George A. Calin, Department of Experimental Therapeutics and Department of Cancer Genetics, University of Texas, MD Anderson Cancer Center, Houston, TX 77030, USA. E-mail: firstname.lastname@example.org
MicroRNAs (miRNAs) have been linked to the initiation and progression of chronic lymphocytic leukaemia (CLL). The main molecular alterations are represented by variations in gene expression, usually mild and with consequences for a vast number of target protein-coding genes. Recent studies have shown that miRNAs are the main candidates for the elusive class of CLL predisposing genes. These discoveries could be exploited for the development of useful markers for diagnosis and prognosis, as well as for the development of new RNA-based cancer therapies.
MicroRNAs (miRNAs) are a class of small non-coding RNAs (ncRNAs) distinct from, but related to small interfering RNAs (siRNAs) regulating both messenger RNA (mRNA) translation and decay (Valencia-Sanchez et al, 2006). MiRNAs are present in a wide variety of organisms and are expressed in a tissue- and/or time-restricted fashion, to ensure the accuracy of gene expression programs involved in the genesis of distinct cell and tissue types [for reviews see (Ambros, 2004; Bartel, 2004)].
Mature miRNAs range in size from 19 to 24 nucleotides (nt), and are initially transcribed as a part of a much longer primary transcript (Lee et al, 2002) called primary microRNA (pri-miRNA). Pri-miRNAs undergo a complex maturation process involving subsequent digestion steps performed by two distinct double-stranded RNA-specific ribonucleases. The first one is completed in the nucleus by Drosha and produces a hairpin RNA of 70–100 nt, the precursor microRNA (Cullen, 2004). The hairpin RNA is then exported to the cytoplasm via an Exportin-5 RanGTP-dependent mechanism, where Dicer carries out the second maturation step. Finally, the mature single-stranded miRNA binds to proteins of the Argonaute family to form the RNA-protein complex known as RISC (RNA-induced silencing complex) (Hammond et al, 2001; Gregory et al, 2005). Mature miRNAs, loaded onto the RISC, bind through the seed regions (the 8 nt starting from the second position of the miRNA 5′-end), to mRNAs, mainly at their 3′-untranslated region (3′-UTR), through a significant but not complete sequence complementarity (Ambros, 2003). By a mechanism that is not fully characterized, the bound mRNA remains un-translated, resulting in reduced levels of the corresponding protein or alternatively, can be degraded, causing reduced levels of both the mRNA transcript and the corresponding protein.
Although increasing numbers of miRNAs are being identified, most have poorly defined biological function. Nevertheless, a few members of the expanding microRNA family have been shown to participate in essential biological eukaryotic cell processes, such as developmental timing, cellular differentiation, proliferation or apoptosis. Among these are to be mentioned haematopoietic B-cell lineage commitment (miR-181), B-cell survival (miR-15a and miR-16-1), cell proliferation control (miR-125b and let-7), brain patterning (miR-430), pancreatic cell insulin secretion (miR-375) and adipocyte development (miR-143) (for review see Harfe, 2005).
Chronic lymphocytic leukaemia, a leukaemia without a deciphered molecular architecture
Chronic lymphocytic leukaemia is the most common leukaemia among adults in western countries, accounting for about 30% of all cases of leukaemia in the US (Chiorazzi et al, 2005). CLL is a disease of the elderly, characterized by the progressive accumulation of morphologically mature but immunologically less mature, CD5-positive B lymphocytes in the blood, marrow and lymphatic tissues. The great majority of these leukaemic cells (>90%) is non-dividing and is arrested at the G0/G1 phase of the cell cycle. CLL cells are also quite resistant to apoptosis. It has been suggested that the excess of B cells is more likely to be the result of decreased apoptosis and deregulation of cell cycle control than of an increased proliferation rate. During its natural course, this indolent disease can evolve in to an aggressive form of lymphoma and/or leukaemia. Because current therapy is not considered curative, treatment generally is delayed until the patient develops symptoms, leukaemia-associated myelosuppression, or until evidence of disease progression and poor prognostic markers manifest.
