Nova-1 regulates neuron-specific alternative splicing and is essential for neuronal viability

We have combined genetic and biochemical approaches to analyze the function of the RNA-binding protein Nova-1, the paraneoplastic opsoclonus-myoclonus ataxia (POMA) antigen. Nova-1 null mice die postnatally from a motor deficit associated with apoptotic death of spinal and brainstem neurons. Nova-1 null mice show specific splicing defects in two inhibitory receptor pre-mRNAs, glycine alpha2 exon 3A (GlyRalpha2 E3A) and GABA(A) exon gamma2L. Nova protein in brain extracts specifically bound to a previously identified GlyRalpha2 intronic (UCAUY)3 Nova target sequence, and Nova-1 acted directly on this element to increase E3A splicing in cotransfection assays. We conclude that Nova-1 binds RNA in a sequence-specific manner to regulate neuronal pre-mRNA alternative splicing; the defect in splicing in Nova-1 null mice provides a model for understanding the motor dysfunction in POMA.     

Role of Nova-1 in regulating alpha2N,a novel glycine receptor splice variant,in developing spinal cord neurons

Inhibitory glycine receptor (GlyR) subunits undergo developmental regulation, but the molecular mechanisms of GlyR regulation in developing neurons are little understood. Using RT-PCR, we investigated the regulation of GlyR alpha-subunit splice forms during the development of the spinal cord of the rat. Experiments to compare the amounts of mRNA for two known splice variants of the GlyR alpha2 subunit, alpha2A and alpha2B, in the developing rat spinal cord revealed the presence of an additional, novel variant that lacked any exon 3, herein named "alpha2N." Examination of the RNA from spinal cords of different-aged rats showed a dramatic down-regulation of alpha2N during prenatal development: alpha2N mRNA formed a significant portion of the alpha2 subunit pool at E14, but its relative level was reduced by 85% by birth and was undetectable in adults. Two proteins previously implicated in regulating the splicing of GlyR alpha2 pre-mRNA, the neurooncological ventral antigen-1 (Nova-1) and the brain isoform of the polypyrimidine tract binding protein (brPTB), underwent small changes over the same period that did not correlate directly with the changes in the level of alpha2N, calling into question their involvement in the developmental regulation of alpha2N. However, treatment of spinal cord neurons in culture with antisense oligonucleotides designed selectively to knock down one of three Nova-1 variants significantly altered the relative level of GlyR alpha2N, showing that Nova-1 isoforms can regulate GlyR alpha2 pre-mRNA splicing in developing neurons. These results provide evidence for a novel splice variant of the GlyR alpha2 subunit that undergoes dramatic developmental regulation, reveal the expression profiles of Nova-1 and brPTB in the developing spinal cord, and suggest that Nova-1 plays a role in regulating GlyR alpha2N in developing neurons.     

Sequence-specific RNA binding by a Nova KH domain: implications for paraneoplastic disease and the fragile X syndrome

The structure of a Nova protein K homology (KH) domain recognizing single-stranded RNA has been determined at 2.4 A resolution. Mammalian Nova antigens (1 and 2) constitute an important family of regulators of RNA metabolism in neurons, first identified using sera from cancer patients with the autoimmune disorder paraneoplastic opsoclonus-myoclonus ataxia (POMA). The structure of the third KH domain (KH3) of Nova-2 bound to a stem loop RNA resembles a molecular vise, with 5'-Ura-Cyt-Ade-Cyt-3' pinioned between an invariant Gly-X-X-Gly motif and the variable loop. Tetranucleotide recognition is supported by an aliphatic alpha helix/beta sheet RNA-binding platform, which mimics 5'-Ura-Gua-3' by making Watson-Crick-like hydrogen bonds with 5'-Cyt-Ade-3'. Sequence conservation suggests that fragile X mental retardation results from perturbation of RNA binding by the FMR1 protein.     

Crystal structures of Nova-1 and Nova-2 K-homology RNA-binding domains.

