CELF1 is an RNA binding protein with regulatory roles in translation, alternative splicing and mRNA degradation. This protein is of particular interest as its upregulation is believed to be involved in the pathogenesis of type 1 myotonic dystrophy. CELF1 functions by binding to a specific sequence in the 3’ untranslated region of its target mRNAs. This sequence has been termed the “EDEN” motif, but the exact requirements for binding of CELF1 were not well defined. In this study we therefore aimed to determine the sequence requirements for an RNA substrate to form a high affinity interaction with CELF1, and characterise the structure of the resulting complex. The CELF1 protein is composed of three structured RNA recognition motifs separated by flexible linkers. Our strategy was to investigate the RNA binding properties of each domain in isolation, and then the requirements for tandem binding of the domains in order to build up the complete “EDEN” motif capable of forming a high affinity complex with the wild type protein. This has been accomplished using NMR spectroscopy to map the chemical shift perturbations in each domain on binding to a range of RNA substrates. ITC was also used to investigate the binding affinities of each domain, and the enhancement of affinity when domains bind in tandem. By these methods we have refined the sequence requirements for simultaneous binding of all domains of CELF1, and designed RNA substrates which will bind with higher affinity than any previously reported. We have also shown the potential involvement of RNA secondary structure in forming the CELF1 binding site, and identified two possible examples of this in natural mRNA targets. CELF1 binding triggers deadenylation of its target mRNAs and this is suspected to be via a mechanism involving recruitment of poly (A) ribonuclease. These two proteins have been shown to interact, but no structural information was available to show which domains were interacting, or whether CELF1 was capable of forming a ternary complex with both RNA and poly(A) ribonuclease. Since the ribonuclease exists as a 146 kDa dimer, the complex of it with CELF1 was an ambitious target for NMR. In this study we demonstrate that high resolution NMR data can be acquired on this key regulatory complex. Using this we go on to confirm the interaction between these two proteins, and that the domains involved in binding suggest a ternary complex is possible.

RNA silencing is a form of post-transcriptional gene regulation, in which a small RNA guides a member of the Argonaute protein family in an effector complex to repress target gene expression. piRNAs, found in germ cells, are the most recently discovered major subset of small RNAs. A key known function of piRNA is to repress the transposable elements in the germline and maintain the germline genome integrity. The defining features of the piRNAs are 1) they are ubiquitously methylated at the 3’-end of the 2’-OH group by methyltransferase Hen1; 2) they associate exclusively with the Piwi subfamily Argonaute proteins. Much is not understood about the biogenesis and the regulation of the piRNA pathway. One of the fundamental questions is how the 3’-end of the piRNA is generated and recognised specifically by Piwi but not by Ago subfamily Argonaute proteins. In this thesis, the high resolution crystal structure of the Aubergine PAZ domain, a domain from a Piwi subfamily Argonaute, bound to a 7 mer single-stranded methylated piRNA ‘mimic’, reveals the mode of recognition for the 3’-end of piRNAs by Piwi subfamily Argonautes. The structure provides the molecular basis for why Piwi but not Ago PAZ domains preferentially bind to RNAs with 2’-O-methylation at the 3’-end, thus conferring substrate specificity. The structural results are confirmed by biochemical studies. Biochemical and biophysical studies on the methyltransferase Hen1 have provided insights into substrate specificity for piRNA 3’-end methylation and revealed a potential regulatory role for the C-terminal region of the protein. Extensive biochemical analysis defined a minimal active Hen1/short RNA complex, though crystallisation screening yielded no crystals for structure determination. Overall, this study provides insights into the generation and molecular recognition of the piRNA in animals.

