Npm1 is a haploinsufficient suppressor of myeloid and lymphoid malignancies in the mouse

Nucleophosmin (NPM1) is a nucleolar phosphoprotein involved in critical cellular processes including ribosome biogenesis, maintenance of genomic stability, regulation of p53 and p19^Arf tumor suppressor stability, modulation of DNA transcription, and response to stress stimuli.¹  The NPM1 gene has been implicated in human cancer as it is frequently a target of genetic alterations primarily in hematopoietic tumors. It is found translocated with distinct partner genes in several diseases such as acute promyelocytic leukemia (APL), anaplastic large cell lymphoma (ALCL), acute myeloid leukemia (AML), and myelodysplasia.^2-4  Moreover, it is mapped to a region of chromosome 5 that is frequently deleted in therapy-related MDS and in nonhematopoietic tumors, such as non–small cell lung carcinoma.^5,6  More recently, NPM1 has been found mutated and aberrantly localized in the cytoplasm of the leukemic blasts in a high proportion of AML patients (approximately 35%), rendering NPM1 the most frequently mutated gene in AML.⁷ 

Characterization of primary cells from an Npm1 hypomorphic series in vitro indicated that Npm1 could exert tumor suppressive functions.⁸ Npm1 loss significantly affected genomic stability and Npm1 was shown to be haploinsufficient for this function. Npm1^+/− mouse embryonic fibroblasts (MEFs) displayed genomic instability, which resulted in increased susceptibility to oncogenic transformation in vitro.⁸  Moreover, Npm1^+/− mice displayed hematologic features similar to those observed in patients with MDS, a preleukemic condition in humans.⁸  However, whether Npm1 functional loss may lead to cancer susceptibility per se remained unclear. In this report, we demonstrate that Npm1 acts indeed as a haploinsufficient tumor suppressor gene in vivo.

The Research Animal Resources Center at Memorial Sloan Kettering Cancer Center and Harvard Medical School Animal Facility approved and monitored all protocols that used animals in this study.

The generation of Npm1 heterozygous mice was previously described.⁸ 

Mouse tissues sections were stained with hematoxylin and eosin (H&E) for histopathological examination. Antibodies for immunohistochemical staining were CD45R/B220 for B-cell lineage, CD3 for T-cell lineage, polyclonal prediluted anti-lysozyme muramidase (Ventana, Tucson, AZ) for myeloid cells, and mouse antinucleophosmin/B23 (Zymed, South San Francisco, CA), used according to the manufacturer's instructions. Pictures of the stained tissue sections were obtained using an Olympus BX41 microscope and a DP20 camera (Olympus, Center Valley, PA). All image acquisition and processing was carried out with Adobe Photoshop 7.0 (Adobe Systems, Mountain View, CA).

Immunofluorescence was performed as previously described⁸  using anti–mouse γ-tubulin antibody (Sigma-Aldrich, St Louis, MO).

Genomic DNA from tumors of 6 Npm1 heterozygous mice was subjected to comparative genomic hybridization (CGH) array analysis at the MSKCC Genomic Core Lab Microarray Facility.

Bone marrow cells were stained with phycoerythrin (PE)-conjugated anti-Mac1 and allophycocyanin (APC)-conjugated anti-Gr1 (BD Pharmingen, San Diego, CA) and analyzed using FACSCalibur (BD Biosciences).

Southern blot analysis was performed using the 5′ probe used to identify the Npm1-null allele as previously described.⁸ 

To determine the susceptibility of Npm1^+/− mice to tumorigenesis, we conducted a 2-year follow-up on a colony of 102 mice (47 wild-type and 55 heterozygotes mice), and have evaluated incidence, latency, and pathological features of neoplasias occurring in these mice compared with wild-type littermates.

Strikingly, Npm1^+/− mice displayed higher susceptibility to develop malignancies than their wild-type counterparts. Of the 55 heterozygous mice under observation, 16 (29%) developed malignancies (Table 1). By comparison, only 3 cases (6.3%) were identified in the wild-type population (P < .01). Tumors were observed after a long latency that ranged from 10 to 24 months.

Pathological analysis revealed that 75% of the diseased Npm1^+/− mice were affected by hematologic malignancies (Figure 1A), while no abnormalities were detected in the hematopoietic organs of wild-type animals analyzed. Interestingly, among the solid tumors developed by the Npm1^+/− mice, 2 were cancers of the lung.

Interestingly, three quarters of the hematologic neoplasms developed by Npm1^+/− mice were myeloid malignancies, with a myeloproliferative disease–like myeloid leukemia (MPD-like ML) having the highest incidence (41.6% of all hematologic malignancies). MPD-like ML was characterized by leukocytosis (white blood cell [WBC] count = 18.1 ± 0.7 × 10⁹/L) and extreme expansion of well-differentiated myeloid cells in bone marrow (Figure 1D), spleen (Figure 1E), and also in other nonhematopoietic tissues, such as liver (Figure 1F), and kidney and intestine (Figure S1C and S1D, respectively, available on the Blood website; see the Supplemental Materials link at the top of the online article).

