Samuel E. Butcher

TDP-43 is an RNA binding protein implicated in neurodegenerative disease. TDP-43 binds to GU dinucleotide repeats, which are highly abundant sequences in human RNA. Here we show TDP-43 has one of the highest affinities and specificities measured for an RNA binding protein. Binding prevents formation of the pUG fold, an intramolecular quadruplex, and conversely pUG fold formation prevents TDP-43 binding. Slow pUG folding and a rapid on-rate enable TDP-43 to capture single-stranded RNA. The protein recognizes the RNA as a 1D lattice, in which overlapping binding sites produce efficient initial binding events that interfere with subsequent interactions. This effect is overcome by RNA facilitated protein-protein interactions, which serve to increase the on-rate of a second TDP-43 molecule, and an auto-inhibitory interaction that increases the off-rate. These data reveal how TDP-43 recognizes GU repeats and identify an interplay between RNA folding and protein recognition that may be relevant to human disease.

Poly(UG) repeats or pUG RNAs can fold into a left-handed parallel quadruplex, the pUG fold. The pUG fold directs the epigenetic amplification of RNAi in C. elegans, and pUG sequences are abundant in eukaryotic transcriptomes. Here, we report the sequence and ionic requirements for pUG folding. The pUG fold preferentially incorporates 12 guanosines but has an otherwise flexible sequence requirement. The uridines can be substituted with other nucleotides, with some sequence variants folding better than pUG RNA. The GA repeat sequence (GA)₁₂ also forms a pUG-like fold, albeit with lower thermodynamic stability than (GU)₁₂. The pUG fold can tolerate multiple deoxyribose substitutions but does not fold when the backbone is entirely deoxyribose. It has a high affinity and specificity for potassium ions (K_1/2 = 6 mM) and does not fold in sodium or ammonium. Addition of 2 mM Mg²⁺ does not further stabilize the pUG fold, and the polyamines spermine and spermidine do not affect its stability. Finally, the pUG fold is sensitive to surrounding sequence context and complementary flanking sequences can stabilize pUG folds, while other sequences can interfere with folding. These data provide an improved ability to understand and predict pUG folds.

TDP-43 is an RNA binding protein implicated in neurodegenerative disease. TDP-43 binds to GU dinucleotide repeats, which are highly abundant sequences in human RNA. Here we show TDP- 43 has one of the highest affinities and specificities measured for an RNA binding protein. Binding prevents formation of the pUG fold, an intramolecular quadruplex, and conversely pUG fold formation prevents TDP-43 binding. A rapid on-rate allows TDP-43 to capture single stranded RNA prior to folding. The protein recognizes the RNA as a 1D lattice, in which overlapping binding sites produce efficient initial binding events that interfere with subsequent interactions. This effect is partially overcome by RNA facilitated protein-protein interactions, which serve to increase the on-rate of a second TDP-43 molecule. In conjunction with all atom models, these data reveal how TDP-43 recognizes RNA repeat sequences and identify an interplay between RNA folding and protein recognition that may be relevant to human disease.

Poly(UG) or “pUG” repeat RNA can fold into a left-handed parallel quadruplex, the pUG fold. The pUG fold directs the amplification of RNAi in C. elegans , and pUG sequences are abundant in eukaryotic transcriptomes. Here, we report the sequence and ionic requirements for pUG folding. The pUG fold requires 12 guanosines but has an otherwise flexible sequence requirement. The uridines can be substituted with other nucleotides, with some sequence variants folding better than pUG RNA. The GA repeat sequence (GA) ₁₂ also forms a pUG-like fold, albeit with lower thermodynamic stability than (GU) ₁₂ . The pUG fold can tolerate multiple deoxyribose substitutions but does not fold when the backbone is entirely deoxyribose. It has a high affinity and specificity for potassium ions (K _1/2 = 6 mM) and does not fold in sodium or ammonium ions. Addition of 2 mM Mg ²⁺ does not further stabilize the pUG fold, and the polyamines spermine and spermidine slightly decrease its stability. Finally, the pUG fold is sensitive to surrounding sequence context and complementary flanking sequences can stabilize pUG folds, while other sequences can interfere with folding. These data expand our understanding of RNA folding and provide a basis for understanding and predicting pUG folds.

