Original Article | Published:

Acute Leukemias

A dose-dependent tug of war involving the NPM1 leukaemic mutant, nucleophosmin, and ARF

Leukemia volume 23, pages 501509 (2009) | Download Citation


In acute myeloid leukaemia (AML), nucleophosmin-1 (NPM1) mutations create a nuclear export signal (NES) motif and disrupt tryptophans at NPM1 C-terminus, leading to nucleophosmin accumulation in leukaemic cell cytoplasm. We investigated how nucleophosmin NES motifs (two physiological and one created by the mutation) regulate traffic and interaction of mutated NPM1, NPM1wt and p14ARF. Nucleophosmin export into cytoplasm was maximum when the protein contained all three NES motifs, as naturally occurs in NPM1-mutated AML. The two physiological NES motifs mediated NPM1 homo/heterodimerization, influencing subcellular distribution of NPM1wt, mutated NPM1 and p14ARF in a ‘dose-dependent tug of war’ fashion. In transfected cells, excess doses of mutant NPM1 relocated completely NPM1wt (and p14ARF) from the nucleoli to the cytoplasm. This distribution pattern was also observed in a proportion of NPM1-mutated AML patients. In transfected cells, excess of NPM1wt (and p14ARF) relocated NPM1 mutant from the cytoplasm to the nucleoli. Notably, this distribution pattern was not observed in AML patients where the mutant was consistently cytoplasmic restricted. These findings reinforce the concept that NPM1 mutants are naturally selected for most efficient cytoplasmic export, pointing to this event as critical for leukaemogenesis. Moreover, they provide a rationale basis for designing small molecules acting at the interface between mutated NPM1 and other interacting proteins.


Nucleophosmin-1 (NPM1) is a phosphoprotein that plays a key role in the regulation of ribosome biogenesis, centrosome duplication and genomic stability.1 NPM1 also interacts with tumour suppressors p14ARF and p53, and influences the cellular apoptotic response, although its exact role in this pathway remains controversial.1 Functions of nucleophosmin are closely related to its chaperone2 and shuttling properties.3 Shuttling and subcellular localization of NPM1 is, in turn, regulated by specific functional domains of the protein and post-translational modifications, such as sumoylation and phosphorylation.4, 5, 6, 7 Moreover, NPM1 partners, such as p14ARF, can influence NPM1 activity, modulating its nucleocytoplasmic shuttling.8, 9 Although functionally important, the shuttling pool of the protein is minimal and at immunocytochemistry, the localization of NPM1 is characteristically restricted to the nucleolus.10

In 2005, we identified heterozygous NPM1 mutations as the most common genetic lesion in adult acute myeloid leukaemia (AML) (about 30% of cases),11 and subsequently provided evidence that AML with mutated NPM1 exhibits distinctive biological and clinical characteristics12, 13, 14 that support its inclusion as a provisional entity in the new WHO classification. Approximately 40 NPM1 mutations have been identified to date,13 and the salient feature of all the mutants they generate is their aberrant localization in leukaemic cell cytoplasm.11

Ectopic distribution of NPM1 mutants in AML cells is due to the concerted action of mutated tryptophan(s) at position 288 or 290 and a new nuclear export signal (NES) motif at the NPM1 protein C-terminus.15 Under normal conditions, the two tryptophans are critical for keeping the C-terminus globular domain of wild-type NPM1 folded,16 which is essential for NPM1 binding to the nucleolus. In leukaemic cells, mutation-induced tryptophan changes unfold the domain and consequently impair NPM1 targeting of the nucleolus.16 Cytoplasmic accumulation of NPM1 leukaemic mutants also depends upon the NES motifs. Two are physiological and are located at the N-terminus (corresponding to residues 42–49 and 94–102).17, 18 The third motif is introduced by the mutational event at the NPM1 C-terminus (residues 287–296).19 However, it is not yet clear how much each of the three NES motifs contributes to mutant protein nuclear export. Moreover, although NPM1 mutants recruit wild-type NPM1 and p14ARF in cytoplasm,20, 21 the dynamics of interactions among these proteins and the functional domains that regulate them are still poorly understood.

In this paper, we explore the role of the three NES motifs in controlling NPM1 mutant traffic and in mediating interaction with wild-type NPM1 and p14ARF proteins. Our findings reinforce the concept that NPM1 mutations in AML are designed to ensure the most efficient cytoplasmic export of mutant proteins.

Materials and methods

Plasmid constructs

To generate NPM1 derivative constructs in which NES(s) were deleted, we used pEGFP-C1-NPM1wt and pEGFP-C1-NPM1mA11 as templates and the QuikChange Multi Site-Directed Mutagenesis Kit (Stratagene, La Jolla, CA, USA) to introduce mutations at codons encoding for two core hydrophobic amino acids (either L → G or V → G) of each NES. The NES starting at position 4218 was mutated at residues 44 and 47, the NES starting at position 9417 was mutated at residues 100 and 102, and the NES starting at position 28715 was mutated at residues 291 and 294 (Figure 1a). Primer sequences are available as Supplementary Information.

