Proceedings of the discoveries on post-transcriptional Bcl-2 deregulation in human leukemias / lymphomas

The Bcl-2 (B-cell lymphoma 2) antiapoptotic gene has been discovered in virtue of its over-expression occurring in B-cell leukemias/lymphomas carrying the 14;18 chromosomal translocation [t(14;18)], which places the Bcl-2 gene next to the immunoglobulin heavy chain (IgH) locus. In this condition, the transcription of the Bcl-2 moiety of the Bcl-2/IgH fusion gene is driven by the four enhancers located in 3’ of the IgH moiety and is, therefore, excessive. This leads to overproduction of Bcl-2 protein, which confers a survival advantage that contributes to neoplastic transformation. Nevertheless, in most malignancies, comprising chronic lymphocytic leukemias, breast, prostate, colorectal and lung cancer, the over-expression of Bcl-2 does not imply chromosomal rearrangements, suggesting that alterations at post-transcriptional level could be involved. Collaborating with the group of Angelo Nicolin (University of Milan, Italy), we first disclosed the existence of a Bcl-2 post-transcriptional control based on interplay among an Adenine and uracil-Rich cis-acting Element (ARE) located in the 3’UTR of Bcl-2 mRNA and several trans-acting ARE-Binding Proteins (AUBPs). We also demonstrated its deregulation in human leukemias/lymphomas. In particular, we have identified some Bcl-2 AUBPs such as AUF-1, TINO/hMex-3D, the Bcl-2 protein itself and ζ-Crystallin and described their qualitative or quantitative alterations in cancer cells. Moreover, in the attempt to correct Bcl-2 deregulation in the human diseases characterized by defects or excesses of apoptosis, we have modulated exogenously Bcl-2 expression by means of different antisense strategies. In this research highlight, we briefly report our proceedings, in which a long non-coding Bcl-2/IgH antisense RNA (Bcl-2/IgH AS) we discovered in a serendipitous manner has played a key role.

The key role of apoptosis defects consequent to overexpression of Bcl-2 (B-cell lymphoma 2) gene in cancer development and therapy has been widely recognized [1] .Bcl-2 has been discovered in B-cell leukemias/lymphomas carrying the 14;18 chromosomal translocation t (14;18), which places the Bcl-2 gene next to the immunoglobulin heavy chain (IgH) locus and is therefore over-transcribed by four enhancers located in 3' of the IgH moiety.This leads to the production of excessive amounts of the Bcl-2 antiapoptotic oncoprotein [2] , responsible for a survival advantage leading to neoplastic transformation [3] .Nevertheless, in most malignancies, such as chronic lymphocytic leukemias, breast, prostate, colorectal and lung cancer, Bcl-2 over-expression can occur in the absence of chromosomal rearrangements, which suggests that it could be caused by alterations at post-transcriptional level.A large amount of evidences indicates that up-and down-regulation of Bcl-2 expression is modulated at transcriptional,

RESEARCH HIGHLIGHT
post-transcriptional, including mRNA stability and translational control, and post-translational levels.Expression of the Bcl-2 gene was known to be regulated transcriptionally by a negative regulatory element [4] and by two estrogen-responsive elements identified within its coding region in a breast cancer cell line [5] .An 11 amino-acidic upstream Open Reading Frame (uORF) located within the 5'UTR of Bcl-2 mRNA inhibits translation of Bcl-2 protein [6]   .In addition, Bcl-2 translation is controlled by the presence of an Internal Ribosome Entry Site (IRES) within the Bcl-2 mRNA 5'UTR.Bcl-2 IRES activity is induced upon cell stress, when cap-dependent translation is repressed, and enables to replenish levels of Bcl-2 protein preventing unwarranted apoptosis induction [7] .A mechanism of post-translational control of Bcl-2 expression has been described to be mediated by phosphorylation of Bcl-2 protein at different amino acid positions [8,9] .Two decades ago, starting from the serendipitous identification of a long non coding Bcl-2/IgH antisense RNA in t(14;18) leukemic cells, in collaboration with the group of Angelo Nicolin (University of Milan, Italy) we identified a complex post-transcriptional mechanism of Bcl-2 regulation, which proceedings are described below.