BACKGROUND: Nova-1 and Nova-2 are related neuronal proteins that were initially cloned using antisera obtained from patients with the autoimmune neurological disease paraneoplastic opsoclonus-myoclonus ataxia (POMA). Both of these disease gene products contain three RNA-binding motifs known as K-homology or KH domains, and their RNA ligands have been identified via binding-site selection experiments. The KH motif structure has been determined previously using NMR spectroscopy, but not using X-ray crystallography. Many proteins contain more than one KH domain, yet there is no published structural information regarding the behavior of such multimers. RESULTS: We have obtained the first X-ray crystallographic structures of KH-domain-containing proteins. Structures of the third KH domains (KH3) of Nova-1 and Nova-2 were determined by multiple isomorphous replacement and molecular replacement at 2.6 A and 2.0 A, respectively. These highly similar RNA-binding motifs form a compact protease-resistant domain resembling an open-faced sandwich, consisting of a three-stranded antiparallel beta sheet topped by three alpha helices. In both Nova crystals, the lattice is composed of symmetric tetramers of KH3 domains that are created by two dimer interfaces. CONCLUSIONS: The crystal structures of both Nova KH3 domains are similar to the previously determined NMR structures. The most significant differences among the KH domains involve changes in the positioning of one or more of the alpha helices with respect to the betasheet, particularly in the NMR structure of the KH1 domain of the Fragile X disease protein FMR-1. Loop regions in the KH domains are clearly visible in the crystal structure, unlike the NMR structures, revealing the conformation of the invariant Gly-X-X-Gly segment that is thought to participate in RNA-binding and of the variable region. The tetrameric arrangements of the Nova KH3 domains provide insights into how KH domains may interact with each other in proteins containing multiple KH motifs.     

pasilla, the Drosophila homologue of the human Nova-1 and Nova-2 proteins, is required for normal secretion in the salivary gland.

From a screen for genes expressed and required in the Drosophila salivary gland, we identified pasilla (ps), which encodes a set of proteins most similar to human Nova-1 and Nova-2. Nova-1 and Nova-2 are nuclear RNA-binding proteins normally expressed in the CNS where they directly regulate splicing. In patients suffering from paraneoplastic opsoclonus myoclonus ataxia (POMA), Nova-1 and Nova-2 proteins are present as auto-antigens. Consistent with a role in splicing, PS is localized to nuclear puncta. The salivary glands of ps mutants internalize normally and maintain epithelial polarity. However, the mutant salivary glands develop irregularities in overall morphology and have defects in apical secretion. The secretory defects in ps mutants provide a potential mechanism for the loss of motor function observed in POMA patients.     

Inhibition of c-erbA mRNA splicing by a naturally occurring antisense RNA.

The rat erbA alpha locus encodes two overlapping mRNAs, alpha 1 and alpha 2, which are identical except for their most 3' exons. alpha 1 mRNA encodes a thyroid hormone receptor, while alpha 2 encodes an altered ligand binding domain of unknown function. Previous studies have shown that the ratio of alpha 1 to alpha 2 is highest in cells expressing a high level of a third RNA, Rev-ErbA alpha mRNA, which is transcribed in the opposite direction and is complementary to alpha 2 but not alpha 1 mRNA. It was hypothesized that base pairing with Rev-ErbA alpha blocks splicing of alpha 2 mRNA, thereby favoring formation of the non-overlapping alpha 1. To test this model, a system was developed in which alpha 2 pre-mRNAs were accurately spliced in vitro. Splicing was inhibited by the addition of a 5-fold excess of antisense RNAs containing the 3' end of Rev-ErbA alpha mRNA. Both an antisense RNA extending across the 3' splice site and a shorter RNA complementary only to exon sequences efficiently blocked splicing. However, splicing was only inhibited by complementary RNAs. These observations are consistent with a mechanism in which base pairing with a complementary RNA regulates alternative processing of alpha 1 and alpha 2 mRNAs.     

RNA annealing activities in HeLa nuclei.

RNA-RNA base pairing plays a critical role in the interactions between pre-mRNAs and trans-acting factors during the processing of pre-mRNAs (hnRNAs) into mRNAs, and it is likely that specific factors are required to promote the annealing of RNAs. To identify particular nuclear components that have such activity, we fractionated HeLa nucleoplasm and assayed for activity which promoted the hybridization of a pre-mRNA with an antisense RNA probe complementary to 60 nucleotides (nt) encompassing the 3' splice site. At least nine major RNA annealing activities were identified and, surprisingly, eight of these copurified partially or to homogeneity with known hnRNP proteins. The activities of three of these proteins, hnRNP A1, C1 and U, were confirmed using purified recombinant proteins. Moreover, we found that the RNA binding domain alone of hnRNP C1/C2 had significant activity, indicating that this RNA annealing may result, at least partly, from chaperone activity: a direct modulation of RNA conformation by hnRNP proteins. The finding that hnRNP proteins have strong RNA annealing activity indicates that they can profoundly affect the interactions of pre-mRNAs with trans-acting factors and suggests this to be an important function of hnRNP proteins in the processing of pre-mRNAs.     