Figure 1
Figure 1

Subcellular localization of NPM1 mutants is influenced at different extents by the three NES motifs. (a) Scheme summarizing how NPM1wt and NPM1mA were mutated at each NES, alone or in combination: the protein sequence is depicted in grey, whereas a red box indicates the presence of the NES. (b) Subcellular localization of each NPM1mA NES-deleted mutant is shown in green. (c) Quantification of the above experiment. Top panel: each bar indicates the percentage of cells with cytoplasmic-restricted positivity for the eGFP-NPM1mA fusion proteins. Middle panel: cells showing both nuclear and cytoplasmic positivity. Bottom panel: percentage of cells showing a nuclear-restricted cytoplasmic positivity. NES, nuclear export signal; NPM1, nucleophosmin-1.

To generate a DsRed-monomer-NPM1wt fusion protein, human NPM1wt cDNA was PCR amplified from pEGFP-C1-NPM1wt and subcloned into pDsRed-monomer-C1 (Clontech, Palo Alto, CA, USA) using standard cloning procedures. Human p14ARF cDNA was PCR amplified from pGEX-2T- p14ARF (a kind gift from Dr Wei Gu, Columbia University, NYC, NY, USA) and subcloned into pDsRed-monomer-C1 (Clontech) using standard cloning procedures. All plasmids were validated by sequencing.

Cell culture and transfection procedures

NIH-3T3 murine fibroblasts were cultured in Dulbecco's modified Eagle's medium supplemented with 10% bovine calf serum, 1% glutamine and antibiotics. For transfection purposes, NIH-3T3 cells were seeded on glass coverslips and transfected 24 h later using Lipofectamine 2000 (Invitrogen, Carlsbad, CA, USA) following the manufacturer's instructions. For co-transfection experiments, two plasmids were co-transfected at a 1:1 molar ratio in the same Lipofectamine 2000 tube. In dose-dependency experiments, the two plasmids were transfected at 1:8, 1:1 and 8:1 molar ratios.

Immunofluorescence analysis

In immunofluorescence and confocal microscope studies, glass coverslips with transfected or co-transfected cells were rinsed in phosphate-buffered saline, fixed in 4% paraformaldehyde pH 7.4 (10 min), washed three times in phosphate-buffered saline, air dried and flipped on to standard glass slides with Mowiol mounting medium. When indicated, nuclei were counterstained with propidium iodide (2.5 μg/ml final concentration) following standard procedures. Confocal microscopy analysis was performed as described earlier.22

To quantify subcellular localization of protein, cells were observed with an Olympus AX70 epifluorescence microscope (Olympus, Center Valley, PA, USA) equipped with a 100/1.4 numerical aperture oil immersion objective and a MNIBA optical filter cube (excitation, 470–495 nm; dichroic mirror, 505 nm; emission, 510–550 nm). A minimum of 200 cells for each slide was examined by two ‘blinded’ independent observers. Cells with or without cytoplasmic fluorescence were counted and their percentage calculated using GraphPad Prism 4 software (GraphPad Software), as described earlier.23 Each experiment was repeated three times. Data showed good interexperiment and interobserver reproducibility.

Western blot and co-precipitation studies

For immunoprecipitation experiments, co-transfected cells were lysed in 1 ml of ice-cold lysis buffer (0.5% NP-40, 150 mM NaCl, 25 mM Tris (pH 7.5), 1 mM EDTA, 1 mM Na3VO4, 1 μg/ml leupeptin, 1 μg/ml aprotinin and 1 mM phenylmethylsulphonyl fluoride). After 20 min of incubation on ice, cell lysates were passed several times through a 27-gauge needle to disrupt nuclei and then centrifuged at 14 000 g (10 min at 4 °C). Lysates were then incubated with 2 μg of mouse anti-green fluorescent protein (GFP) monoclonal antibody (Roche Applied Science, Indianapolis, IN, USA) and 30 μl of Protein A/G Plus-agarose beads (Santa Cruz Biotechnology Inc., Santa Cruz, CA, USA) rocking overnight at 4 °C. Beads were washed with buffer containing 0.1% NP-40, 150 mM NaCl, 25 mM Tris (pH 7.5), 1 mM EDTA and inhibitors. Proteins were separated by SDS-polyacrylamide gel electrophoresis, transferred to polyvinylidene difluoride (Millipore, Billerica, MA, USA) and incubated with either anti-NPM1 (mAb clone 376, pure hybridoma supernatant) or mouse mAb anti-p14ARF clone 14P02 (Abcam Inc., Cambridge, MA, USA; 1 μg/ml). After incubation with the appropriate secondary antibody, the signal was revealed by enhanced chemiluminescence (Amersham Bioscience, Munich, Germany).

Immunohistochemical studies

NPM subcellular expression was assessed in B5-fixed/EDTA-decalcified bone marrow trephines from 100 patients with NPM1-mutated AML. Paraffin sections were stained with an anti-NPM monoclonal antibody (clone 376) using the alkaline phosphatase anti-alkaline phosphatase procedure,11 as described earlier. Monoclonal antibody against nucleolin/C23 was purchased from Santa Cruz Biotechnology.


NPM1 leukaemic mutant subcellular localization is influenced differently by each of the three NES motifs

To assess how each of the three NES motifs contained in the natural leukaemic NPM1mA contributed to regulate its traffic, we generated artificial NPM1mA mutants with one, two or all three NES deleted (Figure 1a).

Transient expression of these artificial mutants fused to eGFP in NIH-3T3 murine fibroblasts revealed that the export signal strength of each NES motif was additive to the others. Export into cytoplasm was most efficient when NPM1mA contained all three NES motifs, as characteristically occurs in patients with NPM1-mutated AML.11 Under these circumstances, localization of the mutant was restricted to the cytoplasm (Figures 1b, top left; Figure 1c).