Long poly(UG) or “pUG” dinucleotide repeats are abundant in eukaryotic transcriptomes. Over a thousand human genes have pUGs longer than 24 repeats, including the cancer-associated lncRNA NEAT1. In C. elegans, enzymatic addition of long pUGs to RNA 3′ ends (pUG tails) marks RNAs as vectors of gene silencing. Gene silencing requires at least one pUG fold, a left-handed quadruplex structure that incorporates 12 repeats, but longer pUG tails are more effective. Here, we investigate the structure, folding kinetics and dynamics of long pUG RNAs. RNAs with 24 or more repeats slowly form compact, double pUG folds. The forward rate of pUG fold formation in vitro is length-dependent with a half-life (t1/2) of 13 min or more, while the unfolding rate is very slow (t1/2 ∼5 days). Long pUG RNAs display biphasic dynamics with an additional, faster unfolding phase (t1/2 ∼30 min). The amplitude of the faster phase indicates partial unfolding. From these data we propose a dynamic model for segmental register exchange and double pUG fold formation in long pUG RNAs. These data broaden our understanding of the structure and dynamics of long pUG RNAs and have implications for understanding the roles of pUG folds in biology and disease.

Long poly(UG) or “pUG” dinucleotide repeats are abundant in eukaryotic transcriptomes. Thousands of human genes have pUGs longer than 24 repeats, including the cancer-associated lncRNA NEAT1. In C. elegans , enzymatic addition of long pUGs to RNA 3′ ends (pUG tails) marks RNAs as vectors of gene silencing. Gene silencing requires at least one pUG fold, a left-handed quadruplex structure that incorporates 12 repeats, but longer pUG tails are more active. Here, we investigate the structure, folding kinetics and dynamics of long pUG RNAs. RNAs with 24 or more repeats slowly form compact, double pUG folds. The forward rate of pUG fold formation in vitro is length-dependent with a half-life (t _1/2 ) of 13 minutes, while the unfolding rate is very slow (t _1/2 ∼5 days). Long pUG RNAs display biphasic dynamics with an additional, faster unfolding phase (t _1/2 ∼30 min). The amplitude of the faster phase indicates partial unfolding. From these data we propose a dynamic model for segmental register exchange and double pUG fold formation in long pUG RNAs. These data broaden our understanding of the structure and dynamics of long pUG RNAs and have implications for understanding the roles of pUG folds in biology and disease.

In eukaryotes, the process of intron removal from nuclear pre-mRNA is performed by the spliceosome, a dynamic molecular machine composed of small nuclear ribonucleoproteins (snRNPs; U1, U2, U4, U5, and U6) and dozens of other protein splicing factors. The U6 snRNP contains the U6 snRNA and the proteins Prp24 and Lsm2-8 heteroheptamer. A key feature of the snRNP is a modified U6 snRNA 3’ end, which in S. cerevisiae (yeast) contains a 3’ phosphate. U6 plays an essential role in splicing, and the U6 snRNP must be completely disassembled for splicing to occur. Once splicing is finished, the snRNP must then be reassembled to participate in a subsequent splicing reaction. While splicing efficiency depends on rapid U6 snRNP assembly, this process has not yet been kinetically characterized. Here, we use colocalization single molecule spectroscopy (CoSMoS) to dissect the kinetic pathways of yeast U6 snRNA association with the Lsm2-8 complex and their dependence on the Prp24 protein and post-transcriptional snRNA modification. In the absence of 3’ end processing, Lsm2-8 association with the RNA is highly dependent on Prp24. However, processed RNAs can rapidly recruit Lsm2-8 in Prp24’s absence. Post-transcriptional processing facilitates Lsm2-8 association while the presence of Prp24 promotes both recruitment and retention of the complex. This suggests that efficient U6 snRNP assembly could depend on kinetic selection of Lsm2-8 binding to 3’-end modified or Prp24 bound U6 snRNAs in order to discriminate against association with other RNAs.

Poly(UG) or ‘pUG’ dinucleotide repeats are highly abundant sequences in eukaryotic RNAs. In Caenorhabditis elegans, pUGs are added to RNA 3′ ends to direct gene silencing within Mutator foci, a germ granule condensate. Here, we show that pUG RNAs efficiently self-assemble into gel condensates through quadruplex (G4) interactions. Short pUG sequences form right-handed intermolecular G4s (pUG G4s), while longer pUGs form left-handed intramolecular G4s (pUG folds). We determined a 1.05 Å crystal structure of an intermolecular pUG G4, which reveals an eight stranded G4 dimer involving 48 nucleotides, 7 different G and U quartet conformations, 7 coordinated potassium ions, 8 sodium ions and a buried water molecule. A comparison of the intermolecular pUG G4 and intramolecular pUG fold structures provides insights into the molecular basis for G4 handedness and illustrates how a simple dinucleotide repeat sequence can form complex structures with diverse topologies.