miR-34b Targets Cyclic AMP–Responsive Element Binding Protein in Acute Myeloid Leukemia Free

MicroRNAs (miRNA) are a class of small noncoding RNAs able to influence gene expression by targeting mRNA. Thirty percent of human genes possess conserved miRNA binding sites and are presumed to be controlled by this regulation mechanism (1). Generally, miRNAs bind mRNA sequences located at the 3′-untranslated region (UTR) with imperfect complementarity. They usually avoid the interaction of a target mRNA with polysomes, block translation, or promote mRNA degradation. Consensus miRNA sequences are described also in 5′-UTRs and in coding sequences, but their role in gene regulation is less efficient (2). Hundreds of miRNAs have been identified to date, but their specific functions and target mRNAs have been assigned for only a few (3). miRNA expression is tissue specific, and to control cellular differentiation, proliferation, and survival, changes in their expression have been associated with many pathologies, including human cancer (4). A role of miRNAs in several tumors has recently been recognized, with intrinsic suppressor or oncogenic functions (5). Little is known about their role in acute myeloid leukemia (AML). There is considerable evidence that supports a crucial role for miRNAs in chronic lymphocytic leukemia (6, 7), and an involvement of miR-223 and miR-155 has already been proposed in the pathogenesis of AML (8, 9). The knowledge and the characterization of novel miRNAs should be expanded to elucidate the pathophysiologic events that cause myeloid transformation, which is considered to accumulate genomic alterations that act as consecutive transforming events in the leukemic clone during leukemic development (10). Multiple mutations have recently been investigated in signaling molecules, such as growth factor receptors and transcription factors, and have been associated to specific leukemia phenotypes and sometimes to treatment response (11). The principal aim is to improve AML knowledge by exploring the mechanism of genetic disruptions to improve the traditional cytogenetic markers in clinical use.

Cyclic AMP (cAMP)–responsive element binding protein (CREB) is a nuclear protein that regulates gene expression principally through the activation of cAMP-dependent cell signal transduction pathways after being phosphorylated at Ser¹³³ usually via protein kinase A (PKA; refs. 12, 13). This modification enhances the transactivation potential of CREB and promotes the recruitment of two major cofactors: CREB-binding protein and p300 (14, 15). CREB recognizes the conserved cAMP-responsive elements (CRE), which occur either as a full palindrome (TGACGTCA) or half-site (CGTCA/TGACG) at gene promoters in a cell type–specific manner, controlling the expression of genes involved in cell proliferation, differentiation, and survival signaling pathways (16). Now, >4,000 human genes are known to contain CRE consensus regions, and the ability of CREB to activate or not target gene transcription depends on recruited cofactors and on the cellular gene regulation program (17). CREB overexpression in leukemia has been shown to cause the up-regulation of its target genes, influencing leukemia phenotypes; in particular, it promotes abnormal proliferation, cell cycle progression, and higher clonogenic potential in vitro and in vivo (18, 19). CREB overexpressing transgenic mice have been shown to develop myeloproliferative disorders, suggesting that CREB plays a role during the leukemogenic process (20).

Here, we report studies of CREB carried out to understand the molecular mechanism that controls its protein overexpression in leukemia. We consider that the inappropriate expression of candidate miRNAs could be a possible mechanism of protein regulation. The CREB 3′-UTR sequence has many miRNA consensus seeds, but this research shows that miR-34b is responsible for CREB expression through the control of its translation.

miR-34b belongs to the evolutionary conserved miRNA family miR-34s (21), known for its role in the p53 tumor suppressor network (22). miR-34s have been found controlled in a tissue-specific manner by p53, to have an antiproliferative potential in cell lines, and to be down-regulated in human tumors (23, 24). Efforts have been made to identify miR-34 target genes, but apart from the few genes known to be miR-34a targets (25, 26), there are no confirmed target genes for miR-34b/miR-34c in leukemia.

In this research, CREB is shown to be a miR-34b target. A series of CREB target proteins and major related pathways are shown to be influenced after miR-34b restoration. The promoter of miR-34b/miR-34c is found methylated, which should explain the lowered miR-34b expression observed in myeloid tissue. Thus, miR-34b suppresses the CREB network, inhibiting tumor growth, and should be further considered during AML development.

