Double knockout Nme1/Nme2 mouse model suggests a critical role for NDP kinases in erythroid development
- First Online:
- Cite this article as:
- Postel, E.H., Zou, X., Notterman, D.A. et al. Mol Cell Biochem (2009) 329: 45. doi:10.1007/s11010-009-0110-9
- 228 Views
Nm23/NDP kinases A and B encoded by the Nme1/Nme2 genes are multifunctional enzymes responsible for the majority of NDP kinase activity in mammals. This review summarizes recent studies on their physiological roles using a mouse model in which both Nme1 and Nme2 genes have been deleted. The double knockout mice are stunted in growth and die perinatally. Additionally, these mice display hematologic phenotypes, including severe anemia, abnormal erythroid cell development, loss of the iron transport receptor molecule TfR1, and reduced iron uptake by Nme1−/−/Nme2−/− erythroid cells. We hypothesize that Nm23/NDP kinases regulate TfR1 gene expression in erythroid cells in some manner, and that defective iron transport into these cells is responsible for the anemia and death. This Nme1/Nme2 mouse model also links nucleotide metabolism with erythropoiesis, suggesting alternative or additional mechanisms that may explain the observed phenomena.
KeywordsErythropoiesis developmentTranscriptional controlMetabolic disorder
NDP kinases (NDPKs) are a textbook case of housekeeping enzymes that have been known for some 60 years to catalyze nonspecific transphosphorylation reactions between nucleoside di- and triphosphates through a conserved histidine residue. It was assumed initially that the primary physiological role of NDPKs is concerned with the general homeostasis of cellular nucleoside triphosphates. However, numerous biochemical and biological studies have concluded subsequently that NDPKs can also serve as regulatory proteins not necessarily connected to nucleoside phosphorylation, e.g., in normal Drosophila melanogaster development , and in the induction and progression of tumors . Indeed, the matriarch of the family, Nm23-1/NDPK-A, was originally cloned as a metastasis suppressor [3, 4]. NDPKs are also implicated in signal transduction [5, 6], transcriptional activation and repression [7–10], DNA repair [11–14], and in regulatory interactions with an assortment of protein partners [11, 15–19].
The human NDPK/NM23 genes NME1 and NME2 encode the NM23-H1/NDPK-A and NM23-H2/NDKP-B proteins, respectively. Both proteins are ubiquitous but are expressed differentially according to the tissue considered, together providing the majority (>95%) of NDPK activity in cells. NME1 and NME2 are also 88% identical in sequence  and are closely linked on chromosome 17q21.3 . Each gene encodes 152 amino acid long peptides, which, when assembled into identical three-dimensional native hexameric structures, carry out identical phosphotransferase reactions [22–24]. Hence, the A and B proteins are presumed to function in a similar manner, acting either independently or together.
The murine NDPK A and B share >98% sequence identity with their human counterparts , thereby offering an excellent model system in which to study for NDPK physiological functions. Gene deletion studies have shown that NDPK-A/Nme1-deficient mice lack overt phenotypic abnormalities and reproduce normally, but have reduced birth weights and manifest delayed mammary gland development . When Nme1−/− mice are crossed with transgenic mice that develop hepatocellular carcinoma (due to liver targeted expression of the SV40 T antigen), an increase in the incidence of lung metastasis is observed, supporting a role for Nme1 in metastasis inhibition . Null mutations in the unique Drosophila melanogaster homolog NDPK gene awd cause lethality in the larval stage . Inactivation of the yeast and bacterial ndk genes produce viable offspring [28–30], albeit with mutator phenotypes that have been attributed to imbalances in nucleotide pools [30, 31] and to DNA repair defects [11, 13]. However, none of the above studies identified the primary physiological roles or the mechanisms of action of these multitasking enzymes.
Here, we review the phenotypic properties of Nm23/NDP kinase A-and B-deficient mice  and offer hypotheses for how these molecules might act to regulate red blood cell development.
Methods and results
Phenotypes of Nme1/Nme2-deficient mice include stunted growth, severe anemia, and perinatal death
Nme1/Nme2-deficient mice were generated by deletion of adjacent coding regions of both genes as described and maintained as heterozygotes in a C57BL/6 background . Genotype analysis by PCR of DNA samples from embryos and newborn mice of heterozygous knockout parents indicated that the number of mice carrying wild type, heterozygous and homozygous knockout alleles was Mendelian. While the heterozygous offspring were morphologically well-developed and lived normal life spans, the homozygous knockout embryos and neonates, albeit morphologically normal, were 3–5 times smaller and profoundly anemic. Although the pups were born alive, they died within minutes of birth, suggesting that NDPK/Nme1/2 provide functions that are essential for the immediate survival of the neonates, while the embryos survive because these functions are partially provided through the cord blood. Assays of mouse embryonic fibroblasts (MEFs) confirmed that NDPK expression was eliminated in the Nme1−/−/Nme2−/− animals while MEFs from heterozygous animals produced the expected haploid amounts of proteins .
