Characterization and Functional Implications of the RNA Binding Properties of Nuclear Factor TDP-43, a Novel Splicing Regulator ofCFTR Exon 9
2001; Elsevier BV; Volume: 276; Issue: 39 Linguagem: Inglês
10.1074/jbc.m104236200
ISSN1083-351X
AutoresEmanuele Buratti, Francisco E. Baralle,
Tópico(s)RNA modifications and cancer
ResumoVariations in a polymorphic (TG)m sequence near exon 9 of the human CFTR gene have been associated with variable proportions of exon skipping and occurrence of disease. We have recently identified nuclear factor TDP-43 as a novel splicing regulator capable of binding to this element in the CFTR pre-mRNA and inhibiting recognition of the neighboring exon. In this study we report the dissection of the RNA binding properties of TDP-43 and their functional implications in relationship with the splicing process. Our results show that this protein contains two fully functional RNA recognition motif (RRM) domains with distinct RNA/DNA binding characteristics. Interestingly, TDP-43 can bind a minimum number of six UG (or TG) single-stranded dinucleotide stretches, and binding affinity increases with the number of repeats. In particular, the highly conserved Phe residues in the first RRM region play a key role in nucleic acid recognition. Variations in a polymorphic (TG)m sequence near exon 9 of the human CFTR gene have been associated with variable proportions of exon skipping and occurrence of disease. We have recently identified nuclear factor TDP-43 as a novel splicing regulator capable of binding to this element in the CFTR pre-mRNA and inhibiting recognition of the neighboring exon. In this study we report the dissection of the RNA binding properties of TDP-43 and their functional implications in relationship with the splicing process. Our results show that this protein contains two fully functional RNA recognition motif (RRM) domains with distinct RNA/DNA binding characteristics. Interestingly, TDP-43 can bind a minimum number of six UG (or TG) single-stranded dinucleotide stretches, and binding affinity increases with the number of repeats. In particular, the highly conserved Phe residues in the first RRM region play a key role in nucleic acid recognition. RNA recognition motif glutathione S-transferase polyacrylamide gel electrophoresis electromobility shift assay short sequence repeats We have recently reported the identification of TDP-43 as a splicing regulator that specifically binds the (UG)m-repeated polymorphic region near the 3′-splice site of CFTRexon 9 and down-regulates its recognition by the splicing machinery (1Buratti E. Dork T. Zuccato E. Pagani F. Romano M. Baralle F.E. EMBO J. 2001; 20: 1774-1784Crossref PubMed Scopus (504) Google Scholar). This region, acting in concert with the adjacent (u)n element, is one of the key cis-acting sequences which regulate the proportion of exon 9 skipping in the mature CFTR mRNA transcript (1Buratti E. Dork T. Zuccato E. Pagani F. Romano M. Baralle F.E. EMBO J. 2001; 20: 1774-1784Crossref PubMed Scopus (504) Google Scholar, 2Niksic M. Romano M. Buratti E. Pagani F. Baralle F.E. Hum. Mol. Genet. 1999; 8: 2339-2349Crossref PubMed Scopus (130) Google Scholar, 3Pagani F. Buratti E. Stuani C. Romano M. Zuccato E. Niksic M. Giglio L. Faraguna D. Baralle F.E. J. Biol. Chem. 2000; 275: 21041-21047Abstract Full Text Full Text PDF PubMed Scopus (155) Google Scholar). Considering that exon 9 skipping produces a non-functional CFTR protein (4Strong T.V. Wilkinson D.J. Mansoura M.K. Devor D.C. Henze K. Yang Y. Wilson J.M. Cohn J.A. Dawson D.C. Frizzell R.A. Collins F.S. Hum. Mol. Genet. 1993; 2: 225-230Crossref PubMed Scopus (94) Google Scholar, 5Delaney S.J. Rich D.P. Thomson S.A. Hargrave M.R. Lovelock P.K. Welsh M.J. Wainwright B.J. Nat. Genet. 