Ribonucleoprotein organization of eukaryotic RNA. XXX. Evidence that U1 small nuclear RNA is a ribonucleoprotein when base-paired with pre-messenger RNA in vivo

Long non-coding RNAs: roles in cellular stress responses and epigenetic mechanisms regulating chromatin

Most of the genome is transcribed into RNA but only 2% of the sequence codes for proteins. Non-coding RNA transcripts include a very large number of long noncoding RNAs (lncRNAs). A growing number of identified lncRNAs operate in cellular stress responses, for example in response to hypoxia, genotoxic stress, and oxidative stress. Additionally, lncRNA plays important roles in epigenetic mechanisms operating at chromatin and in maintaining chromatin architecture. Here, we address three lncRNA topics that have had significant recent advances. The first is an emerging role for many lncRNAs in cellular stress responses. The second is the development of high throughput screening assays to develop causal relationships between lncRNAs across the genome with cellular functions. Finally, we turn to recent advances in understanding the role of lncRNAs in regulating chromatin architecture and epigenetics, advances that build on some of the earliest work linking RNA to chromatin architecture.

RNA as a Major-Groove Ligand: RNA-RNA and RNA-DNA Triplexes Formed by GAA and UUC or TTC Sequences

Friedreich's ataxia is associated with noncanonical nucleic acid structures that emerge when GAA:TTC repeats in the first intron of the FXN gene expand beyond a critical number of repeats. Specifically, the noncanonical repeats are associated with both triplexes and R-loops. Here, we present an in silico investigation of all possible triplexes that form by attaching a third RNA strand to an RNA:RNA or DNA:DNA duplex, complementing previous DNA-based triplex studies. For both new triplexes results are similar. For a pyridimine UUC⁺ third strand, the parallel orientation is stable while its antiparallel counterpart is unstable. For a neutral GAA third strand, the parallel conformation is stable. A protonated GA⁺A third strand is stable in both parallel and antiparallel orientations. We have also investigated Na⁺ and Mg²⁺ ion distributions around the triplexes. The presence of Mg²⁺ ions helps stabilize neutral, antiparallel GAA triplexes. These results (along with previous DNA-based studies) allow for the emergence of a complete picture of the stability and structural characteristics of triplexes based on the GAA and TTC/UUC sequences, thereby contributing to the field of trinucleotide repeats and the associated unusual structures that trigger expansion.

The Mr 70,000 protein of the U1 small nuclear ribonucleoprotein particle binds to the 5' stem-loop of U1 RNA and interacts with Sm domain proteins

The U1 small nuclear ribonucleoprotein (snRNP) particle, a cofactor in mRNA splicing, contains nine proteins, six of which are also present in other U snRNPs and three of which are specific to the U1 snRNP. Here we have used a reconstituted human U1 snRNP together with snRNP monoclonal antibodies to define the RNA binding sites of one of the U1 snRNP-specific proteins. When Sm monoclonal antibody (specific for the B', B, and D proteins of U snRNPs) was bound to U1 snRNPs prior to micrococcal nuclease digestion, the same approximately equal to 24 nucleotide fragment of U1 RNA (corresponding to nucleotides 120-143 and termed the "Sm domain") was protected as when no antibody was bound prior to digestion. In contrast, when RNP monoclonal antibody, which reacts with the U1 snRNP-specific Mr 70,000 protein, was bound, additional U1 RNA regions were protected against nuclease digestion. This phenomenon, which we term "antibody-mediated nuclease protection," was exploited to map the position of the Mr 70,000 protein to stem-loop I of U1 RNA. However, there were also sites of Mr 70,000 protein interaction with more 3'-ward regions of U1 RNA, particularly the Sm domain. This indicates that in the three-dimensional structure of the U1 snRNP, the RNP and Sm antigens are in contact with each other. The proximity of the Mr 70,000 protein's RNA binding site (stem-loop I) to the functionally important 5' end of U1 RNA suggests that this protein may be involved in the recognition of, or stabilization of base pairing with, pre-mRNA 5' splice sites.