Several factors predicting the clinical course of patients with CLL have been defined (Orchard et al, 2004; Rassenti et al, 2004). Lack of somatic mutations in the immunoglobulin heavy-chain variable-region gene (IGHV) or high-level expression of the 70-kDa zeta-associated protein (ZAP-70) by the leukaemia cells are associated with an aggressive clinical course, whereas expression of mutated IGHV or low levels of ZAP-70 are associated with indolent forms of CLL (Kipps, 2001; Chiorazzi et al, 2005). It was also found that genomic aberrations in CLL are important independent predictors of disease progression and survival (Dohner et al, 2000). However, the molecular basis of these correlations was until recently largely unknown, as well as the pathogenic events leading to the initiation and progression of CLL.
Overall, survival for patients with CLL varies widely. Whereas some patients might rapidly progress and/or require therapy, others might enjoy a very long period without leukaemia-associated symptoms or disease-related complications. Conceivably, early treatment of patients with aggressive disease could provide for more effective therapy, particularly for patients with low-tumour burden at diagnosis. Therefore to help clinicians make the correct therapeutic decisions, there is a need for prognostic markers that can provide reliable indication of the probability that a patient will soon develop progressive disease.
Genetic predisposition is thought to exist, as CLL occurs preferentially in patients with at least one first degree relative with a history of CLL or other tumours. To date, no clear predisposing loci have been recognized for CLL (Caporaso et al, 2004), although newer linkage (Sellick et al, 2007) and association studies (Slager et al, 2007) are providing new leads.
The hallmark of the malignant, mostly non-dividing, B cells of CLL is the overexpression the anti-apoptotic protein Bcl-2 (Kitada et al, 1998). BCL2 is a member of the Bcl-2 protein family that, in normal tissues, is responsible for maintaining the delicate homeostasis between proliferation and apoptosis. In particular, the role of Bcl-2 in this genetic program of eukaryotic cells is to favour cell survival by inhibiting cell death (Cory & Adams, 2002). Overexpression of Bcl-2 protein was reported in many types of human cancers including leukaemias, lymphomas and carcinomas (Sanchez-Beato et al, 2003). In follicular lymphomas and in a fraction of diffuse B-cell lymphomas, the mechanism of BCL2 activation was found to be the translocation t(14;18)(q32;q21), which places BCL2 under the control of immunoglobulin heavy chain enhancers, resulting in deregulated expression of the gene (Tsujimoto et al, 1984, 1985). These cases, in which BCL2 is juxtaposed to immunoglobulin loci, constitute <5% of all CLL cases (Adachi et al, 1990) and, for the remaining 95%, no mechanism until recently had been discovered to explain BCL2 deregulation in B-CLL.
MicroRNAs MIRN15a and MIRN16-1 are deleted in the majority of CLLs
The first report linking miRNAs and cancer involved CLL (Calin et al, 2002). Detailed genetic analysis, including extensive loss of heterozygosis (LOH), mutation, and expression studies failed to demonstrate the consistent involvement of any of the 11 protein-coding and non-coding genes located in or close to the deleted region at 13q14.3. One factor hindering the initial analysis of the gene(s) responsible for loss at 13q14 in CLL was the lack of a clear definition of the minimal region of loss. Indeed several relatively large (between 130 and 550 kb) regions were described as being lost in cases that harboured deletions at 13q14 (Fig 2). To overcome this difficulty somatic cell hybrids between mouse LM-TK (tyrosine kinase defective) cells and CLL cells carrying 13q14 translocations and/or deletions [a CLL case carrying a t(2;13)(q32; q14) translocation and another CLL case carrying a t(2;13)(q12; q13) translocation] were generated. PCR screening of resulting hybrid clones allowed the isolation of the two copies of chromosome 13 present in these tumours. A minimal deleted region of 29 kb, containing the 13q14 tumour suppressor gene(s) was found to lie between exons 2 and 5 of DLEU2 (also known as LEU2). DLEU2 has been the object of extensive studies and at that time it had already been excluded as a likely candidate tumour suppressor gene for CLL. A cluster of two miRNAs, MIRN15A and MIRN16-1, was found to be exactly located inside the minimal region of loss at 13q14, and both genes were found to be deleted or downregulated in the majority (∼70%) of CLL samples (Calin et al, 2002) (Fig 2).