1993; 4: 426-431Crossref PubMed Scopus (96) Google Scholar) the study of the RNA binding properties of TDP-43 is of considerable importance to gain further insight concerning the potential disease-causing consequences of its binding in vivo. Indeed, the clinical relevance of these studies is highlighted by the existence of a clear association between certain (TG)m(T)n alleles with distinct forms of Cystic Fibrosis (1Buratti E. Dork T. Zuccato E. Pagani F. Romano M. Baralle F.E. EMBO J. 2001; 20: 1774-1784Crossref PubMed Scopus (504) Google Scholar, 6Chillon M. Casals T. Mercier B. Bassas L. Lissens W. Silber S. Romey M.C. Ruiz-Romero J. Verlingue C. Claustres M. Nunes V. Férec C. Estiril X. N. Engl. J. Med. 1995; 332: 1475-1480Crossref PubMed Scopus (820) Google Scholar, 7Chu C.S. Trapnell B.C. Curristin S. Cutting G.R. Crystal R.G. Nat. Genet. 1993; 3: 151-156Crossref PubMed Scopus (442) Google Scholar, 8Cuppens H. Lin W. Jaspers M. Costes B. Teng H. Vankeerberghen A. Jorissen M. Droogmans G. Reynaert I. Goossens M. Nilius B. Cassiman J.J. J. Clin. Invest. 1998; 101: 487-496Crossref PubMed Scopus (357) Google Scholar, 9Rave-Harel N. Kerem E. Nissim-Rafinia M. Madjar I. Goshen R. Augarten A. Rahat A. Hurwitz A. Darvasi A. Kerem B. Am. J. Hum. Genet. 1997; 60: 87-94PubMed Google Scholar). In addition, the study of (UG)m elements can provide further insight concerning the mRNA splicing process in general because (UG)m sequences have been described to act as splicing regulatory sequences in different genomic contexts. In fact, in addition to theCFTR gene, the presence of simple (UG)m-repeated sequences has been described to influence the splicing process of at least two other genes: the apolipoprotein AII gene (10Shelley C.S. Baralle F.E. Nucleic Acids Res. 1987; 15: 3787-3799Crossref PubMed Scopus (19) Google Scholar) and the human cardiac Na+/Ca2+ exchanger (11Gabellini N. Eur. J. Biochem. 2001; 268: 1076-1083Crossref PubMed Scopus (50) Google Scholar). In the Apo AII gene the UG tract was shown to be functionally equivalent to a polypyrimidine tract and required for efficient splicing of Apo AII exon 2 (10Shelley C.S. Baralle F.E. Nucleic Acids Res. 1987; 15: 3787-3799Crossref PubMed Scopus (19) Google Scholar) while in the human cardiac Na+/Ca2+exchanger (11Gabellini N. Eur. J. Biochem. 2001; 268: 1076-1083Crossref PubMed Scopus (50) Google Scholar) it acts as a strong intronic splicing enhancer situated in intron 2. It should be noted that in contrast with these two genes, the CFTR (UG)m element was found to possess a strong inhibitory effect on CFTR exon 9 splicing, a property that may probably be linked to its peculiar evolutionary history. In fact, sequencing of the mouse CFTR exon 9 genomic region has shown that in the flanking introns, the (TG)m(T)n regulatory elements are absent and that the intron themselves are of very different length when compared with the human introns (2Niksic M. Romano M. Buratti E. Pagani F. Baralle F.E. Hum. Mol. Genet. 1999; 8: 2339-2349Crossref PubMed Scopus (130) Google Scholar). This finding, together with the observation that mouse CFTR exon 9 is not subject to alternative splicing, suggests that the presence in humans of the (UG)m sequence represents a disturbing element, which interferes with the normal maturation process of the CFTR pre-mRNA. This conclusion is also supported by the fact that CFTR exon 9 and its intronic flanking sequences are found co-integrated with characteristic L1 sequences in multiple chromosome locations distinct from theCFTR locus (12Kazazian Jr., H.H. Moran J.V. Nat. Genet. 1998; 19: 19-24Crossref PubMed Scopus (400) Google Scholar, 13Rozmahel R. Heng H.H. Duncan A.M. Shi X.M. Rommens J.M. Tsui L.C. Genomics. 