Nova autoregulation reveals dual functions in neuronal splicing

The Nova family of neuron-specific RNA-binding proteins were originally identified as targets in an autoimmune neurologic disease characterized by failure of motor inhibition. Nova-1 regulates alternative splicing of pre-mRNAs encoding the inhibitory neurotransmitter receptor subunits GABA(A)Rgamma2 and GlyRalpha2 by directly binding intronic elements, resulting in enhancement of exon inclusion. Here we identify exon E4 in the Nova-1 pre-mRNA itself, encoding a phosphorylated protein domain, as an additional target of Nova-dependent splicing regulation in the mouse spinal cord. Nova binding to E4 is necessary and sufficient for Nova-dependent exon exclusion. E4 harbors five repeats of the known Nova-binding tetranucleotide YCAY and mutation of these elements destroys Nova-dependent regulation. Furthermore, swapping of the sites from Nova-1 and GABA(A)Rgamma2 indicates that the ability of Nova to enhance or repress alternative exon inclusion is dependent on the position of the Nova-binding element within the pre-mRNA. These studies demonstrate that in addition to its previously described role as a splicing activator, Nova autoregulates its own expression by acting as a splicing repressor.     

The role of a clinically important mutation in the fold and RNA-binding properties of KH motifs

We have investigated the role in the fold and RNA-binding properties of the KH modules of a hydrophobic to asparagine mutation of clinical importance in the fragile X syndrome. The mutation involves a well-conserved hydrophobic residue close to the N terminus of the second helix of the KH fold (alpha2(3) position). The effect of the mutation has been long debated: Although the mutant has been shown to disrupt the three-dimensional fold of several KH domains, the residue seems also to be directly involved in RNA binding, the main function of the KH module. Here we have used the KH3 of Nova-1, whose structure is known both in isolation and in an RNA complex, to study in detail the role of the alpha2(3) position. A detailed comparison of Nova KH3 structure with its RNA/KH complex and with other KH structures suggests a dual role for the alpha2(3) residue, which is involved both in stabilizing the hydrophobic core and in RNA contacts. We further show by nuclear magnetic resonance (NMR) studies in solution that L447 of Nova-1 in position alpha2(3) is in exchange in the absence of RNA, and becomes locked in a more rigid conformation only upon formation of an RNA complex. This implies that position alpha2(3) functions as a "gate" in the mechanism of RNA recognition of KH motifs based on the rigidification of the fold upon RNA binding.     

Characterization of RNA binding specificity of the Drosophila sex-lethal protein by in vitro ligand selection.

The Drosophila sex-lethal (Sxl) protein, a regulator of somatic sexual differentiation, is an RNA binding protein with two potential RNA recognition motifs (RRMs). It is thought to exert its function on splicing by binding to specific RNA sequences within Sxl and transformer (tra) pre-mRNAs. To examine the Sxl RNA binding specificity in detail, we performed in vitro selection and amplification of ligand RNAs from a random sequence pool on the basis of affinity with Sxl protein. After three cycles of selection and amplification, we cloned and sequenced 17 cDNAs corresponding to the RNAs selected in vitro. Sequencing showed that most of the RNAs selected contain polyuridine stretches surrounded by purine residues. In vitro binding analysis revealed that the sequences of the in vitro selected RNAs with relatively high affinity for Sxl show similarity to that of the Sxl- and tra-regulated acceptor regions, including the invariant AG sequence for splicing. These results suggest that Sxl recognizes and preferentially binds to a polyuridine stretch with a downstream AG sequence.     

A plant viral coat protein RNA binding consensus sequence contains a crucial arginine.

A defining feature of alfalfa mosaic virus (AMV) and ilarviruses [type virus: tobacco streak virus (TSV)] is that, in addition to genomic RNAs, viral coat protein is required to establish infection in plants. AMV and TSV coat proteins, which share little primary amino acid sequence identity, are functionally interchangeable in RNA binding and initiation of infection. The lysine-rich amino-terminal RNA binding domain of the AMV coat protein lacks previously identified RNA binding motifs. Here, the AMV coat protein RNA binding domain is shown to contain a single arginine whose specific side chain and position are crucial for RNA binding. In addition, the putative RNA binding domain of two ilarvirus coat proteins, TSV and citrus variegation virus, is identified and also shown to contain a crucial arginine. AMV and ilarvirus coat protein sequence alignment centering on the key arginine revealed a new RNA binding consensus sequence. This consensus may explain in part why heterologous viral RNA-coat protein mixtures are infectious.     