The RsmA/CsrA family of global post-transcriptional regulators are small RNA-binding proteins involved in the regulation of a large number of genes such as those involved in quorum sensing, virulence factor production, secondary metabolism, motility and biofilm formation. They bind to target mRNAs and hence modulate their stability and translation rates. Their effects are antagonised by small non-coding regulatory RNAs. The control of expression of target genes via this post-transcriptional regulatory network is mostly operated in Pseudomonas spp. via the GacS/GacA two component system. This study aimed to perform a biophysical analysis of RsmA and to obtain a preliminary understanding of the structure, function and regulation of RsmN, a new atypical RsmA homologue from Pseudomonas aeruginosa. RsmA was purified and biophysical analysis confirmed that RsmA exists as a dimer and is highly stable at high temperatures (75 °C) and low pH (5.2). Although RsmN was found to be structurally similar to RsmA, no functional phenotypes have been identified. Consequently, rsmN was mutated and transcriptional fusions to rsmN and its anti-sense transcript were constructed for expression studies. Phenotypic analysis indicated that RsmN was not involved in the control of swarming, pyocyanin, elastase and protease production or glycogen accumulation. Unlike RsmA, RsmN does not have a control on the restriction modification system of P. aeruginosa. Transcriptional fusions revealed RetS, LadS and GacA all appear to have a 21 significant effect as activators of both the rsmN and nmsR promoters. 2-Alkyl-4(1H)-quinolone (AQ) signalling also modulate rsmN expression possibly via the iron chelating properties of 2-alkyl-3-hydroxy-4(1H)-heptyl-quinolone (PQS). RsmN targets identified from Deep Sequencing include those required for structural outer membrane proteins, transcriptional regulators as well as genes involved in motility, secretion, flagellar structure and biofilms. RsmA, RsmZ and RsmY were all identified as targets together with the small RNAs RgsA (indirectly gac-controlled) and the antagonistic RNA CrcZ (represses catabolite repression control protein Crc). Targets common to both RsmN and RsmA include the transcriptional regulators Vfr, PqsR, MvaT and Anr, regulatory RNAs RsmZ and RsmY together with transcripts corresponding to the pqsABCDE operon, LasB, LecA/B, RhlI, LasR/I, Crc and CrcZ. The identification of many mRNA targets for RsmN which are shared with targets of RsmA provides further evidence that RsmN is involved in global-post-transcriptional regulation of gene expression.

The aims of this thesis are to design and synthesize non-viral cationic lipid vectors based on spermine, for the intracellular delivery of siRNA (short interfering RNA) and the subsequent siRNA mediated gene silencing. Two parameters were varied: the type of fatty acid and the cationic head-group. Among the symmetrical spermine conjugates, N4,N9-dierucoyl spermine (DES) resulted in higher siRNA delivery compared to N4,N9-dioleoyl spermine (DOS), while enhanced green fluorescent protein (EGFP) silencing in HeLa cells showed that the unsaturated fatty acid conjugates are more efficient than the saturated fatty acid ones, and cell viability was 75%-85% for conjugates with chain length ≥ 18. Two cationic lipids with guanidine head-groups, N1,N12-diamidino-N4,N9-dioleoylspermine and N1,N12-diamidino-N4-linoleoyl-N9-oleoylspermine, were more efficient in EGFP gene silencing compared to cationic lipids with shorter C12 (lauroyl) and very long C22 (erucoyl) chains, with cell viability (64%-83% for chain length ≥ 18). Changing the cationic headgroup to guanidine did not offer a significant advantage in gene silencing over the conjugates with terminal primary amine groups. The asymmetrical N4-linoleoyl-N9-oleoyl-1,12-diamino-4,9-diazadodecane (LinOS) resulted in the best gene silencing, while LigOS (with one lignoceroyl 24:0 chain) resulted in the best siRNA delivery. Conjugates with two unsaturated fatty chains generally resulted in better EGFP gene silencing, while conjugates with one saturated chain and one unsaturated chain resulted in better siRNA delivery. Increasing the chain length also resulted in increased siRNA delivery (cell viabilities of asymmetrical > 74%, LinOS 88%). siRNA lipoplexes prepared using mixtures of LinOS with either cholesterol or DOPE (1,2-dioleoyl-sn-glycero-3-phosphoethanolamine) resulted in increased siRNA delivery, and enhanced EGFP silencing, with LinOS/Chol mixture (1:2 molar ratio) resulting in the highest siRNA delivery and the best gene silencing (EGFP reduced to 20%). Temperature studies of intracellular entry showed that the majority of lipoplexes are internalized by endocytosis, however the majority of gene-silencing occurs due to lipoplexes internalized via another mechanism.