1997; 45: 554-561Crossref PubMed Scopus (39) Google Scholar). These findings may indicate that the introduction of foreign elements in the CFTR IVS8 and IVS9 sequences may be a consequence of a retrotransposition event, which affected the human CFTR gene early during the course of evolution. In order to better elucidate the role of the CFTR (UG)m element and obtain functional clues regarding the role of TDP-43 in the splicing process we have characterized the RNA/DNA binding properties of TDP-43. Our results have confirmed the existence in this protein of two fully functional RNA recognition motifs (RRM),1 also known as RBD, for RNA binding domains (14Burd C.G. Dreyfuss G. Science. 1994; 265: 615-621Crossref PubMed Scopus (1734) Google Scholar, 15Mattaj I.W. Cell. 1993; 73: 837-840Abstract Full Text PDF PubMed Scopus (221) Google Scholar, 16Biamonti G. Riva S. FEBS Lett. 1994; 340: 1-8Crossref PubMed Scopus (72) Google Scholar, 17Birney E. Kumar S. Krainer A.R. Nucleic Acids Res. 1993; 21: 5803-5816Crossref PubMed Scopus (589) Google Scholar, 18Nagai K. Oubridge C. Ito N. Avis J. Evans P. Trends Biochem. Sci. 1995; 20: 235-240Abstract Full Text PDF PubMed Scopus (202) Google Scholar), which possess distinct binding characteristics. Plasmids pTCTT3 and pTG12 and minigenes lacking the (TG)m/T(n) elements were obtained as previously described (1Buratti E. Dork T. Zuccato E. Pagani F. Romano M. Baralle F.E. EMBO J. 2001; 20: 1774-1784Crossref PubMed Scopus (504) Google Scholar). Plasmids pTG3, pTG6, pTG9 were obtained by annealing the following forward and reverse oligos and ligating them in pBluescript KS (Stratagene) linearized with SmaI: 5′-gaaaattaatgtgtggaaaattaagaaa-3′ (oligo TG3) and 5′-tttcttaattttccacacattaattttc-3′ (oligo AC3) for pTG3, 5′-gaaaattaatgtgtgtgtgtggaaaatt-3′ (oligo TG6) and 5′-aattttccacacacacacattaattttc-3′ (oligo AC6) for pTG6, 5′-gaaaattaatgtgtgtgtgtgtgtgtga-3′ (oligo TG9) and 5′-tcacacacacacacacacattaattttc-3′ (oligo AC9) for pTG9. The plasmid pTAR was obtained by annealing the following primers and ligating them in pBluescript KS (Stratagene) linearized with SmaI: 5′-ctgctttttgcctgtactgggtctctctggttagaccagatctgag-3′ (oligo TARS) as the forward and 5′-ctcagatctggtctaaccagagagacccagtacaggcaaaaagcag-3′ (oligo TARAS) as the reverse primer. The synthetic (UG)12oligo was obtained from MWG Biotech (Firenze, Italy). The GST-TDP43 and GST-TDP43(101–261) fusion proteins were obtained as previously described (1Buratti E. Dork T. Zuccato E. Pagani F. Romano M. Baralle F.E. EMBO J. 2001; 20: 1774-1784Crossref PubMed Scopus (504) Google Scholar). Deletion of the RRM1, RRM2, and 106–111 RNP-2 regions was obtained using a two step polymerase chain reaction extension method with sense and reverse primers spanning RRM1 (5′-aaaacatccgataaacttcctaat-3′ and 5′-attaggaagtttatcggatgtttt-3′), RRM2 (5′-ttgagaagcagatccaatgccgaa-3′ and 5′-ttcggcattggatctgcttctcaa-3′), and 106–111 RNP-2 (5′-aaaacatccgatccatggaaaaca-3′ and 5′-tgttttccatggatcggatgtttt-3′). It should be noted that these regions were identified through a search using the Pfam program at www.sanger.ac.uk. In order to introduce the L106D, V108D, and L111D mutations we used the following primers: 5′-tccgatgatatagatttgggtgatccatgg-3′ and 5′-ccatggatcacccaaatctatatcatcgga-3′. Similarly, the Phe residues (at positions 147 and 149) in the 145–152 RNP-1 motif were mutated to Leu using the following forward and reverse primers: 5′-aaggggttgggcttggttcgtttt-3′ and 5′-aaaacgaaccaagcccaacccctt-3′. The single Phe-194 in the 193–197 RNP-2 motif was mutated to Leu using the following forward and reverse primers: 5′-gtgttggtggggcgctgt-3′ and 5′-acagcgccccaccaacac-3′. The two Phe residues (at positions 229 and 231) in the 227–234 RNP-1 motif were mutated to Leu