Abnormal peripheral blood hematology of Nme1−/−/Nme2−/− late phase embryos
The extreme paleness of the Nme1−/−/Nme2−/− mice and hypocellularity in their peripheral blood suggested severe anemia, which was confirmed by total blood count (CBC) and hematocrit (HCT) data. The significant reduction in red blood cell (RBC) counts closely mirrored decreased hemoglobin (HGB) and HCT values. Consistent with the microscopic data was the increased red cell distribution width (RDW), indicating variation in erythroid cell size. The automated white blood cell counts and differentials, as well as platelet counts, indicated no significant differences among the genotypic groups, confirming microscopic observations. Flow cytometric analysis of CD45+ (leukocyte) and CD61+ (platelet) populations confirmed the CBC findings .
Histological analysis of Nme1−/−/Nme2−/− embryos indicate abnormal erythroid development
The most dramatic histologic finding  in Nme1−/−/Nme2−/− mice was the presence of large numbers of nucleated blood cells within lumens of the aorta and other blood vessels in tissues throughout the body. In contrast, nucleated blood cells were rarely observed within vascular lumens of double heterozygous and wild-type mice. Peripheral blood smears from double knockouts were hypocellular and characterized by increased numbers of nucleated erythroid precursors and decreased numbers of mature, enucleated erythrocytes. Wild type and double heterozygous smears were composed predominantly of mature, enucleated erythrocytes and a few larger, deeply basophilic, nucleated erythroid precursors. Leukocytes, including neutrophils and lymphocytes, and platelets were similarly present in smears from all three genotypes.
Livers from all the three genotypic groups had extramedullary hematopoiesis, which was predominantly erythropoietic. Hematopoietic cells in wild type and double heterozygous livers were arranged as multifocal discrete islands around blood vessels and within sinusoids, whereas in double knockouts these cells were increased in number and widely dispersed throughout the liver .
Sagittal sections of long bones indicated well-defined islands of hematopoietic cells, including myeloid and erythroid cells, as well as megakaryocytes in wild type and double heterozygous mice. Islands of hematopoietic cells in the Nme1−/−/Nme2−/− bone marrow, although present, were much smaller and less cellular. There were no overt differences in other hematolymphoid organs such as the spleen, adrenal gland, and thymus between the genotypic groups .
Flow cytometric analysis of erythroid cell differentiation indicates that the iron transport molecule TfR1 is a downstream target of Nm23/NDPK
Erythroid cell differentiation was studied using flow cytometry to separate erythroid cells into distinctive populations of progressively differentiated cells based on their expression of Ter119, the murine erythroid-specific glycophorin A receptor, and CD71, the non-erythroid specific transferrin receptor-1 (TfR1). Early to late erythroid progenitors normally express CD71 but lack the Ter119 receptor. As erythroid cells continue to divide and differentiate, Ter119 expression is induced by erythropoietin (EPO) while CD71 is downregulated . Analysis of blood erythroblasts from embryonic days E14.5, E16.5, and E18.5 indicated a trend at E18.5 toward loss of the CD71 receptor in Nme1−/−/Nme2−/− blood samples relative to the wild type. An explanation consistent with both the pathological and flow cytometry data is that CD71/TfR1 is downregulated, directly or indirectly, as a consequence of the NDPK/NM23 defect . Expression of additional erythroid membrane proteins including the metastasis-associated cell adhesion molecule CD44, and an integrin-associated red cell marker CD47 indicated that while the expression of the CD71/TfR1 receptor was reduced threefold over wild-type levels, the expression of CD44 and CD47 proteins was not affected by the NDPK/Nm23 deletion .
Altered iron status of Nme1−/−/Nme2−/− blood
As TfR1 expression is a known indicator of iron status, it was expected that the anemia of Nme1−/−/Nme2−/− mice resulted from iron deficiency. Indeed, iron levels were significantly reduced in the Nme1−/−/Nme2−/− blood. Moreover, the distribution of iron between serum and blood cells (~95% of which are erythroid) was also abnormal, with iron levels in serum about 10-fold higher in Nme1−/−/Nme2−/− blood, concomitant with reduced cellular iron. Reduction in heme levels was also observed but was in agreement with other hematologic data, suggesting that iron utilization, at least with respect to heme synthesis, is not affected by the absence of NDPK/Nm23 .