MIRN15A and MIRN16-1 form a small cluster (with <200 bp distance between one and another) at chromosome 13q14.3. Both genes are ubiquitously expressed, as non-coding RNAs (miR-15a and miR-16-1), with highest levels in normal CD5+ lymphocytes, highlighting their relevant part in normal CD5+ B-cell homeostasis. Additionally, in normal tissues, miR-16-1 is consistently expressed at higher levels compared with miR-15a. The majority of CLL patients present a downregulation of MIRN15A and MIRN16-1 expression correlating with homozygous LOH in 68% of the informative cases. Northern blot analysis showed that both miRNAs could be detected in CLL cases with large homozygous deletions at 13q14, albeit at very low levels. This could be because of the presence of other highly similar microRNA genes in the genome that can be expressed in this disease. Indeed, a cluster of very similar micro RNA with different precursors, namely MIRN15B/MIRN16-2, was found on chromosome 3q25-26.1 (Lagos-Quintana et al, 2002). The functional redundancy of these miRNAs indicates a very fine mechanism of regulation for microRNA targeted messenger RNAs. In this particular case however, only a relatively low contribution to the mature microRNA product originates from chromosome 3. By using different probes, designed to specifically recognize miR-16-1 precursors, it was found that, while the premessenger from chromosome 13 is detectable at low levels using its specific probe, no specific hybridization is present in the same samples using the chromosome 3 mir-16 precursor-specific probe. Another possible explanation for the apparently missing compensation by the other family members of the MIRN16 family can be found within the role of microRNA as micromanagers in the fine tuning of gene expression, as a result relatively low alterations in microRNA copy number can lead to major phenotypic consequences (Bartel & Chen, 2004). In this case the MIRN16 family members probably behave as haploinsufficient tumour suppressor genes, where variations in gene expression levels are responsible of the phenotype observed. Nonetheless the two hit inactivation model of tumour-suppressor gene applies unequivocally to the MIRN15A/MIRN16-1 locus per se, as one allele can be deleted by heterozygous LOH and the second may be eliminated by a germ line mutation in a region which compromises its processing (Calin et al, 2005b; Chen, 2005).
How MIRN16-1 and MIRN15a play a pathogenic role in CLL was clarified throughout subsequent in silico, in vitro and in vivo functional studies (Cimmino et al, 2005). First, the two miRNAs presented targeting sites at the BCL2 3′-UTR and were conserved in both human and mouse. In B-CLL samples, MIRN16-1 and MIRN15a expression was inversely correlated with Bcl-2 protein levels. Overexpression of these two miRNAs reduced Bcl-2 expression levels and resulted in apoptosis of the human leukaemia cell line MEG01, which harbours deletions in 13q14. Additionally, xenograft mouse tumours derived from subcutaneously inoculated MEG01, were unable to grow as large as with untreated cells, when inoculated cells were stably transduced with a vector encoding MIRN-15A and MIRN-16-1 (A. Cimmino and G. A. Calin, unpublished observations). Altogether these studies define a novel microRNA mediated mechanism accounting for BCL2 expression and its deregulation in this leukaemia.
How big is the involvement of these small non-coding RNAs in CLL?
MicroRNA expression has also been extensively studied by means of an oligonucleotide miRNA microarray chip, first developed by Croce and collaborators (Liu et al, 2004), containing 368 specific probes for human precursor and mature miRNAs and showed distinct patterns of microRNA expression in human and mouse tissues (tissue-specific microRNA expression signatures). MicroRNA expression profiling delineated sharp differences in microRNA expression between CLL samples and normal CD5+ B cells, the latter representing the normal counterpart of CLL malignant cells (Calin et al, 2004). A unique signature of 13 microRNA (out of 190 analysed) was able to distinguish cases according to their IGHV mutation status and ZAP-70 expression levels, and was also associated with disease progression. Among the miRNAs downregulated in patients with good prognosis, miR16-1 and miR-15a were found expressed at low levels, in accordance with previous reports that 13q14.3 deletions at the locus encoding these genes display a favourable evolution of the disease. The common members for both signatures were miR-15a and miR-16-1, pointing to an important pathogenetic role for these two microRNA with diagnostic and prognostic consequences and possible gene-specific anti-cancer therapeutic applications (Calin et al, 2005b).
The importance of miR-15a and miR-16-1 as critical players in the pathogenesis of CLL is also strengthened by array single nucleotide polymorphism (SNP) studies that confirmed how recurrent 13q14 deletions target MIRN-15A and MIRN-16-1 (Pfeifer et al, 2007). Furthermore, downregulation of miR-15a and 16-1 has been also reported in cases of diffuse large-B-cell lymphomas (Eis et al, 2005), prostate cancer (Porkka et al, 2007), pituitary adenomas (Bottoni et al, 2005) and breast cancer (Volinia et al, 2006) suggesting that the significance of this mechanism may be extended to other human malignancies.
