Final Report Summary - ONCOMIRS (MicroRNAs and Cancer: From Bench to Bedside)
Executive summary:
ONCOMIRS focused on the identification and characterisation of ncRNAs, including microRNAs (miRs) and long non-coding RNAs that play a key role in the aetiology and/or progression of human cancer.
The sequential processing of primary miR transcripts into the mature miRs depends on two main protein complexes, the nuclear microprocessor complex and the Dicer complex. One of the project objectives was the identification of novel key components of the miR-processing machine and examination of their role in carcinogenesis. Through protein complex purification and spectrometry (MS) analyses many factors involved in this process have been identified and reported. Importantly, truncating mutations in TARBP2, encoding for the Dicer-binding protein TRBP, was shown to occur in human cancer. These mutations were shown to affect microRNA processing, dicer function and promote tumorigenesis.
In addition, PRS14, EDC4 and FXR2 have been identified as AGO2-binding proteins and shown to modulate the transforming capacity of various human cancer cell lines. Consistent with the notion that an overall down-regulation of miR levels in cancer contributes significantly to the progression of the disease we showed that while a decrease in Dicer expression increase the transforming potential of cancer cell lines complete silencing is not tolerated by most cancer cells. These data were further supported by genetic experiments in mice. We indeed showed that Dicer functions as a haploinsufficient tumor suppressor gene in a pre-clinical mouse model of retinoblastoma but that it is in the same time required for tumorigenesis. Together these observations have opened exciting new avenues for anticancer therapy. Another important objective is the identification of novel putative cancer-causing ncRNAs.
On a technological point of view, in order to further facilitate and improve sensitivity and accuracy of genome-wide ncRNA profiling, we generated new microarray platforms based on the LNA technology. In addition, promiscuous LNA oligos have been designed in attempts to facilitate the detection or inhibition of entire miRNA families rather than single members of a given family as well as long non-coding RNAs. We also explored the possibility of validating single miRNA-mRNA interactions using LNA oligonucleotides that bind specifically to sequences encompassing and surrounding miRNA target sites within mRNAs. These so-called target site blockers have given very promising results and are now commercially available at Exiqon. Assessment of the biological functions of selected ONCOMIRS by direct experimental identification of target mRNAs is another objective of this project.
In order to define the role of selected ONCOMIRS, lncRNAs and components of the miR-processing machinery in the genesis and progression of cancer, classical gain and loss of function approaches in vivo, using the mouse as a model system were performed. We for instance showed that complete loss of miR17-92 completely prevents retinoblastoma formation in mice (Nittner et al., 2012). In contrast loss of miR17-92 did not affect normal retinogenesis; an observation that has important therapeutic implications. In contrast, however, although upregulated in mouse and human melanoma the miR17-92 cluster was shown to be dispensable for NRas-induced melanoma formation in mice. These data highlighted important differences between tumor types and their dependency to specific ONCOMIRS. Assessment of the therapeutic potential of miR (over)expression and/or inhibition in preclinical mouse models of human cancers was another goal. A new generation of miR mimics and antisense therapeutics have developed. We have tested and optimised the biostability, cytotoxicity and electropulsation-based delivery in vivo of such molecules and demonstrated that electrotransfer of miR inhibitors offers a new avenue for anti-cancer (retinoblastoma and glioblastoma) therapy.
Project context and objectives:
The objectives of the ONCOMIRS' project are presented as they were introduced at the beginning of each period. They remained consistent throughout the whole duration of the project and by and large all objectives have been reached/met.
Period 1
Recent experiments have indicated that alterations in miR levels in cancer could be partly the result of aberrant post-transcriptional processing of precursor miR transcripts. In the last few years, partner 2 has isolated two complexes required for the sequential processing of primary miR transcript into the mature miR in human cells. While the initial processing of primary miR into precursor miR requires the complex composed of RNase III, Drosha, and the double stranded RNA-binding domain protein DGCR8, the formation of the mature miR requires the action of Dicer/TRBP complex. We proposed to actively persue the identification of additional components involved in the regulation of silencing by miRs, through protein complexes purification. One of our main objectives was to assess whether or not components of the Drosha and Dicer complexes are either mutated and/or aberrantly expressed in human cancers.
We proposed to implement the LNA technology present in the group (partner 6) on high-density flexible array platforms to facilitate ncRNA expression analyses. Once this technology will be available, a more long-term goal will be the identification of novel small non-coding RNAs (ncRNAs, including miRs) whose expression is dysregulated in human cancers. In the meantime, we proposed to establish ncRNA gene expression profiles using currently available technologies with the aim to identify genes that are regulated by and therefore act as downstream mediators of the p53 tumor suppressor pathway.
We also proposed to set up / use various gain and loss of function genetic screens with the aim to identify new putative ONCOMIRS based on their abilities to modulate key cellular pathways.
Only a limited number of oncomiR-target pairs have been identified to date. Partner 4 has recently developed a new approach for miR targets identification. We proposed to use this approach in combination with genome-wide gene expression profiling in order to identify the primary (and secondary) targets of a number of selected ONCOMIRS.
One main objective of this application was to study the physiological relevance of a few selected ncRNAs and demonstrate/dissect their active role in cancer formation and progression in vivo. To this end, we proposed to generate classical gene targeting vectors that disrupt expression of the miR(s) of interest in mice. For conditional gain of function studies, we will also knock-in miR gene(s) into the ROSA26 locus by gene targeting. The newly generated alleles will be combined with tumour-causing mutations to study the consequence of aberrant expression of a particular miR in relevant tumor models of human cancers.
RNA interference represents a potential strategy for in vivo target validation, therapeutic product development and clinical new technologies. It is expected to be particularly useful to silence cancer-causing genes that encode targets that are not amenable to conventional therapeutics. Moreover, there is evidence that miR-based molecules enter the RNAi pathway through a more natural route and yield more effective silencing with reduced toxicity and off-target effects. miR delivery in vivo in tumors is achievable by using new physical methods (electropulsation) developed by partner 5 and prone for clinical applications. One of our main objectives is to design miR-based molecules (miR mimics) that either allow potent activation of a 'dormant' tumour suppressor pathway (for instance p53) or alter expression of specific oncogenes. The molecules will be designed to allow efficient delivery, high biostability and low toxicity. Similarly miR antagonists will be designed to target specific ONCOMIRS. The LNA technology developed by partner 6 will be particularly useful in this context. We first planned to test the efficacy and toxicity of these compounds in culture cells.
Period 2
Recent experiments have indicated that alterations in miR levels in cancer could be partly the result of aberrant post-transcriptional processing of precursor miR transcripts. We proposed to actively pursue the identification of additional components involved in the regulation of silencing by miRs, through protein complexes purification. One of our main objectives is to assess whether or not components of the Drosha and Dicer complexes are either mutated and/or aberrantly expressed in human cancers. To this end we proposed to generate monoclonal antibodies against a number of proteins, involved in miRNA function such as Ago proteins, Drosha, DGCR8 and exportin 5.
Human tumours are characterised by widespread reduction in microRNA (miRNA) expression, although it is unclear how such changes come about and whether they have an etiological role in the disease. A defect in miRNA-processing is one possible mechanism for the global down-regulation. To explore this possibility in more detail in vivo we proposed to manipulate Dicer1 gene dosage in a mouse model of retinoblastoma and identify miRs that can either function as tumour suppressors or oncogenes in such a model.
We proposed to implement the LNA technology present in the group (partner 6) on high-density flexible array platforms to facilitate ncRNA expression analyses. A more long-term goal was the identification of novel small non-coding RNAs (ncRNAs, including miRs) whose expression is dysregulated in human cancers. In the meantime, we proposed to establish ncRNA gene expression profiles using currently available technologies with the aim to identify genes that are regulated by and therefore act as downstream mediators of the p53 tumour suppressor pathway.
We also proposed to set up/use various gain and loss of function genetic screens with the aim to identify new putative ONCOMIRS based on their abilities to modulate key cellular pathways.
Only a limited number of oncomiR-target pairs have been identified to date. Partner 4 has recently developed a new approach for miR targets identification. We proposed to use this approach in combination with genome-wide gene expression profiling in order to identify the primary (and secondary) targets of a number of selected ONCOMIRS.
One main objective of this application is to study the physiological relevance of a few selected ncRNAs and demonstrate / dissect their active role in cancer formation and progression in vivo. To this end, we proposed to generate classical gene targeting vectors that disrupt expression of the miR(s) of interest in mice. For conditional gain of function studies, we will also knock-in miR gene(s) into the ROSA26 locus by gene targeting. The newly generated alleles will be combined with tumour-causing mutations to study the consequence of aberrant expression of a particular miR in relevant tumour models of human cancers.
RNA interference represents a potential strategy for in vivo target validation, therapeutic product development and clinical new technologies. It is expected to be particularly useful to silence cancer-causing genes that encode targets that are not amenable to conventional therapeutics. Moreover, there is evidence that miR-based molecules enter the RNAi pathway through a more natural route and yield more effective silencing with reduced toxicity and off-target effects. miR delivery in vivo in tumours is achievable by using new physical methods (electropulsation) developed by partner 5 and prone for clinical applications. One of our main objectives is to design miR-based molecules (miR mimics) that either allow potent activation of a 'dormant' tumour suppressor pathway (for instance p53) or alter expression of specific oncogenes. The molecules will be designed to allow efficient delivery, high biostability and low toxicity. Similarly miR antagonists will be designed to target specific ONCOMIRS. The LNA technology developed by partner 6 will be particularly useful in this context. We first planned to test the efficacy and toxicity of these compounds in culture cells.
Period 3
Recent experiments have indicated that alterations in miR levels in cancer could be partly the result of aberrant post-transcriptional processing of precursor miR transcripts. We proposed to actively pursue the identification of additional components involved in the regulation of silencing by miRs, through protein complexes purification. One of our main objectives is to assess whether or not components of the Drosha and Dicer complexes are either mutated and/or aberrantly expressed in human cancers.
Human tumours are characterised by widespread reduction in microRNA (miRNA) expression, although it is unclear how such changes come about and whether they have an etiological role in the disease. A defect in miRNA-processing is one possible mechanism for the global down-regulation. To explore this possibility in more detail in vivo we proposed to manipulate Dicer1 gene dosage in a mouse model of retinoblastoma and identify miRs that can either function as tumour suppressors or oncogenes in such a model.
We proposed to implement the LNA technology present in the group (partner 6) on high-density flexible array platforms to facilitate ncRNA expression analyses. Another goal is the identification of novel non-coding RNAs (ncRNAs, including miRs) whose expression is dysregulated in human cancers.
We also proposed to set up/use various gain and loss of function genetic screens with the aim to identify new putative ONCOMIRS based on their abilities to modulate key cellular pathways.
Only a limited number of oncomiR-target pairs have been identified to date. Partner 4 and 7 have recently developed new approaches for miR targets identification. We proposed to use this approach in combination with genome-wide gene expression profiling in order to identify the primary (and secondary) targets of a number of selected ONCOMIRS.
One main objective of this application is to study the physiological relevance of a few selected ncRNAs and demonstrate / dissect their active role in cancer formation and progression in vivo. To this end, we proposed to generate miRKO and transgenic mouse lines and combined these alleles with tumour-causing mutations to study the consequence of aberrant expression of selected miRs in relevant tumor models of human cancers.
RNA interference represents a potential strategy for in vivo target validation, therapeutic product development and clinical new technologies. It is expected to be particularly useful to silence cancer-causing genes that encode targets that are not amenable to conventional therapeutics. Moreover, there is evidence that miR-based molecules enter the RNAi pathway through a more natural route and yield more effective silencing with reduced toxicity and off-target effects. miR delivery in vivo in tumours is achievable by using new physical methods (electropulsation) developed by partner 5 and prone for clinical applications. One of our main objectives is to design miR-based molecules (miR mimics) that either allow potent activation of a 'dormant' tumour suppressor pathway (for instance p53) or alter expression of specific oncogenes. The molecules will be designed to allow efficient delivery, high biostability and low toxicity. Similarly miR antagonists will be designed to target specific ONCOMIRS. The LNA technology developed by partner 6 will be particularly useful in this context.
Period 4 (last)
Recent experiments have indicated that alterations in miR levels in cancer could be partly the result of aberrant post-transcriptional processing of precursor miR transcripts. We proposed to actively search for additional components involved in the regulation of silencing by miRs, through protein complexes purification.
Human tumors are characterised by widespread reduction in microRNA (miRNA) expression, although it is unclear how such changes come about and whether they have an etiological role in the disease. A defect in miRNA-processing is one possible mechanism for the global down-regulation. To explore this possibility in more detail in vivo we proposed to manipulate Dicer1 gene dosage in a mouse model of retinoblastoma and identify miRs that can either function as tumor suppressors or oncogenes in such a model.
We proposed to implement the LNA technology present in the group (partner 6) as a tool to knock-down expression of selected non-coding RNAs. Another goal was the identification of novel non-coding RNAs (ncRNAs, including miRs) whose expression is dysregulated in human cancers.
We also proposed to set up/use various gain and loss of function genetic screens with the aim to identify new putative ONCOMIRS based on their abilities to modulate key cellular pathways.