Antibody-nucleic acid interactions. Antibodies to psoralen-modified RNA as probes of RNA structure

Antisera elicited by immunization of rabbits with 4'-aminomethyl-trioxsalen (AMT)-modified poly(A,U) complexed with methylated bovine serum albumin was characterized in competition radioimmunoassays (RIA) and enzyme-linked immunosorbent assays (ELISA). AMT-poly(A,U) was over 10,000-fold more reactive than unmodified poly(A,U) or AMT alone. The antiserum cross-reacted to varying extents with AMT-modified-RNA's and -DNA's. The presence of AMT-uridine usually assured strong reactivity. The amino group of AMT contributed to antibody binding to a small degree. Binding was not significantly affected by high ionic strength, suggesting that binding does not involve ion pair formation. Murine encephalomyocarditis virus replicative intermediates, as well as cellular RNA and DNA were modified by psoralen in intact cells, suggesting that EMCV RNA and cellular RNA's in intact cells possess detectable stretches of base pairs. The antibodies described here will be useful in studying the secondary and tertiary structure of RNA's in vitro and in intact cells.

In-vivo secondary structure analysis of the small nuclear RNA U1 using psoralen cross-linking

Intramolecular base-pairing interactions have been probed in the small nuclear RNA U1 in vivo. HeLa cells were treated with the psoralen derivative aminomethyltrioxsalen, and cross-linking was carried out by irradiating the intact cells with light of 365 nm wavelength. Cross-linking resulted in a discrete shift in electrophoretic mobility of approximately 65 to 70% of the U1. This intramolecularly cross-linked U1 RNA, termed XU1, was purified and shown to co-migrate with uncross-linked U1 upon photo-reversal of psoralen cross-links with light of 254 nm wavelength. XU1 was also generated by the in-vitro cross-linking of deproteinized U1, suggesting that the secondary structure of U1 RNA in solution is similar to that of U1 ribonucleoprotein in the cell. A sequencing analysis was developed, based on partial enzymatic and alkaline cleavage of psoralen-treated RNA, to identify the position of psoralen cross-links and to distinguish between psoralen monoadducts and diadducts (cross-links). Sequencing of 3' and 5' end-labeled XU1 provided direct evidence for the presence of a unique intramolecular cross-link in XU1, located on uridine 116 (U116). This result is consistent with several secondary-structure models for U1 in which U116 is located in a base-paired stem. The proximity of uridine 96 (U96) to U116 on the opposite side of the base-paired stem suggested that U116 was cross-linked to U96. An additional U1 species having an electrophoretic mobility between those of U1 and XU1 was also generated by psoralen treatment. Analysis of this U1 species, termed U1M, revealed a psoralen monoadduct on U96. Further longwave (365 nm) irradiation of purified U1M resulted in its conversion to XU1 by completion of the U96-U116 cross-link. This suggested that cross-linking at the U96-U116 site occurred as a two-step process in which the psoralen first reacted with U96 and then with U116. Sequencing analysis also identified a psoralen monoadduct on uridine 45 (U45) of XU1. Efficient psoralen-adduct formation, which resulted in cross-linking at the U96-U116 site and monoaddition on U45, suggests that these regions are relatively accessible in the native U1 small nuclear ribonucleoprotein particle in vivo.

The effects of covalent additions of a psoralen on transcription by E. coli RNA polymerase

Synthetic DNA substrates containing a site-specifically engineered psoralen monoadduct or diadduct were used to characterize the response of the E. coli RNA polymerase elongation complex to these lesions. The psoralen derivative HMT (4'-hydroxymethyl-4,5', 8-trimethylpsoralen) was site specifically placed into two synthetic double-stranded DNA fragments each of which contained an E. coli RNA polymerase promoter at one end. The HMT molecule was attached to the middle of the DNA fragments as either a furan-side monoadduct or an interstrand diadduct. Transcription off the HMT crosslinked DNA templates showed that E. coli RNA polymerase terminated at the HMT diadduct site, i. e., one nucleotide before the modified thymidine residue on the template strand. The furan-side monoadduct when on the template strand also blocked transcription by the polymerase. However, no effect on transcription was observed when the monoadduct was located on the non-template strand.