Cell culture and transfection. Human AML cell line HL60 (American Type Culture Collection) was cultured in DMEM (Invitrogen-Life Technologies) supplemented with 10% fetal bovine serum (FBS; Invitrogen-Life Technologies). Human myeloid cell lines NOMO1, NB4, and ML2 and human chronic myelogenous leukemia–derived B-cell–like cell line K562 (American Type Culture Collection) were cultured in RPMI 1640 (Invitrogen-Life Technologies) supplemented with 10% FBS. Cells were treated with 2 μmol/L 5-aza-2′-deoxycytidine (DAC; Sigma). Cell transfection was performed using the Nucleofector systems (Amaxa Biosystems) according to the manufacturer's guidelines. Transfection conditions were optimized to result in >70% transfection efficiency with a cell viability of >80%. We analyzed bone marrow samples from 78 patients with newly diagnosed pediatric AML and 17 from healthy pediatric bone marrow. The diagnoses of leukemia were made according to standard morphologic criteria based on immunohistochemical, immunophenotyping, and cytogenetic studies following the AIEOP-2002 AML pediatric protocol. Informed consent in compliance with the Helsinki protocol was obtained.

miRNA target prediction. The analysis of miRNA predicted targets was carried out using the algorithm miRanda.¹

RNA extraction and real-time PCR for miRNA analysis. RNA from cell lines and from bone marrow patients was isolated using a mirVana miRNA Isolation kit (Ambion) according to the manufacturer's instructions. RNA quality was checked using an Agilent 2100 Bioanalyzer (Agilent Technologies) and then used for PCR. Taqman miRNA assays (Applied Biosystems) were carried out using the stem-loop method to detect the expression level of mature miRNAs. For the retro-transcription reaction, 10 ng of total RNA were used in each reaction and mixed with the specific stem-loop primers (Applied Biosystems). All PCRs were run in triplicate and gene expression, relative to U6 small nuclear RNA (RNU6B), was calculated by the comparative ΔC_t method (27).

Methylation analysis. Genomic DNA was extracted according to the manufacturer's instructions (Gentra Autopure LS, Qiagen) from myeloid cell lines. Briefly, 1 μg of DNA was diluted in 50 μL of distilled water and denatured by adding 5.5 μL of 3 mol/L NaOH. On ice, 520 μL of bisulfite solution at pH 5 and 30 μL of 10 mmol/L hydroquinone were mixed. The DNA was recovered and desulfonated by adding 5.5 μL of 3 mol/L NaOH. The solution was neutralized by adding 55 μL of 6 mol/L ammonium acetate at pH 7. The DNA was ethanol precipitated and used with methylation-specific primers and PCR protocol (28).

Cloning of 3′-UTR of CREB1 mRNA. cDNA (100 ng) from myeloid cell line served as template to amplify CREB 3′-UTR (NM_004379, gi 222194459 from nucleotides 1,078 to 1,359). The amplified PCR product was gel purified and subcloned into pCR2.1 (Invitrogen-Life Technologies). The insert was excised with HindIII and SacI restriction enzymes, sequenced, and ligated into the pMIR-REPORT miRNA luciferase (LUC) reporter vector (Ambion) at the 3′-end of the LUC gene. We refer to pMIR-LUC-3′-UTR-CREB in experiments performed using this construct. A mutant 3′-UTR of CREB was also synthesized by PCR.

Transfection and LUC reporter assay. miR-34b oligonucleotide and a miRNA from the Arabidopsis thaliana genome used as negative control (referred to as miR-neg) were purchased from Dharmacon Industries. A mixture of pMIR-LUC-3′-UTR-CREB, Renilla plasmid (REN), and mature miR-34b oligonucleotide was used to cotransfect cell lines. A mixture of pMIR-LUC-3′-UTR-CREB, REN, and miR-neg was used as control. After 24 h, RNA and proteins were extracted. Real-time quantitative PCR (RQ-PCR) was used to test miR-34b expression, and protein lysates were analyzed for LUC and REN activity levels using the Dual Luciferase Assay System (Promega). LUC activity was normalized to REN activity, compensating for variation in transfection efficiency. Experiments were performed in triplicate.

