| Nucleic Acids Research | Pages |
RNA binding specificity of Unr, a protein with five cold shock domains
Introduction
Materials And Methods
Oligonucleotides and DNA templates
Recombinant proteins
In vitro selection
RNA transcription
Determination of RNA-binding affinities
Results
In vitro selection of Unr-binding RNA
Affinity of Unr for the selected RNAs
RNA sequence requirements for binding
RNA binding properties of Unr derivatives
Discussion
Acknowledgements
References
RNA binding specificity of Unr, a protein with five cold shock domains
ABSTRACT
INTRODUCTION
The unr gene (upstream of N-ras) was identified as a transcription unit located immediately upstream of the N-ras gene in the genome of several mammalian species (1-3). The unr gene is ubiquitously expressed in cell lines and tissues (2,3). It encodes Unr, a highly conserved 85 kDa protein (99% amino acid identity between rat and human), which contains five cold shock domains (CSD). Unr is essential for mouse development, since embryos homozygous for a disrupted unr allele die at mid-gestation (O.Boussadia et al., manuscript in preparation).
The CSD is the most evolutionarily conserved nucleic acid-binding protein domain, found in bacteria and eukaryotes (4,5). This domain of ~70 amino acid residues mediates binding to single-stranded DNA and RNA (6,7). Recent studies have revealed similarities between the CSD and the RRM (RNA recognition motif), a motif found in a variety of functionally diverse RNA-binding proteins and which has been characterized in detail. The CSD contains the RNA-binding motifs RNP-1 and RNP-2, which are the hallmark of RRM (8,9). Moreover, the solution and crystal structure of a CSD determined from bacterial cold shock proteins revealed a striking similarity with the RRM, both domains folding in a compact [beta]-barrel, with the basic and aromatic amino acids of RNP-1 and RNP-2 being solvent exposed (10-12). Those residues which make direct contacts with RNA in the RRM protein U1A (13,14) are also involved in nucleic acid-binding of the bacterial protein CspB (15).
CSD proteins, prokaryotic and eukaryotic, have been found up to now to be involved in two processes: transcriptional and translational control. In prokaryotes, most studies have been developed around CspA and CspB, the major cold shock proteins in Escherichia coli and Bacillus subtilis, respectively. CspA and CspB are massively and transiently induced after a temperature downshift and are involved in the adaptation to cold shock (16). These two homologous proteins consist of a unique 67 amino acid CSD and function as transcriptional inducers of genes involved in the cold shock response, most likely by binding to the open transcription complex (17-19). In addition, CspA has been shown to increase translation of its own mRNA, by promoting destabilization of secondary structures (20).
In eukaryotes, the Y-box factors (YB) have been described in vertebrates; they contain a single CSD and an auxiliary basic-aromatic tail domain. Initially, YB proteins were characterized as transcription factors, interacting with a variety of double-stranded or single-stranded DNA sequences in promoters (6). Subsequently, some members of the YB family were identified as major components of messenger ribonucleoprotein particles (mRNPs) in germ cells (FRGY2/mRNP4 and MSY1) and somatic cells (p50) (21-25). In vitro, Y-box factors are involved in translational control, a positive or negative effect depending on the protein concentration (26). Specifically, in Xenopus, FRGY2/mRNP4 plays an essential role in the regulation of translation during early embryonic development and is involved in coupling transcription and translational repression of certain mRNAs (27,28).
Besides the YB proteins, three other CSD proteins have been described in eukaryotes: Drosophila YPS which contains one CSD and one RGG box (29), Caenorhabditis elegans Lin-28 which contains one CSD and two zinc finger motifs (30) and Unr. Lin-28 and YPS are cytoplasmic proteins and are likely to function in post-transcriptional regulation. Unr is also mostly localized in the cytoplasm, in part associated with the endoplasmic reticulum (31), which suggests a role for Unr in cytoplasmic mRNA metabolism. Unr was recently shown to be required for internal initiation of translation of human rhinovirus RNA (32).