Only a limited number of oncomiR-target pairs have been identified to date. Partner 4 and 7 have recently developed new approaches for miR targets identification. We proposed to use this approach in combination with genome-wide gene expression profiling in order to identify the primary (and secondary) targets of a number of selected ONCOMIRS.
One main objective of this application was to study the physiological relevance of a few selected ncRNAs and demonstrate / dissect their active role in cancer formation and progression in vivo. To this end, we generated KO and transgenic mouse lines for miRs and lncRNAs and combined these alleles with tumour-causing mutations to study the consequence of aberrant expression of selected miRs in relevant tumour models of human cancers.
RNA interference represents a potential strategy for in vivo target validation, therapeutic product development and clinical new technologies. It is expected to be particularly useful to silence cancer-causing genes that encode targets that are not amenable to conventional therapeutics. Moreover, there is evidence that miR-based molecules enter the RNAi pathway through a more natural route and yield more effective silencing with reduced toxicity and off-target effects. miR delivery in vivo in tumours is achievable by using new physical methods (electropulsation) developed by partner 5 and prone for clinical applications. One of our main objectives was to design miR-based molecules (miR mimics) that either allow potent activation of a 'dormant' tumour suppressor pathway (for instance p53) or alter expression of specific oncogenes. Such molecules have be designed to allow efficient delivery, high biostability and low toxicity. Similarly miR antagonists were designed to target specific ONCOMIRS. The LNA technology developed by partner 6 has been particularly useful in this context.
Project results:
WP2: miR Biogenesis and Cancer
MiRNAs are transcribed as primary miRNA transcripts (pri-miRNAs), which serve as substrates for the nuclear microprocessor complex. This multi-protein complex contains the RNase III enzyme Drosha, its co-factor DGCR8 as well as a number of so far uncharacterised components. Drosha processing yields miRNA precursors (pre-miRNAs), which are subsequently transported to the cytoplasm by the export receptor exportin 5. In the cytoplasm, the RNase III enzyme Dicer further processes pre-miRNAs yielding mature miRNAs. MiRNAs are incorporated into miRNA-protein complexes (miRNPs), where they interact with a member of the Argonaute (Ago) protein family. In human somatic cells, Ago1, Ago2, Ago3 and Ago4 are expressed. It is likely that all four Ago proteins are incorporated into similar protein complexes. In order to function in gene silencing, miRNPs interact with a member of the GW protein family. In human, the GW proteins TNRC6A, B and C have been implicated in miRNA function. One of our objectives was the identification of novel components of the miR-processing machinery, through protein complexes purification. Partner 2 identified the huntingtin (Htt) protein as a component of the Argonaute complex. Colocalisation studies demonstrated Htt and Ago2 to be present in P bodies, and depletion of Htt showed compromised RNA-mediated gene silencing. Huntington's disease is a dominant autosomal neurodegenerative disorder caused by an extension of polyglutamines in the Htt protein. The data therefore suggest that this disease may be attributed in part to mutant Htt's role in post-transcriptional processes (Savas et al., 2008). A possible role for components of effector complexes such as Ago1-4-containing miRNPs or TNRC6A-C-containing complexes in cancer will also be studied. In close collaboration with Elisabeth Kremmer (Helmholtz center, Munich), beneficiary 7 generated monoclonal antibodies against a number of proteins, involved in miRNA function. Antibodies against individual human Ago proteins have thus been generated. One of these antibodies allows detection of endogenous Ago2 expression on tissue sections. In addition, antibodies against Drosha, DGCR8 and exportin 5 have been generated. Beneficiary 7 isolated a number of interesting Drosha protein partners among them EWS. Partner 7 also established a list of Dicer-dependent and -independent AGO2-binding proteins (Frohn et al., 2012). Beneficiary 7 has confirmed the presence of the helicase p68 in the DROSHA complex; an observation which is consistent with previously published studies. As expected, they also found the Dicer cofactor TARBP2 (TRBP) and the RNA helicase A/DHX9, which has been implicated in siRNA-loading. Proteins such as YBX1, Gemin4, Gemin5, HNRNPC, HNRNPUL1, the ARE binding protein ELAVL1/HuR, the poly-A binding proteins PABPC1 and 4, the mitochondrial protein Matrin3 and the Fragile X mental retardation protein paralog FXR2 have been found in Ago complex purifications before. In addition, they identified the PABPC1-binding protein LARP1, the mRNA binding proteins FAM120A/Ossa and CSDA, which have not been implicated in Ago2 function before. In contrast to the GW protein family members TNRC6A and B, TNRC6C interacts with Ago2 only in the presence of Dicer, suggesting that it requires Ago2 to be loaded onto miRNA target mRNAs. Moreover they found that a high number of proteins associate with Ago2 in a RNA-dependent manner even in the absence of miRNAs. In this group, they found many RNA binding proteins including IGF2BP13, DHX36, DHX30, HNRNPL, and the ribosomal protein RPS14. Strikingly, genetic data in C. elegans demonstrated that PRS14 modulates mature let-7 tumour suppressor.
A set of proteins associates with Ago2 in an mRNA- independent manner. This includes the TNRC6 proteins and Dicer. Other miRNA-independent protein-protein interactors of Ago2 are the HSP90 alpha and beta proteins with their cochaperones PTGES and FKBP5. Among the interactors preferentially binding in the absence of Dicer and miRNAs only Ago3, CLTC and ZNF521 were identified in the datasets and they show a direct binding behaviour. Taken together, this mass spectrometry approach revealed that Ago2 associates with larger RNA species even in the absence of small RNAs. Furthermore, several mRNA-binding proteins are specific to miRNA-free and miRNA-containing Ago2-mRNA complexes. To functionally validate the interaction data, Partner 7 used luciferase-based miRNA reporters. The 3'-UTR of Hmga2, a well-characterised let-7a target, was fused to firefly luciferase and transfected into HeLa, HCT116 and MCF-7 cells in which ZNF521, CSDA and EDC4 were depleted by RNAi. In addition, we employed a reporter containing the Hmga2 3'-UTR with mutated let-7a target sites and normalised the data against each other. As expected, Ago2 knock down led to increased luciferase activity. Knock down of EDC4, which has been implicated in miRNA function in Drosophila, resulted in specific luciferase up-regulation as well, suggesting that EDC4 is indeed involved in silencing of the Hmga2 reporter construct. Similar results were obtained for the mRNP component CSDA. ZNF521, however, is not involved in miRNA-guided gene silencing. Since ZNF521 is a putative transcription factor, it is possible that it cooperates with Ago2 in nuclear Ago functions. The previously recognised role of let-7 family members as tumour suppressors prompted us to assess the role of PRS14 and EDC4 in cellular transformation by performing colony and soft agar assays. Moreover we have also assessed the role of FXR2 in cellular transformation, as other Fragile X mental retardation proteins have recently been implicated in cancer development. These experiments have confirmed that all these factors are able to modulate cellular transformation in these assays. Consistently, depletion of these factors by RNAi in HCT116, MCF-7 and BJ-cells had profound phenotypic consequences on the ability of these cells to grow in vitro and their transforming potential. Further experiments aimed at dissecting the mechanisms underlying these phenotypic consequences are ongoing in the laboratory of partner 1 and 7.
Overall, down-regulation of miR levels in cancer contributes significantly to the progression of the disease. Since this overall-down-regulation is likely to be, at least partly, a consequence of aberrant processing of their precursor transcripts, we proposed to examine the expression levels of Drosha/DGCR8 and Dicer/TRBP components, in cancer cell lines and human primary tumours and assess the contribution of aberrant expression and/or altered function in cellular transformation. Beneficiary 2 -in collaboration with Pr Manel Esteller (CNIO, Madrid, Spain)- has screened various cancer cell lines with microsatellite instability for the presence of mutations in all exonic mononucleotide repeats present in the coding sequences of eight established members of the miRNA processing machinery: Dicer and Drosha, DGCR8, TRBP, PACT and AGO1, AGO2 and AGO4. Wild-type sequences were found for all genes with the exception of TARBP2 encoding for TRBP. Two heterozygous frame-shift mutations were identified in two different cell lines. These mutations were found to cause a significant decrease in TRBP expression. Since TRBP stabilises DICER, these mutations also affected Dicer expression levels and consequently caused a drastic reduction in the efficiency of endogenous processing of miRs. Importantly, ectopic expression of wild-type TRBP, but not of the naturally occurring mutants, induced tumour suppressor-like features in colorectal cancer cell lines. Notably, Dicer overexpression also showed tumour suppressor properties as determined using classical tissue culture assays (MTT and colony assays) and mouse xenograph models (nude mice). Finally, TARBP2 mutations were detected in about 25% of a panel of 282 human primary tumours (Melo et al., 2009). The above findings provide further evidence of a role of loss of function events in the regulation of miRNA processing machinery during tumourigenesis. However, the molecular mechanisms that promote tumourigenesis as a consequence of global repression of miRNA maturation remain elusive. Data obtained by beneficiary 1 suggest that the p53 tumor suppressor pathway is compromised in cancer cells harbouring hypomorphic Dicer expression.
Given the results of the mutational analyses of the various components of the miR-processing machinery in cancer cell lines and primary tumours, it was decided to slightly deviate from the initial plans. Instead of constructing artificial TRBP mutants lacking the dsRNA-binding domains, it was preferred to assess the oncogenic properties of the naturally occurring TRBP mutants. To add relevance to our structure function analysis, the experiments were conducted in colorectal cancer cell lines, instead of NIH-3T3, since these mutations were found to occur in this particular tumour context. Moreover, since no mutations were found in Dicer and AGO2 in human tumours, construction of the initially proposed Dicer and AGO2 mutants has been cancelled. Instead, we have already tested the role of TRBP and the influence of the naturally occurring mutations to cell transformation. In this context, the role of Dicer in cell transformation was also assessed. Moreover, we have demonstrated that the mutations in TRBP affect Dicer levels and consequently the integrity of the RNAi machinery. These data, recently reported in Melo et al., 2009, indicate that:
(i) TRBP and DICER exhibit tumor suppressive activities; and
(ii) mutations that ultimately affect miR-processing occur in human cancer.
Human tumours are characterised by widespread reduction in microRNA (miRNA) expression, although it is unclear how such changes come about and whether they have an etiological role in the disease. Importantly, miRNA-knockdown has been shown to enhance the tumourigenic potential of human lung adenocarcinoma cells. A defect in miRNA-processing is one possible mechanism for the global down-regulation. To explore this possibility in more detail in vivo, we have manipulated Dicer1 gene dosage in a mouse model of retinoblastoma. We show that while monoallelic loss of Dicer1 does not affect normal retinal development it dramatically accelerates tumour formation on a retinoblastoma-sensitised background. Importantly, these tumours retain one wild-type Dicer1 allele and exhibit only partial decrease in miRNA-processing. Accordingly, in silico analysis of human cancer genome data reveals frequent hemizygous, but not homozygous, deletions of DICER1. Strikingly, complete loss of Dicer1 function in mice did not accelerate retinoblastoma formation. miRNA profiling of these tumours identified members of the let-7 and miR-34 families as candidate tumour suppressors in retinoblastoma. We conclude that Dicer1 functions as a haploinsufficient tumour suppressor (Lambertz et al., 2010).
The above finding provides genetic evidence for a causative link between down-regulation of miRNA-processing and cancer progression.
Interestingly, although reduced levels of DICER1 in tumours have been reported, no loss-of-function mutations in DICER1 have been reported to date. Moreover, homozygous loss of Dicer1 appears to be strongly selected against in a K-Ras-induced mouse model of lung cancer and Myc-induced mouse model of B-cell lymphoma. These data raise the possibility that Dicer1 is required for tumour formation. Beneficiary 1's group has recently demonstrated that targeted homozygous loss of Dicer1 completely prevents the formation of retinoblastoma in mice in which the Rb and p53 tumour suppressor pathways are inactivated. Strikingly, it was found that Dicer1 deficiency selectively kills Rb-deficient retinal cells in which p53 is inactivated while sparing cells that retain functional p53. miRNA profiling of mouse and human primary retinoblastomas showed dramatic overexpression of the pro-oncogenic miR17-92 cluster in all samples analysed. High-resolution array-CGH indicates that in about 20 % of human retinoblastoma patients overexpression of miR-17-92 results from copy number alterations. Crucially, functional inactivation of the miRNAs encoded by the miR-17-92 cluster is sufficient to decrease the viability of human retinoblastoma cells and to prevent retinoblastoma formation in mice. The data therefore provide genetic evidence of a synthetic lethal interaction between Dicer and p53 and designate members of the miR-17-92 cluster as a highly selective therapeutic target for the treatment of retinoblastoma (Nittner et al., 2012).