Aimed to reduce Bcl-2 over-expression in t(14;18) cells carrying the Bcl-2/IgH fusion gene by antisense strategies we surprisingly noted that, whilst synthetic antisense oligodeoxyribonucleotides (aODNs) targeting Bcl-2 or IgH RNA did not elicit any effect, the relevant oligodeoxyribonucleotides designed in sense orientation (sODNs) as controls induced a marked decrease of Bcl-2 mRNA and protein [10] .The ability of sODNs in down-regulating Bcl-2 expression suggested that bona fide they could target a natural Bcl-2/IgH antisense RNA that, since its inactivation by sODNs leaded to down-regulation of Bcl-2 expression, could overlap/mask a negative regulative element located in the hybrid Bcl-2/IgH mRNA.We obtained the direct evidence of the actual existence of a long non-coding antisense Bcl-2/IgH RNA (Bcl-2/IgH AS) by a strand-specific PCR analysis, followed by directly sequencing of PCR products.The Bcl-2/IgH AS was present in t(14;18) follicular lymphoma DOHH2 cells while was absent in untranslocated Burk1itt's lymphoma Raji and Acute Lymphatic Leukemia (ALL) Jurkat cells, which indicated that its existence was conditioned by the t(14;18) translocation generating the Bcl-2/IgH hybrid oncogene (Figure 1).Originating in the IgH locus, encompassing the t(14;18) fusion site and spanning at least the complete 3' UTR region of the Bcl-2 mRNA, the hybrid Bcl-2/IgH AS has a certain relationship with another antisense transcript previously identified in Burkitt lymphomas starting in the mu-switch region of the IgH locus and spanning the c-myc gene [11] .The study of the pathophysiological role of the long non-coding RNAs is one of the most intriguing aspects of post-transcriptional control of gene expression [12] .For many years, the fusion sequences arising from chromosomal translocations have been recognized highly tumor-specific molecular targets for ODNs [13,14] .In analogy, the Bcl-2/IgH AS has proven an optimal target for synthetic ODNs, being of more general relevance respect to the single fusion points of each individual t(14;18) cell line we have tested [15] .Indeed, we have targeted the Bcl-2/IgH AS either within the Bcl-2/IgH fusion regions, which have a sequence specificity presumably limited to a single cell line, or within the ectopic Bcl-2 region upstream from the major breakpoint region and the IgH segment, which sequence specificity is extended to all cells carrying the t (14;18).Although all sODNs complementary to the Bcl-2/IgH AS induced a fast reduction of proliferation and a late but massive apoptosis, while the effectiveness of ODNs targeting the Bcl-2/IgH fusion regions was limited to each cell line, the effectiveness of all ODNs targeting the Bcl-2 or IgH regions was extended to all t(14;18) cell lines.The selectivity and efficacy of all sODNs tested provided support for the development of therapeutic ODNs targeting Bcl-2/IgH AS expressed in human follicular lymphomas.
Searching for the negative regulative element we supposed to be harbored in the Bcl-2 mRNA moiety on Bcl-2/IgH RNA, we found that the 3' untranslated region (3'UTR) of Bcl-2 contained a 107-nucleotide Adenine+uracyl Rich Element (ARE) provided with a series of AUUUA repeats similar to others elements endowed with mRNA negative regulative functions [16] .Besides its impressive evolutionary conservation (from C. elegans to humans), the Bcl-2 ARE had all the features of a typical ARE, included a particular distribution of the AUUUA pentamers near an UUAUUUAUU nonamer, which let it ascribe to the class II AREs according to the classification proposed by Shyu et al. [17, 18]   .The class II AREs usually impart a biphasic kinetic of degradation to their relevant mRNA, are sensitive to actinomycin D treatment, and do not necessarily act on translation.Bona fide, the Bcl-2/IgH AS could stabilize the Bcl-2 mRNA in t(14;18) cells by overlapping its ARE.
