No description available.

Small RNAs, and in particular microRNAs, are currently receiving a great deal of attention due to their important roles in gene regulation and organism development. Recently, new high-throughput technologies have made it possible to sequence hundreds of thousands of small RNAs from a single experimental sample. In this thesis we develop new computational tools to process such high-throughput small RNA datasets in order to identify microRNAs and other biologically interesting small RNA candidates and to predict their target genes. We apply these tools to a variety of plant and animal datasets and present some novel discoveries including miRNAs involved in fruit development in tomato (Solanum lycopersicon).

This thesis contains two separate projects. The first project involves the study of Argonaute 1 (Ago l) which is a member of the ArgonautelPIWI protein family involved in small RNA-mediated gene regulation. In Drosophila, Agol plays a specific role in microRNA (miRNA) function. Previous studies have demonstrated that Ago 1 regulates the fate of germline stem cells. However, the function of Ago 1 in other aspects of oogenesis is still elusive. Here, the function of Ago 1 was analysed in the female germline and follicle cells. The Ago 1 protein is enriched in the oocyte and also highly expressed in the cytoplasm of follicle cells. Clonal analysis of multiple ago1 mutant alleles shows that mutant egg chambers often contain only eight nurse cells and lack an oocyte which is phenocopied in dicer-I , pasha and drosha mutants. This implies a general role of miRNAs in this process. Further analysis of the clones shows that Agol, Dicer-l, Drosha and miR-l24a play a role in maintaining follicle cell integrity. Additionally, it was found that Ago 1 protein levels were lowered when miRNAs were not present and that overexpression of miR-124a increased the steady state level of Ago 1 proteins. These were believed to be the effect of Ago 1 protein stability. In the meantime, a second project was carried studying a novel filamentous structure called cytoophidia which contains CTP synthase (CTPS) was analysed. Previous studies have shown that the cytoophidia is prominent in the egg chambers but how and why CTPS forms cytoophidia is still unclear. Here, the different CTPS isoforms were analysed and only CTPS isoform C was found to form cytoophidia. This isoform could also induce cytoophidia formation in other tissues that does not have obvious cytoophidia formation. Then, it was identified that the first 56 amino acid of CTPS isoform C is the important region for cytoophidia formation. Furthermore, ctps mutants were generated and analysed. CTPS mutations caused lethality during the larval stage and when maternal contribution was removed, lethality was seen during embryogenesis.

Mammalian expression systems, particularly Chinese hamster ovary (CHO) cells, are used industrially for the production of biopharmaceuticals. Despite improvements over the last few decades with regard to the maximum cell concentrations obtainable and the amount of recombinant protein produced from such expression systems, there is still the potential to engineer and manipulate mammalian cells to generate more robust host cells that grow faster, for longer and produce higher product yields with less heterogeneity. A number of recent studies have reported that mRNA translation is a key limitation in terms of defining the product yield from in vitro cultured mammalian cells. One mechanism by which mRNA translation is controlled is via the expression of microRNAs (miRNA). miRNA are endogenous single stranded non-coding pieces of RNA of approximately 22 nt in length that can either stall mRNA translation or activate cleavage of target mRNAs. This study set out to investigate the role of specific miRNAs on determining CHO cell growth and recombinant protein productivity. Initially, Lonza Biologics undertook an extensive miRNA profiling study using the mercury LNA microarray, and found that -56 miRNA were significantly differentially expressed during the course of a state-of-the-art fed-batch bioreactor culture. In this study we determined the profile of four target miRNA (miR-15b, miR- 16, miR-21 and miR-34c) throughout batch culture in model monoclonal antibody producing CHO cell lines and the effect of transiently over-expressing the corresponding pre-miRNA individually or together. The different miRNA were expressed differentially throughout batch culture, their profile correlating with either growth or productivity characteristics. There was no relationship between pri-microRNA amounts and the mature microRNA amounts observed in the model cell lines investigated. Overexpressing/ knocking-up of the human pre-microRNAs in CHO cells resulted in an increase in the observed mature microRNA amounts. In some cases this over-expression resulted in increased recombinant protein expression and enhanced cell specific productivity when transiently expressed in model cB72.3 monoclonal antibody producing cells. Thus, the transient over-expression of pre-microRNAs resulted in increased product yields and/or changes in growth characteristics. In particular the over-expression of microRNA-34c resulted in an increase in transgene expression. Interestingly, when multiple microRNA were over-expressed transiently together an additive or synergistic effect in growth or product titre was observed in a number of cases. This is one of the first such studies to investigate the effect of manipulating multiple miRNA on recombinant protein synthesis from mammalian cells. When individual pre-microRNAs were stably expressed the amount of mature micro RNA observed in the resulting stable pools was increased. However, the amount of mature microRNA observed could be further increased by transient transfection of the stable pools with a plasmid for the expression of the target pre-microRNA suggesting that the maturation machinery for the generation of mature microRNA was not saturated. Ultimately this work has provided an improved understanding of specific microRNAs in CHO cells and their subsequent effect on industrially relevant phenotypes, providing informed strategies to manipulate microRNAs in order