It is important to add here that while the heterozygous knockout mice are physically and histologically indistinguishable from wild type, their hematologic indices and heme levels were somewhat abnormal, implying that the expression of both Nme alleles is necessary for normal erythroid development.
Nme1−/−/Nme2−/− mice are defective in definitive erythropoiesis
The abnormal peripheral blood hematology of Nme1−/−/Nme2−/− mice, including an unusually large number of circulating immature nucleated erythroblasts and abnormal hematopoiesis in the liver and bone marrow  suggest a defect in definitive, rather than primitive, erythropoiesis. This conclusion is based on the facts that primitive erythropoiesis in the mouse normally begins around E7, at which time nucleated red blood cells originate in the blood islands of the yolk sac, while definitive erythropoiesis sets in around E10, the time when the site of red blood cell production shifts to the fetal liver [34, 35]. Evidence of erythropoiesis in double knockout embryos at E18.5 was present in the liver, spleen, and adrenal glands but was essentially absent in the bone marrow, which is normally seeded around E16 . Therefore, the Nme1/Nme2 requiring events are likely to be associated with the onset of the last trimester (around day 14) of gestation, a time when the switch from embryonic to adult globin gene expression, i.e., from primitive to definitive erythropoiesis, has already taken place. The types of hemoglobin accumulated can distinguish primitive and definitive erythropoiesis. Our analyses of adult α- and ß-major globin and ßh-1 fetal globin expression by reverse transcripton of RNA indicated that hemoglobin abnormalities are not the cause of anemia, iron transport deficiency, and differentiation defects in the Nme1−/−/Nme2−/− animals.
TfR1 is downregulated in Nme1−/−/Nme2−/− erythrocytes concomitant with changes in blood iron parameters
Following the expression of CD71/TfR1, the ubiquitous transferrin receptor-1 molecule TfR1 on Ter119-positive cells revealed a surprisingly premature and significant loss of CD71/TfR1 between E16.5 and E18.5 in the Nme1−/−/Nme2−/− blood. However, because of the preponderance of immature nucleated erythroblasts in the E18.5 blood, loss of CD71/TfR1 was interpreted as a downstream effect of the absence of NDPK/Nm23 rather than as a consequence of accelerated differentiation. The expression of two additional red cell receptors, CD44 and CD47, was not altered by the NDPK/Nm23 deletion , suggesting that the TfR1 gene may be a specific target of the Nm23/NDPK molecules. Whether this is a direct or an indirect effect of Nm23/NDPK and/or the participation of additional target genes and their downstream effectors also contribute to the hematologic observations remains to be determined.
The most important role for TfR1 in the development of erythroid precursors is that of importing iron for hemoglobin synthesis . Consistent with the loss of TfR1 in Nme1−/−/Nme2−/− mice is their abnormal blood iron status as discussed above. As proper iron metabolism is necessary for hemoglobin synthesis and oxygenation, these findings suggest that iron deficiency in Nme1−/−/Nme2−/− erythroid precursors may be causing the anemia, growth retardation, and perinatal death. The significance of TfR1 to iron homeostasis in erythroid cells has already been demonstrated in Tfr1 null mice, which are also small and anemic but die earlier in utero .
Molecular models of erythropoiesis regulation by NDPK/Nm23
Because NDPK/Nm23/Awd proteins act as molecules of many different functions in ex vivo experiments, their effect on erythropoiesis could occur on several different levels and by several different mechanisms, including transcriptional and post-transcriptional regulation, control by iron levels, and/or by nucleotide pool sizes and signaling. The finding that the active site for nucleoside triphosphate synthesis and for DNA transactions of NM23-H2 are shared , suggests that these two activities are part of a larger scheme.
In the course of erythroid development TfR1 is regulated predominantly at the transcriptional level [39, 40], and active transcription of TfR1 is one of the essential molecular mechanisms for efficient hemoglobin synthesis. These requirements, together with observations that TfR1 expression in the Nme1/Nme2 null erythrocytes is attenuated might suggest that the TfR1 gene is an NM23/NDPK downstream target. This would also be consistent with previous observations that NM23/NDPK can function as transcriptional activators [7–10]. The sequence specific DNA-binding elements required for transcription by Nm23/NDPK [7–10, 41] are conserved in the TfR1 promoter . However, little is known about the transcriptional regulatory pathways to TfR1 expression in developing erythroid cells. Assuming that Nm23/NDPK participates in regulating TfR1 expression, one pathway that might be utilized is phosphoinositol 3-kinase (P13 K) signaling, since P13 K is involved both in the downstream activation of TfR1  and in the regulation of the mouse NDPK-B/Nm23-M2 gene . Regulation through interactions with other proteins in the pathway is another possible mechanism.