AREs modulate the fate of relevant mRNAs (in terms of stability, localization and translation) by interacting with a series of trans-acting factors included RNA-binding proteins, namely ARE-binding proteins (AUBPs) and miRNAs.On this basis, the exhaustive clarification of the ARE dependent post-transcriptional control of Bcl-2 expression required identification and functional analysis of the Bcl-2 AUBPs.For this purpose, we firstly demonstrated in Jurkat cells that the ARE of Bcl-2 ARE bound to several cytoplasmic proteins, which molecular weights were from 35 to 100 kDa, and whose pattern underwent modifications in response to apoptotic stimuli.We hypothesized that these proteins must be trans-acting regulators in the ARE mediated degradation of Bcl-2 mRNA during apoptosis [19] .Considering the antiapoptotic activity of Bcl-2, our observations strongly suggested that possible alterations of Bcl-2 AUBPs could contribute to carcinogenesis and neoplastic progression.
The first analyses of the Bcl-2 AUBPs in Bcl-2 over-expressing cell lines demonstrated significant alterations with respect to the normal counterpart.In particular, the observation that proteins ranging from 30-50 kDa underwent the most noticeable increase led us to hypothesize that AU-rich element RNA-binding protein 1 (AUF1) could be a Bcl-2 ARE-binding protein.Indeed, AUF1, first identified as an RNA-binding protein with selective affinity for AREs located within mRNAs such as c-myc, c-fos, and GM-CSF [20,21] , is comprised of four isoforms of 37, 40, 42, and 45 kDa.We demonstrated that AUF1 bound to the Bcl-2 mRNA both in vitro and in vivo and that potentially all its isoforms constituted complexes with the Bcl-2 ARE in Jurkat cells [22] .At doses able to induce apoptosis, UVC irradiation induced an increase of cytoplasmic levels of the p45 AUF1 isoform, which paralleled an enhancement of a Bcl-2 mRNA/AUF1 complex and subtended a mechanism requiring caspase activation.These results indicated that ARE-mediated Bcl-2 mRNA down-regulation during apoptosis involved AUF1 and suggested different roles for its four isoforms.
By using a non-radioactive cell-free mRNA decay system we observed that the degradation of Bcl-2 mRNA was related to the amount of Bcl-2 protein expressed by different cell types at steady state, was lost upon Bcl-2 depletion and was reconstituted by adding recombinant Bcl-2.This clearly indicated that Bcl-2 was necessary to activate the degradation complex on the relevant RNA target [23] .Successively, in the context of a AUBPs silencing approach, we demonstrated that Human antigen R (HuR) knockdown reduced the expression of endogenous Bcl-2, whereas increased significantly a Bcl-2 ARE-reporter transcript, which suggested that HuR expression has opposite effects on endogenous and ectopic Bcl-2 ARE [24] .Having also demonstrated that Bcl-2 protein had a specific and dose dependent role in regulating its own mRNA degradation and that its activity overcame the activity of HuR, we suggested that Bcl-2 was the main determinant of Bcl-2 mRNA turnover [24] .Confirming our observations, Ishimaru D. et al. demonstrated that HuR plays a positive role in Bcl-2 mRNA stability and translation regulation in HL60 leukemia and A431 epidermoid carcinoma cells [25] .We have also shown by UV cross-linking that KH-type splicing regulatory protein (KSRP) and Tristetraprolin (TTP) bound in vitro to the Bcl-2 mRNA [26] .While the functional role of KSRP on Bcl-2 remains to be disclosed, Park SB et al. have very recently demonstrated the ability of TTP to down-regulate Bcl-2 expression in head and neck cancer cells in response to cisplatin [27] .