Quantitative experiments showed that each NES motif made a different contribution to overall mutant export (Figures 1b and c). Specifically, the artificial NPM1mA mutant lacking either or both NES motifs 42 and 94 was exported into cytoplasm less efficiently than NPM1mA, with most of the cells showing cytoplasmic and nucleoplasmic positivity (Figure 1b, top right and middle panels; Figure 1c). On the other hand, the artificial NPM1mA mutant containing both NES motifs at position 42 and 94 (that is, lacking only NES 287) showed no cytoplasmic export and re-gained access to the nucleolus (Figure 1b, bottom left; Figure 1c) in spite of disruption of nucleolar localization signals (NoLSs) (due to mutations of tryptophans 288 and 290). The NPM1mA mutant lacking all the three NES motifs was not exported into cytoplasm either, but localized mainly to the nucleoplasm (Figure 1b, bottom right; Figure 1c).

The above findings strongly suggest that the two N-terminal NES motifs influence not only the cytoplasmic export of NPM1 mutants but also their capability to bind to the nucleolus. As the two N-terminal NES motifs are embedded in the NPM1 oligomerization domain, we hypothesized that they could influence mutant subcellular traffic by interfering with its ability to form heterodimers with NPM1 wild-type (NPM1wt) protein.

Changes in nucleophosmin N-terminal NES motifs interfere with NPM1wt//NPM1 mutant heterodimerization and subcellular distribution

To study the interactions between NPM1wt and NPM1mA proteins, we created a DsRed-Monomer-NPM1wt fusion protein and transiently co-expressed it together with eGFP-NPM1mA at different molar ratios. Different amounts of transfected plasmid resulted in proportionally different exogenous protein levels. At equimolar ratios, NPM1mA was mainly cytoplasmic, whereas NPM1wt was mainly nucleolar (Figure 2a, top panel, middle column), as is usually observed in bone marrow samples from patients with NPM1-mutated AML.24 Interestingly, excess doses of NPM1mA pulled almost all NPM1wt protein out of the nucleoli into nucleoplasm and cytoplasm (Figure 2a, top panel, left column), whereas excess doses of NPM1wt relocated NPM1mA from the cytoplasm to the nucleoli (Figure 2a, top panel, right column). These findings clearly indicated a dose-dependent reciprocal influence of NPM1mA and NPM1wt on the other's subcellular distribution.

Figure 2
Figure 2

Changes in the NPM N-terminal NES motifs influence NPM hetero- and homodimerization. (a) NPM1mA (in green) and NPM1wt (in red) influence each other's subcellular localization. Top panel: an excess of NPM1mA (left column) dislocates NPM1wt to the cytoplasm, whereas an excess of NPM1wt relocates NPM1mA to the nucleoli (right column). At equimolar doses (middle column), the situation is intermediate. This dose-dependent tug of war is lost when NPM1mA mutants lacking NES motifs at residues 42 and 94 are challenged against NPM1wt (second and third panels from the top), and restored with an oligomeric NPM1mA mutant (NPM1mA Δ287). (b) Co-immunoprecipitation experiment showing how NPM1mA and NPM1wt mutants lacking NES motifs 42 and 94 can not bind endogenous NPM1 in NIH3T3 cells. (c) NPM1wt (in green) is partially or totally dislocated from nucleoli upon loss of either or both (respectively) NES motifs at residues 42 and 94. The nucleus is counterstained in red by propidium iodide. (d) Quantification of the experiment shown in (c): each bar indicates the percentage of nucleolar-positive cells for each NPM1wt NES-deleted mutant. NES, nuclear export signal; NPM1, nucleophosmin-1.

We then used this dose-dependent co-transfection assay to determine whether changes in the N-terminal NES motifs 42–49 and 94–102 may influence the NPM1 traffic across nucleoplasmic and nucleolar compartments. Interestingly, when NPM1wt was co-transfected with the mutant lacking both NES motifs 42 and 94 (NPM1mA Δ42–94), the two recombinant proteins did not influence each other's nucleoplasmic vs nucleolar localization no matter what the molar ratio of the co-transfection assay was (Figure 2a, second panel from the top); a similar pattern was observed with artificial mutants lacking all three NES motifs (NPM1mA Δ42–94–287) (Figure 2a, third panel from the top). In contrast, NPM1wt and the mutant retaining the two NES motifs 42 and 94 (NPM1mA Δ287), reciprocally influenced each other's subnuclear distribution in a dose-dependent manner. Indeed, excess doses of NPM1wt relocated NES NPM1mA Δ287 from the nucleoplasm to the nucleoli (Figure 2a, bottom panel, right column) and vice versa (Figure 2a, bottom panel, left column). These findings suggested that the deletion of the two N-terminal NESs altered the mutant traffic by disrupting the heterodimerization interface between NPM1wt and NPM1 mutant proteins.

To address this issue, we carried out co-immunoprecipitation experiments in transiently transfected NIH-3T3 fibroblasts. As expected, the NPM1mA immunoprecipitated endogenous NPM1wt (Figure 2b), whereas the mutant lacking NES motifs 42 and 94 (NPM1mA Δ42–94) did not (Figure 2b). These findings provide evidence that the deletion of two core amino acids in the two N-terminal NPM1 NES motifs (residues 44 and 47; residues 100 and 102) alters the dimerization interface between NPM1wt and NPM1 mutants, and impacts upon the subcellular distribution of each.