WP3: Identification of novel ONCOMIRS (and long non-coding RNAs)
The data described above imply that a set of miRNAs might act to prevent full-blown retinoblastoma formation on the Chx10Cre;Rblox/lox;p107-/- background. To identify such candidate tumor suppressor miRNAs beneficiary 1 determined the expression profile of the entire miRNome in P20 retinae from wild-type (Cre-negative), Chx10Cre; Rblox/lox; p107-/-; Dic+/+ and Chx10Cre;Rblox/lox;p107-/-;Diclox/+ using LNA-based microarray (in collaboration with Patrner 6) and RT-qPCR approaches. Both types of analyses identified a common set of 11 miRNAs that are consistently up-regulated between wild-type and Chx10Cre; Rblox/lox; p107-/-; Dic+/+. Most interestingly, among them are 2 members of the let-7 family (let-7c and let7-i), the up-regulation of which and of another let-7 member, let-7b, was confirmed by independent Q-RT-PCR analysis. Given that this up-regulation was significantly attenuated in the Dic heterozygous retinae and the recognised role of let-7 family members in tumor suppression, this observation raises the possibility that let-7 have a causal role in retinoblastoma formation as critical regulators of the switch from retinomas to retinoblastoma. Our list of 11 differentially expressed miRNAs also included miR-34c. There was also a clear upregulation of miR-34b-3p in both microarray and RT-q-PCR analyses and a moderate, but reproducible, up-regulation of miR-34a was evident from the micro-array data. Interestingly, the miR-34 family members have been identified as p53 targets and key mediators of its tumor suppressor function.
The data from WP2 also implied that a set of miRNAs is required for retinoblastoma formation despite the combined mutation in Rb, p107 and p53. In search of such pro-oncogenic miRNAs we profiled the miRNome by stem-loop RT-qPCR in P21 retinae from wild-type (Cre-negative), Chx10Cre; Rblox/lox; p107-/- and from Laser-Captured-Microdissected tumor materials from 4 Chx10Cre;Rblox/lox;p107-/-;p53lox/lox (TKO) mice. This analysis identified a set of 102 miRNAs that are significantly up-regulated in the TKO tumours. To find correlates of the mouse data in human tumours, we also profiled miRNA expression in 30 different human primary retinoblastomas. 68 miRNAs were significantly up-regulated in retinoblastoma compared to normal human retinae. Cross-species comparison identified 25 miRNAs that were up-regulated in both mouse and human tumours. Strikingly, 12 of them are members of the known oncogenic miR-17-92 and 106b-25/miR-106a-92 paralogue clusters. Consistently, hierarchical clustering of all RB cases and normal retinae based on miRNA expression singled out all members of these clusters as being dramatically up-regulated in all mouse and human tumours analyzed. To explore the potential causes of miRNA deregulation in human retinoblastoma beneficiary 1 looked for genomic aberrations using a 44K oligonucleotide array, which was specifically designed to include regions harbouring miRNA genes. In addition to identifying previously reported retinoblastoma-associated genomic aberrations (1q gain and 6p22 gain were frequently seen in our cohort) focal amplification of the miR-17-92 locus, which lies on chromosome 13, was found in one patient. Another patient had a whole chromosome 13 gain and 3 patients had copy number gains including the miR-17-92 cluster but, importantly, not the closely linked Rb-1 locus. miR-17-92 copy number gains were found in 17 % of the patients (5 out of 29 cases analysed). Moreover, while the Rb-1 locus was deleted in 21 % of cases (6/29) this deletion never included the closely linked miR-17-92 locus. This analysis therefore indicates that up-regulation of the miR-17-92 cluster is, at least in a proportion of retinoblastoma cases, a direct consequence of increased gene copy number. Since transcription of miR-17-92 is positively regulated by the E2Fs and negatively regulated by p53, deregulation of their transcriptional activities may also account for miR-17-92 overexpression in retinoblastoma. Regardless of the underlying mechanism, our data demonstrate that the miR17-92 cluster is overexpressed in 100 % of retinoblastomas analysed. Together these data raise the possibility that up-regulation of expression of the miR17-92 cluster is a critical event for the development retinoblastoma; this was later shown to be case (see WP6).
We also proposed to establish miR gene expression profiles using currently available technologies with the aim to identify genes that are regulated by and therefore act as downstream mediators of the p53 tumour suppressor pathway. In a first set of experiments, we have activated the p53 pathway chemically using the Mdm2-inhibitor Nutlin3a. BJ primary human diploid fibroblasts undergo p53-mediated irreversible cell cycle arrest or senescence upon such treatment. Using a Q-RT-PCR-based profiling method, we have identified miRs that are induced and repressed under these conditions. As control, quiescent -serum starved- cells were also profiled. MiR-34a, b and c were all specifically induced in senescent cells. In addition, miR-24, miR-23b and miR-27b, all encoded by the same locus, were also found induced in Nutlin-treated cells. In addition, miR-195, miR-7 and miR-452# were also all found dramatically up-regulated whereas miR-302b and miR-586 were both strongly down-regulated in senescence cells but not in quiescent cells.
Beneficiary 3 identified a panel of miRs whose expression is repressed by p53 during replicative senescence of human diploid fibroblasts; this study has now been published (Brosh et al, 2009). In subsequent work, the regulation of miR-372 and miR-373 during oncogene-induced senescence has been addressed. Specifically, it was found that these two miRNAs are downregulated in response to overexpression of oncogenic H-Ras. This correlates with the upregulation of the Lats2 gene by oncogenic Ras. Of note, Lats2 is required for p53 activation in human fibroblasts in response to oncogenic Ras, and is thus a critical mediator of oncogene-induced senescence in this system. These observations are consistent with a potential role for the miRNAs in inhibition of oncogene-induced senescence, and provide a mechanistic explanation for earlier observations of Agami and co-workers (Aylon et al, 2010). Beneficiary 3 also investigated the notion that some p53-regulated miRNAs may be epigenetically silenced in tumours that retain wild-type p53, rendering these tumours less susceptible to killing by genotoxic chemotherapy. To that end, conditions for the synergistic killing of A549 lung cancer cells by a combination of cisplatin and agents that 'erase' epigenetic silencing marks (TSA+ 5-Aza-cytidine) have been optimised. miRNA profiling from such cells with and without various treatment combinations will be performed by beneficiary 6. The data will be subjected to validation by qRT-PCR in the subsequent months. To identify miRNAs that are regulated by p53 in vivo, possibly in a tissue-specific manner, beneficiary 2 carried out miRNA profiling on different tissues of wild type and p53 knockout mice. Following validation of the most promising hits by qRT-PCR, it was decided to focus on miR-223 and miR-449. Both of these miRNAs are positively regulated by p53 in some cultured cell types but not in others. miR-223 was previously reported to play a role in the regulation of differentiation in the myeloid lineage.
Interestingly, using a myeloid leukemia cell line carrying an inducible (temperature sensitive) p53, a modest but reproducible increase in miR-223 levels upon p53 induction was observed. The effect of p53 activation on the differentiation of those cells (which can be induced by either IL6 or dexamethazone) and the contribution of miR-223 to this effect will be further investigated. Beneficiary 2 has also performed miRNA profiling using an experimental array developed by beneficiary 6, which combines probe sets for mature miRNA and for pre-miRNA. The input RNA was derived from H1299 lung cancer-derived cells containing a temperature sensitive p53 (the same experimental model that was employed earlier to discover that miR-34a was a p53 target). Surprisingly, the results suggested that there may be profound differences between the impact of p53 activation on some mature miRNAs relative to their precursors. The data still need to be validated by more quantitative methods; if validated, this may suggest an effect of p53 on the processing of specific pre-miRNAs by Dicer, and will merit further investigation. An additional observation provided by this miRNA profiling experiment was that p53 seemed to up-regulate the levels of miR-372 and miR-373*; subsequent preliminary data suggest that these miRNAs may also be positively regulated by p53 in A549 cells, which harbor endogenous wtp53.
Given that the expression of miR-372 was found by us earlier to be negatively regulated by Ras, it will be interesting to investigate the cross-talk between p53 and Ras in regulation of the expression of the miR-372/373 cluster. In parallel, we initiated a new line of investigation, not included in the original workplan. This is based on the recent realisation that p53 plays a role in regulation of autophagy, a process that is turning out to be of great importance in cancer biology. In a collaborative study with Prof. Z. Elazar at the Weizmann Institute, we searched for miRNAs whose expression is induced in HeLa cells under conditions of nutrient starvation that trigger autophagy. One of the top hits was miR-492, primate-specific miRNA that highly expressed in retinoblastoma. Interestingly, miR-492 was also one of the top hits in another screen carried out by beneficiary 3, which looked for miRNAs whose expression is altered in a p53-dependent manner in HCT116 colorectal cancer cells treated with the mitotic spindle poison nocodazole. Follow-up experiments validated that miR-492 is induced by nutrient starvation (24-48h) in p53-proficient HCT116 cells but not in their derivative p53 knockout cells (HCT116 p53-/-).
We further found that HCT116 p53-/- cells have an increased propensity to become polyploid upon nutrient starvation, a feature not seen in wt HCT116 cells and thus indicative of a role of p53 in preventing starvation-induced polyploidisation. On note, overexpression of miR-492 in the p53-/- cells could reduce significantly this starvation-induced polyploidy.
Beneficiary 7 searched for miRNAs whose expression is dependent on the p53 family members p63 and p73. A list of validated p63- and p73-regulated miRNAs has been available to the members of the consortium.
The ultra conserved regions (UCRs) are a subset of conserved regions that are absolutely conserved (100 %) between orthologous regions of the human, rat and mouse genomes and are longer than 200bp (Bejerano et al., 2004). Because of the very high degree of conservation, the UCRs are likely to have fundamental functional importance for the ontogeny and normal physiological functioning of cells and tissues, including gene regulation and chromosome biology of the organisms in which they are found. The majority of UCRs (about 90 %) are transcribed (T-UCRs) in normal human tissues. While a minority is expressed ubiquitously, most are tissue specific (Calin et al., 2007). As with miRs (Liu et al., 2004), hierarchical clustering of T-UCR expression in various hematopoietic and non-hematopoietic tissues from different individuals suggests that T-UCRs can be used as tissue specific markers. The same types of tissues from different individuals clustered as closest neighbours. Transcription of UCRs may be regulated by specific transcription factors and/or initiated from polyadenine rich genomic regions, as was recently proposed for several long ncRNAs in the mouse (Furuno et al., 2006). Importantly, levels of UCRs may be regulated post-transcriptionally by miRs. Calin et al. (2007) has indeed provided evidence that specific miRs interact with complementary T-UCRs and that this interaction results in down-regulation of T-UCRs expression. T-UCRs are frequently located at fragile sites and other genomic regions that have been implicated in cancers (Calin et al., 2007). Genome-wide profiling revealed that UCRs, just like miRs, have distinct signatures in human leukemias and carcinomas and therefore can be used to differentiate between human cancers (Calin et al., 2007). Moreover, about 10 % of T-UCRs are differentially expressed in at least one type of tumor cells when compared to normal cells of the same origin. Among the most significant differentially expressed T-UCRs in leukemias and carcinomas are uc.29 uc.73 (high in colorectal cancers), uc111, uc135 (low in CLL), ... Interestingly, a significant negative correlation between 87 miRs (out of 285 analyzed) and T-UCRs expression levels was found at a genome-wide level in 50 patients with leukemias (CLL) (Calin et al., 2007). This observation is consistent with the possibility that T-UCRs are targeted by miRs and these interactions may have biological and prognostic significance for cancer patients. Strikingly, preliminary functional studies suggest that uc.73 (P), which is significantly up-regulated in colon cancers, behaves like an oncogene in tissue culture by promoting both proliferative and anti-apoptotic activities (Calin et al., 2007). We have therefore decided - even if not planned originally - to profile the T-UCRnome (481 different UCRs) using a Q-RT-PCR-based approach in senescent and quiescent BJ cells. A list of UCRs specifically induced and repressed in senescent cells is now available. Of note, among the most differentially regukated genes are uc135 (up 30-fold) and uc73 (down more than 100-fold). Equally interesting is the observation that uc.177 which is drastically down-regulated in senescent cells is putatively regulated by all miR-34 family members. Further experiments are currently ongoing in the laboratory of partner 1 to test this hypothesis and assess the relevance of both miRs and UCRs that are specifically induced in Nutlin-treated cells for the senescence phenotype.
Partner 3 has identified a panel of miRs whose expression is altered during replicative senescence of human diploid fibroblasts. A substantial fraction of those miRs were found to belong to 3 paralogous polycistronic clusters, including the miR17-92 cluster, the miR-106b/93/25 cluster and the miR-17-5p cluster; levels of many miRs encoded within these clusters were shown to decrease markedly in senescent cells. Furthermore, this downregulation was p53-dependent, whereas the expression of these miRs in non-senescent cells was largely driven by E2F. Further analysis revealed a feed-forward loop that drives E2F-dependent transcription of those miRs. Functionally, overexpression of these miRs could promote excessive proliferation and delay senescence, consistent with a proposed oncogenic role of at least some of them. Indeed, expression of many of those miRs was found to be upregulated in breast cancer tumors that carry p53 mutations. These findings have recently been published (Brosh et al., 2008).
Partner 4 has identified a set of miRNAs regulated by oncogenic B-RAF in TIG3 primary human fibroblasts. Of the about 10 miRNAs significantly altered following 4 days of B-RAF induction miR-34a and miR-146a were the most significantly upregulated. We focused on miR-34a, as the absolute level of this miRNAs was much higher. Using a combination of siRNA experiments, chromatin-immunoprecipitations and heterologous promoter assays we have identified ELK1 as the transcription factor responsible for miR-34a upregulation downstream B-RAF. In addition, we have established that p53 plays little or no role in miR-34a induction during oncogene-induced senescence.
