Separate functions of the Paf1 and Ski complexes in transcription and RNA decay

Each stage of the mRNA life cycle, from its transcription and processing in the nucleus, to its export into the cytoplasm where it is translated and eventually degraded, is subject to elaborate control. Although commonly viewed as discrete and independent events, the different stages of gene expression are often physically and functionally linked. The first chapter of this thesis is focused on the conserved Polymerase Associated Factor 1 complex (Paf1C), which regulates multiple steps of the RNA polymerase II transcription cycle as well as events downstream of transcript synthesis such as polyadenylation of mRNAs and export of nascent transcripts. Recent studies in the fission yeast Schizosaccharomyces pombe uncovered an additional role of Paf1C as an antagonist of RNAi-directed transcriptional gene silencing. In Paf1C mutant strains, ectopically expressed small RNAs (siRNAs) mediate gene repression by inducing de novo heterochromatin formation at the target locus in the nucleus. Although euchromatic loci in wild type cells seem refractory to this repressive mechanism, the induced silent gene state in Paf1C mutants can be inherited across generations even in the absence of the original siRNA source. In mammalian somatic cells, the existence of a nuclear RNAi pathway that can similarly modulate gene expression by modifying the underlying chromatin environment is highly debatable. However, the discovery of Paf1C as a negative regulator of this process in fission yeast raised the question of whether the complex performs similar repressive functions in higher organisms. Therefore, one of the goals of my PhD project was to test if in the absence of the Paf1 complex, siRNAs can initiate de novo heterochromatin formation in mammals, using mouse embryonic stem cells (mESCs) as a model system. However, by examining Paf1C conditional knock out cells expressing endogenous siRNAs, I did not find any supporting evidence for the existence of a nuclear RNAi silencing mechanism that can be regulated by the mammalian Paf1 complex.
Although Paf1C is conserved from yeast to humans, the complex in higher eukaryotes has an additional subunit, Ski8/Wdr61, which is also part of the helicase Ski complex functioning in cytoplasmic RNA decay. In human cells, Ski8/Wdr61 is thought to bridge the two complexes, suggesting that Paf1C might affect other stages of RNA processing or decay in the nucleus via its association with the Ski complex. Thus, another goal of my PhD project was to elucidate the functional role of the interaction between the Paf1 and Ski complexes in mammalian cells. In contrast to previous studies, my data suggest that in mESC the two complexes are not physically or functionally linked, even though they contain a common subunit. Further investigation of the mammalian Ski complex provided additional support for independent roles of the two complexes in transcription and cytoplasmic RNA decay.
The above findings became the basis for my second PhD project, described in the second chapter of the thesis. This was part of a collaborative effort examining the two major cytoplasmic mRNA degradation pathways in mES cells. In the cytoplasm, mRNAs can be degraded in the 5′-3′ direction by the exoribonuclease Xrn1 or in the 3′-5′ direction by the RNA exosome. The latter pathway requires the function of the Ski complex, comprising the scaffold protein Ski3/Ttc37, two copies of Ski8/Wdr61 and the RNA helicase Ski2/Skiv2l. The Ski complex is suggested to facilitate substrate passage through the exosome channel during 3′-5′ decay. RNA degradation has been extensively studied in the yeast Saccharomyces cerevisiae, where Xrn1 seems to be the predominant route for mRNA decay, whereas the 3′-5′ Ski-exosome pathway is thought to function redundantly with Xrn1, and contribute more significantly to RNA surveillance. In mammalian cells, the specific endogenous targets of the two pathways are poorly defined and it remains unclear if certain mRNAs can be preferentially degraded via one route, and if so, what factors could mediate such specificity. In addition, cytoplasmic RNA degradation is widely influenced by translation, consistent with recent cryo-EM structures capturing the yeast Ski complex or Xrn1 bound directly to translating ribosomes. It is currently unknown whether this physical link between the translation and RNA degradation machineries is conserved in mammals. Furthermore, it is unclear to what extent each decay pathway interacts with translation, and what factors might influence this process.
We combined crosslinking and analysis of cDNA (CRAC) with ribosome profiling in mES cells to examine the two major cytoplasmic RNA decay pathways and how they are linked to translation. Our approach allowed us to determine the direct transcriptome-wide RNA targets of Xrn1 and the Ski complex helicase Skiv2l in unperturbed cells and identify a subset of transcripts whose steady-state levels depend on the 3′-5′ pathway. Strikingly, although we found that both pathways are physically linked to translation, Skiv2l binding to RNA was exclusively dictated by ribosome occupancy and was heavily dependent on the translational status of the substrate. Our data reveal diverse triggers of RNA decay, including specific amino acid codons and RNA sequences that seem to impede ribosome elongation. We further identified a novel interaction between the Ski complex and the higher-eukaryote-specific RNA binding protein Aven. We showed that Aven and Skiv2l function closely to oppose aberrant translation, with Aven helping to prevent ribosome stalling at structured regions, while Skiv2l eliminates transcripts when these events accumulate. Interestingly, the Aven-Skiv2l pathway acts on a wide range of substrates, including mRNAs, uORFs and most surprisingly on small-ORF-containing RNAs derived from transcription of non-coding regions. As Aven is conserved from Drosophila to humans, this work uncovered a higher-eukaryote-specific pathway that coordinates cytoplasmic 3′-5′ RNA decay.