These findings served to elucidate the differences in NPM1 mutant subcellular localization shown in Figure 1b. The artificial NPM1mA mutants lacking both or either NES motifs 42 and 94 are exported somehow more efficiently into cytoplasm because they cannot form heterodimers with NPM1wt. On the other hand, the artificial NPM1mA mutant containing both NESs at positions 42 and 94 (ie, NPM1mA Δ287) re-gains access to the nucleolus (Figure 1b), despite having a disrupted NoLS, because it can form heterodimers with NPM1wt.

The above results suggest that, when NPM1wt forms a heterodimer with NPM1mA, it drives the complex to the nucleolus and opposes its nuclear export. In AML with mutated NPM1, this equilibrium is disturbed, resulting in an aberrant cytoplasmic export of nucleophosmin.11, 13

Changes in nucleophosmin N-terminal NES motifs influence endogenous NPM1wt homodimerization and subcellular localization

To explore the role of N-terminal NES motifs in NPM1wt nucleolar localization, we determined whether this protein, with its intact NoLS, entered nucleoli upon mutation at NES motifs 42–49 and 94–102. We expressed GFP-NPM1wt mutants carrying mutations at either or both NES motifs 42 and 94 in NIH-3T3 cells and found that the disruption of one NES greatly impaired the ability of NPM1wt to target nucleoli. Disruption of both the NES motifs prevented the protein from entering the nucleoli, and NPM1wt Δ42–94 localized only in the nucleoplasm (Figure 2c; quantification in Figure 2d).

Co-immunoprecipitation experiments showed that NPM1wt immunoprecipitated endogenous NPM1 (Figure 2b), whereas the mutant with altered NES motifs 42 and 94 (NPM1wt Δ42–94) did not (Figure 2b). These results further support the hypothesis that intact N-terminal NES motifs are a requisite for entering the nucleoli, probably because they mediate NPM1wt oligomerization.

Variable subcellular distribution of wild-type NPM1 in AML with mutated NPM1

We then asked whether the subcellular distribution of endogenous and mutant NPM1 proteins observed in transfection experiments was also seen in leukaemic cells from cases with NPM1-mutated AML. To address this issue, we looked at the subcellular localization of nucleophosmin in 100 patients with NPM1-mutated AML after immunostaining with a monoclonal antibody (clone 376) that recognized both wild-type and mutant NPM1 proteins.11 Using this test, we were able to assess nucleophosmin distribution across cytoplasmic and nuclear compartments but not to distinguish between nucleoplasmic and nucleolar nucleophosmin as this is not possible in fixed samples.11 Leukaemic cells from all the 100 NPM1-mutated AML showed the expected aberrant cytoplasmic positivity, which is due to both NPM1 mutant and wild-type NPM1 proteins (the latter being recruited by the mutant into the cytoplasm)15 (Figure 3). In contrast, nuclear staining for NPM1 (which is exclusively due to NPM1wt22) was quite heterogeneous, ranging from cases (about 80%) in which the totality of leukaemic cells showed a nuclear positivity for NPM1 (Figure 3, top left) to others (about 5%) in which no leukaemic cell showed nuclear positivity for NPM1 as the protein was totally dislocated to the cytoplasm (Figures 3, bottom left and right), with all gradations in between (Figure 3, top right).

Figure 3
Figure 3

Leukaemic cells from bone marrow samples of NPM1-mutated AML show heterogeneous NPM1 nuclear expression. (Top left) NPM1 staining pattern most commonly observed in AML with mutated NPM1: the totality of the leukaemic cells show both cytoplasmic and nuclear positivity for nucleophosmin (bone marrow biopsy). Cells showing nucleus-restricted positivity in the centre of the field represent normal residual haemopoietic cells. (Top right) In this case of AML with mutated NPM1, the totality of leukaemic cells show an aberrant cytoplasmic expression of nucleophosmin, whereas the expression of the protein at the nuclear level is heterogeneous, with some leukaemic cells being completely negative (double arrows). Cells showing nucleus-restricted positivity (arrowhead) represent normal residual haemopoietic cells. (Bottom left) This is a representative example of the NPM1 staining pattern less commonly observed in AML with mutated NPM1. The totality of the leukaemic cells (myeloid blasts indicated by double arrows; leukaemic megakaryocytes indicated by a single arrow) shows cytoplasmic-restricted positivity for nucleophosmin, whereas no expression of the protein is observed at the nuclear level. Cells showing nucleus-restricted positivity (arrowhead) represent normal residual haemopoietic cells. (Bottom right) This is a case of NPM1-mutated myeloid sarcoma (skin biopsy) showing the same staining pattern for NPM1 as Figure 3, bottom left. The single arrows indicate leukaemic cells with cytoplasmic-restricted NPM1 positivity. Cells showing nucleus-restricted positivity (double arrows) represent normal residual stromal cells. AML, acute myeloid leukaemia; NPM1, nucleophosmin-1.

The above immunohistological findings in AML patients clearly demonstrate that the amount of nuclear wild-type NPM1 may vary from one case to another, possibly because different expression ratios between the two NPM1 alleles dictate the amount of NPM1wt that is recruited by the mutant in the cytoplasm. In contrast, the expression of mutated NPM1 is consistently restricted to the cytoplasm, and its nuclear relocation (documented in cells transfected at high NPM1wt/NPM1mA ratio) is not observed in AML patients.