Western blot. Western blot analyses were carried out as previously described (19). Briefly, 20 μg of the total protein fraction (Buffer-Biosource International) isolated from transfected cells were used. Protein concentration was measured using a bicinchoninic acid protein assay kit (Pierce). Antibodies used were anti-phospho-CREB (Upstate Biotechnology) and anti-β-actin (Sigma-Aldrich). The secondary antibody was horseradish peroxidase–conjugated goat anti-rabbit or mouse IgG (Upstate Biotechnology). Proteins were detected using enhanced chemiluminescence and films (GE Healthcare).

RNA isolation and SYBR Green quantitative real-time reverse transcription-PCR assays. Total RNA was isolated from cell lines (2 to 5 × 10⁶ per sample) using Trizol (Invitrogen). RNA (1 μg) was transcribed using the SuperScript II system (Invitrogen) in 25 μL final volume following the manufacturer's instructions. RQ-PCR was performed with 1 μL cDNA in 20 μL using the SYBR Green method (Invitrogen) and analyzed on an ABI PRISM 7900HT Sequence Detection System (Applied Biosystems). Experiments were carried out in triplicate and were analyzed with respect to standard curves in a relative quantity study. Expression values for target genes (Supplementary Table S1) were normalized to the concentration of GUS, which showed the least variation among reference genes in our cell model.

Reverse-phase protein microarrays. The experiments were performed as described by Espina and colleagues (29, 30). Briefly, cell lysates were diluted in a mixture of 2× Tris-glycine SDS sample buffer (Invitrogen-Life Technologies) plus 5% β-mercaptoethanol and loaded onto a 384-well plate and serially diluted with lysis buffer into four-point dilution curves ranging from undiluted to 1:8 dilution. Samples were printed in duplicate onto nitrocellulose-coated slides (FAST slides, Whatman Schleicher & Schuell) with a 2470 Arrayer (Aushon BioSystems). Slides were stained with Fast Green FCF (Sigma) according to the manufacturer's instruction and visualized (ScanArray 4000, Packard). Arrays were stained with antibodies (Cell Signaling) on an automated slide stainer (Dako Autostainer Plus, DakoCytomation) using a Catalyzed Signal Amplification System kit (CSA kit, DakoCytomation) according to the manufacturer's recommendations. Antibody staining was revealed using 3,3′-diaminobenzidine. The TIF images of antibody-stained and Fast Green FCF–stained slides were analyzed using MicroVigene software (VigeneTech, Inc.). For each sample, the signal of the negative control array (stained with the secondary antibody only) was subtracted from the antibody slide signal, and then the resulting value was normalized to the total protein value. The data processing generated a single value for each leukemia sample relative to each protein.

Cell cycle analysis. Cell lines were transfected with oligonucleotide miR-34b or miR-neg. After 24 h of incubation, 5 × 10⁵ cells were washed twice with PBS, lysed, and treated with 50 μg/mL propidium iodide in 1 mL PBS overnight at 4°C. Cells were analyzed using Cytomics FC500 (Beckman Coulter). Cycle analyses were performed using Multicycle Wincycle software (Phoenix Flow Systems).

Small interfering RNA experiments. Exogenous small interfering RNAs (siRNA) specific for the CREB gene (Dharmacon Industries) were introduced in myeloid cell lines (100 nmol/L in 2 mL of medium; ref. 18). A scramble of all four siRNAs was also used.

Anchorage-independent assay. To determine anchorage-independent growth of transfected cells, a total of 2.5 × 10³ cells were seeded in a methylcellulose semisolid medium not supplemented with nutrients and cytokines after miRNA and siRNA transfection (StemCell Technologies). Colony evaluation and enumeration was done in situ by light microscopy after 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) incorporation.

Data analysis. Statistical analyses were performed with Mann-Whitney or unpaired two-tailed t test. A P value of <0.05 was considered significant.

miRNA expression in cell lines. We studied the expression of five mature miRNAs predicted to target the 3′-UTR region of the CREB gene by informatic tools (Fig. 1A). Expression analyses revealed that miR-124a and miR-367 were not detectable in our samples, whereas miR-25, miR-32, and miR-34b were down-regulated compared with healthy sorted myeloid subpopulations (CD3⁻/CD19⁻) in all cell lines (Fig. 1B). Among them, miR-34b was significantly less expressed (10⁻¹²–fold) and was selected to be studied in vitro.