Unusual dimeric Zn(II)-cytosine complexes: new models of the interaction of Zn(II) with DNA and RNA

Synthesis and crystal structure of two Zn(II) dimer complexes with 1-methylcytosine (1-MeC) are reported. In complex [Zn(2)Cl(4)(mu-1-MeC-O2,N3)(2)] (1), two 1-MeC ligands are bridging two ZnCl(2) moieties. In [Zn(2)(1-MeC-N3)(4)(mu-SO(4))(2)].2H(2)O (2), the sulfates act as bridging ligands and 1-MeC are linked via N3 to Zn(II) as terminal ligands. Both complexes represent the first examples of Zn(II)-pyrimidine dimers. The potential biological significance of 1 and 2 is discussed.     

A specific RNA hairpin loop structure binds the RNA recognition motifs of the Drosophila SR protein B52.

B52, also known as SRp55, is a member of the Drosophila melanogaster SR protein family, a group of nuclear proteins that are both essential splicing factors and specific splicing regulators. Like most SR proteins, B52 contains two RNA recognition motifs in the N terminus and a C-terminal domain rich in serine-arginine dipeptide repeats. Since B52 is an essential protein and is expected to play a role in splicing a subset of Drosophila pre-mRNAs, its function is likely to be mediated by specific interactions with RNA. To investigate the RNA-binding specificity of B52, we isolated B52-binding RNAs by selection and amplification from a pool of random RNA sequences by using full-length B52 protein as the target. These RNAs contained a conserved consensus motif that constitutes the core of a secondary structural element predicted by energy minimization. Deletion and substitution mutations defined the B52-binding site on these RNAs as a hairpin loop structure covering about 20 nucleotides, which was confirmed by structure-specific enzymatic probing. Finally, we demonstrated that both RNA recognition motifs of B52 are required for RNA binding, while the RS domain is not involved in this interaction.     

A modified fluorescent intercalator displacement assay for RNA ligand discovery

Fluorescent intercalator displacement (FID) is a convenient and practical tool for identifying new nucleic acid-binding ligands. The success of FID is based on the fact that it can be fashioned into a versatile screening assay for assessing the relative binding affinities of compounds to nucleic acids. FID is a tagless approach; the target RNAs and the ligands or small molecules under investigation do not need to be modified in order to be examined. In this study, a modified FID assay for screening RNA-binding ligands was established using 3-methyl-2-((1-(3-(trimethylammonio)propyl)-4-quinolinylidene)methyl)benzothiazolium (TO-PRO) as the fluorescent indicator. Electrospray ionization mass spectrometry (ESI-MS) results provide direct evidence that correlates the reduction in fluorescence intensity observed in the FID assay with displacement of the dye molecule from RNA. The assay was successfully applied to screen a variety of RNA-binding ligands with a set of small hairpin RNAs. Ligands that bind with moderate affinity to the chosen RNA constructs (A-site, TAR [transactivation response element], h31 [helix 31], and H69 [helix 69] were identified.     

Mutations within the agonist-binding site convert the homomeric alpha1 glycine receptor into a Zn2+-activated chloride channel

The divalent cation Zn2+ has been shown to regulate inhibitory neurotransmission in the mammalian CNS by affecting the activation of the strychnine-sensitive glycine receptor (GlyR). In spinal neurons and cells expressing recombinant GlyRs, low micromolar (<10 microM) concentrations of Zn2+ enhance glycine currents, whereas higher concentrations (>10 microM) have an inhibitory effect. Mutational studies have localized the Zn2+ binding sites mediating allosteric potentiation and inhibition of GlyRs in distinct regions of the N-terminal extracellular domain of the GlyR alpha-subunits. Here, we examined the Zn2+ sensitivity of different mutations within the agonist binding site of the homomeric alpha(1)-subunit GlyR upon heterologous expression in Xenopus oocytes. This revealed that six substitutions within the ligand-binding pocket result in a total loss of Zn2+ inhibition. Furthermore, substitution of the positively charged residues arginine 65 and arginine 131 by alanine (alpha(1)(R65A), alpha(1)(R131A), or of the aromatic residue phenylalanine 207 by histidine (alpha(1)(F207H)), converted the alpha(1) GlyR into a chloride channel that was activated by Zn2+ alone. Dose-response analysis of the alpha(1)(F207H) GlyR disclosed an EC(50) value of 1.2 microM for Zn2+ activation; concomitantly the apparent glycine affinity was 1000-fold reduced. Thus, single point mutations within the agonist-binding site of the alpha(1) subunit convert the inhibitory GlyR from a glycine-gated into a selectively Zn2+-activated chloride channel. This might be exploited for the design of metal-specific biosensors by modeling-assisted mutagenesis.     