using the following forward and reverse primers: 5′-agggccttggccttggttacattt-3′ and 5′-aaatgtaaccaaggccaaggccct-3′. Double and triple mutants were obtained using the same methodology on single- and double-mutated proteins. All fusion proteins were expressed in Escherichia coli DH5α cells by overnight induction at room temperature in the presence of 0.1–0.3 mm IPTG. Cells were then resuspended in phosphate-buffered saline, 1% Triton X-100, and sonicated. The supernatant was recovered after centrifugation at 3000 × g for 30 s in an Eppendorf 5810R centrifuge and incubated with glutathione S-Sepharose 4B beads (Amersham Pharmacia Biotech). The absorbed proteins were then eluted according to the manufacturer's instructions. Purified proteins were quantitated on an SDS-PAGE gel using bovine serum albumin standards (Sigma). Plasmids were linearized by digestion with HindIII and transcription was performed with T7 RNA polymerase (Stratagene) in the presence of labeled [α-32P]UTP, DNase-treated according to standard protocols, and purified on a Nick column (Amersham Pharmacia Biotech) according to the manufacturer's instructions. The labeled RNAs were then precipitated and resuspended in RNase-free water. The UV cross-linking assay was performed by adding [α-32P]UTP-labeled RNA probes (1 × 106 cpm per incubation) in a water bath for 15 min at 30 ° C with 200 ng of each different purified protein in a 20-μl final volume. Binding conditions were 20 mm Hepes pH 7.9, 72 mm KCl, 1.5 mm MgCl2, 0.78 mm magnesium acetate, 0.52 mmdithiothreitol, 3.8% glycerol, 0.75 mm ATP, and 1 mm GTP. In the competition experiments, cold RNA and DNA were also added as competitors 5 min before addition of the labeled RNAs (the molar excess of the unlabeled competitor used in the different experiments is stated in each figure legend). Samples were then transferred in the wells of an HLA plate (Nunc, InterMed) and irradiated with UV light on ice (0.8 joules, ∼5 min) using a UV Linker (Euroclone). Unbound RNA was then digested with 30 μg of RNase A (Sigma) and 6 units of RNase T1 (Sigma) by incubating at 37 °C for 30 min in a water bath and then adding SDS-PAGE sample buffer. Samples were then analyzed on a 10% SDS-PAGE gel followed by autoradiography with autoradiographic XAR film (Kodak). Films were then scanned on a Macintosh G3 work station using Adobe Photoshop and printed using a Phaser 400 printer. Oligonucleotides (200 ng, ∼25 pmols) were labeled by phosphorylation with [γ-32P]ATP and T4 polynucleotide kinase (PNK, Stratagene) for 1 h at 37 °C and then precipitated in 0.3m sodium acetate, pH 5.2 and three volumes of ethanol. After centrifugation and a washing step with 70% ethanol the labeled oligos were resuspended in 400 μl of water. Each binding reaction was performed at room temperature for 15 min by mixing the purified protein with the labeled oligo (or RNA) in a 20-μl final volume. The reactions were performed in 1× bind shift binding buffer (20 mm Hepes pH 7.9, 2 mm MgCl2) and electrophoresed on a 5% polyacrylamide gel at 100 V for 1 h in 0.5× Tris borate/EDTA buffer at 4 °C. The gel was then dried on 3 MM Whatman filter paper and exposed for 20 min with autoradiographic XAR film (Kodak). For quantitation gels were measured with an InstantImager (Packard). In our search to identify proteins that recognize the splicing regulatory elements of CFTR exon 9, we recently isolated TDP-43 as a protein that binds specifically to the splicing regulatory (UG)m element found near the 3′-splice site of this exon (1Buratti E. Dork T. Zuccato E. Pagani F. Romano M. Baralle F.E. EMBO J. 2001; 20: 1774-1784Crossref PubMed Scopus (504) Google Scholar). Up to now, the only other described cellular function of TDP-43 was the ability to bind a HIV-1 TAR DNA polypyrimidinic sequence motif leading to the inhibition of HIV-1 transcription (19Ou S.H. Wu F. Harrich D. Garcia-Martinez L.F. Gaynor R.B. J. Virol. 