In addition to their transcriptional roles, both NDPK/NM23-H1 and H2 possess DNAse activities [12, 44, 45]. If this nuclease activity were to be necessary for chromosomal fragmentation during terminal stages of differentiation, it could explain the failure of Nme1−/−/Nme2−/− erythroid cells to enucleate.
Given the highly conserved albeit nonspecific NDP kinase activity of Nme1/Nme2 proteins, it is possible that the double knockout phenotype is somehow caused by a defect in nucleotide metabolism involving nucleotide phosphorylation and/or binding. Insufficient nucleotide pool sizes would be expected to have the effect of limiting rapid growth of erythropoietic cells necessary for expansion. In this scenario, however, one would expect failure of growth and cellular expansion in other organs as well. While we did not observe any gross anatomical defects in the knockout neonates/embryos nor did we discover any significant effects on other hematopoietic cell lineages, it is highly likely that upon further examination additional developmental defects and additional targets will be discovered. One such target may be the nervous system, where TfR1-mediated iron uptake is required for neuronal development .
Linkage between nucleotide and iron homeostasis
In an NDP kinase-deficient yeast mutant with iron phenotypic defects, overexpression of human mitochondrial NDPK-D/NM23-H4 targeted to the matrix resulted in correction of this defect . The authors suggested that GDP phosphorylation to GTP by ATP and mitochondrial NDP kinase/Nm23-H4 somehow restored normal iron regulation and guanine nucleotide transport involving GTPases. Another possible interpretation might be that NDPK-D/NM23-H4, capable of binding DNA (Postel and Lacombe, unpublished observations), has restored iron and heme levels in mitochondria through transcriptional upregulation of a hypothetical TfR1-like receptor. NDPK may also locally modulate the GTP pool in erythroid cells and thereby regulate GTPases  in concert with chemokines and cytokines influencing signal transduction. As a source of guanine nucleotides, NDPK could influence TfR1 turnover by controlling endocytosis, not unlike the dynamin-dependent synaptic vesicle recycling process that requires GTP . NDPK has also been suggested as an activator of the GTPase dynamin, thereby facilitating endocytosis of growth factor receptors and attenuating growth factor signaling .
Defects in colonization and seeding
Increased hematopoiesis in the liver and decreased hematopoiesis in the bone marrow in Nme1−/−/Nme2−/− embryos may indicate that TfR1, iron levels, and/or nucleotide composition somehow affect colonization and the seeding of erythroid precursors. Indeed, NDPK has been linked to homing as a supplier of UTP, which, acting as a chemotactic molecule, is a potent inducer of hematopoietic stem cell migration .
Many more questions remain, including whether under physiological conditions NDPK-B alone is regulating erythropoiesis, or both the A and B enzymes are necessary and to what degree do they cooperate. Single knockout Nme1−/− mice do not exhibit hematologic phenotypes . On the other hand, both A and B human isoforms have been implicated in in vitro erythropoiesis and in other hematopoietic tissue development [51, 52], and the Drosophila NDPK/Awd protein in a hematopoietic regulatory pathway . While it is likely that in vivo NDPK A is also concerned with erythropoiesis, single knockout of the Nme2 gene should be an obvious next consideration.
Summary and conclusions
Recent studies using double knockout Nme1/Nme2 NDP kinase A and B-deficient mice  offer some new insights into the physiological roles of these multitasking enzymes. The findings can be summarized as follows: (1) Nme1/Nme2-deficient mice are not viable and display severe anemia and additional hematologic phenotypes; (2) the iron transport receptor TfR1 is downregulated in Nme1−/−/Nme2−/− erythroid cells; and (3) these cells have considerably reduced iron levels. We suggest that these observations are related and that dysregulation of the TfR1 gene, a downstream effector of NDPK/Nm23, is responsible for at least some of the phenotypic observations. The linkage established by this novel Nme1/Nme2 knockout mouse model between erythropoiesis and nucleotide metabolism remains to be elucidated, as well as the pathogenesis of dyserythropoiesis and the detailed mechanisms by which NDPK/Nm23 molecules function in vivo.
This work is supported by NIH/NCI grant RO1 CA76496 (to EHP). The double heterozygous knockout mice were a gift from Amgen Inc. The authors acknowledge helpful discussions with Drs. Achille Iolascon and Stefano Rivella.