An introductory chapter discusses the steric block, RNase H and RNA interference antisense mechanisms and the application of antisense nucleic acids as therapeutic agents. Examples of existing chemical modifications of the sugar and backbone regions of nucleic acids are given, followed by the introduction of the vinylphosphonate modification. The vinylphosphonate has previously been examined in DNA and has been synthesised by either Pd(0) catalysed cross-coupling of an H-phosphonate with a vinyl bromide, or by the cross-metathesis of a vinylphosphonate with a terminal olefin. This thesis details the first examples of the vinylphosphonate modification in RNA. The initial aim of this project was the synthesis of a range of nucleosides where the 5'-C-O was replaced by a vinyl bromide carbon-carbon double bond. Starting from alpha-D-glucose, acid catalysed formation of the 1,2:5,6-diisopropylidene alpha-D-glucofuranose was carried out followed by protection of the 3-OH as an acetate. The 5,6-isopropylidene was then subjected to H5IO6 mediated one-pot hydrolysis-oxidative cleavage to obtain the 5-aldehyde. Wittig olefination using CBr4 and Ph3P led to the dibromo olefin which was then stereoselectively reduced using dimethyl phosphite and diisopropylamine to obtain the pure trans-vinyl bromide. Following hydrolysis of the acetate, the stereochemistry of the 3-OH was then inverted by sequential oxidation and reduction. With the correct stereochemistry, the 3-OH was protected as the 2-methylnaphthyl ether. The 1,2-isopropylidene moiety was then hydrolysed and acetylated to the bis-acetate which was subjected to Vorbruggen conditions obtaining the uridine (93%), adenosine (77%), cytidine (30) and guanosine (63%) vinyl bromide nucleosides. The 2'-OAc of the nucleosides were hydrolysed to the 2'-OH in yields of 74-92%. The uridine 2'-OH was protected as the 2'-OTBS ether (98%), analogous to the commercially available phosphoramidites used in automated oligonucleotide synthesis. Similarly, the adenosine and uridine nucleosides could also be blocked as the 2'-OMe (59% and 73% respectively). In the case of the uridine vinyl bromide, the 3'-O-(2-methylnaphthyl) protecting group was cleaved using DDQ, this then enabled the vinylphosphonate-linked uridine dinucleotides to be functionalised at the 3'-OH as the cyanoethyl phosphoramidite using N,N-diisopropyl-2-cyanoethyl-chlorophosphoramidite, DIPEA and DMAP in dichloromethane (2'-OTBS 74%, 2'-OMe 41%). These could then be used in automated solid phase oligonucleotide synthesis. The H-phosphonates were prepared in a single step form the commercially available phosphoramidites using a tetrazole. These were then coupled to the vinyl bromide nucleosides using standard conditions of Pd(OAc)2 (0.2 eq.), dppf (0.4 eq.) and propylene oxide (20 eq.) in THF at 70 oC in a sealed vial for 6 hours. A range of vinylphosphonate-linked dinucleotides were accessed in yields of 61-99%. A detailed experimental section at the end of this thesis describes the procedures used in the synthesis and the analysis of the structures obtained.