Mutated NPM1 and p14ARF proteins reciprocally influence each other's subcellular distribution

As wild-type and mutant NPM1 proteins influence each other's subcellular distribution in a dose-dependant manner, we asked whether these reciprocal traffic alterations also occur when NPM1 mutants interact with protein partners other than endogenous NPM1. We selected the p14ARF protein as interacting partner, as it forms high-molecular weight complexes with NPM1 within the nucleoli,9 and is known to be delocalized by NPM1 leukaemic mutant.20, 21 We co-transfected NIH-3T3 cells (which lack the endogenous p19Arf genetic locus) with pEGFP-C1-NPM1mA and pDsRed-Monomer-C1-p14ARF at different p14ARF-to-NPM1mA molar ratios. Different amounts of transfected constructs resulted in proportionally different protein expression levels. As observed in the NPM1wt–NPM1mA interaction, the NPM1 mutant and p14ARF proteins appeared to exert a dose-dependent, reciprocal influence on each other's subcellular localization. In fact, a high NPM1mA/p14ARF ratio resulted in p14ARF accumulation in the nucleoplasm and cytoplasm (Figure 4a, top panel, left column), whereas high p14ARF levels relocated most NPM1mA protein to the nucleoplasm and nucleolus (Figure 4a, top panel, right column). Again, as observed in the interaction between mutant and wild-type NPM1, non-oligomeric NPM1 mutants (that is, NPM1mA Δ42–94 and NPM1mA Δ42–94–287) were unable to dislocate p14ARF from the nucleoli, and neither protein influenced the subcellular localization of the other, no matter what the expression level was (Figure 4a, second and third panels from the top). This observation suggests that changes in the two N-terminal NPM1 NES motifs not only impair NPM1 homodimerization but also blunt the NPM1 functional interaction with p14ARF. As a positive control, NPM1–p14ARF interaction was restored with an oligomeric NPM1 mutant: high doses of NPM1mA Δ287 were able to pull p14ARF outside the nucleoli, whereas high p14ARF levels were able to relocate most of NPM1mA Δ287 to the nucleoli (Figure 4a, bottom panel).

Figure 4
Figure 4

Mutated NPM1 proteins and Arf reciprocally influence each other's subcellular distribution. (a) NPM1mA (in green) and p14ARF (in red) influence each other's subcellular localization. Top panel: an excess of p14ARF (left column) dislocates NPM1mA to the cytoplasm, whereas an excess of NPM1mA relocates p14ARF to the nucleoli (right column). At equimolar doses (middle column), the situation is intermediate. This dose-dependent tug of war is lost when NPM1mA mutants lacking NES motifs at residues 42 and 94 are challenged against p14ARF (second and third panels from the top), and restored with an oligomeric NPM1mA mutant (NPM1mA Δ287). (b) Co-immunoprecipitation experiment showing how NPM1mA Δ42–94 can bind p14ARF in NIH3T3 cells without forming a ternary complex with endogenous NPM. On the contrary, NPM1wt, and NPM1mA Δ287 bind p14ARF in a complex involving endogenous NPM. NES, nuclear export signal; NPM1, nucleophosmin-1.

To demonstrate that the NPM1 mutants interacted physically with p14ARF proteins, we carried out co-immunoprecipitation experiments in cells co-expressing GFP-tagged NPM1 mutants and DsRed-monomer-p14ARF. Anti-GFP immunoprecipitation followed by anti-p14ARF western blot confirmed that wild-type and mutant NPM1 bind p14ARF (Figure 4b). Interestingly, NPM1mA Δ42–94 also binds p14ARF although to a lesser extent (Figure 4b), without immunoprecipitating wild-type NPM1, providing evidence that the NPM1 mutant can directly bind p14ARF. Indeed, this appeared to be in contrast with the inability of NPM1mtA Δ42–94 to pull p14ARF out of the nucleoli and vice versa, and needs to be further clarified experimentally. Interestingly, the existence of an NPM1 C-terminal binding site for p14ARF apparent only upon disruption of the oligomerization domain has already been postulated.25 We speculate that this latter site could have lower affinity for p14ARF and therefore is proved to be insufficient to pull p14ARF out of the nucleoli. Moreover, as shown above, NPM1 can enter the nucleoli only as part of an oligomeric complex, and there it binds to nucleolar p14ARF. As NPM1mtA Δ42–94 does not bind NPM1 wild-type protein and is excluded from nucleoli, we speculate that the p14ARF band seen in the NPM1mtA Δ42–94 co-immunoprecipitation experiment could come from the less-represented p14ARF nucleoplasmic pool, which does not contribute to the nucleolar/nucleoplasmic tug of war between the two proteins.


This study further clarifies the role of the NES motifs in regulating the traffic of wild-type and mutant NPM1 proteins. It shows that their export into the cytoplasm is most efficient when NPM1 contains all three NES motifs, as characteristically occurs with natural NPM1 mutants found in leukaemic cells from patients with NPM1-mutated AML.11 Moreover, it provides evidence that each of the three NES motifs in NPM1 mutants contributes differently to aberrant mutant export into cytoplasm and that core amino acids in the two physiological N-terminal NES motifs are involved in mediating NPM1 homo- and heterodimerization. Ability to form oligomers, in turn, appears to influence the subcellular distribution of not only wild-type and mutated NPM1 protein but also protein p14ARF, as non-oligomeric NPM1 forms could not enter the nucleolus. As a consequence, in the cell, endogenous NPM1, NPM1 leukaemic mutant and p14ARF protein seem to co-exist in a dynamic equilibrium. Depending on the expression levels, each is capable of dislocating the others from its main subcellular compartment. We have defined this behaviour pattern as a ‘dose-dependent tug of war’.