Transcription boundaries of U1 small nuclear RNA

Transcription-proximal stages of U1 small nuclear RNA biosynthesis were studied by 32P labeling of nascent chains in isolated HeLa cell nuclei. Labeled RNA was hybridized to nitrocellulose-immobilized, single-stranded M13 DNA clones corresponding to regions within or flanking a human U1 RNA gene. Transcription of U1 RNA was inhibited by greater than 95% by alpha-amanitin at 1 microgram/ml, consistent with previous evidence that it is synthesized by RNA polymerase II. No hybridization to DNA immediately adjacent to the 5' end of mature U1 RNA (-6 to -105 nucleotides) was detected, indicating that, like all studied polymerase II initiation, transcription of U1 RNA starts at or very near the cap site. However, in contrast to previously described transcription units for mRNA, in which equimolar transcription occurs for hundreds or thousands of nucleotides beyond the mature 3' end of the mRNA, labeled U1 RNA hybridization dropped off sharply within a very short region (approximately 60 nucleotides) immediately downstream from the 3' end of mature U1 RNA. Also in contrast to pre-mRNA, which is assembled into ribonucleoprotein (RNP) particles while still nascent RNA chains, the U1 RNA transcribed in isolated nuclei did not form RNP complexes by the criterion of reaction with a monoclonal antibody for the small nuclear RNP Sm proteins. This suggests that, unlike pre-mRNA-RNP particle formation, U1 small nuclear RNP assembly does not occur until after the completion of transcription. These results show that, despite their common synthesis by RNA polymerase II, mRNA and U1 small nuclear RNA differ markedly both in their extents of 3' processing and their temporal patterns of RNP assembly.

Antibodies against small nuclear ribonucleoproteins immunoprecipitate complexes containing ts110 Moloney murine sarcoma virus genomic and messenger RNAs

Small nuclear ribonucleoproteins (snRNPs) are believed to play a role in processing premessenger RNAs. In this study, snRNPs were immunoprecipitated from extracts of cells infected with ts110 Moloney murine sarcoma virus (ts110 MoMuSV). Both the unspliced 4.0 kb and the spliced 3.5-kb ts110 MoMuSV specific RNA species were found in the immunoprecipitates obtained with monoclonal antibody anti-Sm and polyclonal anti-Sm, anti-(U1) RNP and anti-La sera. Although only a portion of the total ts110 RNAs was present in these immunoprecipitates, immune recognition by the anti-snRNPs was specific and not due to contaminating anti-RNA (at least for the anti-Sm sera) or, to anti-viral protein activities. Genomic 8.3-kb RNA and subgenomic 3.0-kb spliced env mRNA from Moloney murine leukemia virus (MoMuLV) infected cells as well as the cellular actin mRNA were also detected in immunoprecipitates obtained with the same antisera. The fact that pre-mRNAs and mature mRNAs of different origin can be recovered from immunoprecipitates formed with anti-snRNP sera establishes their tight association and confirms the role of snRNPs in mRNA processing.

Identification of proteins that bind tightly to pre-mRNA during in vitro splicing

Incubation of a human beta-globin pre-mRNA in a HeLa cell nuclear extract under conditions permissive for efficient splicing resulted in the assembly of the RNA into ribonucleoprotein (RNP) complexes. This RNP formation occurred largely within the characteristic lag period that precedes splicing. Two classes of RNP were detected by the criterion of their stability in Cs2SO4 gradients. One was unstable and contained mainly aberrant RNA cleavage products. The other class of RNP complexes comprised 50-85% of the beta-globin RNA, formed only under splicing-permissive conditions, was stable in Cs2SO4 gradients, and contained both unspliced pre-mRNA molecules and the lariat intron 1-exon 2 splicing intermediate. This latter class of RNP complexes banded at approximately equal to 1.30 g/cm3, a density very similar to that of native heterogeneous nuclear RNP particles that contain pre-mRNA. RNA-protein crosslinking revealed major proteins of Mr approximately equal to 38,000 and 41,000 in the stable class of RNP. The use of antibodies specific for heterogeneous nuclear RNP core proteins and for small nuclear RNA-associated proteins, in conjunction with [32P]RNA-protein crosslinking, revealed polypeptides having the molecular weights of both sets of antigens. These results show that both heterogeneous nuclear RNP particle core proteins and small nuclear RNA-associated proteins bind tightly to pre-mRNA during splicing in vitro.