Partner 2 has recently compared induction of miRs in different mouse organs following exposure of wild type and p53 knock-out mice to whole body ionising radiation (IR). We observed substantial differences between different organs: while some miRs (e.g. miR-34a) were more widely induced following p53 activation by IR, others were induced only in one or two organs (e.g. miR-223 and miR-449a), attesting to the tissue-specificity of p53-dependent miR induction. The results of this analysis, obtained by miR array hybridisation, are presently being subjected to validation by qPCR-RT-PCR analysis.
We explored the regulation of miR199a by p53. Initial results suggested that this particular microRNA is selectively induced by several p53 mutants. Further analysis revealed a more complex picture: depending on cell type, miR-199a could be induced by either mutant p53 and/or WT p53. However, we could not find any experimental model in which the extent of induction was more than 2-3 fold, rendering the subsequent functional analysis insufficiently definitive. This line of research was therefore discontinued. In parallel, we sought to identify more comprehensively the spectrum of miRs regulated by mutant p53. To that end, we performed miR expression profiling on DU-145 cells (human prostate cancer, harbouring 2 mutant p53 alleles) before and after knockdown of their endogenous mutant p53 proteins. The results are still being validated, but overall the differences were rather minimal and disappointing, suggesting that this cell line is probably not an optimal choice for discovery of mutant p53-regulated miRs. We are presently investigating the possible regulation of miR-214 by WT and mutant p53. Preliminary results suggest that WT p53 can repress the expression of this miR, whose excessive expression has been implicated in ovarian cancer. Further analysis will therefore focus on cell lines derived from this type of cancer.
The setting up of several automated forward genetic screens as a mean to identify cancer-causing miRs is in progress. To this end, Partner 6 (Exiqon) has optimised Tm and LNA spiking pattern of LNA oligos to obtain miRNA inhibitors with increased potency and stability relative to previous versions. LNA based miRNA inhibitors have been designed for all human miRNAs listed in miRBASE 11.0 and the inhibitors have been assembled in a library consisting of 96-well microtiter plates with the inhibitors. In addition to these, oligos have been designed in attempts to facilitate miRNA family specific inhibition. The inhibitor oligos target three families: 1) miR-20a-b, -18a-b, -17, -106, -93, 2) miR-34a-c, and 3) miR-25,-92, -363.
Towards setting up functional screens for miRNA in various cellular pathways, we optimised the protocols needed for efficient reverse transfections of different cell lines including HeLa, Nmumg and TIG3/hTERT, for automated immunostainings of endogenous proteins and for visualisation and enzymatic readouts using reporters such as GFP and luciferase. Several successful screenings have been conducted by partner 3. miR101 was shown to modulate autophagy (Frenkel et al., 2011). miR-339 was for instance identified as a modulator of p53 transcriptional activity and overexpression of miR-339 was shown to induce senescence in BJ-cells. In collaboration Partner 3 and 4 have further dissected the miR-339 mechanism of action. Partner 3 and 4 also collaborated on the functional characterisation of miR-661, which also regulates p53 functionality. miR-661 expression was shown to decrease in response to DNA damage and Nutlin-3-induced p53. Expression of both miR-661 and miR-339 enhance p53 levels and activity and sensitise cancer cell lines to DNA-damaging agents. Several putative binding sites for both of these miRs had been identified in the 3'-UTR of MDM2 and MDMX. Recent biochemical data confirm the targeting of MDM2 and MDM2 and MDMX by miR-339 and miR-661, respectively.
Interestingly, the predicted targets of miR-661 within the mRNAs of both Mdm2 and Mdm4 are mainly located within Alu repeats. This prompted us to study the functionality of miR targets within human Alu repeats, as described below. In the case of Mdm2 and Mdm4, we are still in the process of determining which of the several potential miR-661 binding sites within each transcript are actively employed in miR-661 binding and are required for the ability of that miRNA to suppress the expression of the relevant protein.
As regards cancer, one would suspect that a miRNA such as miR-661, which positively regulates the stability and function of p53 by suppressing its two main inhibitors, will be selected against in tumours as a means to quench the tumour suppressor activity of wild type p53. In contrast, in tumours that harbour p53 mutations, particularly mutations that can confer upon the mutant p53 protein an oncogenic gain-of-function, miR661 expression will actually benefit the tumour by upregulating the levels of the oncogenic mutant p53, and therefore will be selected for rather than against. Indeed, analysis of available cancer-derived expression data revealed that in oestrogen receptor-positive breast cancer, where p53 is usually wild type, low levels of miR-661 (which are expected to result in attenuation of the resident wild type p53) are a predictor of bad prognosis. In contrast, in high grade serous ovarian cancer, where p53 is almost unanimously mutant and p53 mutations are perhaps the earliest genetic alteration, high miR-661 is associated with bad prognosis.
Prompted by the realisation that some of the predicted miR-661 targets are hosted within Alu repeats, we undertook a computational analysis of the functionality of putative miRNA targets located within Alu repeats throughout the human transcribed genome. Using a comprehensive dataset of miRNA overexpression assays, we could show that mRNAs with miRNA targets within Alus are significantly less responsive to the miRNA effects as compared to mRNAs that have the same targets outside Alus. Using Ago2-binding mRNA profiling data, we confirmed that the miRNA machinery avoids miRNA targets within Alus, as opposed to the highly efficient binding of targets outside Alus. Based on our analysis, we proposed three features that prevent potential miRNA sites within Alus from being recognised by the miRNA machinery:
(i)Alu repeats that contain miRNA targets and genuine functional miRNA targets appear to reside in distinct mutually-exclusive territories within 3'UTRs;
(ii)Alus have tight secondary structure that may limit access to the miRNA machinery;
(iii)A-to-I editing of Alu-derived mRNA sequences may divert miRNA targets.
The combination of these features is proposed to allow toleration of Alu insertions into mRNAs. Nonetheless, a subset of miRNA targets within Alus appear not to possess any of the above features, and thus may represent cases where Alu insertion in the genome has introduced novel functional miRNA targets, as probably exemplified by Mdm2 and Mdm4 mRNA in the context of miR-661 (Hoffman et al., 2013).
Partner 1 has recently focused his attention to miR-605 which is also predicted to target MDM2 and for which evidence of a role in the regulation of p53 has been obtained. A recently published paper by a competitor group indicated that miR-605 is regulated by p53 and in turn directly targets MDM2 for degradation. Although we could confirm that miR-605 regulates p53 function, we could not confirm MDM2 as a bone fide target. Target identification experiments have been performed to resolve this conundrum.
Through in silico and expression profiling experiments, partner 1 identified miR-29 as putative modulators of melanomagenesis. The miR-29 family of microRNAs has three mature members, miR-29a, miR-29b, and miR-29c. miR-29s are encoded by two gene clusters. The miR-29 family members share a common seed region sequence and are predicted to target largely overlapping sets of genes. miR-29s directly target at least 16 extracellular matrix genes, providing a dramatic example of a single microRNA targeting a large group of functionally related genes. miR-29s have also been shown to be proapoptotic and involved in the regulation of cell differentiation. The physiological and pathological roles of miR29s has been further explored using mouse models.
WP4: ONCOMIRS targets identification
Only few oncomiR-target pairs have been identified to date. Beneficiary 4 has developed an approach for miR targets identification. This technique has already been used to, in a first step, predict targets for the cancer-causing miRs miR10a, miR21 and miR34a and, in a second step, validate several of these targets. Some of these data have been published (Ørom et al., Mol Cell, 2008, Christoffersen et al., 2010). We proposed to use this approach in combination with genome-wide gene expression profiling in order to identify the primary (and secondary) targets of a number of selected ONCOMIRS.
miR-145 is reported to be down-regulated in several cancers, but knowledge of its targets in colon cancer remains limited. To investigate the role of miR-145 in colon cancer, beneficiary 4 has employed the microarray-based approach to identify miR-145 targets. Based on seed site enrichment analyses and unbiased word analyses, they found a significant enrichment of miRNA binding sites in the 3'-untranslated regions (UTRs) of transcripts down-regulated upon miRNA overexpression. Gene ontology analysis showed an overrepresentation of genes involved in cell death, cellular growth and proliferation, cell cycle, gene expression and cancer. A number of the identified miRNA targets have previously been implicated in cancer, including YES, FSCN1, ADAM17, BIRC2, VANGL1 as well as the transcription factor STAT1. Both YES and STAT1 were verified as direct miR-145 targets. The study identifies and validates new cancer-relevant direct targets of miR-145 in colon cancer cells and hereby adds important mechanistic understanding of the tumor-suppressive functions of miR-145. This work has recently been published (Gregersen et al., 2010).
miR-34a is an established p53-target gene and a large body of evidence suggests a critical role for this particular miR in tumour suppression. Consistent with this notion miR34a was found up-regulated in senescence cells and thus both in Nutlin-treated BJ cells or following RAF activation in TIG3 fibroblasts. Even if in the latter, induction appears largely p53-independent. To further gain insight into the mode of action of miR-34a, partner 4 searched for putative miR-34a targets and identified a number of putative miR-34a targets. Among this list is the prominent proto-oncogene Myc.
Two studies -one in the lab of partner 4 and one in the lab of partner 7- focused on miR-9:
Cancer stem cells or cancer initiating cells are believed to contribute to cancer recurrence after therapy. MicroRNAs (miRNAs) are short RNA molecules with fundamental roles in gene regulation. The role of miRNAs in cancer stem cells is only poorly understood. Partner 7 has recently reported miRNA expression profiles of glioblastoma stem cell-containing CD133(+) cell populations. They find that miR-9, miR-9(*) (referred to as miR-9/9(*)), miR-17 and miR-106b are highly abundant in CD133(+) cells. Furthermore, inhibition of miR-9/9(*) or miR-17 leads to reduced neurosphere formation and stimulates cell differentiation. Calmodulin-binding transcription activator 1 (CAMTA1) is a putative transcription factor, which induces the expression of the anti-proliferative cardiac hormone natriuretic peptide A (NPPA). They identify CAMTA1 as an miR-9/9(*) and miR-17 target. CAMTA1 expression leads to reduced neurosphere formation and tumour growth in nude mice, suggesting that CAMTA1 can function as tumour suppressor. Consistently, CAMTA1 and NPPA expression correlate with patient survival.
WP5: Expression of miRs and miR-processors in cancer
To further facilitate and improve sensitivity and accuracy of genome-wide ncRNA profiling, we proposed to generate new microarray platforms based on the LNA technology developed by partner 6. In order to incorporate LNA monomer into the synthesis of oligonucleotides on the array, the LNA phosphoramidites have to have appropriate protecting groups for the light directed synthesis. This means a protecting group, which can be cleaved by light at the 5'-position. The protecting groups on the nucleobase should be photo-stable, but easy to cleave off by hydrolysis. The standard protected LNA amidites do not fulfil these requirements; therefore LNA phopshoramidites with new protecting groups were synthesised. The initial strategy for synthesizing the desired LNA phoshphoramidites was to use the same synthetic pathways as for the DNA counterparts. This proved to be unsuccessful, due to slightly different reactivity of the hydroxyl groups of the LNA compared with the DNA counterparts.
WP6: OncomiR mouse models
In order to define the physiological function and assess the role of selected ncRNAs in the genesis and progression of cancer, we proposed to perform classical gain and loss of function experiments in cultured cells and in vivo, using the mouse as a model system. For conditional gain of function studies, we knock-in the gene of interest into the ROSA26 locus by gene targeting. The newly generated alleles are combined with tumour-causing mutations to study the consequence of aberrant expression of a particular ncRNA or component of the miR-machinery in relevant tumour models of human cancers.
WP7: miR-based therapy
RNA interference represents a potential strategy for in vivo target validation, therapeutic product development and clinical new technologies. It is expected to be particularly useful to silence cancer-causing genes that encode targets that are not amenable to conventional therapeutics. Moreover, there is evidence that miR-based molecules enter the RNAi pathway through a more natural route and yield more effective silencing with reduced toxicity and off-target effects. miR delivery in vivo in tumours is achievable by using new physical methods (electropulsation) developed by partner 5 and prone for clinical applications. Towards this goal, a new generation of miR mimics and antisense therapeutics have been developed and tested. The synthesis of chemically modified LNA-based oligos as miR inhibitors and miR-like oligos is in progress. Initiatives made in collaboration with Partner 3 continue to evaluate performance of both miRNA specific and miRNA family specific inhibitor oligos.
Potential impact:
Scientific impact
The importance of non-coding RNA on normal development and cellular processes such as proliferation, differentiation and apoptosis can no longer be under appreciated. The results from the efforts of this consortium unveiled novel functions of non-coding RNAs in normal and pathological processes.
Community impact
The relevance and impact for the European community is evident at several levels. Scientifically: The research in the consortium is of outmost importance for understanding the roles of ncRNAs in pathological processes such as cancer.
Educationally: The consortium created a perfect environment for research training, both at PhD and post-doc levels. These PhDs and post-docs will continue being very attractive for biotech and pharmaceutical companies, as well as other research institutions.
Economically: The research led to commercially interesting results and thus enable patent applications covering novel targets for anticancer therapy. Moreover, the consortium attracted international competitive funding and high-level non-EU researchers, thereby boosting European research. Finally, the research led to the identification of novel targets for cancer therapy and development of novel drugs or clinical tools for diagnostics and cancer treatment.