RNA aptamers with 2'-fluoro-pyrimidine chemistry were previously selected by in vitro evolution to bind to monomeric HIV-1 gp120 from the R5 strain BaL. A group of 36 novel aptamers were cloned and sequenced from the heterogeneous pool and were tested for their ability to bind to gp120. The diversity of the RNA secondary structure of these, and 27 aptamers isolated previously, was analysed using a bioinformatics approach. This showed that eight aptamers contain a common branched motif, and RNA mutagenesis indicated that this structure is probably required for gp120 binding. Chemically synthesised derivatives of one such aptamer, B40, were designed and tested for binding to gp120. Truncation was found to decrease their binding, but the introduction of point mutations to stabilise the branched conformation and 2'-O-dimethylallyl-modified residues to stabilise helices increased binding to levels greater than that of the parental aptamer. The aptamer epitope on gp120 was mapped by testing aptamer binding to alanine-scanning mutants and deletion mutants of gp120 using a novel plate-based assay. This study showed that the aptamer binding site overlaps with the CCR5 epitope and is confined to four key residues at the base of the V3 loop, one of which is highly conserved. This finding may account for the observation that a number of aptamers were shown previously to neutralise a range of HIV-1 R5 clinical isolates in PBMC cultures. Interestingly however, the aptamer was unable to neutralise HIV-1 pseudovirus in a cell line, which is most likely due to the increased levels of cell-surface CCR5 in cell lines compared to PBMC. Future work should focus on identifying the structure and epitopes of other anti-gp120 aptamers as well as testing neutralisation of HIV-1, HIV-2 and SIV by the B40-derived aptamers. These aptamers can be used as tools to investigate the HIV-1 entry pathway and also have the potential to be developed as anti-HIV-1 microbicides.

Recent work has demonstrated that microRNAs (miRNAs) are involved in critical biological processes by suppressing the translation of coding genes. This work develops an integrated database, miRNAMap, to store the known miRNA genes, the putative miRNA genes, the known miRNA targets and the putative miRNAtargets. The knownmiRNAgenes in fourmammalian genomes such as human, mouse, rat and dog are obtained from miRBase, and experimentally validated miRNA targets are identified in a survey of the literature. Putative miRNA precursors were identified by RNAz, which is a non-coding RNA prediction tool based oncomparative sequence analysis. The mature miRNA of the putative miRNA genes is accurately determined using a machine learning approach, mmiRNA. Then, miRanda was applied to predict the miRNAtargets within the conserved regions in 30-UTR of the genes in the four mammalian genomes. The miRNAMap also provides the expression profiles of the known miRNAs, cross-species comparisons, gene annotations and cross-links to other biological databases. Both textual and graphical web interface are provided to facilitate the retrieval of data from the miRNAMap.

Background: MicroRNAs have been identified as crucial regulators in both animals and plants.Here we report on a comprehensive comparative study of all known miRNA families in animals.We expand the MicroRNA Registry 6.0 by more than 1000 new homologs of miRNA precursorswhose expression has been verified in at least one species. Using this uniform data basis we analyzetheir evolutionary history in terms of individual gene phylogenies and in terms of preservation ofgenomic nearness across species. This allows us to reliably identify microRNA clusters that arederived from a common transcript. Results: We identify three episodes of microRNA innovation that correspond to majordevelopmental innovations: A class of about 20 miRNAs is common to protostomes anddeuterostomes and might be related to the advent of bilaterians. A second large wave ofinnovations maps to the branch leading to the vertebrates. The third significant outburst of miRNAinnovation coincides with placental (eutherian) mammals. In addition, we observe the expectedexpansion of the microRNA inventory due to genome duplications in early vertebrates and in anancestral teleost. The non-local duplications in the vertebrate ancestor are predated by local(tandem) duplications leading to the formation of about a dozen ancient microRNA clusters. Conclusion: Our results suggest that microRNA innovation is an ongoing process. Majorexpansions of the metazoan miRNA repertoire coincide with the advent of bilaterians, vertebrates,and (placental) mammals.