1995; 69: 3584-3596Crossref PubMed Google Scholar). Interestingly, no binding to TAR RNA had been reported (19Ou S.H. Wu F. Harrich D. Garcia-Martinez L.F. Gaynor R.B. J. Virol. 1995; 69: 3584-3596Crossref PubMed Google Scholar). However, our recent observation that TDP-43 can efficiently bind to (UG)m sequences (1Buratti E. Dork T. Zuccato E. Pagani F. Romano M. Baralle F.E. EMBO J. 2001; 20: 1774-1784Crossref PubMed Scopus (504) Google Scholar) is consistent with the presence of two putative full-length RRM domains (Fig. 1A) located between residues 106 and 175 (RRM1) and 193 and 257 (RRM2) of its coding sequence according to the output of the Pfam program (available at www.sanger.ac.uk). This finding provided a functional basis that accounted for the ability of TDP-43 to bind RNA sequences, and in this study we provide a detailed analysis of their functionality and importance. Initially, using a GST fusion protein containing the TDP-43 full coding sequence (GST-TDP43), we confirmed the binding specificities of recombinant GST-TDP43 toward different RNAs in UV-cross-linking analysis. Fig. 1B shows that increasing amounts of unlabeled (UG)12 RNA were very efficient competitors. Because (UG)m sequences have been described as efficient polypyrimidinic sequences (10Shelley C.S. Baralle F.E. Nucleic Acids Res. 1987; 15: 3787-3799Crossref PubMed Scopus (19) Google Scholar) we also tested the possibility that recognition could be extended generally to this type of sequences. However, addition of a cold polypyrimidinic RNA (UCUU)3 did not have any effect on the binding of GST-TDP43 to (UG)12 (Fig. 1B), confirming the high specificity of TDP-43 binding to UG-repeated motifs. This result is consistent with our previously reported pull-down assay, which did not result in any TDP-43 being recognized by (UCUU)3 RNA (1Buratti E. Dork T. Zuccato E. Pagani F. Romano M. Baralle F.E. EMBO J. 2001; 20: 1774-1784Crossref PubMed Scopus (504) Google Scholar). The high sequence binding specificity of TDP-43 is also highlighted by the fact that addition of cold TAR RNA was incapable of competing with the binding of GST-TDP43 to the (UG)12 sequence, a finding that had already been described in the original isolation of TDP-43 (19Ou S.H. Wu F. Harrich D. Garcia-Martinez L.F. Gaynor R.B. J. Virol. 1995; 69: 3584-3596Crossref PubMed Google Scholar) and which we confirm here. It should be noted that at present no other RNA-binding protein has been described to bind UG-repeated sequences although CUG-BP (CUG-binding protein) has been recently described to bind UG repeats in a yeast three-hybrid system (20Takahashi N. Sasagawa N. Suzuki K. Ishiura S. Biochem. Biophys. Res. Commun. 2000; 277: 518-523Crossref PubMed Scopus (67) Google Scholar). It was then of interest to analyze the minimum length of (UG) repeats that could be specifically bound by this protein. Therefore, cold RNAs competitors containing different lengths of (UG) repeats (3, 6, 9, and 12) were incubated in the presence of GST-TDP43 and labeled (UG)12 RNA. The results shown in Fig. 1Cdemonstrate that efficient competition can be observed only when the number of (UG) repeats is equal or above six and that there is a relationship between the number of (UG) repeats and the efficiency of binding. The fact that this protein was originally described to bind TAR DNA sequences raised the possibility that the binding characteristics of this protein might include (TG)-repeated sequences as well as (UG)-repeated sequences. Therefore, we performed competition analysis using GST-TDP43 bound to (UG)12 RNA in the presence of cold single-stranded DNA oligos (see TableI). As shown in Fig.2A the most efficient competitor was represented by the oligo TG12 carrying twelve (TG) repeats followed by the TARS oligo, a result that is consistent with what had been observed by Ou et al. (19Ou S.H. Wu F. Harrich D. Garcia-Martinez L.F. Gaynor R.B. J. Virol. 