Project website: http://med.kuleuven.be/cme-mg/ONCOMIRS/index.html
ONCOMIRS focused on the identification and characterisation of ncRNAs, including microRNAs (miRs) and long non-coding RNAs that play a key role in the aetiology and/or progression of human cancer.
The sequential processing of primary miR transcripts into the mature miRs depends on two main protein complexes, the nuclear microprocessor complex and the Dicer complex. One of the project objectives was the identification of novel key components of the miR-processing machine and examination of their role in carcinogenesis. Through protein complex purification and spectrometry (MS) analyses many factors involved in this process have been identified and reported. Importantly, truncating mutations in TARBP2, encoding for the Dicer-binding protein TRBP, was shown to occur in human cancer. These mutations were shown to affect microRNA processing, dicer function and promote tumorigenesis.
In addition, PRS14, EDC4 and FXR2 have been identified as AGO2-binding proteins and shown to modulate the transforming capacity of various human cancer cell lines. Consistent with the notion that an overall down-regulation of miR levels in cancer contributes significantly to the progression of the disease we showed that while a decrease in Dicer expression increase the transforming potential of cancer cell lines complete silencing is not tolerated by most cancer cells. These data were further supported by genetic experiments in mice. We indeed showed that Dicer functions as a haploinsufficient tumor suppressor gene in a pre-clinical mouse model of retinoblastoma but that it is in the same time required for tumorigenesis. Together these observations have opened exciting new avenues for anticancer therapy. Another important objective is the identification of novel putative cancer-causing ncRNAs.
On a technological point of view, in order to further facilitate and improve sensitivity and accuracy of genome-wide ncRNA profiling, we generated new microarray platforms based on the LNA technology. In addition, promiscuous LNA oligos have been designed in attempts to facilitate the detection or inhibition of entire miRNA families rather than single members of a given family as well as long non-coding RNAs. We also explored the possibility of validating single miRNA-mRNA interactions using LNA oligonucleotides that bind specifically to sequences encompassing and surrounding miRNA target sites within mRNAs. These so-called target site blockers have given very promising results and are now commercially available at Exiqon. Assessment of the biological functions of selected ONCOMIRS by direct experimental identification of target mRNAs is another objective of this project.
In order to define the role of selected ONCOMIRS, lncRNAs and components of the miR-processing machinery in the genesis and progression of cancer, classical gain and loss of function approaches in vivo, using the mouse as a model system were performed. We for instance showed that complete loss of miR17-92 completely prevents retinoblastoma formation in mice (Nittner et al., 2012). In contrast loss of miR17-92 did not affect normal retinogenesis; an observation that has important therapeutic implications. In contrast, however, although upregulated in mouse and human melanoma the miR17-92 cluster was shown to be dispensable for NRas-induced melanoma formation in mice. These data highlighted important differences between tumor types and their dependency to specific ONCOMIRS. Assessment of the therapeutic potential of miR (over)expression and/or inhibition in preclinical mouse models of human cancers was another goal. A new generation of miR mimics and antisense therapeutics have developed. We have tested and optimised the biostability, cytotoxicity and electropulsation-based delivery in vivo of such molecules and demonstrated that electrotransfer of miR inhibitors offers a new avenue for anti-cancer (retinoblastoma and glioblastoma) therapy.
Project context and objectives:
The objectives of the ONCOMIRS' project are presented as they were introduced at the beginning of each period. They remained consistent throughout the whole duration of the project and by and large all objectives have been reached/met.
Period 1
Recent experiments have indicated that alterations in miR levels in cancer could be partly the result of aberrant post-transcriptional processing of precursor miR transcripts. In the last few years, partner 2 has isolated two complexes required for the sequential processing of primary miR transcript into the mature miR in human cells. While the initial processing of primary miR into precursor miR requires the complex composed of RNase III, Drosha, and the double stranded RNA-binding domain protein DGCR8, the formation of the mature miR requires the action of Dicer/TRBP complex. We proposed to actively persue the identification of additional components involved in the regulation of silencing by miRs, through protein complexes purification. One of our main objectives was to assess whether or not components of the Drosha and Dicer complexes are either mutated and/or aberrantly expressed in human cancers.
We proposed to implement the LNA technology present in the group (partner 6) on high-density flexible array platforms to facilitate ncRNA expression analyses. Once this technology will be available, a more long-term goal will be the identification of novel small non-coding RNAs (ncRNAs, including miRs) whose expression is dysregulated in human cancers. In the meantime, we proposed to establish ncRNA gene expression profiles using currently available technologies with the aim to identify genes that are regulated by and therefore act as downstream mediators of the p53 tumor suppressor pathway.
We also proposed to set up / use various gain and loss of function genetic screens with the aim to identify new putative ONCOMIRS based on their abilities to modulate key cellular pathways.
Only a limited number of oncomiR-target pairs have been identified to date. Partner 4 has recently developed a new approach for miR targets identification. We proposed to use this approach in combination with genome-wide gene expression profiling in order to identify the primary (and secondary) targets of a number of selected ONCOMIRS.
One main objective of this application was to study the physiological relevance of a few selected ncRNAs and demonstrate/dissect their active role in cancer formation and progression in vivo. To this end, we proposed to generate classical gene targeting vectors that disrupt expression of the miR(s) of interest in mice. For conditional gain of function studies, we will also knock-in miR gene(s) into the ROSA26 locus by gene targeting. The newly generated alleles will be combined with tumour-causing mutations to study the consequence of aberrant expression of a particular miR in relevant tumor models of human cancers.
RNA interference represents a potential strategy for in vivo target validation, therapeutic product development and clinical new technologies. It is expected to be particularly useful to silence cancer-causing genes that encode targets that are not amenable to conventional therapeutics. Moreover, there is evidence that miR-based molecules enter the RNAi pathway through a more natural route and yield more effective silencing with reduced toxicity and off-target effects. miR delivery in vivo in tumors is achievable by using new physical methods (electropulsation) developed by partner 5 and prone for clinical applications. One of our main objectives is to design miR-based molecules (miR mimics) that either allow potent activation of a 'dormant' tumour suppressor pathway (for instance p53) or alter expression of specific oncogenes. The molecules will be designed to allow efficient delivery, high biostability and low toxicity. Similarly miR antagonists will be designed to target specific ONCOMIRS. The LNA technology developed by partner 6 will be particularly useful in this context. We first planned to test the efficacy and toxicity of these compounds in culture cells.
Period 2
Recent experiments have indicated that alterations in miR levels in cancer could be partly the result of aberrant post-transcriptional processing of precursor miR transcripts. We proposed to actively pursue the identification of additional components involved in the regulation of silencing by miRs, through protein complexes purification. One of our main objectives is to assess whether or not components of the Drosha and Dicer complexes are either mutated and/or aberrantly expressed in human cancers. To this end we proposed to generate monoclonal antibodies against a number of proteins, involved in miRNA function such as Ago proteins, Drosha, DGCR8 and exportin 5.
Human tumours are characterised by widespread reduction in microRNA (miRNA) expression, although it is unclear how such changes come about and whether they have an etiological role in the disease. A defect in miRNA-processing is one possible mechanism for the global down-regulation. To explore this possibility in more detail in vivo we proposed to manipulate Dicer1 gene dosage in a mouse model of retinoblastoma and identify miRs that can either function as tumour suppressors or oncogenes in such a model.
We proposed to implement the LNA technology present in the group (partner 6) on high-density flexible array platforms to facilitate ncRNA expression analyses. A more long-term goal was the identification of novel small non-coding RNAs (ncRNAs, including miRs) whose expression is dysregulated in human cancers. In the meantime, we proposed to establish ncRNA gene expression profiles using currently available technologies with the aim to identify genes that are regulated by and therefore act as downstream mediators of the p53 tumour suppressor pathway.
We also proposed to set up/use various gain and loss of function genetic screens with the aim to identify new putative ONCOMIRS based on their abilities to modulate key cellular pathways.
Only a limited number of oncomiR-target pairs have been identified to date. Partner 4 has recently developed a new approach for miR targets identification. We proposed to use this approach in combination with genome-wide gene expression profiling in order to identify the primary (and secondary) targets of a number of selected ONCOMIRS.
One main objective of this application is to study the physiological relevance of a few selected ncRNAs and demonstrate / dissect their active role in cancer formation and progression in vivo. To this end, we proposed to generate classical gene targeting vectors that disrupt expression of the miR(s) of interest in mice. For conditional gain of function studies, we will also knock-in miR gene(s) into the ROSA26 locus by gene targeting. The newly generated alleles will be combined with tumour-causing mutations to study the consequence of aberrant expression of a particular miR in relevant tumour models of human cancers.
RNA interference represents a potential strategy for in vivo target validation, therapeutic product development and clinical new technologies. It is expected to be particularly useful to silence cancer-causing genes that encode targets that are not amenable to conventional therapeutics. Moreover, there is evidence that miR-based molecules enter the RNAi pathway through a more natural route and yield more effective silencing with reduced toxicity and off-target effects. miR delivery in vivo in tumours is achievable by using new physical methods (electropulsation) developed by partner 5 and prone for clinical applications. One of our main objectives is to design miR-based molecules (miR mimics) that either allow potent activation of a 'dormant' tumour suppressor pathway (for instance p53) or alter expression of specific oncogenes. The molecules will be designed to allow efficient delivery, high biostability and low toxicity. Similarly miR antagonists will be designed to target specific ONCOMIRS. The LNA technology developed by partner 6 will be particularly useful in this context. We first planned to test the efficacy and toxicity of these compounds in culture cells.
Period 3
Recent experiments have indicated that alterations in miR levels in cancer could be partly the result of aberrant post-transcriptional processing of precursor miR transcripts. We proposed to actively pursue the identification of additional components involved in the regulation of silencing by miRs, through protein complexes purification. One of our main objectives is to assess whether or not components of the Drosha and Dicer complexes are either mutated and/or aberrantly expressed in human cancers.
Human tumours are characterised by widespread reduction in microRNA (miRNA) expression, although it is unclear how such changes come about and whether they have an etiological role in the disease. A defect in miRNA-processing is one possible mechanism for the global down-regulation. To explore this possibility in more detail in vivo we proposed to manipulate Dicer1 gene dosage in a mouse model of retinoblastoma and identify miRs that can either function as tumour suppressors or oncogenes in such a model.
We proposed to implement the LNA technology present in the group (partner 6) on high-density flexible array platforms to facilitate ncRNA expression analyses. Another goal is the identification of novel non-coding RNAs (ncRNAs, including miRs) whose expression is dysregulated in human cancers.
We also proposed to set up/use various gain and loss of function genetic screens with the aim to identify new putative ONCOMIRS based on their abilities to modulate key cellular pathways.
Only a limited number of oncomiR-target pairs have been identified to date. Partner 4 and 7 have recently developed new approaches for miR targets identification. We proposed to use this approach in combination with genome-wide gene expression profiling in order to identify the primary (and secondary) targets of a number of selected ONCOMIRS.
One main objective of this application is to study the physiological relevance of a few selected ncRNAs and demonstrate / dissect their active role in cancer formation and progression in vivo. To this end, we proposed to generate miRKO and transgenic mouse lines and combined these alleles with tumour-causing mutations to study the consequence of aberrant expression of selected miRs in relevant tumor models of human cancers.
RNA interference represents a potential strategy for in vivo target validation, therapeutic product development and clinical new technologies. It is expected to be particularly useful to silence cancer-causing genes that encode targets that are not amenable to conventional therapeutics. Moreover, there is evidence that miR-based molecules enter the RNAi pathway through a more natural route and yield more effective silencing with reduced toxicity and off-target effects. miR delivery in vivo in tumours is achievable by using new physical methods (electropulsation) developed by partner 5 and prone for clinical applications. One of our main objectives is to design miR-based molecules (miR mimics) that either allow potent activation of a 'dormant' tumour suppressor pathway (for instance p53) or alter expression of specific oncogenes. The molecules will be designed to allow efficient delivery, high biostability and low toxicity. Similarly miR antagonists will be designed to target specific ONCOMIRS. The LNA technology developed by partner 6 will be particularly useful in this context.
Period 4 (last)
Recent experiments have indicated that alterations in miR levels in cancer could be partly the result of aberrant post-transcriptional processing of precursor miR transcripts. We proposed to actively search for additional components involved in the regulation of silencing by miRs, through protein complexes purification.
Human tumors are characterised by widespread reduction in microRNA (miRNA) expression, although it is unclear how such changes come about and whether they have an etiological role in the disease. A defect in miRNA-processing is one possible mechanism for the global down-regulation. To explore this possibility in more detail in vivo we proposed to manipulate Dicer1 gene dosage in a mouse model of retinoblastoma and identify miRs that can either function as tumor suppressors or oncogenes in such a model.
We proposed to implement the LNA technology present in the group (partner 6) as a tool to knock-down expression of selected non-coding RNAs. Another goal was the identification of novel non-coding RNAs (ncRNAs, including miRs) whose expression is dysregulated in human cancers.
We also proposed to set up/use various gain and loss of function genetic screens with the aim to identify new putative ONCOMIRS based on their abilities to modulate key cellular pathways.
Only a limited number of oncomiR-target pairs have been identified to date. Partner 4 and 7 have recently developed new approaches for miR targets identification. We proposed to use this approach in combination with genome-wide gene expression profiling in order to identify the primary (and secondary) targets of a number of selected ONCOMIRS.
