NPM1 Deletion Is Associated with Gross Chromosomal Rearrangements in Leukemia

Background NPM1 gene at chromosome 5q35 is involved in recurrent translocations in leukemia and lymphoma. It also undergoes mutations in 60% of adult acute myeloid leukemia (AML) cases with normal karyotype. The incidence and significance of NPM1 deletion in human leukemia have not been elucidated. Methodology and Principal Findings Bone marrow samples from 145 patients with myelodysplastic syndromes (MDS) and AML were included in this study. Cytogenetically 43 cases had isolated 5q-, 84 cases had 5q- plus other changes and 18 cases had complex karyotype without 5q deletion. FISH and direct sequencing investigated the NPM1 gene. NPM1 deletion was an uncommon event in the “5q- syndrome” but occurred in over 40% of cases with high risk MDS/AML with complex karyotypes and 5q loss. It originated from large 5q chromosome deletions. Simultaneous exon 12 mutations were never found. NPM1 gene status was related to the pattern of complex cytogenetic aberrations. NPM1 haploinsufficiency was significantly associated with monosomies (p<0.001) and gross chromosomal rearrangements, i.e., markers, rings, and double minutes (p<0.001), while NPM1 disomy was associated with structural changes (p = 0.013). Interestingly, in complex karyotypes with 5q- TP53 deletion and/or mutations are not specifically associated with NPM1 deletion. Conclusions and Significance NPM1/5q35 deletion is a consistent event in MDS/AML with a 5q-/-5 in complex karyotypes. NPM1 deletion and NPM1 exon 12 mutations appear to be mutually exclusive and are associated with two distinct cytogenetic subsets of MDS and AML.


Introduction
In humans, nucleophosmin (NPM1) is implicated in the genesis of different cancers [1]. We previously found NPM1 exon 12 somatic mutations in approximately 60% of adult acute myeloid leukemia (AML) and normal karyotype [2]. NPM1 fusion proteins are also specifically involved: t(5;17)(q35;q21)/NPM1-RARA causes acute promyelocytic leukemia; t(2;5)(p23;q35)/NPM1-ALK underlies anaplastic large cell lymphoma; t(3;5)(q26;q35)/NPM1-MLF1 is associated with myelodysplastic syndromes (MDS) and AML with trilineage dysplasia and poor prognosis (http://AtlasGeneticsOncology.org). In the last cytogenetic subgroup of MDS/AML a cryptic loss of NPM1 gene may occur at the 5q35 breakpoint of the translocation [3]. Despite of this sporadic observation the role of NPM1 deletion in hematological malignancies remains to be elucidated. In human leukemia an interstitial isolated deletion at chromosome 5q identifies a myelodysplastic syndrome with a benign clinical course (the ''5q-syndrome'') [4]. Moreover a partial or complete 5q loss (5q-/-5) may occur as part of a complex karyotype in high risk MDS/AML, including cases arising after radio-and/or chemo-therapy [5]. Chromosomes 5q-are heterogeneous in size as deleted segments vary in extension and have different centromeric and telomeric breakpoints [6,7]. Experimental evidences indicate that the 5q-syndrome originates from haploinsufficiency of more than one critical pathogenetic gene included in the hemizygous chromosome deletion [8][9][10][11][12]. Thus identification of genes that go lost in different 5q-chromosomes of MDS and AML is very important to understand the pathophysiology of the associated malignant disorders. Focusing on NPM1 gene we investigated a large series of MDS and AML with and without a 5q deletion.

NPM1 expression by Real Time Quantitative PCR
RT-qPCR was performed on cryopreserved bone marrow RNA samples from 9 patients with NPM1+/and 5q-, 10 patients with NPM1+/+ and 5q-, 5 patients with NPM1+/+ without 5q-and 11 healthy controls. Total RNA was extracted using Trizol (Invitrogen) according to the manufacturer's protocol with minor modifications. Two micrograms of total RNA were treated with 4 Units of Dnase I (Deoxyribonuclease I Amplification Grade, Sigma) in a total volume of 11 ml. One microgram of treated RNA was reverse transcribed to single strand cDNA according to the EAC protocol [15] and 1 mg was used in a control reaction (-RT) using the same procedure without the M-MLV reverse transcriptase enzyme (Perkin Elmer Applied Biosystems, Foster City, USA). The RT-qPCR mixture reaction contained 12.5 ml Taq Man Universal PCR Master Mix (Life Technologies), 300 nM primers, 200 nM probe and 5 ml cDNA (1/10 of RT product), in a total volume of 25 ml. Amplification conditions were: 2 minutes at 50uC, 10 minutes at 95uC followed by 40 cycles at 95uC for 15 seconds and at 60uC for 1 minute [16]. RT-qPCR reactions and fluorescence measurements were performed on an ABI PRISM 7700 Sequence Detection System (Life Technologies). Gene expression levels of NPM1 and ABL1 were detected using specific assays on Demand (Life Technologies) following the manufacturer's instructions: ID: Hs02339479_g1 (exon 4-5) for NPM1 and