Transposable elements account for the almost half of the sequence encoded by mammalian genomes, which become silenced during early embryonic development. This thesis sought to explore the hypothesis of the involvement of the RNAi pathway in the silencing of transposable elements in mammals, predominantly through the identification of transposon-associated RNA of -20-25nt using a gel blotting technique. Initially cell lines of embryonic and tumour origin were analysed. This lead to the identification of several previously unreported transposon-associated RNA ranging from 70-90nt. However, it was not until a more detailed analysis of the embryonic cell lines, with the induction of differentiation in cell culture that several discrete RNA of -20nt were detected for the mouse transposons L1 and B2. The differentiation of embryonic cell lines in culture also serendipitously lead to the detection of several short 55 rRNA of -22-26nt, these were also later detected in several human cell lines of breast cancer origin and healthy breast tissue. Intriguingly the -20-26nt repeat-associated identified were predominantly observed after two-days of differentiation in cell culture in several cell lines and often coincided with an ethidium bromide stainable band of -19nt. The latter may indicate a large proportion of these RNAs. Further analysis of the B2 and 55 rRNA repeat-associated short RNA revealed both to have reduce accumulation in Dicer-null embryonic stem cells, implicating a possible association with a known component of the RNAi pathway. Dicer was also observed to process the longer -50-80nt 55rRNA to -20-26nt in vitro. BLAST was also used to identify possible mRNA targets for the short B2 and 55 rRNA. One of these, the mRNA encoding sialic acid acetylesterase (SIAE) was consistently observed to be reduced with the accumulation of the short RNA using end-point RT-PCR, consistent with targeting through the RNAi or a similar pathway. However, no further links with the RNAi pathway were established, with no targeting detected for the short B2 or 55 rRNA using dual luciferase sensor assays. The repeat-associated RNA identified in this thesis are among the first of their type and further work will be required to establish what relevance they have to the RNAi pathway and transposon regulation.

The archaeal RNA polymerase (RNAP) is similar to the eukaryotic RNAP-II in terms of subunit composition and overall protein structure. Despite its similarity, a new archaeal-specific Rpo13 subunit has been identified. Rpo13 occupies a position in the enzyme which, in RNAP-II, is filled by the eukaryotic-specific Rpb5 jaw domain. It has therefore been proposed to contact DNA, where the positively charged C-terminal tail might mediate protein-DNA interactions. Furthermore, analysis of archaeal genomes has identified a homologue of the eukaryotic RNAP-III-specific RPC34 subunit. RPC34 may associate with the single archaeal RNAP, modulating the specificity of the archaeal RNAP and re-directing it to a subset of genes such as non-coding genes, in analogy to the RNAP-III/RPC34 eukaryotic system involved in the transcription of 5S rRNA, tRNAs and other small RNAs. More importantly, the association of RPC34 with the single archaeal RNAP would define an archaeal enzyme which acts as a precursor of the eukaryotic RNAP-III. Electrophoretic mobility shift assay (EMSA) analysis of purified Rpo13 protein by recombinant means subsequently incubated with a double-stranded (ds)DNA sequence reveals the formation of protein-DNA complexes, where Rpo13’s binding to DNA is non-sequence specific but discriminatory to dsDNA, as no binding is observed in the presence of single-stranded (ss)DNA. Also, it is found that the ma jor determinant of DNA binding is the Rpo13’s positively charged C-terminal tail, since DNA binding is abolished with a Rpo13 mutant deficient in this tail. Furthermore, neither ma jor groove nor minor groove interacting compounds have a major impact on Rpo13’s binding to DNA, suggesting that Rpo13 may associate with the negatively charged DNA phosphate backbone. Moreover, in vitro transcription assays indicate that a transcription product is observed upon RNAP incubation with a bubble DNA oligo shown to make Rpo13 contacts in the RNAP-DNA crystal structure. In addition, while a GST-pulldown experiment suggests the existence of an interaction between the archaeal RNAP and RCP34 in vitro, co-immunoprecipitation assays argue against the existence of such interaction from an in vivo point of view. Finally, a chromatin immunoprecipitation (ChIP)-sequencing approach to analyse Rpo13’s genomic distribution versus the one of the bulk RNAP was undertaken. While the gel filtration elution profile analysis of Rpo13 in the S. acidocaldarius cell extract versus the one of recombinant Rpo13 suggests that there is a free Rpo13 pool in the cell extract, indicating that Rpo13 may be acting as a transiently-associated RNAP subunit displaying a factor-like function, the ChIP-sequencing approach reveals that Rpo13 is a bona fide RNAP subunit since it co-localises from a genomic point of view with the bulk RNAP.