An increasing number of RNA targeted gene repression mechanisms are being discovered operating in mammalian cells. Antisense-RNA mediated silencing of CpG island-associated genes is not limited to imprinting and X inactivation, but can occur at autosomal non-imprinted loci leading to disease. An antisense silencing phenomenon observed in a patient with an inherited form of anaemia was recently recreated in differentiating mouse embryonic stem cells, demonstrating that the silencing and methylation of the oppositely transcribed tissue specific alpha globin CpG island exclusively occurs in the presence of antisense RNA. The aim of this thesis is to use differentiating mouse embryonic stem cells to further characterise this silencing mechanism through analysis of the histone modifications associated with repression of the alpha globin gene upon differentiation, and to further define the developmental period in which repression is established. Also, the requirement for antisense RNA transcription through the alpha globin gene is examined through insertion of a transcriptional terminator to truncate antisense RNA transcription. The gene silencing mechanism is tested with other genes in place of the alpha globin gene to investigate whether repression is limited to the alpha globin gene or is also applicable to other tissue specific or ubiquitously expressed genes. Analysis of the tissue specific gene MYOD and the ubiquitously expressed gene UBC demonstrated that the silencing mechanism could effect other genes in a similar manner, though not all genes were susceptible to repression as the ubiquitously expressed gene ACTB remained unsilenced by antisense RNA expression. These observations establish that this silencing mechanism can play a more general role in repression, thereby suggesting its potential role as a constituent of the multivariate repressive pathways within the mammalian cell.

At the imprinted mouse Gnas locus, expression of Nesp in the growing oocyte is required for methylation of the downstream Nespas-Gnasxl DMR and ExonlA DMR; resulting in silencing of Gnasxl and ExonlA, and expression of Gnas on the maternal allele. On the paternal allele, the somatic Nesp promoter is silenced and Gnasxl and Gnas show the opposite pattern of expression. Using mutants in which Nesp is ectopically expressed on the paternal allele, I show that Nesp expression correlates with gain of DNA methylation at the ExonlA DMR, and with de-repression of Gnas; just as it does on the maternal allele. This DNA methylation is acquired post-fertilisation on the paternal allele, as opposed to the maternal allele where the ExoniA DMR is methylated in the growing oocyte. My subsequent work was focussed on understanding how Nesp expression is silenced on the paternal allele. I investigated if RNA interference is involved in silencing of Nesp, the developmental profile of expression of transcripts at the Gnas cluster, and the acquisition of DNA methylation and histone marks at the Nesp DMR. Nesp is expressed predominantly from the maternal allele at 6.5dpc, before the Nesp DMR is fully methylated on the paternal allele. The maternally inherited Nesp DMR is tri-methylated at H3K4 and H3K27 between the ICM and the 10.5dpc stages of development. Preferential enrichment of H3K4me3 and H3K27me3 at the maternally inherited Nesp DMR may playa role in establishing monoallelic expression of Nesp. Expression of Nespas, a noncoding RNA on the paternal allele protects the Nesp DMR from gaining H3K4me3 and H3K27me3 marks in cis. My work supports an emerging model of establishment of imprinted expression at the Gnas cluster, in which imprinting is regulated by concerted expression of two RNAs: a noncoding RNA (Nespas) on the paternal allele and, a protein coding RNA (Nesp) on the maternal allele.

Human haemoglobin is made up of two types of proteins called alpha and beta-globin. The β-Globin locus, located on chromosome 11, is 100kb in length and comprises five globin genes (HBE, HBG1, HBG2, HBBPl [a pseudogene], HBD and HBB). These genes are arranged in the same order as their expression occurs during the course of development from an embryo to neonate. However the mechanism that controls the switching between globin genes during development is unknown. A zinc finger transcription factor called Bcllla is known to reduce the levels of HbF but it is unclear how this effect is ... coding RNA's (ncRNA), such as XIST, HOTAIR and HOTTIP ... been shown to control the activity of genes by forming complexes with proteins that either increase or decrease gene expression. Bcllla does not bind directly to any of the genes in the β-globin locus, but it does appear to bind to a region called BGL3, situated between HBGl and HBBP1, which encodes a non-coding RNA. This raises the question of whether the BGL3 transcript plays a role in globin switching. The work presented in this thesis aims to characterise this ncRNA and test whether it plays a direct functional role in Bcllla-mediated HbF repression. siRNA-mediated knockdowns of BGL3 in an erythroid cell line, K562, led to a statistically significant reduction in the steady state levels of BGL3 (0.0053), HBG2 (0.0314) and the pseudogene HBBPl (0.0003). Over-expression of Bcllla in the same cell system led to a dramatic reduction in BGL3, HBG2 and HBBP1. Based on these data a new model of HbF to Hba switching is proposed: In the HbF stage BGL3 is expressed, recruiting chromatin modifying complexes, which activate the HBG1, HBG2 and HBBPl genes. During the switch to Hba, Bcllla