Intranuclear localization of snRNP antigens

Anti-Sm antibodies recognize a group of small, nuclear RNA-protein complexes (snRNPs) containing U1, U2, U4, U5, and U6 snRNAs. Anti-RNP antibodies only react with U1 snRNA-containing complexes. The intranuclear distribution of snRNP particles was studied by double immunofluorescence staining of human fibroblasts. Mouse monoclonal anti-Sm antibodies and polyclonal patient sera reacting with different peptides in the snRNP complexes were used. The immunofluorescence patterns obtained with fluorescein isothiocyanate-conjugated anti-mouse Ig and tetramethylrhodamine isothiocyanate-conjugated anti-human Ig second antibodies were examined using computer analysis of digitized images. With this approach the similarity of different patterns could be visualized and estimated with mathematical methods. It was found that human anti-Sm serum as well as three different anti-RNP sera produced speckled patterns overlapping with the anti-Sm monoclonal pattern. Thus, Sm antigenic intranuclear domains also reacted with anti-RNP antibodies, suggesting a high degree of co-localization of the antigenic structures. A partial overlap was found between speckles detected by mouse anti-Sm antibodies and a human La-antiserum. No significant co-localization occurred between speckles detected by mouse anti-Sm antibodies and speckles detected by human antisera reacting with Scl-70 and centromeric antigens. As the U1 snRNP complex is believed to play a role in the splicing of RNA polymerase II transcripts, it appears that the speckles detected by Sm and RNP antibodies may be regions of hnRNA synthesis and mRNA processing. Although no function has been demonstrated for the U2, U4, U5, and U6 snRNPs, the co-localization with the U1 RNA complexes shown in this report indicate that they too participate in some aspect of mRNA processing. The results suggest that computer-assisted analysis of nuclear immunofluorescence patterns will be a useful tool in studies of the spatial and functional organization of the interphase nucleus.

Transcription boundaries of U1 small nuclear RNA

Transcription-proximal stages of U1 small nuclear RNA biosynthesis were studied by 32P labeling of nascent chains in isolated HeLa cell nuclei. Labeled RNA was hybridized to nitrocellulose-immobilized, single-stranded M13 DNA clones corresponding to regions within or flanking a human U1 RNA gene. Transcription of U1 RNA was inhibited by greater than 95% by alpha-amanitin at 1 microgram/ml, consistent with previous evidence that it is synthesized by RNA polymerase II. No hybridization to DNA immediately adjacent to the 5' end of mature U1 RNA (-6 to -105 nucleotides) was detected, indicating that, like all studied polymerase II initiation, transcription of U1 RNA starts at or very near the cap site. However, in contrast to previously described transcription units for mRNA, in which equimolar transcription occurs for hundreds or thousands of nucleotides beyond the mature 3' end of the mRNA, labeled U1 RNA hybridization dropped off sharply within a very short region (approximately 60 nucleotides) immediately downstream from the 3' end of mature U1 RNA. Also in contrast to pre-mRNA, which is assembled into ribonucleoprotein (RNP) particles while still nascent RNA chains, the U1 RNA transcribed in isolated nuclei did not form RNP complexes by the criterion of reaction with a monoclonal antibody for the small nuclear RNP Sm proteins. This suggests that, unlike pre-mRNA-RNP particle formation, U1 small nuclear RNP assembly does not occur until after the completion of transcription. These results show that, despite their common synthesis by RNA polymerase II, mRNA and U1 small nuclear RNA differ markedly both in their extents of 3' processing and their temporal patterns of RNP assembly.

