TDP-43 encodes an alternative-splicing regulator with tandem RNA-recognition motifs (RRMs). The protein regulates cystic fibrosis transmembrane regulator (CFTR) exon 9 splicing through binding to long UG-rich RNA sequences and is found in cytoplasmic inclusions of several neurodegenerative diseases. We solved the solution structure of the TDP-43 RRMs in complex with UG-rich RNA. Ten nucleotides are bound by both RRMs, and six are recognized sequence specifically. Among these, a central G interacts with both RRMs and stabilizes a new tandem RRM arrangement. Mutations that eliminate recognition of this key nucleotide or crucial inter-RRM interactions disrupt RNA binding and TDP-43-dependent splicing regulation. In contrast, point mutations that affect base-specific recognition in either RRM have weaker effects. Our findings reveal not only how TDP-43 recognizes UG repeats but also how RNA binding-dependent inter-RRM interactions are crucial for TDP-43 function.

Nuclear factor TDP-43 has been reported to play multiple roles in transcription, pre-mRNA splicing, mRNA stability and mRNA transport. From a structural point of view, TDP-43 is a member of the hnRNP protein family whose structure includes two RRM domains flanked by the N-terminus and C-terminal regions. Like many members of this family, the C-terminal region can interact with cellular factors and thus serve to modulate its function. Previously, we have described that TDP-43 binds to several members of the hnRNP A/B family through this region. In this work, we set up a coupled minigene/siRNA cellular system that allows us to obtain in vivo data to address the functional significance of TDP-43-recruited hnRNP complex formation. Using this method, we have finely mapped the interaction between TDP-43 and the hnRNP A2 protein to the region comprised between amino acid residues 321 and 366. Our results provide novel details of protein–protein interactions in splicing regulation. In addition, we provide further insight on TDP-43 functional properties, particularly the lack of effects, as seen with our assays, of the disease-associated mutations that fall within the TDP-43 321-366 region: Q331K, M337V and G348C.

Disease-causing splicing mutations described in the literature primarily produce changes in splice sites and, to a lesser extent, variations in exon-regulatory sequences such as the enhancer elements. The gene ATM is mutated in individuals with ataxia-telangiectasia; we have identified the aberrant inclusion of a cryptic exon of 65 bp in one affected individual with a deletion of four nucleotides (GTAA) in intron 20. The deletion is located 12 bp downstream and 53 bp upstream from the 5' and 3' ends of the cryptic exon, respectively. Through analysis of the splicing defect using a hybrid minigene system, we identified a new intron-splicing processing element (ISPE) complementary to U1 snRNA, the RNA component of the U1 small nuclear ribonucleoprotein (snRNP). This element mediates accurate intron processing and interacts specifically with U1 snRNP particles. The 4-nt deletion completely abolished this interaction, causing activation of the cryptic exon. On the basis of this analysis, we describe a new type of U1 snRNP binding site in an intron that is essential for accurate intron removal. Deletion of this sequence is directly involved in the splicing processing defect.

In monosymptomatic forms of cystic fibrosis such as congenital bilateral absence of vas deferens, variations in the TG m and T n polymorphic repeats at the 3 end of intron 8 of the cystic fibrosis transmembrane regulator (CFTR) gene are associated with the alternative splicing of exon 9, which results in a nonfunctional CFTR protein. Using a minigene model system, we have previously shown a direct relationship between the TG m T n polymorphism and exon 9 splicing. We have now evaluated the role of splicing factors in the regulation of the alternative splicing of this exon. Serine-arginine-rich proteins and the heterogeneous nuclear ribonucleoprotein A1 induced exon skipping in the human gene but not in its mouse counterpart. The effect of these proteins on exon 9 exclusion was strictly dependent on the composition of the TG m and T n polymorphic repeats. The comparative and functional analysis of the human and mouse CFTR genes showed that a region of about 150 nucleotides, present only in the human intron 9, mediates the exon 9 splicing inhibition in association with exonic regulatory elements. This region, defined as the CFTR exon 9 intronic splicing silencer, is a target for serine-arginine-rich protein interactions. Thus, the nonevolutionary conserved CFTR exon 9 alternative splicing is modulated by the TG m and T n polymorphism at the 3 splice region, enhancer and silencer exonic elements, and the intronic splicing silencer in the proximal 5 intronic region. Tissue levels and individual variability of splicing factors would determine the penetrance of the TG m T n locus in monosymptomatic forms of cystic fibrosis. Cystic fibrosis (CF),1 the most common life-shortening autosomal recessive disorder in Caucasians, is caused by mutations in the CF transmembrane regulator (CFTR) gene and is characterized by pathological features of variable severity at the level of lungs, pancreas, sweat glands, testis, ovaries, and intestine (1). Monosymptomatic forms of the disease such as congenital bilateral absence of vas deference (CBAVD), pancreatitis, nasal polyposis, disseminated bronchiectasies, and broncopulmonary allergic aspergillosis frequently present a peculiar allele at the polymorphic CFTR intron 8-exon 9 junction (2-8). At this locus a variable number of dinucleotide TG repeats (from 9 to 13) followed by a T repeat (T5, T7, or T9) can be found in the normal population, and it has been suggested that the T5 allele is a disease mutation with incomplete penetrance that could be modulated by the simultaneous presence of other mutations and/or polymorphisms (3). The pathologic effect of the T5 allele has been associated to the alternative splicing of the CFTR exon 9, which is extremely variable in humans among different individuals (9). Interestingly, exon 9 skipping is absent in mouse, and it has been reported not to be evolutionary conserved (10). This exon encodes part of the functionally important first nucleotide-binding domain, and its skipping produces a nonfunctional CFTR protein (10, 11). In CBAVD patients ...

Transactive response DNA-binding protein 43 kDa (TDP-43) is an RNA transporting and processing protein whose aberrant aggregates are implicated in neurodegenerative diseases. The C-terminal domain of this protein plays a key role in mediating this process. However, the N-terminal domain (residues 1-77) is needed to effectively recruit TDP-43 monomers into this aggregate. In the present study, we report, for the first time, the essentially complete 1 H, 15N and 13C NMR assignments and the structure of the N-terminal domain determined on the basis of 26 hydrogen-bond, 60 torsion angle and 1058 unambiguous NOE structural restraints. The structure consists of an a-helix and six b-strands. Two b-strands form a b-hairpin not seen in the ubiquitin fold. All Pro residues are in the trans conformer and the two Cys are reduced and distantly separated on the surface of the protein. The domain has a well defined hydrophobic core composed of F35, Y43, W68, Y73 and 17 aliphatic side chains. The fold is topologically similar to the reported structure of axin 1. The protein is stable and no denatured species are observed at pH 4 and 25°C. At 4 kcalÁmol À1 , the conformational stability of the domain, as measured by hydrogen/deuterium exchange, is comparable to ubiquitin (6 kcalÁmol À1 ).The b-strands, a-helix, and three of four turns are generally rigid, although the loop formed by residues 47-53 is mobile, as determined by model-free analysis of the 15 N{ 1 H}NOE, as well as the translational and transversal relaxation rates.