1995; 69: 3584-3596Crossref PubMed Google Scholar). It should also be noted that other oligos (in particular TCTTS) display a weak but significant ability to compete for this protein, a characteristic not found in the (UCUU)3 RNA (see Fig. 1). This finding also confirms the original observations of Ou et al. (19Ou S.H. Wu F. Harrich D. Garcia-Martinez L.F. Gaynor R.B. J. Virol. 1995; 69: 3584-3596Crossref PubMed Google Scholar) who, using a polymerase chain reaction-based site selection procedure, found that recombinant TDP-43 preferably bound DNA stretches of eight contiguous pyrimidine residues. Nonetheless, Fig. 2B shows that the use of oligos containing different numbers of TG repeats yielded results very similar to those obtained in Fig. 1Cusing (UG)-repeated sequences. Also in this case, the minimum number of (TG) repeats needed to efficiently compete for GST-TDP43 binding is six, and there is a relationship between the number of (TG) repeats and the efficiency of competition.Table IssDNA oligonucleotides used for competition analysisNucleotide sequence (5′-3′)OligogaaaattaatgtgtggaaaattaagaaaTG3gaaaattaatgtgtgtgtgtggaaaattTG6gaaaattaatgtgtgtgtgtgtgtgtgaTG9tgtgtgtgtgtgtgtgtgtgtgtgTG12tttcttaattttccacacattaattttcAC3aattttccacacacacacattaattttcAC6tcacacacacacacacacattaattttcAC9acacacacacacacacacacacacAC12tcctcctccttcttcttcttcaggTCTTScctgaagaagaagaaggaggaggaTCTTASctgctttttgcctgtactgggtctctctggttagaccagatctgagTARSctcagatctggtctaaccagagagacccagtacaggcaaaaagcagTARASgAAAATTAACAATTTAAAmEX9AS Open table in a new tab Finally, oligos containing (AC) repeats can function as efficient competitors for the binding of GST-TDP43 to (UG)12 (data not shown). However, in this case competition was caused by the (AC) repeats annealing directly to the (UG)12-labeled sequence and inhibiting binding of the protein rather than by binding directly to GST-TDP43. In fact, EMSA analysis shows that there is little if any direct binding of GST-TDP43 to AC6, AC9, or AC12 end-labeled oligos (Fig. 2C, left panel) while direct binding of GST-TDP43 efficiently occurs for single-stranded (TG)-repeated sequences containing 6, 9, and 12 (TG) repeats (Fig. 2C,central panel). Notably, the fact that the oligo bearing three tg-repeats (TG3) could not efficiently bind GST-TDP43 confirms the previous competition data by UV-cross-linking (Fig. 2B). Thus, in order to establish whether TDP-43 could bind double-stranded oligos we then annealed labeled (AC) oligos with equal amounts of complementary and unlabeled (TG) oligos and then repeated the EMSA analysis. The results confirm that double-stranded oligos containing TG repeats do not bind TDP-43 (Fig. 2C, right panel). The DNA and RNA binding efficiencies of TDP-43 were assumed to depend on the two RRM regions. In order to establish whether there was no other protein domain involved in RNA/DNA recognition and to compare the two binding efficiencies we produced a construct coding for a truncated TDP-43 protein lacking the N- and C-terminal regions (Fig.3A). The DNA and RNA binding efficiencies of GST-TDP43 and GST-TDP43-(101–261) were then compared using as substrate a 5′-labeled (TG)12 or (UG)12 oligo (at a fixed concentration of 6 nm). The results, shown in Fig. 3B, demonstrate that deletion of the N- and C-terminal regions of TDP-43 does not appear to affect the RNA binding efficiency of the central RRM-containing region and may even slightly enhance