Opposing forces determine the subcellular localization of endogenous NPM1 and leukaemic NPM1 mutants.13, 15, 26 In normal and neoplastic cells without NPM1 mutations, nuclear localization signals and NoLSs drive endogenous NPM1 to the nucleus and nucleolus, respectively.13, 15 Two physiological NES motifs at the NPM1 N-terminus oppose these forces, causing NPM1 nuclear export. However, because of their weak export efficiency26 or because of post-translational regulation of their activity, the two NES motifs do not counterbalance the nuclear localization signal-dependent nuclear import and the NoLS-dependent force that drives NPM1 to the nucleolus. Consequently, at immunocytochemistry, endogenous NPM1 shows a nucleolar-restricted localization. In AML with mutated NPM1, this shuttling pattern is reversed through the creation of a C-terminus NES motif19 that increases the export efficiency and through the mutation of tryptophans 288 and 290 (or 290 only), which disrupts the NoLS,11, 13 and therefore the capability of the mutant to bind to the nucleolus. The final result is that the NPM1 mutants are aberrantly exported into the cytoplasm of leukaemic cells.15, 24

This study highlights another force, in addition to NoLS,23 that drives NPM1 towards the nucleolus, that is the capability of nucleophosmin to form homo- or heterodimers. In fact, the wild-type and mutant NPM1 proteins that were unable to form oligomers did not enter the nucleolus. This observation concurs with the finding that endogenous NPM1 has been mainly isolated from nucleoli in the form of oligomers,27 principally hexamers.25, 28 Core amino acids in the two physiological N-terminal NES motifs (residues 44 and 47; residues 100 and 102) appear to be critical for NPM1 homo- and heterodimerization, although the mechanism through which this occurs is unclear. How NPM1 oligomerization favours nucleolar targeting of the proteins also remains to be defined. The simplest explanation is that a NPM1 oligomer (possibly a hexamer) could result in a ‘high load’ of NoLSs that reaches the nucleolar targeting threshold. Alternatively, oligomerization could confer the whole molecular complex with structural properties that facilitate its targeting to the nucleolus.

In transfected cells, wild-type and mutant NPM1 reciprocally influenced each other's subcellular distribution in a dose-dependent manner. This process was again dependent upon the integrity of the two N-terminal NES motifs, consistently with the fact that they are involved in the formation of heterodimers between wild-type and mutated NPM1 proteins. Correlation of dose-dependant subcellular distribution of wild-type and mutated NPM1 proteins in transfected cells with the immunostaining pattern of nucleophosmin in leukaemic cells from patients with NPM1-mutated AML is of interest.

In transfected cells, nucleolar wild-type NPM1 levels closely depended on the NPM1wt/mutant expression ratio. With an excess of the mutant, the nucleolus was completely devoid of wild-type NPM1. Interestingly, the analysis of bone marrow biopsies from patients with NPM1-mutated AML revealed wild-type NPM1 levels in leukaemic cell nuclei ranging from the strongly positive (the most common pattern) to the completely negative, with a variable grading in between. This heterogeneous nuclear staining pattern for NPM1wt may reflect different NPM1wt/mutant expression ratios (with consequent variations in the amount of NPM1wt dislocated by the mutant in the cytoplasm), although the influence of other interacting molecules on endogenous NPM1 subcellular distribution in AML cells cannot be excluded. No functional studies could be performed in these cases as only paraffin-embedded material was available for analysis.

In transfected cells, an excess of endogenous NPM1 recruited the leukaemic NPM1 mutant into the nucleolus and hindered its accumulation in the cytoplasm. This pattern has never been observed in bone marrow biopsies from AML patients with mutated NPM124, 29 or in the OCI-AML3 cell line, which carries NPM1 mutation A.22 In fact, in all these circumstances, NPM1 mutant protein localizes only in the cytoplasm, as shown by immunostaining with specific NPM1 mutant antibodies.29 This observation suggests that the relocation of NPM1 leukaemic mutant in the nucleolus by an excess of wild-type NPM1 (or p14ARF) may blunt the mutant's gain of function and impair its transforming abilities.

The above findings further support the concept that NPM1 mutants are ‘born to be exported’,26 as it is suggested by the findings that (i) rare truncated mutants generated by mutations at exon 9 or 11 also acquire a new C-terminus NES and localize in cytoplasm;30, 31, 32, 33 (ii) NPM1 leukaemic mutants retaining tryptophan 288 harbour a stronger C-terminus NES motif than mutants lacking both tryptophans 288 and 290, to ensure the most efficient cytoplasmic export;26 and (iii) the capability of NPM1 mutants to dislocate to the cytoplasm is stable over time.34, 35 Thus, all the evidences to date point to natural selection of a mutational event on the basis of its ability to promote cytoplasmic dislocation of the resulting protein, pointing to this event as critical for leukaemogenesis.13 On the other hand, as NPM1 mutations in AML patients are always heterozygous and never co-exist with uniparental disomy13, 36 or deletion of the wild-type allele, a fraction of wild-type NPM1 seems needed for the survival of leukaemic cells. This is supported by the evidence that in murine models, homozygous NPM1 loss is lethal in early embryonic development,37 and that mouse embryonic fibroblast cells in these mice undergo senescence after a few passages. Consequently, the NPM1wt/mutant expression ratio must be tightly regulated for the mutation to produce its effects. The cytoplasmic NPM1 leukaemic mutant has been shown to be an oncogene with paradoxical functions in vitro.38 Our results suggest that only a conditional knock-in animal model mimicking the real consequences of NPM1 mutations in human leukaemic blasts will prove to be a valid disease model.

