MDM4 (MDMX) is overexpressed in chronic lymphocytic leukaemia (CLL) and marks a subset of p53wild-type CLL with a poor cytotoxic response to Nutlin-3.

Although about one-third of patients with B-cell chronic lymphocytic leukaemia (CLL) never require treatment, in other patients the disease progresses at a variable rate and requires treatment regimens by alkylating agents, purine analogs, monoclonal antibodies or combinations thereof. Despite improvements in response rates using chemo-immunotherapy combinations, patients refractory to fludarabine and/or monoclonal antibodies or patients that have suffered multiple disease relapses have a poor outlook. In cells with functional p53, the p53 activity is primarily inhibited through direct and tonic interaction with the human homolog of the murine double minute 2 (MDM2) protein. After the original description of the small molecule Nutlin-3 as a potent and selective inhibitor of p53/MDM2 interaction, subsequent studies first demonstrated that Nutlin-3 induced ex-vivo cytotoxic cell death of most p53 wild-type (p53wt) CLL samples and is able to determine a comparable accumulation of p53 and degree of apoptosis in CLL cells from bad prognosis CLL subsets, i.e. bearing an unmutated IGHV status, or high levels of the 70 kDa f-associated protein (ZAP- 70). However, we have recently demonstrated that cells of a minority of p53wt CLL cases do not respond to Nutlin-3 treatment and fail to display the canonical signature associated with in-vitro Nutlin-3 exposure. According to this background, the present study sought to identify and validate the relevant genes associated with resistance of p53wt CLL cells to the in-vitro Nutlin-3 effects. To do this, the constitutive gene expression signature associated with the reduced response of p53wt CLL cells to in-vitro Nutlin-3 exposure was investigated by comparing the gene expression profile (GEP) of CLL cells from 13 Nutlin-3 ‘responder’ versus 3 Nutlin-3 ‘non-responder’ cases, as identified by previous studies, all of which displayed a p53wt phenotype. The main clinical and biological features of CLL samples utilized for this purpose are listed in Table SI. Patients provided informed consent in accordance with the local institutional review board requirements and the declaration of Helsinki. Problems associated with bioinformatic analyses of such an unbalanced datasets (13 vs. 3 cases) were overcome by applying a purposely set-up modified version of significance analysis of microarrays (SAM), named multi-SAM. By means of this approach, we were able to select 278 differentially expressed genes, of which 149 up-regulated and 129 down-regulated in ‘non-responder’ p53wt CLL, correlated with the lack of response to Nutlin-3. A hierarchical clustering using the 278 genes separated the three ‘non-responder’ p53wt CLL cases from all but one the 13 ‘responder’ p53wt CLL cases (Fig 1A). Among the genes expressed at higher levels in the ‘non-responder’ p53wt CLL category, we focused on MDM4 (MDMX), given the potential interest of its related gene product in our setting. In fact, MDM4 is a structural homologue of MDM2 which, like MDM2, is able to bind p53, preventing p53-dependent transcription as opposed to MDM2, however, interaction of MDM4 with p53 cannot be inhibited by Nutlin-3 treatment .Therefore, it is conceivable that, for ‘non-responder’ p53wt CLL cases, the lack of response to the in-vitro treatment with Nutlin-3 could at least partially depend on higher relative levels of MDM4. Similarly, basal levels of MDM4 transcripts have been shown to be critical for the cytotoxic response to Nutlin-3 in solid tumour cell lines. In agreement with GEP, quantitative Real Time-polymerase chain reaction (qRT-PCR) evaluation of MDM4 in ‘non-responder’ versus ‘responder’ p53wt CLL, confirmed the higher expression levels of MDM4 in the former category (P = 0Æ0004) (Fig 1B). These results prompted us to investigate, by utilizing the same qRT-PCR approach, the constitutive levels of MDM4 transcripts in a different series of 43 CLL samples, in which information on the principal biological prognosticators, i.e. IGHV gene status, fluorescence in situ hybridization (FISH) groups, and CD38, CD49d or ZAP-70 expression, were also available; in these experiments, MDM4 transcript levels were also compared to those found in normal B cells purified from peripheral blood samples of 15 healthy donors (kindly provided by the blood bank of the Centro di Riferimento Oncologico, Aviano, Italy). By considering the whole cohort of CLL samples irrespective of the principal prognosticators, overall higher expression levels of MDM4 in CLL compared to normal B cells were detected, although with a great case-tocase variation (P < 0Æ0001; Fig 2). Of note, when CLL patients were divided according to the above reported prognosticators, no differences were detected between the various CLL categories in terms of MDM4 expression levels, which however remained higher than in normal B cells (data not shown). In the present study, moving from the gene expression signature that identified a a subgroup of p53wt CLL showing a lack of response to the in-vitro treatment with Nutlin-3, we eventually identified MDM4 as a gene that, despite its peculiar over-expression in a subset of Nutlin-3 ‘non-responder’ p53wt CLL, turned out to be universally over-expressed by CLL cells compared to normal B cells. This observation is in keeping with studies describing the over-expression of MDM4 in primary samples from several solid tumours, including glioblastoma, retinoblastoma, breast, colon, and lung cancers, and may contribute to identify MDM4 as a potentially useful new therapeutic target also for CLL. Preclinical studies indicating the anti-neoplastic effects of MDM4 down-regulation in murine lymphoma models are of further support for this hypothesis.