A recent report provides evidence that mRNA decay, mediated by the AU-rich elements (ARE) present in the 3′-untranslated region (UTR) of unstable mRNAs is dependent on the presence of miR-16 (Jing et al, 2005). The role of miR-16 in ARE-RNA decay is proven to be sequence-specific and to require the ARE binding protein tristetraprolin (TTP). This interaction is indirect and needs TTP association with Ago/eiF2C family members to complex with miR-16 and assist in the targeting of ARE and emphasizes how miRNA targeting ARE is an essential step in ARE-mediated mRNA degradation (Jing et al, 2005).
Further proof of the role played by miR-16 in tumorigenesis is shown by a genome wide study performed in search for targets of the MIRN16 family, exploiting gene expression profiles (Linsley et al, 2007). Investigation of transcripts with expression patterns inversely correlated to the introduced microRNA and with a 3′-UTR enriched in motifs complementary to the microRNA seed regions revealed that many miR-16 targets are important regulators of the G0/G1 cell cycle transition and therefore miR-16 downregulation may enable tumour cells to avoid cell cycle control mechanisms.
To explain gene alteration in cancer by a microRNA-dependent mechanism, besides looking for mutations inside or surrounding microRNA genetic loci, one can also search for mutations altering the 3′-UTR-site targeted by the miRNA. These changes can lead to deregulation of the oncogenes and/or tumour suppressor genes with the mutated 3′-UTR (Clop et al, 2006). This is the case for miR-206, which showed a 3·3-fold increase in its ability to downregulate the estrogen receptor 1 (hER-1) in breast cancer when it bears the C–T SNP-9 (Adams et al, 2007). It cannot be ruled out that SNPs occurring on the 5′-UTR region of a target gene involved in cancer may also affect its expression. A regulatory polymorphism has been localized to the promoter region of BCL2 and was associated with increased protein expression levels and correlated with a poor prognosis (Nuckel et al, 2007) in CLL patients. It is important to consider sequencing CLL samples to identify polymorphic variants responsible for creating or destroying conserved or non-conserved microRNA target sites at loci regulating BCL2 expression (UTR regions), potentially representing a new mechanism for altered gene expression by microRNA.
Several other miRNAs have been shown to be involved in human cancer, particularly in haematological cancers. High expression of human miR-155 was reported in Hodgkin lymphoma (Kluiver et al, 2005) and accumulation of miR-155 and BIC RNA was also reported in other B-cell lymphomas (Eis et al, 2005). Recently, it has also been shown that enforced expression of the MIRN17-92 cluster acts in concert with MYC expression to accelerate tumour development in a mouse B-cell lymphoma model (He et al, 2005). Collectively, this evidence strongly suggests a role for miRNAs in haematological cancers of B-cell origin.
The first genome-wide systematic search for correlations between the genomic positions of miRNAs and cancer-associated genomic regions (CAGRs) in humans was performed and demonstrated the significance of miRNAs in human cancers (Calin et al, 2004). Of 186 miRNAs mapped to the human genome, 19% (35 of 186) were located inside or near fragile sites (FRAs) and the relative incidence of miRNAs inside FRAs occurred at a rate 9-times higher than in non-FRAs (P <0·001 by mixed effect Poisson regression). A significant proportion, 52·5% (98 of 186) of miRNAs are in the CAGRS described in a variety of tumours, such as lung, breast, ovarian, colon, gastric and hepatocellular carcinoma, as well as leukaemias and lymphomas (P <0·001 by mixed effect Poisson regression).
The TCL1A transgenic mice develop a disease very similar to human CLL and present extensive miRNAs alterations
T-cell leukaemia 1 (TCL1) gene (TCL1A) is over expressed in about the 25-35% (90% positive by immunochemistry) of CLL cases (G. A. Calin and C. M. Croce, unpublished observations) and its high expression correlates with expression of unmutated IGHV and ZAP-70, supporting the idea that TCL1A-driven CLL constitutes an aggressive form of CLL.