In an attempt to search for other Bcl-2 mRNA binding proteins, we used the yeast RNA three-hybrid system assay and identified a novel human protein we named TINO, which interacted with Bcl-2 ARE [28] .Upon binding, TINO had a post-transcriptional negative activity on Bcl-2 expression, demonstrated by its ability to destabilize a chimeric reporter construct containing the Bcl-2 ARE.The predicted sequence of TINO had two heterogeneous nuclear ribonucleoprotein K homology (KH) amino-terminal motifs for the nucleic acid binding and a carboxy-terminal RING domain endowed with a putative protein E3 ubiquitin ligase activity.In addition, using the protein-protein BLAST analysis, we observed that the novel protein was evolutionarily conserved.In particular, we identified Posterior End Mark-3 (PEM-3) of Ciona savignyi and Muscle EXcess protein-3 (MEX-3) of Caenorhabditis elegans as TINO orthologous proteins.More recently, TINO has been recognized as a variant form of hMex-3D [29] .Compared to hMex-3D, TINO is truncated in its N-terminal region, beginning at the first KH domain.This form of hMex-3D also differs at its C-terminal end, with 19 amino acids, encoded by a potential alternative exon, replacing the four last terminal amino acids of hMex-3D.We did not succeed in identifying TINO protein in any experimental model and conditions we have explored.Although the function of TINO/hMex-3D remains to be clarified, some data reported in the literature combined with a series of our preliminary results strongly suggest that it could be involved in a conserved circuit comprising Quaking/GLD-1 and ζ-Crystallin, which alteration leads to stem cell polarity disruption and acquisition of cancer phenotype [30] .
Five years ago, by means of a bidimensional SDS-PAGE carried out on bcl-2 AUBPs from phytoemoagglutinin (PHA)-activated T lymphocytes or Jurkat T-cells followed by mass spectrometry analysis, we have identified ζ-crystallin as a new Bcl-2 AUBP, which augmented binding to the Bcl-2 mRNA in ALL T-cells and increased Bcl-2 expression by enhancing the stability of its mRNA [31] .The specific association of ζ-crystallin to the Bcl-2 ARE was significantly higher in T cells of ALL patients respect to normal T cells, which accounted for the higher stability of Bcl-2 mRNA and suggested a possible contribution of ζ-crystallin to Bcl-2 overexpression occurring in this leukemia.Surprisingly, we found that the cytoplasmic levels  Nucleolin, which overexpression and altered subcellular localization in Chronic Lymphatic Leukemia (CLLs) leads to excessive Bcl-2 mRNA stability [32] ; EBP1, endowed with destabilizing activity on a chimeric construct harboring the Bcl-2 ARE in HL-60 leukemia cells [33] ; ZFP36L1, which Bcl-2 mRNA destabilizing activity has been demonstrated in leukemia, lymphoma and renal carcinoma cell lines [34] (Figure 2).Furthermore, some miRNA have also been demonstrated to be involved in Bcl-2 post-transcriptional control and in its alterations in human leukemias [35][36][37][38] .Very recently, Díaz-Muñoz MD et al. have disclosed the ability of Bcl-2 AUBP/ARE association to stabilize in vivo Bcl-2 mRNA, contributing to Bcl-2 protein over-production and B cell survival [39] .
The complex of the past and current literature clearly indicate that the pathogenesis of most human diseases underlies either defects or excesses of apoptosis and that Bcl-2 deregulation plays a key role in apoptosis execution.On this basis, Bcl-2 still represent a preferred target for innovative cancer therapies [40][41][42] and some clinical trials have also been recently described [43] .For more than two decades, we have used the antisense strategy as potential therapeutic tool, and (besides the Bcl-2/IgH AS in t(14;18) cells), we have chosen the Bcl-2 ARE as rationally preferred oligonucleotide target to down or up regulate Bcl-2 expression.Indeed, the simulated folding of Bcl-2 ARE by the MUFOLD program [44] indicates that it forms a relatively wide loop and is therefore an optimal target both for natural endogenous molecules (the Bcl-2 AUBPs) and for synthetic exogenous molecules (such as, antisense oligonucleotides and ribozymes).
