Gene expression is the conduit by which genetic information is connected into cellular phenotypes. Recently, it was shown that gene expression in mammalian cells is governed, at least in part, by the expression of short double stranded RNA (dsRNA). This mode of gene regulation is influenced by a large group of dsRNA binding proteins that could either stabilize or trigger the degradation of dsRNA. Indeed, double stranded RNA (dsRNA) specific ribonucleases (RNases) play an important role in regulating gene expression. In most eukaryotes, members of the dsRNA specific RNase III family trigger RNA degradation and initiate cellular immune response. Disruption of human . RNase III (Dicer) deregulates fetal gene expression and promotes the development of cancer. However, very little is known about the housekeeping function of eukaryotic RNase III and the mechanism by which they distinguish between exogenous and endogenous cellular RNA species. This thesis elucidates how dsRNAs are selected for cleavage and demonstrates their contribution to RNA metabolism in yeast as model eukaryote. Initially, the reactivity determinants of yeast RNase III (Rnt1p) were identified in vitro and used to study the global impact of Rnt1p on the processing of non-coding RNA. The results indicate that Rnt1p is required for the processing of all small nucleolar RNAs (snoRNAs) involved in rRNA methylation and identify a new role of Rnt1p in the processing of intronic snoRNAs. It was shown that Rnt1p cleavage helps to coordinate the expression of some ribosomal protein genes hosting intronic snoRNAs. Direct snoRNA processing from the pre-mRNA blocks the expression of the host gene, while delayed snoRNA processing from the excised intron allows the expression of both genes. In this way, the cell can carefully calibrate the amount of snoRNA and ribosomal proteins required for ribosome biogenesis. In addition, a global analysis of snoRNA processing identified new forms of Rnt1p cleavage signals that do not exhibit a conserved sequence motif but instead use a new RNA fold to recruit the enzyme to the cleavage site. This finding led to the conclusion that Rnt1p may use a wide combination of structural motifs to identify its substrates and thus increases the theoretical number of potential degradation targets in vivo . To evaluate this possibility, a new search for snoRNA independent Rnt1p cleavage targets was performed. Interestingly, many Rnt1p cleavage signals were identified in intergenic regions devoid of known RNA transcripts. In vivo , it was shown that Rnt1p induce the termination of non-polyadenylated transcripts and functions as a surveillance mechanism for transcription read-through. This finding directly links Rnt1p to the transcription machinery and provides a new mechanism for polyadenylation independent transcription termination. Together the work described in this thesis presents an example of how eukaryotic RNase III may identify its substrates and present a case study where transcription, RNA processing and stability are linked.

Poly(A) tail addition to pre-mRNAs is a highly coordinated and essential step in mRNA maturation involving multiple cis- and trans-acting factors. The trans-acting factor, poly(A) polymerase (PAP) plays an essential role in the polyadenylation of mRNA precursors. The Arabidopsis thaliana genome contains four putative PAP genes. We have found, using in silico analysis and transgenic plants expressing GUS under the control of the four PAP promoters, that each of these genes is expressed in overlapping, yet unique patterns. This gives rise to the possibility that these genes are not redundant and may be essential for plant survival. To further test this, inducible RNAi and T-DNA mutagenized plants were obtained and analyzed. Plants lacking all, or most, of each PAP gene product, due to RNAi induction, were not viable at any of the stages of plant growth tested. Furthermore, T-DNA PCR analysis determined that no plants containing a homozygous mutation, were viable. This data reveals that lack of any of the four PAP gene products has a significant effect on the plants ability of survive, thus indicating that each PAP gene is essential. Finally, transient expression experiments with each of the full length PAP cDNAs fused to GFP showed that the PAP I, PAP II and PAP IV gene products are localized throughout the nucleus and within nuclear speckles. The cellular localization of PAP III could not be determined.

Plant polyadenylation factors have proven difficult to purify and characterize, owing to the presence of excessive nuclease activity in plant nuclear extracts, thereby precluding the identification of polyadenylation signal-dependent processing and polyadenylation in crude extracts. As an alternative approach to identifying such factors, a screen was conducted for activities that inhibit the non-specific activity of plant poly(A) polymerases (PAP). One such factor (termed here as Putative Polyadenylation Factor B, or PPF-B) was identified in a screen of DEAE-Sepharose column fractions using a partially purified preparation of a plant nuclear poly(A) polymerase. This factor was purified to near homogeneity. Surprisingly, in addition to being an effective inhibitor of the nuclear PAP, PPF-B inhibited the activity of a chloroplast PAP. In contrast, this factor stimulated the activity of the yeast PAP. Direct assays of ATPase, proteinase, and nuclease activities indicated that inhibition of PAP activity was not due to depletion of substrates or degradation of products of the PAP reaction. The major polypeptide component of PPF-B proved to be a novel linker histone (RSP), which copurified with inhibitory activity by affinity chromatography on DNA-cellulose. The association of inhibitory activity with a linker histone and the spectrum of inhibitory activity, raise interesting possibilities regarding the role of PPF-B in nuclear RNA metabolism. These include a link between DNA damage and polyadenylation, as well as a role for limiting the polyadenylation of stable RNAs in the nucleus and nucleolus. The Arabidopsis genome possesses genes encoding probable homologs of most of the polyadenylation subunits that have been identified in mammals and yeast. Two of these reside on chromosome III and V and have the potential to encode a protein that is related to the yeast and mammalian Fip1 subunit (AtFip1-III and AtFip1-V). These genes are universally expressed in Arabidopsis tissues. AtFip1-V stimulates the non-specific activity of at least one Arabidopsis nuclear PAP, binds RNA, and interacts with other polyadenylation homologs AtCstF77 and AtCPSF30. These studies suggest that AtFip1- V is an authentic polyadenylation factor that coordinates other subunits and plays a role in regulating the activityof PAP in plants.

No description available.

Cleavage and polyadenylation are essential pre-mRNA processing reactions maturing the 3’end of almost all protein encoding eukaryotic mRNAs. Analysis of the sequences required for cleavage and polyadenylation in the human melanocortin 4 receptor (MC4R) and the human transcription factors JUNB and JUND pre-mRNAs revealed that, at least for some mammalian genes, 3’end processing of the primary transcript is independent of previously described auxiliary sequence elements located upstream or downstream of the core poly(A) sequences. The analysis of the MC4R poly(A) site, contrary to the current understanding of mammalian poly(A) sites, showed that mutations of the AUUAAA hexamer sequence had no effect on 3’end processing levels while mutations in the short DSE severely reduced cleavage efficiency. The MC4R poly(A) site uses a potent DSE and to direct maximal cleavage efficiency requires only a short upstream adenosine rich sequence. Furthermore, analysis of the endogenous A-rich human JUNB poly(A) signal validated upstream A-rich core sequences as genuine 3’end formation directing sequences in human non-canonical 3’end formation signals. The results show that a minimal human poly(A) site, similar to yeast and plants, can be defined by an adenosine rich sequence adjacent to a U/GU-rich sequence element and a cleavage site. These findings further imply that some human non-canonical poly(A) sites may be recognised via a similar DSE-dependent mechanism and may not require additional auxiliary sequence elements. Finally, results on the analysis of the EDF1 poly(A) signal show that, in a spliced environment, A-rich sequences are also 3’end formation effectors but depend on an competent upstream splicing reaction for efficient definition of the 3’end processing site.