The remaining mice affected by myeloid malignancies presented with myeloid leukemia (ML). ML was characterized by severe anemia (hemoglobin [HGB] level = 69 ± 8 g/L), mild leukocytosis (WBC count = 17.4 ± 0.5 ×10⁹/L), and thrombocytopenia (platelet [PLT] count = 290 ± 60 × 10⁹/L). Both peripheral blood and bone marrow contained a significant percentage of leukemic blasts (20% ± 4% and 36% ± 2%, respectively). Morphologically, blasts presented like immature myeloid cells with increased nucleus-cytoplasm ratio and highly vacuolated cytoplasm (Figure 1B arrowheads). Hepatomegaly and splenomegaly were invariably detected in these leukemic mice (Figure S1A). Pathologically, spleen and liver were filled with a large amount of leukemic blasts that compromised the architecture of the tissues (Figures 1C, S1B). Immunohistologic analysis of spleen and liver sections identified cells whose cytoplasm was positive for muramidase (lysozyme) staining, and positivity is usually associated with monocytic differentiation in human and mouse leukemias⁹  (Figure 1C inset). These are all features distinctive of myeloid leukemia in the mouse.¹⁰ 

In addition to an increased rate of myeloid malignancies, Npm1^+/− mice also developed B- and T-cell lymphomas (one quarter of all hematologic malignancies observed). These mice presented with enlarged lymph nodes with atypical lymphoid infiltrates affecting the normal architecture (Figure 1G). Lymphadenopathy was accompanied by hepatosplenomegaly with both spleen pulps and liver parenchyma replaced by nodules of atypical lymphoid cells (Figure 1H left panel; Figure S1F,G). In two thirds of lymphomas, the atypical lymphoid infiltrates were positive for CD45R staining (Figure 1G inset), supporting the diagnosis of B-cell lymphoma. In the other cases, neoplastic lymphoid infiltrates displayed strong positivity for CD3 staining (Figure 1H right panel), suggestive of T-cell lymphomagenesis.

We next analyzed the status of the Npm1 gene in genomic DNA extracted from blasts of focal homogeneous infiltrates localized in the spleen of ML and MPD-like ML affected mice (6 mice in total), using the corresponding tail DNA as reference. Loss of the wild-type Npm1 allele was never observed (Figure 1I). The nuclear expression of Npm1 in these tumors was confirmed by immunohistochemistry (Figure S2A).

As Npm1 function is required for maintaining genomic stability, we have analyzed centrosome number and the presence of chromosomal abnormalities in Npm1^+/− mice tumors. Bone marrow cytospins from MPD-like ML affected mice were immunostained for the centrosome marker γ-tubulin. Two independent cytospins clearly showed an increased number of cells displaying multiple centrosomes (Figure S2B). On average, 70% of cells displayed amplified centrosomes in these MPD-like ML, compared with 6% of bone marrow cells from nonleukemic Npm1^+/− mice. These data demonstrate that centrosome amplification is associated with Npm1 loss and may result in leukemia pathogenesis in Npm1^+/− mice.

Genomic DNA analysis of 6 hematologic malignancies performed by comparative genomic hybridization invariably revealed the presence of numeric and structural chromosomal abnormalities (Figure S2C).

Although Npm1^+/− mice display features resembling some of those identified in human MDS, it was not clear whether this represented a genuine preleukemic situation as is the case for the human disease. Here, we addressed this question by following a large cohort of Npm1^+/− mice to establish whether Npm1 heterozygosity results in an increased susceptibility to leukemogenesis. Indeed, a number of these mice were found to develop a variety of hematologic neoplasms, with myeloid malignancies having the highest incidence. These data confirm our hypothesis that the Npm1 phenotype originally characterized has the potential to progress to overt leukemia.

The absence of loss of heterozygosity (LOH) indicates that Npm1 haploinsufficiency, rather than biallelic inactivation, is promoting the development of tumors in these mice. This suggests that while Npm1 heterozygosity may render the cells more susceptible to oncogenic transformation, the complete somatic loss of the Npm1 gene is neither required for the transformation nor selected for by the transformed cell itself.

In conclusion, we demonstrate that Npm1 is a haploinsufficient suppressor of myeloid and lymphoid malignancies in the mouse. Npm1^+/− mice are prone to spontaneous tumor formation and in particular to the development of disease of myeloid origin.

We thank Linda Disantis and Owen Milbury for editing the paper, Manoj Kumar of BIDMC Genomics Center for help with CGH analysis, and members of the Pandolfi laboratory for helpful discussions.

This work was supported by National Institutes of Health grant (CA-71692) to P.P.P. P.S. was supported in part by the Associazione Umbra Leucemie e Linfomi (AULL).

National Institutes of Health

Contribution: P.S. and S.G. designed and performed experiments, analyzed data, and wrote the paper; S.M.M. performed experiments; K.C. and J.G.C. performed experiments and edited the paper; A.V. performed the CGH analysis; J.T.-F. performed the pathological analysis of tumor samples; P.P.P. designed research, analyzed data, and wrote the paper.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Pier Paolo Pandolfi, Beth Israel Deaconess Medical Center, 330 Brookline Avenue, NRB 1038A, Boston, MA 02215; e-mail: ppandolf@bidmc.harvard.edu.