it. It should be noted that the retarded complexes formed by each protein do not migrate at the same level, an indication of the higher molecular weight of the GST-TDP43/nucleic acid complex as opposed to the GST-TDP43(101–261)/nucleic acid complex. Interestingly, binding of TDP-43 to (UG)12 as opposed to (TG)12 presents some differences as well. In fact, Fig.3C shows that in UV cross-linking analysis only one RNA-protein complex of 50–52 kDa is formed when GST-TDP43-(101–261) is bound to (UG)12, as previously described (1Buratti E. Dork T. Zuccato E. Pagani F. Romano M. Baralle F.E. EMBO J. 2001; 20: 1774-1784Crossref PubMed Scopus (504) Google Scholar). On the other hand, at least two major DNA-protein complexes with altered mobility can be detected when the same protein is bound to (TG)12. The formation of multiple complexes in SDS-PAGE following UV cross-linking analysis has already been described for another RRM protein, Gbp1p (21Johnston S.D. Lew J.E. Berman J. Mol. Cell. Biol. 1999; 19: 923-933Crossref PubMed Scopus (30) Google Scholar), as a result of the formation of multiple covalent cross-links between protein and nucleic acid. The fact that the migrating complexes are different when using (UG)12 as opposed to (TG)12 represents a further indication that TDP-43 RNA and DNA binding characteristics may not be identical. In order to test the importance of each TDP-43 RRMs we then made a series of deletion mutants and analyzed their binding to (UG)12 RNA. Fig.4A shows a schematic representation of two mutants in which we selectively deleted the first RRM motif (GST-TDP43/ΔRRM1) and the second RRM motif (GST-TDP43/ΔRRM2). In addition, in order to confirm the presence of the first RNP-2 motif (residues 106–111), which had not been previously detected in the original work by Ou et al. (19Ou S.H. Wu F. Harrich D. Garcia-Martinez L.F. Gaynor R.B. J. Virol. 1995; 69: 3584-3596Crossref PubMed Google Scholar) we prepared two mutants, the first containing a deletion of the entire 6-amino acid region (GST-TDP43/Δ(106–111)) and the second introducing an Asp residue in substitution for the three amino acids (Leu-106, Val-108, and Leu-111), which were predominantly conserved in the corresponding RNP-2 motifs of well characterized RNA-binding proteins (see Fig. 5). The four mutants were then analyzed by EMSA analysis using labeled (UG)12RNA. The results show that deletion of RRM1 and deletion (or mutation) of the 106–111 RNP-2 motif completely abolished the ability of TDP-43 to bind the RNA (Fig. 4B, first and second upper panels). This result not only confirms the presence of a functional RNP-2 motif localized in position 106–111 but also suggests that the RRM1 sequence spanning residues 106–175 is of fundamental importance for the binding to RNA.Figure 5Comparison of TDP-43 RRM regions with similarRRM motifs of known tertiary structure. This figure shows a comparison of the two TDP-43 RRM motifs with homologous RRM motifs belonging to proteins for which their crystallographic structure is known, hnRNP A1, Sxl, and PABP. Sequence identities with the TDP-43 sequence are marked in bold lettering while key residues involved in direct stacking interactions with the RNA, as confirmed by structural analysis are highlighted with open circles. The highly conserved RNP-1 and RNP-2 consensus sequences areunderlined. The asterisk marks the position of the conserved TDP-43 Phe residues with respect to the other RRM sequences.View Large Image Figure ViewerDownload Hi-res image Download (PPT) It should be noted that deletion of RRM2 (leaving the RRM1 sequence intact) does not completely abolish the RNA binding capability of TDP-43 but leads to the appearance of a super-shifted RNA-protein complex (Fig. 4B, first