Small nuclear RNAs in the ciliate Tetrahymena

We have isolated and partially characterized a family of small nuclear RNAs (snRNAs) from three different species of the protozoan Tetrahymena. We find six distinct snRNAs ranging in size from 100 to 250 nucleotides. The two largest snRNAs, as well as an abundant, heterogenous group of smaller snRNAs are found in the nucleolar RNA fraction. None of the snRNAs are transcription products of the ribosomal RNA gene or its flanking regions, as shown by hybridization tests. The snRNAs are metabolically stable as determined by pulse/chase experiments and several of them contain a number of modified nuclotides. The snRNAs from Tetrahymena all have slightly different sizes from mammalian snRNAs. The cap structure of the snRNAs from Tetrahymena differs from that of the snRNAs from mammalian cells, but has not yet been fully characterized. The relative amount of snRNAs to total RNA is less in Tetrahymena (greater than 0.1%) than in mammalian cells (2%).

Ribonucleoprotein organization of eukaryotic RNA. XXXII. U2 small nuclear RNA precursors and their accurate 3' processing in vitro as ribonucleoprotein particles

Biosynthetic precursors of U2 small nuclear RNA have been identified in cultured human cells by hybrid-selection of pulse-labeled RNA with cloned U2 DNA. These precursor molecules are one to approximately 16 nucleotides longer than mature U2 RNA and contain 2,2,7-trimethylguanosine "caps". The U2 RNA precursors are associated with proteins that react with a monoclonal antibody for antigens characteristic of small nuclear ribonucleoprotein particles. Like previously described precursors of U1 and U4 small nuclear RNAs, the pre-U2 RNAs are recovered in cytoplasmic fractions, although it is not known if this is their location in vivo. The precursors are processed to mature-size U2 RNA when cytoplasmic extracts are incubated in vitro at 37 degrees C. Mg2+ is required but ATP is not. The ribonucleoprotein structure of the pre-U2 RNA is maintained during the processing reaction in vitro, as are the 2,2,7-trimethylguanosine caps. The ribonucleoprotein organization is of major importance, as exogenous, protein-free U2 RNA precursors are degraded rapidly in the in vitro system. Two lines of evidence indicate that the conversion of U2 precursors to mature-size U2 RNA involves a 3' processing reaction. First, the reaction is unaffected by a large excess of mature U2 small nuclear RNP, whose 5' trimethylguanosine caps would be expected to compete for a 5' processing activity. Second, when pre-U2 RNA precursors are first stoichiometrically decorated with an antibody specific for 2,2,7-trimethylguanosine, the extent of subsequent processing in vitro is unaffected. These results provide the first demonstration of a eukaryotic RNA processing reaction in vitro occurring within a ribonucleoprotein particle.

In vitro reconstitution of messenger ribonucleoprotein particles from globin messenger RNA and cytosol proteins

Deproteinized globin poly(A) + mRNAs reassociate readily in vitro with soluble RNA-binding proteins of the cytosol; reconstituted messenger ribonucleoprotein complexes are obtained which are very similar to native globin polyribosomal-mRNP as far as bouyant density in Cs2SO4 and the composition of proteins which can be crosslinked to the mRNA are concerned. Proteins thus identified bind specifically to mRNA and not to ribosomal RNA or any synthetic oligonucleotides, with one exception: a 78-kDa protein could be cross-linked to poly(A).

The small nuclear RNAs of Drosophila

We have investigated the sequences of the major small nuclear RNAs of Drosophila cultured cells, with the objective of elucidating phylogenetically conserved primary and secondary structures by comparison of the data with previously determined sequences of these RNAs in vertebrate species. Our results reveal striking degrees of conservation between each Drosophila RNA and its vertebrate cognate, and also demonstrate blocks of homology among the Drosophila small nuclear RNAs, as previously described for vertebrates. The most conserved features include the 5' terminal region of U1 RNA, though to function in pre-mRNA splicing, most of the regions of U4 RNA recently implicated in 3' processing of pre-mRNA, and the major snRNP protein binding site ("domain A") that is also shared by vertebrate U1, U2, U4 and U5 RNAs. Several other conserved features have been revealed, suggesting additional regions of functional significance in these RNAs and also providing further insights into the evolutionary history of the small nuclear RNAs.