miR-34b promoter region is methylated in leukemia. The miR-34b promoter region was studied in leukemic cell lines for hypermethylation of the CpG island, as recently described in colon cancer (28). After treatment with DAC, miR-34b expression by RQ-PCR was found to have increased in the cell lines used, from 3.34-fold to 19.67-fold induction, confirming that miR-34b expression might be controlled by methylation in myeloid cancer cells (Fig. 1C). Methylation-specific PCR (MSP) revealed that the miR-34b/miR-34c was methylated in all cell lines tested. By contrast, no methylation was detected in the sample made up of three normal bone marrow samples collected from healthy donors, indicating that methylation of the miR-34b/miR-34c region is a tumor-specific phenomenon (Fig. 1D).

miR-34b suppresses CREB expression and inhibits cell growth. To assess the role of miR-34b in CREB expression and cell growth, HL60 cell line was transfected with the mimic oligo miR-34b with either control miRNA (miR-neg) at 60, 120, and 180 nmol/L up to 72 hours. The MTT-based cell proliferation assay revealed that miR-34b inhibits cell growth in a dose-dependent manner. Using 180 nmol/L miR-34b, cell proliferation was reduced to 58% at 48 hours with respect to miR-neg, whereas at 120 nmol/L, cell growth was always up to 70% after 72 hours of treatment (n = 3; P < 0.05). miR-34b concentration (120 nmol/L) was the treatment that gave the maximum effect on CREB translation inhibition (Supplementary Fig. S1).

miR-34b targets CREB 3′-UTR. To validate CREB as a miR-34b target gene in different myeloid cell lines, the potential base pairing between miR-34b and CREB 3′-UTR was investigated. A portion of CREB 3′-UTR was cloned and introduced into myeloid cell lines along with the miR-34b oligonucleotide. The ectopic expression of miR-34b was confirmed by RQ-PCR, which revealed a higher miR-34b expression, up to 10⁴-fold induction, with respect to the basal condition (data not shown). A decrease of 40% to 25% of LUC activity was observed compared with that of miR-neg introduction, suggesting that CREB 3′-UTR carries a miR-34b regulatory element. In contrast, exogenous miR-34b expression did not reduce LUC activity with the mutated CREB 3′-UTR, further suggesting that CREB expression is typically regulated by miR-34b (Fig. 2A). To discover the miR-34b role in translational CREB, Western blots were performed. Results confirmed that miR-34b, not miR-neg, inhibited CREB protein expression, which was found drastically reduced in all myeloid cell lines studied. The active phospho-CREB protein was shown to be reduced after miR-34b expression, suggesting a cascade effect on its transcriptional activity (Fig. 2B). Moreover, the expression of CREB was studied after DAC treatment. The demethylating treatment provoked a decrease of CREB protein (Fig. 2C) as a result of miR-34b increased expression (Fig. 1C), confirming CREB as a miR-34b target gene and its promoter to be epigenetically controlled.

miR-34b influences CREB transcriptional activity, which controls leukemia phenotype. To evaluate if restored miR-34b expression influences CREB transcriptional activity, we monitored the mRNA expression of a series of CREB target genes. Results revealed that the transcription of cyclin A1, cyclin B1, cyclin D1, BCL-2, signal transducer and activator of transcription 3 (STAT3), Janus-activated kinase 1 (JAK1), and nuclear factor-κB (NF-κB) was decreased in our model (n = 3; P < 0.05; Fig. 3A). Further, we used protein array to evaluate if CREB target gene repression was also able to influence protein expression levels. The HL60 cell line, after being transfected with the miR-34b oligonucleotide, showed a decreased expression of a series of CREB target proteins (BCL-2, cyclin A1, cyclin B, cyclin D1, STAT3, NF-κB, and JAK1) and a series of kinases and molecules that control cell proliferation [extracellular signal-regulated kinase (ERK), AKT, mammalian target of rapamycin (mTOR), PKA, Smac-Diablo, and SMAD1; n = 2; P < 0.05; Fig. 3B]. The results of the gene and protein expression alteration described above were sought in cell growth and in cell clonogenic potential in an anchorage-independent manner, which are hallmarks of tumorigenicity. Results showed that cell growth (Supplementary Fig. S1) and proliferation were compromised, as a lowered number of cells in S and G₂-M phases were observed (Fig. 3C). Furthermore, clonogenicity assay revealed a significant lowered number of colonies for NB4 and HL60, whereas K562 colonies were significantly larger in dimension and morphology (n = 3; P < 0.05; Fig. 3D).