panel), suggesting that the binding characteristics of RRM2 are quite different from those of RRM1. The specificity of this complex formation is confirmed by a competition analysis (Fig. 4C) in which we added increasing amounts of each mutant (GST-TDP43/ΔRRM1 and GST-TDP43/ΔRRM2) to a reaction mix that contained GST-TDP43 and labeled (UG)12RNA. The results show that the GST-TDP43/ΔRRM2 mutant is capable of actively competing with GST-TDP43 for the formation of the supershifted complex, but no change is observed following addition with GST-TDP43/ΔRRM1. In order to better characterize how TDP-43 binds to RNA we compared the sequence of TDP-43 RRMs domains with that of other well characterized RRMs found in proteins whose structure has been solved by crystallography: hnRNP A1 (22Shamoo Y. Krueger U. Rice L.M. Williams K.R. Steitz T.A. Nat. Struct. Biol. 1997; 4: 215-222Crossref PubMed Scopus (119) Google Scholar), Sxl (23Handa N. Nureki O. Kurimoto K. Kim I. Sakamoto H. Shimura Y. Muto Y. Yokoyama S. Nature. 1999; 398: 579-585Crossref PubMed Scopus (318) Google Scholar), PABP (24Deo R.C. Bonanno J.B. Sonenberg N. Burley S.K. Cell. 1999; 98: 835-845Abstract Full Text Full Text PDF PubMed Scopus (412) Google Scholar) and U1A spliceosomal protein (18Nagai K. Oubridge C. Ito N. Avis J. Evans P. Trends Biochem. Sci. 1995; 20: 235-240Abstract Full Text PDF PubMed Scopus (202) Google Scholar, 25Oubridge C. Ito N. Evans P.R. Teo C.H. Nagai K. Nature. 1994; 372: 432-438Crossref PubMed Scopus (784) Google Scholar). Fig. 5 shows the RRM domains found in these proteins, which are most similar to RRM1 and RRM2 of TDP-43. It should be noted that very little homology was detected between U1A RRMs and TDP-43 RRMs (data not shown). Overall, the highest amino acid identity between the different RRMs can be found in correspondence with the highly conserved RNP-1 and RNP-2 consensus motifs. In particular, several key aromatic residues that have been reported to be responsible for direct stacking interactions with RNA bases in these different proteins (marked with open circles) are conserved in TDP-43 RRMs. The only exception is represented by the first putative TDP-43 RNP-2 motif (residues 106–111) in which none of the aromatic residues reported to make direct stacking interactions with the RNA are conserved. It should be noted that mutation of these aromatic residues in hnRNPA1 (26Merrill B.M. Stone K.L. Cobianchi F. Wilson S.H. Williams K.R. J. Biol. Chem. 1988; 263: 3307-3313Abstract Full Text PDF PubMed Google Scholar) and U1A (27Jessen T.H. Oubridge C. Teo C.H. Pritchard C. Nagai K. EMBO J. 1991; 10: 3447-3456Crossref PubMed Scopus (166) Google Scholar, 28Hoffman D.W. Query C.C. Golden B.L. White S.W. Keene J.D. Proc. Natl. Acad. Sci. U. S. A. 1991; 88: 2495-2499Crossref PubMed Scopus (166) Google Scholar) has long been known to severely affect the RNA binding capability of these proteins. To further investigate these similarities we then prepared a series of GST-TDP43-(101–261) mutants in which the conserved Phe residues in each RNP motif were mutated to leucine residues (Fig.6A). The rationale for this change in residue resides in the fact that a Phe to Leu single amino acid mutation has been previously described to abolish the functionality of the RNP motif in the case of the nucleolin protein (29Serin G. Joseph G. Ghisolfi L. Bauzan M. Erard M. Amalric F. Bouvet P. J. Biol. Chem. 1997; 272: 13109-13116Abstract Full Text Full Text PDF PubMed Scopus (85) Google Scholar). Each mutant was expressed in E. coli (Fig.6B), and its ability to bind (UG)12 RNA and (TG)12 DNA was assayed by EMSA (Fig. 6, C andD). In both cases, the mutations that most reduced binding to the nucleic acid were the F147L and F14
Referência(s)