TCL1A was originally discovered as the oncogene located at the 14q32.1 breakpoint region in T-cell prolymphocytic leukaemia (T-PLL) (Virgilio et al, 1994). While inversions or reciprocal translocation at this site juxtaposing the T-cell receptor (TCR) a/d or TCR-b promoter/enhancer elements with TCL1A explain its overexpression in T-PLL, no chromosomal aberration was found to constitutively activate TCL1A in CLL (Narducci et al, 2000). To elucidate its role in B-cell development, differentiation and transformation, transgenic mice under the control of a promoter and enhancer expressing specifically TCL1 expression in the B-cell compartment were produced. At first, the mice developed a preleukaemic state evident in blood, spleen, marrow, peritoneal cavity and peripheral lymphoid tissue, and later a frank leukaemia with all the characteristics of CLL (Bichi et al, 2002). These mice presented a very aggressive disease, resembling the ZAP-70 positive/IgVH negative, treatment resistant human form of CLL (Yan et al, 2006). Comparative genomic hybridization, performed on CD5+ tumour cells from the peritoneal cavity as well as from the spleen of sick mice, identified several regions of genomic deletion and amplification. One of the most commonly deleted regions (present in seven of 12 mice) was a stretch of about 20 Mb located on the long arm of chromosome 14, with high resemblance to the 13q14.3 deletion found in human CLL. In fact the centromeric extremity of this region contains the mouse homologues of MIRN15A and MIRN16-1. Half of the tumours with these genomic deletions have lost completely the normal homolog, harbouring homozygous deletions. Northern blotting experiments confirmed the effect of genomic deletions on the expression of these two miRNAs, with downregulation of MIRN15a and of MIRN16-1 in more than 50% of samples. Of important note, members of MIRN29 family, known to be downregulated in patients with poor prognosis CLL (Calin et al, 2005b), directly target TCL1A mRNA (Pekarsky et al, 2006), making this family of miRNAs attractive ‘drugs’ for the potential treatment of the disease in TCL1 transgenics. The indolent, aggressive and aggressive with chromosome 11 deletions subtypes of human CLL, characterized by different TCL1 expression levels can be also identified by distinct microRNA expression signatures found through microRNA expression profiling approaches (Pekarsky et al, 2006).
MicroRNAs and CLL predisposition: the missed link?
Chronic lymphocytic leukaemia predisposition has been studied by a multitude of scientific groups in the last two decades, but the genetic basis for the majority of familial cases as well as the hereditary contribution remains largely unknown. miRNAs represent ideal candidates for cancer predisposing genes as, despite being so small, they are able to target a large number of important protein coding genes. Small variations in the expression of a specific microRNA can affect tens to hundreds of target mRNAs with possible significant functional consequences. Therefore, inherited variations in miRNAs expression could represent the basis for the predisposition to various types of familial cancers with unknown pathogeny, such as familial CLL or familial prostate cancers (Calin & Croce, 2006). Among the examples supporting this hypothesis are mutations, such as the one found in the primary transcript of MIRN16-1, that can account for the reduced expression of the mature microRNA in a patient with familial CLL and familial breast cancer (Calin et al, 2005b). Linkage of (MOUSE) Mirn15A/Mirn16-1 mutations to the development of a CLL-like disease in a mouse model for spontaneous tumours suggests that the altered expression of the MIRN-15A/MIRN-16-1 is the molecular lesion in CLL (Raveche et al, 2007).
miRNAs may play a role as genes involved in the development and penetrance of tumours, as tumour susceptibility genes and are candidate molecular markers for assessing cancer risk. Moreover, the genome positions of mouse tumour susceptibility loci known to influence the development of solid tumours in inbred strains display a statistically significant association with the chromosomal locations of mouse miRNAs (Sevignani et al, 2007). Additionally, several miRNAs located at or near susceptibility loci in inbred strains with different tumour susceptibilities present distinct patterns of flanking DNA sequences.
A model of CLL initiation and progression in which miR-15a and miR-16-1 act as possible tumour suppressors can be proposed (Fig 3). The homozygous deletions or a combination of mutation and deletion of the MIRN15A/MIRN16-1 cluster at 13q14.3 may be responsible for the inactivation of these miRNAs. The final effect would be the overexpression of oncogenes (such as BCL2 or other to be defined as targets) or genes not yet related to cancer (such as the arginyl-tRNA synthetase gene RARS) or the downregulation of yet to be identified tumour suppressor genes.
Dr Calin was supported, in part, as a University of Texas System Regents Scholar and as a Fellow of the University of Texas MD Anderson Research Trust and by the CLL Global Research Foundation. We apologize to our colleagues whose work was not cited because of space limitations.