CREB silencing suppresses myeloid cell proliferation. We used siRNAs to down-regulate expression of endogenous CREB in HL60 and K562. The addition of different CREB siRNAs, but not the siRNA-neg used as control, significantly inhibited CREB expression and myeloid cell line proliferation. CREB mRNA expression significantly decreased from 35% to 75% in HL60 (n = 2; P < 0.05). In the K562 cell line, mRNA decreased CREB was found lower with respect to HL60; in fact, the decrease was from 10% to 60% (Fig. 4A). The effect of CREB silencing strongly influenced protein expression, which was found reduced for all siRNAs used. The siRNA3 for HL60 and siRNA1 for K562 gave the strongest CREB translation inhibition (n = 2; P < 0.05; Fig. 4B). Finally, cell line proliferation and tumorigenicity assays revealed that CREB silencing induced a significantly reduced cell proliferation and clonogenicity (n = 2; P < 0.05; Fig. 4C), as previously shown by miR-34b restoration. These results suggested that CREB regulates growth and survival of myeloid leukemia cells.

miR-34b expression in AML patients. CREB pathologic overexpression was previously reported for a large percentage of AML patients at diagnosis (19). The distribution of miR-34b expression in 78 AML patients at diagnosis (Supplementary Table S2) was heterogeneous but always significantly down-regulated with respect to 17 healthy bone marrow patients, which revealed a higher miR-34b expression (P < 0.001; Fig. 5A). The mean of miR-34b expression in patients was 1.07 ± 0.23 compared with 12.48 ± 4.26 of the healthy bone marrow patients. To explain miR-34b heterogeneity in patients, we considered their CREB protein levels as previously discussed (19). Results showed that patients with higher CREB protein levels (CREB+) had the lowest miR-34b expression (mean, 0.41 ± 0.17), whereas patients with lower CREB protein levels (CREB−) had a heterogeneous and highly miR-34b expression (0.95 ± 0.43), confirming a strong relationship between CREB and miR-34b (P < 0.001; Fig. 5B). Next, we considered miR-34b expression among different cytogenetic groups of AML patients at diagnosis; in particular, patients with rearranged mixed lineage leukemia (MLL) and without cytogenetic markers (NEG) were observed for the lowest detected level of miR-34b (P < 0.001; Fig. 5C) as previously described (31).

The mRNA partners of miRNA, their function, and their tissue specificity are being continuously investigated in normal and diseased samples to increase our understanding of tumorigenesis and improve cancer therapy. The extent of miRNA regulation is under scrutiny. miRNAs have been shown to control a variety of cellular pathways by influencing the expression of specific target genes (32, 33) and are considered here in an effort to explore the mechanism that causes CREB protein overexpression. CREB has recently been defined as an oncogene (34), and it was found overexpressed in acute leukemia, as well as proving able to induce myeloproliferative syndrome in transgenic mice (19, 20).

In silico and gene expression analyses conducted in this research focus on miR-34b as a possible candidate to target CREB. miR-34b is documented to be less expressed in myeloid leukemia cell lines compared with healthy bone marrow, in agreement with other experiments that described low levels of the expression of the miR-34 family of miRNA in other human cancers (21). The forced expression of miR-34s has previously been shown to decrease cell growth and induce senescence in mouse embryonic fibroblasts (35) as well as to control cell proliferation and clonogenic potential in ovarian surface epithelial cells (25), indicating that miR-34 expression plays an important role in influencing tumorigenesis in diverse cell types, probably through the control of different mRNA targets. However, their p53 dependence and target genes remain to be determined (36).

In this regard, our study establishes CREB as one of the targets of miR-34b in myeloid leukemia. miR-34b binds directly to the 3′-UTR region of CREB mRNA with specificity to the seed region, as mutation in this region eliminates this phenomenon. Furthermore, the restoration of miR-34b expression changes the leukemia phenotype, confirming its possible role as tumor suppressor. Until now, there is no genomic evidence of miR-34b down-regulation, apart from the frequent deficiency of functional p53 that drives their transcription in several cancer cells. A recent investigation in colon-rectal cancer found a methylated miR-34b/miR-34c promoter at chromosome 11q23, which might exclude p53 transcriptional activity on this region (28). Our findings about increased miR-34b expression in leukemic cell lines after treatment with a demethylation agent, and the study of the CpG island at the miR-34b/miR-34c promoter, suggest that this mechanism might be considered in myeloid leukemia for explaining miR-34b down-regulation (37). Moreover, myeloid cell lines have been shown to increase miR-34b expression and decrease CREB protein levels after demethylation treatment. These data lead us to suppose that myeloid leukemia cells might epigenetically maintain miR-34b down-regulated to sustain CREB protein overexpression as a possible hypothesis for observed leukemia progression.