The mechanism through which cytoplasmic nucleophosmin contributes to leukaemogenesis remains unknown. Recruitment of the p14ARF tumour suppressor by the NPM1 mutant from the nucleolus into the nucleoplasm and cytoplasm20, 21 could play a role in leukaemogenesis. In this study, we found that the NPM1 mutant interacts in a similar manner with p14ARF and NPM1wt proteins to reciprocally influence subcellular distribution of each. In transfected cells, high p14ARF levels relocated the NPM1 mutant to the nucleolus. As p14ARF is known to inhibit NPM1 shuttling by targeting it to the nucleoli (which is hypothesized to be one of the p14ARF p53-independent tumour suppressor activities), it is tempting to speculate that leukaemic cells are selected on the basis of ‘excess’ NPM1 mutant as otherwise NPM1 mutant transforming activity would be blunted. Indeed, in cells transfected with the NPM1 mutant, reduced p14ARF levels and function were demonstrated and attributed to increased p14ARF degradation following cytoplasmic dislocation.20 Unfortunately, as available anti-Arf antibodies are not suitable for immunohistochemistry, no information is yet available on levels of p14ARF and its subcellular distribution in leukaemic cells from patients with NPM1-mutated AML.

Finally, our results have a potential clinical impact, as they suggest that small molecules acting at the interface between NPM1 mutants and other interacting proteins may serve as new potential therapeutic agents in AML carrying NPM1 mutations.


  1. 1.

    , , , . Nucleophosmin and cancer. Nat Rev Cancer 2006; 6: 493–505.

  2. 2.

    , . Nucleolar protein B23 has molecular chaperone activities. Protein Sci 1999; 8: 905–912.

  3. 3.

    , , , . Major nucleolar proteins shuttle between nucleus and cytoplasm. Cell 1989; 56: 379–390.

  4. 4.

    , . Effects of interphase and mitotic phosphorylation on the mobility and location of nucleolar protein B23. J Cell Sci 2006; 119: 3676–3685.

  5. 5.

    , , , , , et al. Thr199 phosphorylation targets nucleophosmin to nuclear speckles and represses pre-mRNA processing. FEBS Lett 2006; 580: 399–409.

  6. 6.

    , , , , , et al. Sumoylation of nucleophosmin/B23 regulates its subcellular localization, mediating cell proliferation and survival. Proc Natl Acad Sci USA 2007; 104: 9679–9684.

  7. 7.

    . The structure and functions of NPM1/nucleophsmin/B23, a multifunctional nucleolar acidic protein. J Biochem 2008; 143: 441–448.

  8. 8.

    , , , . ARF impedes NPM/B23 shuttling in an Mdm2-sensitive tumor suppressor pathway. Mol Cell Biol 2004; 24: 9327–9338.

  9. 9.

    . Divorcing ARF and p53: an unsettled case. Nat Rev Cancer 2006; 6: 663–673.

  10. 10.

    , , , , , et al. Detection of normal and chimeric nucleophosmin in human cells. Blood 1999; 93: 632–642.

  11. 11.

    , , , , , et al. Cytoplasmic nucleophosmin in acute myelogenous leukemia with a normal karyotype. N Engl J Med 2005; 352: 254–266.

  12. 12.

    , , , , , et al. Acute myeloid leukemia bearing cytoplasmic nucleophosmin (NPMc+ AML) shows a distinct gene expression profile characterized by up-regulation of genes involved in stem-cell maintenance. Blood 2005; 106: 899–902.

  13. 13.

    , , , . Acute myeloid leukemia carrying cytoplasmic/mutated nucleophosmin (NPMc+ AML): biologic and clinical features. Blood 2007; 109: 874–885.

  14. 14.

    , , , , , et al. Distinctive microRNA signature of acute myeloid leukemia bearing cytoplasmic mutated nucleophosmin. Proc Natl Acad Sci USA 2008; 105: 3945–3950.

  15. 15.

    , , , , , et al. Both carboxy-terminus NES motif and mutated tryptophan(s) are crucial for aberrant nuclear export of nucleophosmin leukemic mutants in NPMc+ AML. Blood 2006; 107: 4514–4523.

  16. 16.

    , , , , . Structural consequences of nucleophosmin mutations in acute myeloid leukaemia. J Biol Chem 2008; 34: 23326–23332.

  17. 17.

    , , , . Temporal and spatial control of nucleophosmin by the Ran–Crm1 complex in centrosome duplication. Nat Cell Biol 2005; 7: 823–830.

  18. 18.

    , , , , , et al. Nucleophosmin is essential for ribosomal protein L5 nuclear export. Mol Cell Biol 2006; 26: 3798–3809.

  19. 19.

    , , . Nucleophosmin in acute myelogenous leukemia (letter). N Engl J Med 2005; 352: 1819–1820.

  20. 20.

    , , , , , et al. Delocalization and destabilization of the Arf tumor suppressor by the leukemia-associated NPM mutant. Cancer Res 2006; 66: 3044–3050.