To downregulate Bcl-2 expression in apoptosis-defective Bcl-2 overexpressing Raji cells, we targeted the Bcl-2 ARE with a synthetic hammerhead ribozyme [45] , designed relying on in vitro results obtained by probing RNA accessibility to antisense ODNs.The cellular uptake of this lipotransfected ribozyme resulted in a marked reduction of Bcl-2 mRNA and Bcl-2 protein levels and a dramatic increase of cell death by apoptosis.Although the Bcl-2 ARE is not a tumour specific target, we proposed to evaluate such ribozyme as potential therapeutic tool for the treatment of Bcl-2 overexpressing tumors.
Symmetrically, we have attempted to prevent Bcl-2 dowregulation, thereby inhibiting apoptosis in pathological conditions characterized by apoptosis excesses, by targeting the Bcl-2 ARE with three 26-mer 2'-O-methyl oligoribonucleotides (ORNs) homologous to the core region of the Bcl-2 ARE used as decoy-aptamers.Sense-oriented ORNs competed with the Bcl-2 ARE for the interaction with both destabilizing and stabilizing AUBPs in cell-free systems and in cell lines [26] .Moreover, ORNs induced mRNA stabilization and therefore up regulated both Bcl-2 mRNA and protein levels.Furthermore, Bcl-2 ORNs stabilized other ARE containing transcripts and up regulated their expression.We also demonstrated that treatment of the SHSY-5Y neuronal cells with Bcl-2 ORNs prevented Bcl-2 down-regulation in response to apoptotic stimuli, such as glucose/growth factor starvation or oxygen deprivation, inhibited cell cycle entry and induced a markedly increase of cellular neurite number and length, a hallmark of neuronal differentiation resulting from Bcl-2 up-regulation [46] .Enhancement of apoptotic threshold and induction neuronal differentiation by Bcl-2 ORNs suggested evaluating their potential application to prevent pathological apoptosis and neuronal degenerations.
The previously described results confirmed that the destabilizing activity of Bcl-2 mRNA ARE, we discovered thanks to the Bcl-2/IgH AS, underlies a new mechanism of post-transcriptional control of Bcl-2 expression, whose disruption could contribute to the oncogenicity of this antiapoptotic gene.Furthermore, they demonstrated that the Bcl-2 ARE represents an optimal target for antisense strategies aimed to down-or up-regulate Bcl-2 expression, thereby modulating apoptosis in apoptosis deregulation-related diseases.

Figure 1 .
Figure 1.Schematic structure of t(14;18) translocation juxtaposing the Bcl-2 gene to the IgH locus.(A) The hybrid Bcl-2/IgH gene showing UTRs of the Bcl-2 moiety (purple hatched) flanking the coding region (red hatched) and the IgH moiety (orange hatched).Intronic sequences are reported as lines.(B) The hybrid Bcl-2/IgH mRNA showing its ORF (red hatched) and its ARE (green hatched).(C) The hybrid Bcl-2/IgH antisense RNA showing the extension of the portion revealed (blue line) by the strand specific RT-PCR, which includes the ARE overlapping stretch.
of ζ-crystallin did not differ in normal PHA-activated T-lymphocytes with respect to leukemia T-cells, indicating that the different binding of ζ-crystallin to the Bcl-2 ARE in leukemia T-cells did not depend on its concentration and might be explained by other mechanisms, which remain to be disclosed.We propose two alternative scenarios: the first predicts that qualitative alterations of the ζ-crystallin protein in ALL T-cells allow increased Bcl-2 ARE binding; the second proposes that modifications of Bcl-2 AUBPs pattern in ALL T-cells could advantage ζ-crystallin interaction with the Bcl-2 ARE.Others have identified further Bcl-2 AUBPs.They are

Figure 2 .
Figure 2. Scheme of the Bcl-2 mRNA.The position and sequence of the Bcl-2 mRNA AU-Rich Element (ARE) and its stabilizing or destabilizing ARE-Binding proteins (AUBPs) are shown.References are listed in brackets.