The process of pre-mRNA splicing involves the removal of intronic sequences from the pre-mRNA and it is directed by intronic cis acting elements know as the 5' and 3' splice sites that mark the boundaries of the exons. Over the two decades, however, it has become clear that exons encode for auxiliary splicing signals that either enhance or perturb their inclusion in the final mRNA product. It is possible that the evolution of mRNA sequences could be conditioned by the presence of these exonic cis-acting splicing regulatory elements and not mainly by the selection of optimal protein function. To explore this hypothesis, I have investigated how the need for ESE influences the gene evolution of a paralogous gene family, specifically the human Alkaline Phosphatases (ALPs). In this work, I have identified in correspondence to a weak 3'splice site, two ESE sequences in the placental ALP exon 4, and demonstrate that the ESE are necessary for the exon inclusion in the mRNA due to the weak 3 'splice sites. Furthermore, I show that they are absent in the corresponding exon of the non-tissue specific ALP transcript, specifically exon 5 that carries a strong 3' splice site. Most importantly, the localization of the ESEs correspond to an area that in the paralogous non-tissue specific ALP gene differs in amino , acid composition with respect, not only to the placental ALP where I mapped the ESEs but also to the other members of the family, where this area is well conserved. These amino acid changes may represent a possible evolutionary constraint on enzymatic activity, in keeping with this hypothesis, substituting the amino acids in the region of the ESE for those of the paralogous non-tissue specific ALP gene increases the enzymatic activity. Thus splicing-related constraints challenge the primacy of biochemical function in rates of protein evolution.

Splicing and pri-miRNA cropping events occur co-transcriptionally, but the precise relationship between these two processes on pre-mRNA is not completely understood. I have studied the relationship between the spliceosome and the Microprocessor complex (MPC) activities in a novel class of miRNAs, named Splice site Overlapping (SO) miRNAs, whose hairpins overlap with splice sites. I focused on the evolutionarily conserved SO miR-34b, whose hairpin is in a non-coding transcript, juxtaposed to an acceptor splice site. In vivo, the transcript has a tissue specific, evolutionarily conserved pattern of alternative splicing, with variable amounts of the intron retention isoform, and different quantities of miR-34b are synthesised in the tissues. In minigene systems, I identified two indispensable elements for the recognition of the SO miR•34b 3' splice site: a branch point located in the hairpin and a downstream purine-rich enhancer (ESE). Splicing inhibition due to ESE or AG 3' splice site mutations increase miR-34b levels. Moreover, siRNA-mediated knock down of Drosha and DGCR8 improves splicing efficiency and abolishes miR-34b biosynthesis. Thus, the processing of 3' SO miR- 34b is regulated in an antagonistic manner by the MPC and the spliceosome due to competition between these two machineries on the nascent transcript. This novel competitive mechanism that discriminates between juxtaposed splice sites and pri-miRNA structures may be commonly used to regulate the relative amount of SO mi RNAs and mRNAs produced from a precursor RNA transcript.

The beta adducin gene has a very complex architecture, with tissue-specific use of promoters and polyadenylation sites (PAS). Its expression is restricted to neuronal and hematopoietic tissues. In mice, usage of a distal brain-specific PAS (PAS4) generates an 8.3 kb long mRNA with an unusually long 3'UTR of about 5.7 kb. Instead, the use of proximal erythroid-specific PASs results in short 3.1-3.7 kb mRNA isoforms. The presence and use of the different alternative PASs in the beta adducin gene are very well conserved among species suggesting an important and specific role for each of the generated 3'UTRs. To study the regulatory mechanisms involved in beta adducin polyadenylation, minigene constructs containing the mouse beta adducin polyadenylation signals were used. The constructs were transfected into HeLa cells and their expression and PAS selection determined by Northern blot analysis. The exclusive usage of the PAS4 in HeLa cells has been observed. All the elements defining the core polyadenylation signal were characterized: the hexanucleotide motif (Hm), the upstream and downstream sequence elements (USE and DSE, respectively) elements, and the cleavage site. Interestingly, I have detected the presence of two novel non-canonical cis-acting elements regulating 3'end processing at the PAS4. Both elements act on long-distance and to the best of our knowledge, long-distance upstream polyadenylation regulatory elements have not previously been described for the non-viral eukaryotic transcripts. The first of these elements was essential to enable polyadenylation at the PAS4. It was located in a region spanning 355 nucleotides and includes the stop codon of beta adducin. The second non-canonical upstream polyadenylation regulatory element seems to inhibit processing at the PAS4. It was located in the region close to the second proximal PAS, about 4.5 kb upstream of the PAS4. These results highlight the complexity of the regulatory mechanisms directing beta adducin pre-mRNA 3'end processing.