One main objective of this application was to study the physiological relevance of a few selected ncRNAs and demonstrate / dissect their active role in cancer formation and progression in vivo. To this end, we generated KO and transgenic mouse lines for miRs and lncRNAs and combined these alleles with tumour-causing mutations to study the consequence of aberrant expression of selected miRs in relevant tumour models of human cancers.
RNA interference represents a potential strategy for in vivo target validation, therapeutic product development and clinical new technologies. It is expected to be particularly useful to silence cancer-causing genes that encode targets that are not amenable to conventional therapeutics. Moreover, there is evidence that miR-based molecules enter the RNAi pathway through a more natural route and yield more effective silencing with reduced toxicity and off-target effects. miR delivery in vivo in tumours is achievable by using new physical methods (electropulsation) developed by partner 5 and prone for clinical applications. One of our main objectives was to design miR-based molecules (miR mimics) that either allow potent activation of a 'dormant' tumour suppressor pathway (for instance p53) or alter expression of specific oncogenes. Such molecules have be designed to allow efficient delivery, high biostability and low toxicity. Similarly miR antagonists were designed to target specific ONCOMIRS. The LNA technology developed by partner 6 has been particularly useful in this context.
Project results:
WP2: miR Biogenesis and Cancer
MiRNAs are transcribed as primary miRNA transcripts (pri-miRNAs), which serve as substrates for the nuclear microprocessor complex. This multi-protein complex contains the RNase III enzyme Drosha, its co-factor DGCR8 as well as a number of so far uncharacterised components. Drosha processing yields miRNA precursors (pre-miRNAs), which are subsequently transported to the cytoplasm by the export receptor exportin 5. In the cytoplasm, the RNase III enzyme Dicer further processes pre-miRNAs yielding mature miRNAs. MiRNAs are incorporated into miRNA-protein complexes (miRNPs), where they interact with a member of the Argonaute (Ago) protein family. In human somatic cells, Ago1, Ago2, Ago3 and Ago4 are expressed. It is likely that all four Ago proteins are incorporated into similar protein complexes. In order to function in gene silencing, miRNPs interact with a member of the GW protein family. In human, the GW proteins TNRC6A, B and C have been implicated in miRNA function. One of our objectives was the identification of novel components of the miR-processing machinery, through protein complexes purification. Partner 2 identified the huntingtin (Htt) protein as a component of the Argonaute complex. Colocalisation studies demonstrated Htt and Ago2 to be present in P bodies, and depletion of Htt showed compromised RNA-mediated gene silencing. Huntington's disease is a dominant autosomal neurodegenerative disorder caused by an extension of polyglutamines in the Htt protein. The data therefore suggest that this disease may be attributed in part to mutant Htt's role in post-transcriptional processes (Savas et al., 2008). A possible role for components of effector complexes such as Ago1-4-containing miRNPs or TNRC6A-C-containing complexes in cancer will also be studied. In close collaboration with Elisabeth Kremmer (Helmholtz center, Munich), beneficiary 7 generated monoclonal antibodies against a number of proteins, involved in miRNA function. Antibodies against individual human Ago proteins have thus been generated. One of these antibodies allows detection of endogenous Ago2 expression on tissue sections. In addition, antibodies against Drosha, DGCR8 and exportin 5 have been generated. Beneficiary 7 isolated a number of interesting Drosha protein partners among them EWS. Partner 7 also established a list of Dicer-dependent and -independent AGO2-binding proteins (Frohn et al., 2012). Beneficiary 7 has confirmed the presence of the helicase p68 in the DROSHA complex; an observation which is consistent with previously published studies. As expected, they also found the Dicer cofactor TARBP2 (TRBP) and the RNA helicase A/DHX9, which has been implicated in siRNA-loading. Proteins such as YBX1, Gemin4, Gemin5, HNRNPC, HNRNPUL1, the ARE binding protein ELAVL1/HuR, the poly-A binding proteins PABPC1 and 4, the mitochondrial protein Matrin3 and the Fragile X mental retardation protein paralog FXR2 have been found in Ago complex purifications before. In addition, they identified the PABPC1-binding protein LARP1, the mRNA binding proteins FAM120A/Ossa and CSDA, which have not been implicated in Ago2 function before. In contrast to the GW protein family members TNRC6A and B, TNRC6C interacts with Ago2 only in the presence of Dicer, suggesting that it requires Ago2 to be loaded onto miRNA target mRNAs. Moreover they found that a high number of proteins associate with Ago2 in a RNA-dependent manner even in the absence of miRNAs. In this group, they found many RNA binding proteins including IGF2BP13, DHX36, DHX30, HNRNPL, and the ribosomal protein RPS14. Strikingly, genetic data in C. elegans demonstrated that PRS14 modulates mature let-7 tumour suppressor.
A set of proteins associates with Ago2 in an mRNA- independent manner. This includes the TNRC6 proteins and Dicer. Other miRNA-independent protein-protein interactors of Ago2 are the HSP90 alpha and beta proteins with their cochaperones PTGES and FKBP5. Among the interactors preferentially binding in the absence of Dicer and miRNAs only Ago3, CLTC and ZNF521 were identified in the datasets and they show a direct binding behaviour. Taken together, this mass spectrometry approach revealed that Ago2 associates with larger RNA species even in the absence of small RNAs. Furthermore, several mRNA-binding proteins are specific to miRNA-free and miRNA-containing Ago2-mRNA complexes. To functionally validate the interaction data, Partner 7 used luciferase-based miRNA reporters. The 3'-UTR of Hmga2, a well-characterised let-7a target, was fused to firefly luciferase and transfected into HeLa, HCT116 and MCF-7 cells in which ZNF521, CSDA and EDC4 were depleted by RNAi. In addition, we employed a reporter containing the Hmga2 3'-UTR with mutated let-7a target sites and normalised the data against each other. As expected, Ago2 knock down led to increased luciferase activity. Knock down of EDC4, which has been implicated in miRNA function in Drosophila, resulted in specific luciferase up-regulation as well, suggesting that EDC4 is indeed involved in silencing of the Hmga2 reporter construct. Similar results were obtained for the mRNP component CSDA. ZNF521, however, is not involved in miRNA-guided gene silencing. Since ZNF521 is a putative transcription factor, it is possible that it cooperates with Ago2 in nuclear Ago functions. The previously recognised role of let-7 family members as tumour suppressors prompted us to assess the role of PRS14 and EDC4 in cellular transformation by performing colony and soft agar assays. Moreover we have also assessed the role of FXR2 in cellular transformation, as other Fragile X mental retardation proteins have recently been implicated in cancer development. These experiments have confirmed that all these factors are able to modulate cellular transformation in these assays. Consistently, depletion of these factors by RNAi in HCT116, MCF-7 and BJ-cells had profound phenotypic consequences on the ability of these cells to grow in vitro and their transforming potential. Further experiments aimed at dissecting the mechanisms underlying these phenotypic consequences are ongoing in the laboratory of partner 1 and 7.
Overall, down-regulation of miR levels in cancer contributes significantly to the progression of the disease. Since this overall-down-regulation is likely to be, at least partly, a consequence of aberrant processing of their precursor transcripts, we proposed to examine the expression levels of Drosha/DGCR8 and Dicer/TRBP components, in cancer cell lines and human primary tumours and assess the contribution of aberrant expression and/or altered function in cellular transformation. Beneficiary 2 -in collaboration with Pr Manel Esteller (CNIO, Madrid, Spain)- has screened various cancer cell lines with microsatellite instability for the presence of mutations in all exonic mononucleotide repeats present in the coding sequences of eight established members of the miRNA processing machinery: Dicer and Drosha, DGCR8, TRBP, PACT and AGO1, AGO2 and AGO4. Wild-type sequences were found for all genes with the exception of TARBP2 encoding for TRBP. Two heterozygous frame-shift mutations were identified in two different cell lines. These mutations were found to cause a significant decrease in TRBP expression. Since TRBP stabilises DICER, these mutations also affected Dicer expression levels and consequently caused a drastic reduction in the efficiency of endogenous processing of miRs. Importantly, ectopic expression of wild-type TRBP, but not of the naturally occurring mutants, induced tumour suppressor-like features in colorectal cancer cell lines. Notably, Dicer overexpression also showed tumour suppressor properties as determined using classical tissue culture assays (MTT and colony assays) and mouse xenograph models (nude mice). Finally, TARBP2 mutations were detected in about 25% of a panel of 282 human primary tumours (Melo et al., 2009). The above findings provide further evidence of a role of loss of function events in the regulation of miRNA processing machinery during tumourigenesis. However, the molecular mechanisms that promote tumourigenesis as a consequence of global repression of miRNA maturation remain elusive. Data obtained by beneficiary 1 suggest that the p53 tumor suppressor pathway is compromised in cancer cells harbouring hypomorphic Dicer expression.
Given the results of the mutational analyses of the various components of the miR-processing machinery in cancer cell lines and primary tumours, it was decided to slightly deviate from the initial plans. Instead of constructing artificial TRBP mutants lacking the dsRNA-binding domains, it was preferred to assess the oncogenic properties of the naturally occurring TRBP mutants. To add relevance to our structure function analysis, the experiments were conducted in colorectal cancer cell lines, instead of NIH-3T3, since these mutations were found to occur in this particular tumour context. Moreover, since no mutations were found in Dicer and AGO2 in human tumours, construction of the initially proposed Dicer and AGO2 mutants has been cancelled. Instead, we have already tested the role of TRBP and the influence of the naturally occurring mutations to cell transformation. In this context, the role of Dicer in cell transformation was also assessed. Moreover, we have demonstrated that the mutations in TRBP affect Dicer levels and consequently the integrity of the RNAi machinery. These data, recently reported in Melo et al., 2009, indicate that:
(i) TRBP and DICER exhibit tumor suppressive activities; and
(ii) mutations that ultimately affect miR-processing occur in human cancer.
Human tumours are characterised by widespread reduction in microRNA (miRNA) expression, although it is unclear how such changes come about and whether they have an etiological role in the disease. Importantly, miRNA-knockdown has been shown to enhance the tumourigenic potential of human lung adenocarcinoma cells. A defect in miRNA-processing is one possible mechanism for the global down-regulation. To explore this possibility in more detail in vivo, we have manipulated Dicer1 gene dosage in a mouse model of retinoblastoma. We show that while monoallelic loss of Dicer1 does not affect normal retinal development it dramatically accelerates tumour formation on a retinoblastoma-sensitised background. Importantly, these tumours retain one wild-type Dicer1 allele and exhibit only partial decrease in miRNA-processing. Accordingly, in silico analysis of human cancer genome data reveals frequent hemizygous, but not homozygous, deletions of DICER1. Strikingly, complete loss of Dicer1 function in mice did not accelerate retinoblastoma formation. miRNA profiling of these tumours identified members of the let-7 and miR-34 families as candidate tumour suppressors in retinoblastoma. We conclude that Dicer1 functions as a haploinsufficient tumour suppressor (Lambertz et al., 2010).
The above finding provides genetic evidence for a causative link between down-regulation of miRNA-processing and cancer progression.
Interestingly, although reduced levels of DICER1 in tumours have been reported, no loss-of-function mutations in DICER1 have been reported to date. Moreover, homozygous loss of Dicer1 appears to be strongly selected against in a K-Ras-induced mouse model of lung cancer and Myc-induced mouse model of B-cell lymphoma. These data raise the possibility that Dicer1 is required for tumour formation. Beneficiary 1's group has recently demonstrated that targeted homozygous loss of Dicer1 completely prevents the formation of retinoblastoma in mice in which the Rb and p53 tumour suppressor pathways are inactivated. Strikingly, it was found that Dicer1 deficiency selectively kills Rb-deficient retinal cells in which p53 is inactivated while sparing cells that retain functional p53. miRNA profiling of mouse and human primary retinoblastomas showed dramatic overexpression of the pro-oncogenic miR17-92 cluster in all samples analysed. High-resolution array-CGH indicates that in about 20 % of human retinoblastoma patients overexpression of miR-17-92 results from copy number alterations. Crucially, functional inactivation of the miRNAs encoded by the miR-17-92 cluster is sufficient to decrease the viability of human retinoblastoma cells and to prevent retinoblastoma formation in mice. The data therefore provide genetic evidence of a synthetic lethal interaction between Dicer and p53 and designate members of the miR-17-92 cluster as a highly selective therapeutic target for the treatment of retinoblastoma (Nittner et al., 2012).