Identification of a residue in the gamma-aminobutyric acid type A receptor alpha subunit that differentially affects diazepam-sensitive and -insensitive benzodiazepine site binding

GABAA receptors that contain either the alpha4- or alpha6-subunit isoform do not recognize classical 1,4-benzodiazepines (BZDs). However, other classes of BZD site ligands, including beta-carbolines, bind to these diazepam-insensitive receptor subtypes. Some beta-carbolines [e.g. ethyl beta-carboline-3-carboxylate (beta-CCE) and methyl 6,7-dimethoxy-4-ethyl-beta-carboline-3-carboxylate (DMCM)] display a higher affinity for alpha4- compared to alpha6-containing receptors. In order to identify the structural determinants that underlie these affinity differences, we constructed chimeric alpha6/alpha4 subunits and co-expressed these with wild-type rat beta2 and gamma2L subunits in tsA201 cells for radioligand binding analysis. After identification of candidate regions, site-directed mutagenesis was used to narrow the ligand selectivity to a single amino acid residue (alpha6N204/alpha4I203). Substitutions at alpha6N204 did not alter the affinity of the imidazobenzodiazepine Ro15-4513. A homologous mutation in the diazepam-sensitive alpha1 subunit (S205N) resulted in a 7-8-fold reduction in affinity for the beta-carbolines examined. Although the binding of the classical agonist flunitrazepam was relatively unaffected by this mutation in the alpha1 subunit, the affinity for Ro15-1788 and Ro15-4513 was decreased by approximately 19-fold and approximately 38-fold respectively. The importance of this residue, located in the Loop C region of the extracellular N-terminus of the subunit protein, emphasizes the differential interaction of ligands with the alpha subunit in diazepam-sensitive and -insensitive receptors.     

Binding and cleavage of unstructured RNA by nuclear RNase P

Ribonuclease P (RNase P) is an essential endoribonuclease for which the best-characterized function is processing the 5' leader of pre-tRNAs. Compared to bacterial RNase P, which contains a single small protein subunit and a large catalytic RNA subunit, eukaryotic nuclear RNase P is more complex, containing nine proteins and an RNA subunit in Saccharomyces cerevisiae. Consistent with this, nuclear RNase P has been shown to possess unique RNA binding capabilities. To understand the unique molecular recognition of nuclear RNase P, the interaction of S. cerevisiae RNase P with single-stranded RNA was characterized. Unstructured, single-stranded RNA inhibits RNase P in a size-dependent manner, suggesting that multiple interactions are required for high affinity binding. Mixed-sequence RNAs from protein-coding regions also bind strongly to the RNase P holoenzyme. However, in contrast to poly(U) homopolymer RNA that is not cleaved, a variety of mixed-sequence RNAs have multiple preferential cleavage sites that do not correspond to identifiable consensus structures or sequences. In addition, pre-tRNA(Tyr), poly(U)(50) RNA, and mixed-sequence RNA cross-link with purified RNase P in the RNA subunit Rpr1 near the active site in "Conserved Region I," although the exact positions vary. Additional contacts between poly(U)(50) and the RNase P proteins Rpr2p and Pop4p were identified. We conclude that unstructured RNAs interact with multiple protein and RNA contacts near the RNase P RNA active site, but that cleavage depends on the nature of interaction with the active site.     

The RNA binding domain of the hantaan virus N protein maps to a central, conserved region

The nucleocapsid (N) protein of hantaviruses encapsidates both viral genomic and antigenomic RNAs, although only the genomic viral RNA (vRNA) is packaged into virions. To define the domain within the Hantaan virus (HTNV) N protein that mediates these interactions, 14 N- and C-terminal deletion constructs were cloned into a bacterial expression vector, expressed, and purified to homogeneity. Each protein was examined for its ability to bind the HTNV S segment vRNA with filter binding and gel electrophoretic mobility shift assays. These studies mapped a minimal region within the HTNV N protein (amino acids 175 to 217) that bound vRNA. Sequence alignments made from several hantavirus N protein sequences showed that the region identified has a 58% identity and an 86% similarity among these amino acid sequences. Two peptides corresponding to amino acids 175 to 196 (N1) and 197 to 218 (N2) were synthesized. The RNA binding of each peptide was measured by filter binding and competition analysis. Three oligoribonucleotides were used to measure binding affinity and assess specificity. The N2 peptide contained the major RNA binding determinants, while the N1 peptide, when mixed with N2, contributed to the specificity of vRNA recognition.