Ribonucleoprotein organization of eukaryotic RNA. XXXI. Structure of the U1 small nuclear ribonucleoprotein

A small nuclear ribonucleoprotein, U1 snRNP, has been implicated in mRNA processing. In this investigation sites of protein binding on U1 RNA were mapped by nuclease protection and RNA sequencing. Partially purified human U1 snRNP was sequentially digested with Escherichia coli RNAase III and S1 nuclease. The resistant ribonucleoprotein fragments were deproteinized, preparatively hybridized to the U1 RNA--complementary DNA strand of a human U1 gene cloned in bacteriophage M13, and displayed by electrophoresis. The nuclease-resistant U1 RNA fragments were between 23 and 63 nucleotides in length. Most of these fragments were not obtained when protein-free U1 RNA was similarly digested, whereas others were obtained in low yield from U1 RNA and much higher yield from U1 snRNP. RNA sequencing of the fragments revealed that the protein-protected sites in U1 snRNP correspond to base-paired stems I and II, loop a, and portions of stems III and IV (secondary structure nomenclature of Branlant et al., 1981). Single, "bulged" pyrimidines are present within the protein-covered helical regions of stems I and III. Most interestingly, the single-stranded 5' end of U1 RNA, implicated in mRNA splicing, was also highly protected by protein. These results demonstrate that the great majority of U1 RNA is covered by protein in U1 snRNP. The association of protein with the 5' end of U1 RNA is in agreement with recent evidence that snRNP proteins potentiate the binding of this region of U1 RNA with pre-mRNA splice sites.

Transcription-dependent localization of U1 and U2 small nuclear ribonucleoproteins at major sites of gene activity in polytene chromosomes

The location and dynamics of small nuclear ribonucleoproteins (snRNPs) were studied in salivary gland polytene chromosomes of Chironomus tentans by immunofluorescence with specific snRNP antibodies. Monoclonal antibody against the snRNP Sm antigens reacted at all sites of transcription (puffs and Balbiani rings). The amount of snRNP immunofluorescence was strictly dependent on transcription, increasing in parallel with gene activation and decreasing upon repression. Identical patterns of localization and transcriptional dependence were observed with antibodies specific for U1 or U2 snRNPs. These latter results show that the involvement of U1 and U2 snRNPs in transcription-related processes involves a high proportion, rather than small subsets, of active gene loci. In addition, the colocalization of U1 and U2 snRNPs at loci known to contain only one messenger RNA transcription unit (e.g. Balbiani ring 2) raises the possibility that both of these snRNPs interact with the same transcript. Finally, the lack of immunofluorescence at repressed loci indicates that snRNPs are not structural components of the chromatin (DNP) fiber, and also shows that unused snRNPs are not stored in chromatin. These latter points, and the growing evidence for the involvement of U1 snRNP in splicing, suggest that nascent pre-mRNA is the major chromosomal binding site for snRNPs.

Precursors of U4 small nuclear RNA

The processing and ribonucleoprotein assembly of U4 small nuclear RNA has been investigated in HeLa cells. After a 45-min pulse label with [3H]uridine, a set of apparently cytoplasmic RNAs was observed migrating just behind the gel electrophoretic position of mature U4 RNA. These molecules were estimated to be one to at least seven nucleotides longer than mature U4 RNA. They reacted with Sm autoimmune patient sera and a monoclonal Sm antibody, indicating their association with proteins characteristic of small nuclear ribonucleoprotein complexes. The same set of RNAs was identified by hybrid selection of pulse-labeled RNA with cloned U4 DNA, confirming that these are U4 RNA sequences. No larger nuclear precursors of these RNAs were detected. Pulse-chase experiments revealed a progressive decrease in the radioactivity of the U4 precursor RNAs coincident with an accumulation of labeled mature U4 RNA, confirming a precursor-product relationship.

Ultraviolet-crosslinking reveals specific affinity of eukaryotic initiation factors for Semliki Forest virus mRNA

Eukaryotic initiation factors (eIF) associate readily with 32P-labeled Semliki Forest virus (SFV) mRNA in vitro, forming complexes which can be crosslinked by 254 nm ultraviolet irradiation. After ribonuclease digestion, the initiation factors were released and analysed by gel electrophoresis. Autoradiography revealed proteins by virtue of crosslinked 32P-labeled mRNA fragments. eIF-4A, -4B and -4C as well as three subunits of eIF-3 could be crosslinked with SFV mRNA. None of these proteins bound to ribosomal RNAs.