WP3: Identification of novel ONCOMIRS (and long non-coding RNAs)
The data described above imply that a set of miRNAs might act to prevent full-blown retinoblastoma formation on the Chx10Cre;Rblox/lox;p107-/- background. To identify such candidate tumor suppressor miRNAs beneficiary 1 determined the expression profile of the entire miRNome in P20 retinae from wild-type (Cre-negative), Chx10Cre; Rblox/lox; p107-/-; Dic+/+ and Chx10Cre;Rblox/lox;p107-/-;Diclox/+ using LNA-based microarray (in collaboration with Patrner 6) and RT-qPCR approaches. Both types of analyses identified a common set of 11 miRNAs that are consistently up-regulated between wild-type and Chx10Cre; Rblox/lox; p107-/-; Dic+/+. Most interestingly, among them are 2 members of the let-7 family (let-7c and let7-i), the up-regulation of which and of another let-7 member, let-7b, was confirmed by independent Q-RT-PCR analysis. Given that this up-regulation was significantly attenuated in the Dic heterozygous retinae and the recognised role of let-7 family members in tumor suppression, this observation raises the possibility that let-7 have a causal role in retinoblastoma formation as critical regulators of the switch from retinomas to retinoblastoma. Our list of 11 differentially expressed miRNAs also included miR-34c. There was also a clear upregulation of miR-34b-3p in both microarray and RT-q-PCR analyses and a moderate, but reproducible, up-regulation of miR-34a was evident from the micro-array data. Interestingly, the miR-34 family members have been identified as p53 targets and key mediators of its tumor suppressor function.
The data from WP2 also implied that a set of miRNAs is required for retinoblastoma formation despite the combined mutation in Rb, p107 and p53. In search of such pro-oncogenic miRNAs we profiled the miRNome by stem-loop RT-qPCR in P21 retinae from wild-type (Cre-negative), Chx10Cre; Rblox/lox; p107-/- and from Laser-Captured-Microdissected tumor materials from 4 Chx10Cre;Rblox/lox;p107-/-;p53lox/lox (TKO) mice. This analysis identified a set of 102 miRNAs that are significantly up-regulated in the TKO tumours. To find correlates of the mouse data in human tumours, we also profiled miRNA expression in 30 different human primary retinoblastomas. 68 miRNAs were significantly up-regulated in retinoblastoma compared to normal human retinae. Cross-species comparison identified 25 miRNAs that were up-regulated in both mouse and human tumours. Strikingly, 12 of them are members of the known oncogenic miR-17-92 and 106b-25/miR-106a-92 paralogue clusters. Consistently, hierarchical clustering of all RB cases and normal retinae based on miRNA expression singled out all members of these clusters as being dramatically up-regulated in all mouse and human tumours analyzed. To explore the potential causes of miRNA deregulation in human retinoblastoma beneficiary 1 looked for genomic aberrations using a 44K oligonucleotide array, which was specifically designed to include regions harbouring miRNA genes. In addition to identifying previously reported retinoblastoma-associated genomic aberrations (1q gain and 6p22 gain were frequently seen in our cohort) focal amplification of the miR-17-92 locus, which lies on chromosome 13, was found in one patient. Another patient had a whole chromosome 13 gain and 3 patients had copy number gains including the miR-17-92 cluster but, importantly, not the closely linked Rb-1 locus. miR-17-92 copy number gains were found in 17 % of the patients (5 out of 29 cases analysed). Moreover, while the Rb-1 locus was deleted in 21 % of cases (6/29) this deletion never included the closely linked miR-17-92 locus. This analysis therefore indicates that up-regulation of the miR-17-92 cluster is, at least in a proportion of retinoblastoma cases, a direct consequence of increased gene copy number. Since transcription of miR-17-92 is positively regulated by the E2Fs and negatively regulated by p53, deregulation of their transcriptional activities may also account for miR-17-92 overexpression in retinoblastoma. Regardless of the underlying mechanism, our data demonstrate that the miR17-92 cluster is overexpressed in 100 % of retinoblastomas analysed. Together these data raise the possibility that up-regulation of expression of the miR17-92 cluster is a critical event for the development retinoblastoma; this was later shown to be case (see WP6).
We also proposed to establish miR gene expression profiles using currently available technologies with the aim to identify genes that are regulated by and therefore act as downstream mediators of the p53 tumour suppressor pathway. In a first set of experiments, we have activated the p53 pathway chemically using the Mdm2-inhibitor Nutlin3a. BJ primary human diploid fibroblasts undergo p53-mediated irreversible cell cycle arrest or senescence upon such treatment. Using a Q-RT-PCR-based profiling method, we have identified miRs that are induced and repressed under these conditions. As control, quiescent -serum starved- cells were also profiled. MiR-34a, b and c were all specifically induced in senescent cells. In addition, miR-24, miR-23b and miR-27b, all encoded by the same locus, were also found induced in Nutlin-treated cells. In addition, miR-195, miR-7 and miR-452# were also all found dramatically up-regulated whereas miR-302b and miR-586 were both strongly down-regulated in senescence cells but not in quiescent cells.
Beneficiary 3 identified a panel of miRs whose expression is repressed by p53 during replicative senescence of human diploid fibroblasts; this study has now been published (Brosh et al, 2009). In subsequent work, the regulation of miR-372 and miR-373 during oncogene-induced senescence has been addressed. Specifically, it was found that these two miRNAs are downregulated in response to overexpression of oncogenic H-Ras. This correlates with the upregulation of the Lats2 gene by oncogenic Ras. Of note, Lats2 is required for p53 activation in human fibroblasts in response to oncogenic Ras, and is thus a critical mediator of oncogene-induced senescence in this system. These observations are consistent with a potential role for the miRNAs in inhibition of oncogene-induced senescence, and provide a mechanistic explanation for earlier observations of Agami and co-workers (Aylon et al, 2010). Beneficiary 3 also investigated the notion that some p53-regulated miRNAs may be epigenetically silenced in tumours that retain wild-type p53, rendering these tumours less susceptible to killing by genotoxic chemotherapy. To that end, conditions for the synergistic killing of A549 lung cancer cells by a combination of cisplatin and agents that 'erase' epigenetic silencing marks (TSA+ 5-Aza-cytidine) have been optimised. miRNA profiling from such cells with and without various treatment combinations will be performed by beneficiary 6. The data will be subjected to validation by qRT-PCR in the subsequent months. To identify miRNAs that are regulated by p53 in vivo, possibly in a tissue-specific manner, beneficiary 2 carried out miRNA profiling on different tissues of wild type and p53 knockout mice. Following validation of the most promising hits by qRT-PCR, it was decided to focus on miR-223 and miR-449. Both of these miRNAs are positively regulated by p53 in some cultured cell types but not in others. miR-223 was previously reported to play a role in the regulation of differentiation in the myeloid lineage.
Interestingly, using a myeloid leukemia cell line carrying an inducible (temperature sensitive) p53, a modest but reproducible increase in miR-223 levels upon p53 induction was observed. The effect of p53 activation on the differentiation of those cells (which can be induced by either IL6 or dexamethazone) and the contribution of miR-223 to this effect will be further investigated. Beneficiary 2 has also performed miRNA profiling using an experimental array developed by beneficiary 6, which combines probe sets for mature miRNA and for pre-miRNA. The input RNA was derived from H1299 lung cancer-derived cells containing a temperature sensitive p53 (the same experimental model that was employed earlier to discover that miR-34a was a p53 target). Surprisingly, the results suggested that there may be profound differences between the impact of p53 activation on some mature miRNAs relative to their precursors. The data still need to be validated by more quantitative methods; if validated, this may suggest an effect of p53 on the processing of specific pre-miRNAs by Dicer, and will merit further investigation. An additional observation provided by this miRNA profiling experiment was that p53 seemed to up-regulate the levels of miR-372 and miR-373*; subsequent preliminary data suggest that these miRNAs may also be positively regulated by p53 in A549 cells, which harbor endogenous wtp53.
Given that the expression of miR-372 was found by us earlier to be negatively regulated by Ras, it will be interesting to investigate the cross-talk between p53 and Ras in regulation of the expression of the miR-372/373 cluster. In parallel, we initiated a new line of investigation, not included in the original workplan. This is based on the recent realisation that p53 plays a role in regulation of autophagy, a process that is turning out to be of great importance in cancer biology. In a collaborative study with Prof. Z. Elazar at the Weizmann Institute, we searched for miRNAs whose expression is induced in HeLa cells under conditions of nutrient starvation that trigger autophagy. One of the top hits was miR-492, primate-specific miRNA that highly expressed in retinoblastoma. Interestingly, miR-492 was also one of the top hits in another screen carried out by beneficiary 3, which looked for miRNAs whose expression is altered in a p53-dependent manner in HCT116 colorectal cancer cells treated with the mitotic spindle poison nocodazole. Follow-up experiments validated that miR-492 is induced by nutrient starvation (24-48h) in p53-proficient HCT116 cells but not in their derivative p53 knockout cells (HCT116 p53-/-).
We further found that HCT116 p53-/- cells have an increased propensity to become polyploid upon nutrient starvation, a feature not seen in wt HCT116 cells and thus indicative of a role of p53 in preventing starvation-induced polyploidisation. On note, overexpression of miR-492 in the p53-/- cells could reduce significantly this starvation-induced polyploidy.
Beneficiary 7 searched for miRNAs whose expression is dependent on the p53 family members p63 and p73. A list of validated p63- and p73-regulated miRNAs has been available to the members of the consortium.
The ultra conserved regions (UCRs) are a subset of conserved regions that are absolutely conserved (100 %) between orthologous regions of the human, rat and mouse genomes and are longer than 200bp (Bejerano et al., 2004). Because of the very high degree of conservation, the UCRs are likely to have fundamental functional importance for the ontogeny and normal physiological functioning of cells and tissues, including gene regulation and chromosome biology of the organisms in which they are found. The majority of UCRs (about 90 %) are transcribed (T-UCRs) in normal human tissues. While a minority is expressed ubiquitously, most are tissue specific (Calin et al., 2007). As with miRs (Liu et al., 2004), hierarchical clustering of T-UCR expression in various hematopoietic and non-hematopoietic tissues from different individuals suggests that T-UCRs can be used as tissue specific markers. The same types of tissues from different individuals clustered as closest neighbours. Transcription of UCRs may be regulated by specific transcription factors and/or initiated from polyadenine rich genomic regions, as was recently proposed for several long ncRNAs in the mouse (Furuno et al., 2006). Importantly, levels of UCRs may be regulated post-transcriptionally by miRs. Calin et al. (2007) has indeed provided evidence that specific miRs interact with complementary T-UCRs and that this interaction results in down-regulation of T-UCRs expression. T-UCRs are frequently located at fragile sites and other genomic regions that have been implicated in cancers (Calin et al., 2007). Genome-wide profiling revealed that UCRs, just like miRs, have distinct signatures in human leukemias and carcinomas and therefore can be used to differentiate between human cancers (Calin et al., 2007). Moreover, about 10 % of T-UCRs are differentially expressed in at least one type of tumor cells when compared to normal cells of the same origin. Among the most significant differentially expressed T-UCRs in leukemias and carcinomas are uc.29 uc.73 (high in colorectal cancers), uc111, uc135 (low in CLL), ... Interestingly, a significant negative correlation between 87 miRs (out of 285 analyzed) and T-UCRs expression levels was found at a genome-wide level in 50 patients with leukemias (CLL) (Calin et al., 2007). This observation is consistent with the possibility that T-UCRs are targeted by miRs and these interactions may have biological and prognostic significance for cancer patients. Strikingly, preliminary functional studies suggest that uc.73 (P), which is significantly up-regulated in colon cancers, behaves like an oncogene in tissue culture by promoting both proliferative and anti-apoptotic activities (Calin et al., 2007). We have therefore decided - even if not planned originally - to profile the T-UCRnome (481 different UCRs) using a Q-RT-PCR-based approach in senescent and quiescent BJ cells. A list of UCRs specifically induced and repressed in senescent cells is now available. Of note, among the most differentially regukated genes are uc135 (up 30-fold) and uc73 (down more than 100-fold). Equally interesting is the observation that uc.177 which is drastically down-regulated in senescent cells is putatively regulated by all miR-34 family members. Further experiments are currently ongoing in the laboratory of partner 1 to test this hypothesis and assess the relevance of both miRs and UCRs that are specifically induced in Nutlin-treated cells for the senescence phenotype.
Partner 3 has identified a panel of miRs whose expression is altered during replicative senescence of human diploid fibroblasts. A substantial fraction of those miRs were found to belong to 3 paralogous polycistronic clusters, including the miR17-92 cluster, the miR-106b/93/25 cluster and the miR-17-5p cluster; levels of many miRs encoded within these clusters were shown to decrease markedly in senescent cells. Furthermore, this downregulation was p53-dependent, whereas the expression of these miRs in non-senescent cells was largely driven by E2F. Further analysis revealed a feed-forward loop that drives E2F-dependent transcription of those miRs. Functionally, overexpression of these miRs could promote excessive proliferation and delay senescence, consistent with a proposed oncogenic role of at least some of them. Indeed, expression of many of those miRs was found to be upregulated in breast cancer tumors that carry p53 mutations. These findings have recently been published (Brosh et al., 2008).
Partner 4 has identified a set of miRNAs regulated by oncogenic B-RAF in TIG3 primary human fibroblasts. Of the about 10 miRNAs significantly altered following 4 days of B-RAF induction miR-34a and miR-146a were the most significantly upregulated. We focused on miR-34a, as the absolute level of this miRNAs was much higher. Using a combination of siRNA experiments, chromatin-immunoprecipitations and heterologous promoter assays we have identified ELK1 as the transcription factor responsible for miR-34a upregulation downstream B-RAF. In addition, we have established that p53 plays little or no role in miR-34a induction during oncogene-induced senescence.