The molecular mechanism by which miR-34 family miRNAs suppress tumors is currently under consideration in many cancers (38, 39). Until now, gene expression analyses have been performed, suggesting that the cause of tumor suppression might be the ability to target genes related to the cell cycle pathway (40). CREB is the first direct miR-34b target gene identified. Reverse-phase protein microarray analysis allows (29, 30) the elucidation of the complex cellular signaling that would be influenced by exogenous miR-34b expression and the consequent CREB down-regulation. As a transcription factor, CREB controls thousands of genes that are known to contribute to healthy cell life (17). The observed decreased expression of CREB target genes, such as cyclin A1, cyclin B, and cyclin D1, might explain cell cycle abnormalities found in myeloid cell lines. Furthermore, the down-regulation of the antiapoptotic BCL-2, as well as AKT/mTOR proteins, reads out for the antisurvival signaling pathway caused by the miR-34b restored expression. In addition, the documented down-regulation of STAT3, NF-κBp65, and JAK1, which are directly controlled by CREB, might be considered a critical event in cancer suppression, principally for the disruption of their downstream pathways. The ERK1/2, PKA, AKT, SMADs, and Smac lowered protein expression might be explained as a cellular response to miR-34b suppressor activity on CREB. For the first time, protein arrays account for direct signaling pathways and targets to be further studied for leukemogenesis and targeted therapy (41, 42).

To better understand whether miR-34b down-regulation of CREB might be considered a primary pathway involved in leukemia cell proliferation, we silenced CREB using siRNAs targeting CREB gene and showed that myeloid cell growth and tumorigenicity were strongly compromised. In this sense, we proved that CREB deregulation is a critical phenomenon for leukemia progression and that the role of miR-34b on the CREB protein is a direct and fundamental mechanism of tumor regulation.

The relevance of our in vitro experiments is substantiated by data obtained in AML patients. The fact that CREB protein was previously documented as overexpressed in a large number of AML patients at diagnosis (18, 19) suggests the inverse correlation between CREB and miR-34b expression. The expression levels of miR-34b are shown here to be lowered in patients when compared with healthy bone marrow samples. Moreover, miR-34b is found to be always at the lowest degree of expression when CREB protein levels are high. We also documented a heterogeneous expression of miR-34b in distinct cell types and during myeloid differentiation (31), suggesting that miR-34b may influence hematopoiesis and play a role in different leukemia phenotypes. Its role as a potential marker might be further evaluated (43, 44). In addition, the observation of reduced miR-34b levels in samples that have higher CREB protein levels opens future investigation of miR-34b and CREB as possible diagnostic indicators (45, 46), especially when a broad spectrum of miRNAs involved in AML is available.

Based on the current study, miR-34b targets CREB, mediating biological activity in normal and leukemic tissue. The process of methylation that deregulates miR-34b expression in hematopoietic development leads to pathologic outcomes, mainly through CREB protein up-regulation, which leads to a combinatorial overexpression of a large number of targets provoking cell proliferation and survival. The silencing of miR-34b/miR-34c genes by methylation in leukemia supports the possibility that a large series of other target oncogenes that would contribute to disease are up-regulated (47–49); these remain to be discovered. Therefore, the ectopic miR-34b expression or the use of demethylating agents in myeloid leukemic cells could reactivate the control on CREB expression, contributing to the reduction of malignancy. miR-34b restoration turns out to be a fundamental step in treating myeloid cells, prompting consideration in AML pathogenesis and for new therapeutic strategies.

The authors declare no competing financial interest.

Grant support: University of Padua, “Fondazione Città della Speranza,” and Murst PRIN, Associazione Italiana per la Ricerca sul Cancro.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

We thank Benedetta Accordi, Manuela Sciro, Emanuela Giarin, Grazia Giacometti, Sabrina Gelain, and Alessandra Beghin and Nancy Jenkins for editing.