Statistical analysis
Inter-group differences were analyzed by non parametric tests using first analysis of variance (ANOVA) with the Kruskal-Wallis test to compare three groups and the Mann-Whitney U test to compare two. If significant, multiple comparisons were carried out using the Mann-Whitney U test with the Bonferroni correction (i.e. 0.05/numbers of comparisons). The multivariable Poisson regression model analyzed the adjusted effect of NPM1 and TP53 on gross chromosomal changes and monosomies.
Contingency tables studied associations among categorical variables which were analyzed by Fisher's Exact test for 2X2 tables or the chi-square test. Unless otherwise indicated, significance was set at ,0.05. SPSS version 17.0 software (Chicago, IL) was used for statistical analysis.

Ethics statement
This study was carried out solely on encoded archival samples that were originally taken for diagnostic purposes. The residual of these samples was used after all diagnostic procedures had been completed. Our specific study was approved by the Institutional Review Board of the Hematology Department of the University of Perugia (IRB 00003450) which guidelines allow the use of anonymous residual diagnostic samples for research purposes. Written informed consent was obtained. The ''Demokritos'' Cancer Research Centre Ethics Committee (Athens, Greece) allows the use of anonymous or encoded residual diagnostic samples for research purposes, unless the patient expresses its refusal to participate in research studies. The authors declare that all samples used in this study were encoded anonymously and that none of the patients had explicitly refused to participate in research.

Results and Discussion
One hundred forty five patients with MDS/AML (60 males and 85 females; median age = 68 years, range 9-94) were included in the study ( Table S1).
The mono-allelic NPM1 deletion was a very infrequent lesion in the ''5q-syndrome'' because it was not detected in 42/43 cases. It identified two molecular sub-groups, here named NPM1+/2 and NPM1+/+, since it was present in 38/84 (45%) cases with complex karyotypes and 5q-. Interestingly Lessard et al. [18] demonstrated loss of the 5q35 chromosome band in 30% of de novo and therapy related AML. NPM1 loss may have occurred as an early step in MDS/AML as it remained stably mono-allelic in 4 NPM1+/2 patients and did not develop in 6 NPM1+/+ over a median of 9 months follow-up (range: 2-156 months) ( Table 2).
Since NPM1 is related to TP53 [19,21223], a master gene of chromosome stability that is frequently involved in MDS/AML with complex karyotype [24], 57 cases from this series were investigated ( Table 3 and TableS1). Univariate analysis on TP53 and chromosome rearrangements showed a significant association between the number of monosomies and TP53 deletion and/or mutation (median 2 range 028 vs median 0 range 023; P = 0.006) ( Table S3). However, both NPM1 and TP53 gene status maintained significant association with monosomies in a multivariate analysis (P = 0.014 and P = 0.013, respectively) (Poisson regression model), suggesting each gene plays an independent role, but may interact synergistically, in determining chromosomal abnormalities.
NPM1 loss was not associated with copy number variations of 14 genes other than TP53 that are putatively involved in the complex regulatory network of genetic stability. Gains and losses of clones containing BUB1A, ATR, FBW7, c-MYC, CDKN2A-2B, PTEN, CHEK1, ATM, MDM2, BRCA2, RB1, BUB1B, BRCA, CHEK2 were equally distributed in NPM1+/+ and NPM1+/2 cases (Figure 3), suggesting that whatever roles these genes play in MDS/AML  with complex karyotypes, they are independent of the NPM1 deletion.
In conclusion these findings prove that haploinsufficiency of the NPM1 gene, is present in more than 40% of patients with high risk MDS/AML characterised by chromosome 5q-and complex karyotypes. NPM1 haploinsufficiency and mutation appear to exert very different effects (Figure 4) as patients with NPM1 mutated-AML have normal karyotype, rare TP53 involvement, good prognosis if FLT3-ITD mutations are absent, and a predominance of females [2,25,26]. NPM1 haploinsufficiency was not linked to female gender (Table S4). It significantly reduced NPM1 expression and, in synergy but independently of TP53 deletion and/or mutation, was associated with a strong tendency to perturb chromosomal stability. Consequently NPM1 gene status appears to indicate diverse genetic backgrounds in two different myeloid leukemias.