Partner 2 has recently compared induction of miRs in different mouse organs following exposure of wild type and p53 knock-out mice to whole body ionising radiation (IR). We observed substantial differences between different organs: while some miRs (e.g. miR-34a) were more widely induced following p53 activation by IR, others were induced only in one or two organs (e.g. miR-223 and miR-449a), attesting to the tissue-specificity of p53-dependent miR induction. The results of this analysis, obtained by miR array hybridisation, are presently being subjected to validation by qPCR-RT-PCR analysis.
We explored the regulation of miR199a by p53. Initial results suggested that this particular microRNA is selectively induced by several p53 mutants. Further analysis revealed a more complex picture: depending on cell type, miR-199a could be induced by either mutant p53 and/or WT p53. However, we could not find any experimental model in which the extent of induction was more than 2-3 fold, rendering the subsequent functional analysis insufficiently definitive. This line of research was therefore discontinued. In parallel, we sought to identify more comprehensively the spectrum of miRs regulated by mutant p53. To that end, we performed miR expression profiling on DU-145 cells (human prostate cancer, harbouring 2 mutant p53 alleles) before and after knockdown of their endogenous mutant p53 proteins. The results are still being validated, but overall the differences were rather minimal and disappointing, suggesting that this cell line is probably not an optimal choice for discovery of mutant p53-regulated miRs. We are presently investigating the possible regulation of miR-214 by WT and mutant p53. Preliminary results suggest that WT p53 can repress the expression of this miR, whose excessive expression has been implicated in ovarian cancer. Further analysis will therefore focus on cell lines derived from this type of cancer.
The setting up of several automated forward genetic screens as a mean to identify cancer-causing miRs is in progress. To this end, Partner 6 (Exiqon) has optimised Tm and LNA spiking pattern of LNA oligos to obtain miRNA inhibitors with increased potency and stability relative to previous versions. LNA based miRNA inhibitors have been designed for all human miRNAs listed in miRBASE 11.0 and the inhibitors have been assembled in a library consisting of 96-well microtiter plates with the inhibitors. In addition to these, oligos have been designed in attempts to facilitate miRNA family specific inhibition. The inhibitor oligos target three families: 1) miR-20a-b, -18a-b, -17, -106, -93, 2) miR-34a-c, and 3) miR-25,-92, -363.
Towards setting up functional screens for miRNA in various cellular pathways, we optimised the protocols needed for efficient reverse transfections of different cell lines including HeLa, Nmumg and TIG3/hTERT, for automated immunostainings of endogenous proteins and for visualisation and enzymatic readouts using reporters such as GFP and luciferase. Several successful screenings have been conducted by partner 3. miR101 was shown to modulate autophagy (Frenkel et al., 2011). miR-339 was for instance identified as a modulator of p53 transcriptional activity and overexpression of miR-339 was shown to induce senescence in BJ-cells. In collaboration Partner 3 and 4 have further dissected the miR-339 mechanism of action. Partner 3 and 4 also collaborated on the functional characterisation of miR-661, which also regulates p53 functionality. miR-661 expression was shown to decrease in response to DNA damage and Nutlin-3-induced p53. Expression of both miR-661 and miR-339 enhance p53 levels and activity and sensitise cancer cell lines to DNA-damaging agents. Several putative binding sites for both of these miRs had been identified in the 3'-UTR of MDM2 and MDMX. Recent biochemical data confirm the targeting of MDM2 and MDM2 and MDMX by miR-339 and miR-661, respectively.
Interestingly, the predicted targets of miR-661 within the mRNAs of both Mdm2 and Mdm4 are mainly located within Alu repeats. This prompted us to study the functionality of miR targets within human Alu repeats, as described below. In the case of Mdm2 and Mdm4, we are still in the process of determining which of the several potential miR-661 binding sites within each transcript are actively employed in miR-661 binding and are required for the ability of that miRNA to suppress the expression of the relevant protein.
As regards cancer, one would suspect that a miRNA such as miR-661, which positively regulates the stability and function of p53 by suppressing its two main inhibitors, will be selected against in tumours as a means to quench the tumour suppressor activity of wild type p53. In contrast, in tumours that harbour p53 mutations, particularly mutations that can confer upon the mutant p53 protein an oncogenic gain-of-function, miR661 expression will actually benefit the tumour by upregulating the levels of the oncogenic mutant p53, and therefore will be selected for rather than against. Indeed, analysis of available cancer-derived expression data revealed that in oestrogen receptor-positive breast cancer, where p53 is usually wild type, low levels of miR-661 (which are expected to result in attenuation of the resident wild type p53) are a predictor of bad prognosis. In contrast, in high grade serous ovarian cancer, where p53 is almost unanimously mutant and p53 mutations are perhaps the earliest genetic alteration, high miR-661 is associated with bad prognosis.
Prompted by the realisation that some of the predicted miR-661 targets are hosted within Alu repeats, we undertook a computational analysis of the functionality of putative miRNA targets located within Alu repeats throughout the human transcribed genome. Using a comprehensive dataset of miRNA overexpression assays, we could show that mRNAs with miRNA targets within Alus are significantly less responsive to the miRNA effects as compared to mRNAs that have the same targets outside Alus. Using Ago2-binding mRNA profiling data, we confirmed that the miRNA machinery avoids miRNA targets within Alus, as opposed to the highly efficient binding of targets outside Alus. Based on our analysis, we proposed three features that prevent potential miRNA sites within Alus from being recognised by the miRNA machinery:
(i)Alu repeats that contain miRNA targets and genuine functional miRNA targets appear to reside in distinct mutually-exclusive territories within 3'UTRs;
(ii)Alus have tight secondary structure that may limit access to the miRNA machinery;
(iii)A-to-I editing of Alu-derived mRNA sequences may divert miRNA targets.
The combination of these features is proposed to allow toleration of Alu insertions into mRNAs. Nonetheless, a subset of miRNA targets within Alus appear not to possess any of the above features, and thus may represent cases where Alu insertion in the genome has introduced novel functional miRNA targets, as probably exemplified by Mdm2 and Mdm4 mRNA in the context of miR-661 (Hoffman et al., 2013).
Partner 1 has recently focused his attention to miR-605 which is also predicted to target MDM2 and for which evidence of a role in the regulation of p53 has been obtained. A recently published paper by a competitor group indicated that miR-605 is regulated by p53 and in turn directly targets MDM2 for degradation. Although we could confirm that miR-605 regulates p53 function, we could not confirm MDM2 as a bone fide target. Target identification experiments have been performed to resolve this conundrum.
Through in silico and expression profiling experiments, partner 1 identified miR-29 as putative modulators of melanomagenesis. The miR-29 family of microRNAs has three mature members, miR-29a, miR-29b, and miR-29c. miR-29s are encoded by two gene clusters. The miR-29 family members share a common seed region sequence and are predicted to target largely overlapping sets of genes. miR-29s directly target at least 16 extracellular matrix genes, providing a dramatic example of a single microRNA targeting a large group of functionally related genes. miR-29s have also been shown to be proapoptotic and involved in the regulation of cell differentiation. The physiological and pathological roles of miR29s has been further explored using mouse models.
WP4: ONCOMIRS targets identification
Only few oncomiR-target pairs have been identified to date. Beneficiary 4 has developed an approach for miR targets identification. This technique has already been used to, in a first step, predict targets for the cancer-causing miRs miR10a, miR21 and miR34a and, in a second step, validate several of these targets. Some of these data have been published (Ørom et al., Mol Cell, 2008, Christoffersen et al., 2010). We proposed to use this approach in combination with genome-wide gene expression profiling in order to identify the primary (and secondary) targets of a number of selected ONCOMIRS.
miR-145 is reported to be down-regulated in several cancers, but knowledge of its targets in colon cancer remains limited. To investigate the role of miR-145 in colon cancer, beneficiary 4 has employed the microarray-based approach to identify miR-145 targets. Based on seed site enrichment analyses and unbiased word analyses, they found a significant enrichment of miRNA binding sites in the 3'-untranslated regions (UTRs) of transcripts down-regulated upon miRNA overexpression. Gene ontology analysis showed an overrepresentation of genes involved in cell death, cellular growth and proliferation, cell cycle, gene expression and cancer. A number of the identified miRNA targets have previously been implicated in cancer, including YES, FSCN1, ADAM17, BIRC2, VANGL1 as well as the transcription factor STAT1. Both YES and STAT1 were verified as direct miR-145 targets. The study identifies and validates new cancer-relevant direct targets of miR-145 in colon cancer cells and hereby adds important mechanistic understanding of the tumor-suppressive functions of miR-145. This work has recently been published (Gregersen et al., 2010).
miR-34a is an established p53-target gene and a large body of evidence suggests a critical role for this particular miR in tumour suppression. Consistent with this notion miR34a was found up-regulated in senescence cells and thus both in Nutlin-treated BJ cells or following RAF activation in TIG3 fibroblasts. Even if in the latter, induction appears largely p53-independent. To further gain insight into the mode of action of miR-34a, partner 4 searched for putative miR-34a targets and identified a number of putative miR-34a targets. Among this list is the prominent proto-oncogene Myc.
Two studies -one in the lab of partner 4 and one in the lab of partner 7- focused on miR-9:
Cancer stem cells or cancer initiating cells are believed to contribute to cancer recurrence after therapy. MicroRNAs (miRNAs) are short RNA molecules with fundamental roles in gene regulation. The role of miRNAs in cancer stem cells is only poorly understood. Partner 7 has recently reported miRNA expression profiles of glioblastoma stem cell-containing CD133(+) cell populations. They find that miR-9, miR-9(*) (referred to as miR-9/9(*)), miR-17 and miR-106b are highly abundant in CD133(+) cells. Furthermore, inhibition of miR-9/9(*) or miR-17 leads to reduced neurosphere formation and stimulates cell differentiation. Calmodulin-binding transcription activator 1 (CAMTA1) is a putative transcription factor, which induces the expression of the anti-proliferative cardiac hormone natriuretic peptide A (NPPA). They identify CAMTA1 as an miR-9/9(*) and miR-17 target. CAMTA1 expression leads to reduced neurosphere formation and tumour growth in nude mice, suggesting that CAMTA1 can function as tumour suppressor. Consistently, CAMTA1 and NPPA expression correlate with patient survival.
WP5: Expression of miRs and miR-processors in cancer
To further facilitate and improve sensitivity and accuracy of genome-wide ncRNA profiling, we proposed to generate new microarray platforms based on the LNA technology developed by partner 6. In order to incorporate LNA monomer into the synthesis of oligonucleotides on the array, the LNA phosphoramidites have to have appropriate protecting groups for the light directed synthesis. This means a protecting group, which can be cleaved by light at the 5'-position. The protecting groups on the nucleobase should be photo-stable, but easy to cleave off by hydrolysis. The standard protected LNA amidites do not fulfil these requirements; therefore LNA phopshoramidites with new protecting groups were synthesised. The initial strategy for synthesizing the desired LNA phoshphoramidites was to use the same synthetic pathways as for the DNA counterparts. This proved to be unsuccessful, due to slightly different reactivity of the hydroxyl groups of the LNA compared with the DNA counterparts.
WP6: OncomiR mouse models
In order to define the physiological function and assess the role of selected ncRNAs in the genesis and progression of cancer, we proposed to perform classical gain and loss of function experiments in cultured cells and in vivo, using the mouse as a model system. For conditional gain of function studies, we knock-in the gene of interest into the ROSA26 locus by gene targeting. The newly generated alleles are combined with tumour-causing mutations to study the consequence of aberrant expression of a particular ncRNA or component of the miR-machinery in relevant tumour models of human cancers.
WP7: miR-based therapy
RNA interference represents a potential strategy for in vivo target validation, therapeutic product development and clinical new technologies. It is expected to be particularly useful to silence cancer-causing genes that encode targets that are not amenable to conventional therapeutics. Moreover, there is evidence that miR-based molecules enter the RNAi pathway through a more natural route and yield more effective silencing with reduced toxicity and off-target effects. miR delivery in vivo in tumours is achievable by using new physical methods (electropulsation) developed by partner 5 and prone for clinical applications. Towards this goal, a new generation of miR mimics and antisense therapeutics have been developed and tested. The synthesis of chemically modified LNA-based oligos as miR inhibitors and miR-like oligos is in progress. Initiatives made in collaboration with Partner 3 continue to evaluate performance of both miRNA specific and miRNA family specific inhibitor oligos.
Potential impact:
Scientific impact
The importance of non-coding RNA on normal development and cellular processes such as proliferation, differentiation and apoptosis can no longer be under appreciated. The results from the efforts of this consortium unveiled novel functions of non-coding RNAs in normal and pathological processes.
Community impact
The relevance and impact for the European community is evident at several levels. Scientifically: The research in the consortium is of outmost importance for understanding the roles of ncRNAs in pathological processes such as cancer.
Educationally: The consortium created a perfect environment for research training, both at PhD and post-doc levels. These PhDs and post-docs will continue being very attractive for biotech and pharmaceutical companies, as well as other research institutions.
Economically: The research led to commercially interesting results and thus enable patent applications covering novel targets for anticancer therapy. Moreover, the consortium attracted international competitive funding and high-level non-EU researchers, thereby boosting European research. Finally, the research led to the identification of novel targets for cancer therapy and development of novel drugs or clinical tools for diagnostics and cancer treatment.
Project website: http://med.kuleuven.be/cme-mg/ONCOMIRS/index.html