  21. 21.

    , , , . Myeloid leukemia-associated nucleophosmin mutants perturb p53-dependent and independent activities of the Arf tumor suppressor protein. Cell Cycle 2005; 4: 1593–1598.

  22. 22.

    , , , , , et al. Cell line OCI/AML3 bears exon-12 NPM gene mutation-A and cytoplasmic expression of nucleophosmin. Leukemia 2005; 19: 1760–1767.

  23. 23.

    , , , . Tryptophans 286 and 288 in the C-terminal region of protein B23.1 are important for its nucleolar localization. Biosci Biotechnol Biochem 2002; 66: 2239–2242.

  24. 24.

    , , , , , et al. Immunohistochemistry predicts nucleophosmin (NPM) mutations in acute myeloid leukemia. Blood 2006; 108: 1999–2005.

  25. 25.

    , , , , . Essential role of the B23/NPM core domain in regulating ARF binding and B23 stability. J Biol Chem 2006; 281: 18463–18472.

  26. 26.

    , , , , , et al. Born to be exported: COOH-terminal nuclear export signals of different strength ensure cytoplasmic accumulation of nucleophosmin leukemic mutants. Cancer Res 2007; 67: 6230–6237.

  27. 27.

    , . Nucleophosmin/B23 (NPM) oligomer is a major and stable entity in HeLa cells. Biochim Biophys Acta 1995; 1262: 37–42.

  28. 28.

    , . Identification and characterization of a hexameric form of nucleolar phosphoprotein B23. Biochim Biophys Acta 1987; 925: 74–82.

  29. 29.

    , , , , , et al. Mutated nucleophosmin detects clonal multilineage involvement in acute myeloid leukemia: impact on WHO classification. Blood 2006; 108: 4146–4155.

  30. 30.

    , , , , , et al. Cytoplasmic localization of NPM in myeloid leukemias is dictated by gain-of-function mutations that create a functional nuclear export signal. Oncogene 2006; 25: 4376–4380.

  31. 31.

    , , , , , et al. Identification and functional characterization of a cytoplasmic nucleophosmin leukaemic mutant generated by a novel exon-11 NPM1 mutation. Leukemia 2007; 21: 1099–1103.

  32. 32.

    , , , , , et al. Aberrant cytoplasmic expression of C-terminal-truncated NPM leukaemic mutant is dictated by tryptophans loss and a new NES motif. Leukemia 2007; 21: 2052–2054.

  33. 33.

    , , , , , et al. A new type of NPM1 gene mutation in AML leading to a C-terminal truncated protein. Leukemia 2007; 21: 1564–1566.

  34. 34.

    , , , , , et al. Cytoplasmic mutated nucleophosmin is stable in primary leukemic cells and in a xenotransplant model of NPMc+ acute myeloid leukemia in SCID mice. Haematologica 2008; 93: 775–779.

  35. 35.

    , , , , , . Late relapse of AML with mutated NPM1 after 8 years: evidence of NPM1 mutation stability. Haematologica 2008 (in press).

  36. 36.

    , , , , , et al. Translocations and mutations involving the nucleophosmin (NPM1) gene in lymphomas and leukemias. Haematologica 2007; 92: 519–532.

  37. 37.

    , , , , , et al. Role of nucleophosmin in embryonic development and tumorigenesis. Nature 2005; 437: 147–153.

  38. 38.

    , , , , , et al. The leukemia-associated cytoplasmic nucleophosmin mutant is an oncogene with paradoxical functions: Arf inactivation and induction of cellular senescence. Oncogene 2007; 26: 7391–7400.

Download references


This study was supported by the Associazione Italiana per la Ricerca sul Cancro (AIRC), Milan, and by the Fondazione Cassa di Risparmio di Perugia. We are grateful to Dr GA Boyd for editorial assistance. B Falini has applied for a patent on the clinical use of NPM1 mutants. N Bolli is recipient of a Fellowship from FIRC (Federazione Italiana per la Ricerca sul Cancro).

Author information


  1. Section of Haematology and Immunology, Department of Clinical and Experimental Medicine, University of Perugia, IBiT Foundation, Fondazione IRCCS Biotecnologie nel Trapianto, Perugia, Italy

    • N Bolli
    • , M F De Marco
    • , M P Martelli
    • , B Bigerna
    • , A Pucciarini
    • , R Rossi
    • , N Manes
    • , V Pettirossi
    •  & B Falini
  2. Institute of Internal Medicine, Department of Clinical and Experimental Medicine, University of Perugia, Perugia, Italy

    • R Mannucci
    •  & I Nicoletti
  3. Haematopathology and Haematology Units, Department of Haematology and Oncology ‘L and A Seràgnoli’, S Orsola-Malpighi Hospital, University of Bologna, Bologna, Italy

    • S A Pileri


  1. Search for N Bolli in:

  2. Search for M F De Marco in:

  3. Search for M P Martelli in:

  4. Search for B Bigerna in:

  5. Search for A Pucciarini in:

  6. Search for R Rossi in:

  7. Search for R Mannucci in:

  8. Search for N Manes in:

  9. Search for V Pettirossi in:

  10. Search for S A Pileri in:

  11. Search for I Nicoletti in:

  12. Search for B Falini in:

Corresponding author

Correspondence to B Falini.

Supplementary information

About this article

Publication history






Supplementary Information accompanies the paper on the Leukemia website (http://www.nature.com/leu)

Further reading