expression in several plasma types and in all cells of hematopoietic origin. This article is mainly focused on the nm23 isoforms nm23-H1 and nm23-H2. They have a molecular mass of 21 and 19 kDa, respectively, are encoded by separate genes, colocalized on chromosome 17q21.3, which are probably independently regulated (28,29). Recently, two new isoforms, DR-nm23 and nm23-H4, have been characterized and localized on chromosome 16q13 and 16p13, respectively (4,30).
An interesting extracellular property of nm23-H1 and nm23-H2 is their ability to inhibit the induced differentiation of hematopoietic cell lines, independently of their enzymatic activity (13)(14)(15). In this study we demonstrated the presence of NDPK in plasma from peripheral blood, umbilical cord blood, and bone marrow. The highest activity was measured in bone marrow plasma. However, the observed correlation between the hemoglobin content and the NDPK phosphotransferase activity in the plasma samples indicated that the presence of nm23 in plasma is probably due to lysis of red blood cells. In the latter cells, a relatively high level of NDPK was detected. The high activity observed in bone marrow plasma is probably due to the harsher treatment of cells during bone marrow sampling. However, even in vivo, a limited lysis of senescent red blood cells occurs and causes release of hemoglobin and NDPK in the plasma. In addition, a basal level of 33 Ϯ 5 milliunits/ml was estimated in human peripheral blood plasma. These data suggest that in vivo, NDPK is present in plasma, and may have a physiological significance. We also demonstrated the presence of NDPK in several protein solutions, often used in cell cultivation. These findings emphasize that studies on the effect of extracellular nucleotides or on the differentiation inhibitory effect of exogenous NPDK should be performed in serum-free conditions, and with NDPK-free serum substitutes.
Previously, extracellular NDPK has been reported on the cell surface of several cell lines (23). These results were confirmed in our study, although the observed cell surface immunoreactivity was always small and close to the flow cytometric detection limit (Ϯ1000 molecules/cell). In contrast, the presence of ecto-NDPK on platelets (22) was not confirmed in our study. Moreover, we could not detect any nm23 surface expression on normal hematopoietic cells. Apparently, malignant cells seem to express more surface NDPK than normal cells. Probably this expression is not due to the proliferating state of the cell, since PHA-stimulated lymphocytes or CD34 ϩ progenitors did not express detectable surface-bound NDPK. In addition, nm23 surface expression seems to occur on tumor cells in a lineage specific way, since we detected this expression on all myeloid cell lines and could not demonstrate any on lymphoid cells. Whether nm23 surface expression can be used as a marker for certain hematopoietic malignancies remains to be determined.
Our data demonstrated the presence of NDPK in all investigated cells of hematopoietic origin, indicating the constitutive expression of nm23-H1 and nm23-H2 in these cells in agreement with the hypothesis that nm23 is ubiquitously expressed in all tissues (20). However, the main finding of our study is the inverse correlation between nm23 expression and hematopoietic maturation. In fact, hematopoietic CD34 ϩ progenitors express high levels of both nm23-H1 and -H2, and their expression decreases in more differentiated cell types. Previously, high levels of DR-nm23 mRNA have also been reported in CD34 ϩ progenitors (3). Taken together, these data point to an important role for nm23 in the early stages of hematopoiesis. The CD34 ϩ compartment represents a population of proliferation inducible and early proliferating cells, therefore the expression of NDPK might reflect the proliferative potential of hematopoietic cells. The data generated with PHA-stimulated lymphocytes also supported this hypothesis and demonstrated that during the G 0 /G 1 transition, peripheral blood lymphocytes up-regulate nm23-H1 and nm23-H2 approximately 10-fold, a level which is maintained during further cell cycle progression.
Although our findings indicate a correlation between nm23 expression and the proliferation of nucleated hematopoietic cells, the specific function of nm23 in this process remains unknown. Several non-enzymatic functions have been proposed for NDPK. Nm23-H2 has been identified as PuF, a transcription factor which positively regulates c-myc expression in Burkitt's lymphoma (6,31). This proto-oncogene is well known to be overexpressed in proliferating cells (32), and in vitro and in vivo terminal differentiation of hematopoietic cells is accompanied by down-regulation of c-myc expression (33,34). Therefore, the high nm23 expression in proliferating hematopoietic cells could be responsible for the constitutive expression of c-myc. However, in contrast to nm23, c-myc is not ubiquitously expressed in all cells. This apparent contradiction indicates that the transcriptional activity of nm23 must be strictly regulated. Regulatory mechanisms may include their subcellular localization, post-translational modifications, or changes in the oligomeric structure (35)(36)(37). Another important hypothesis concerning the level of nm23 expression is the necessity for cells of guanosine 5Ј-triphosphate (GTP) or other triphosphate nucleotides. In the promyelocytic leukemia cell line HL-60, a down-regulation of the intracellular GTP level after vitamin D 3 -induced granulocyte differentiation was demonstrated (38). In this cell line, the down-regulation of nm23 expression upon granulocyte differentiation has also been reported (39). Similarly, the high expression of nm23 in normal proliferating hematopoietic cells may reflect their necessity for a high intracellular GTP level. As proposed previously, NDPK may be involved in providing GTP for the activation of G-proteins, for the activation of the translation apparatus or for the polymerization of microtubuli (5,40). As induction of proliferation requires a reorganization of the cytoskeleton, it is not unlikely that the nm23 up-regulation is correlated with the latter process. Research is in progress to elucidate the function of NDPK in the G 0 /G 1 transition upon induction of proliferation in lymphocytes.