In our initial characterization of the human Unr protein, we have determined that it has the capacity to interact in vitro with single-stranded DNA and RNA. Competition analysis of Unr-RNA interaction indicated that among simple polymers poly(A/G) was the best competitor, providing a first indication of its sequence specificity. To further characterize the RNA-binding specificity of Unr and eventually identify RNA ligands, we have used an in vitro selection/amplification approach (SELEX; 33). As previous studies with Unr (31) and FRGY2/mRNP4 (34) indicated that their interaction with RNA was inhibited by MgCl2, we performed two independent SELEX experiments, in the presence or absence of MgCl2. In this study we isolated high affinity binding sites for Unr and through mutagenesis of selected RNAs, we analyzed sequence requirement for Unr-RNA interaction. We also determined that a single CSD exhibits the same sequence specificity as Unr, although with a 5-fold lower affinity.
MATERIALS AND METHODS
Oligonucleotides and DNA templates
Synthetic oligodeoxynucleotides (Oligo Express) used for SELEX experiments were identical (Rev) or similar (T7 and Random N20) to those used by Tsaï et al. (35): T7, 5[prime]-TGCATGGATCCTAATACGACTCACTATAGGGGCCACCAACGACATT-3[prime]; Random N20, 5[prime]-CCCGGTGGTTGCTGTAA(N)20CAACTATATTTATCACGGGT-3[prime]; Rev, 5[prime]-CAACTATATTTATCACGGGTACTTAAGCTGTC-3[prime]. Restriction sites for BamHI and EcoRI were introduced in oligos T7 and Rev, respectively (underlined), for cloning. T7 primer contained a T7 RNA polymerase promoter sequence.
The mutant DNA templates were synthesized by PCR as described for the SELEX pool C0, using T7 and Rev as ampli-fication primers. The oligonucleotide templates for mutagenesis have the same fixed sequences as oligo N20, but the 20 nt internal sequence was substituted by the indicated sequences in Table
Sequencing primer: 5[prime]-GCTATGACCATGATTACGCC-3[prime].
Recombinant proteins
Constructs. The human cDNA encoding full-length Unr (767 amino acids), in which the ATG initiating codon was substituted by a BamHI site, was subcloned as a BamHI-SmaI fragment into the BamHI and PvuII sites of E.coli expression vector pRSETA (Invitrogen). The fusion protein derived from this plasmid contains a 36 amino acid tag that includes six consecutive histidine residues. Similarly, a cDNA with a stop codon at position +381 of the open reading frame, encoding a 127 amino acid protein (CSD1, amino acids 2-128), was subcloned in the same way in pRSETA (Invitrogen). The Unr derivatives CSD1-2-3 (amino acids 15-387) and CSD5 (amino acids 629-718) were generated by PCR using Pfu turbo DNA polymerase (Stratagene) and human unr cDNA as template. The PCR primers introduced an in-frame stop codon, as well as BamHI and HindIII sites at the 5[prime]- and 3[prime]-ends of the amplified fragment, respectively. The Unr mutant protein CSD1*2345 was a gift from Dr R Jackson; this protein was purified from E.coli as a C-terminal histidine-tagged protein (32). In this mutant protein, the essential phenylalanine of the RNP-1 motif of the first CSD (amino acid 39) was mutated to alanine to prevent RNA-binding activity of this CSD.Protein expression and purification. The His-tagged proteins were expressed in E.coli BL21 cells using the pET system (Novagen) and purified on a Ni2+-NTA-agarose column. Briefly, ~0.5-1 mg of protein equilibrated in urea-Tris buffer (9 M urea, 20 mM Tris-HCl, pH 7.4) was incubated with 2 ml of Ni2+-NTA-agarose (Qiagen) for 40 min. The beads were transferred into a column and successively washed with decreasing urea concentrations (from 9 M to none) in the same buffer, allowing renaturation of the proteins. Proteins were eluted with 2 ml of 500 mM imidazole and dialyzed against TNG buffer (20 mM Tris-HCl, pH 7.4, 150 mM NaCl, 10% glycerol). For Unr, purification by SDS-polyacrylamide gel electrophoresis was performed as previously described (31) prior to loading on the Ni2+-NTA column. The length and protein concentrations were estimated from SDS-PAGE electrophoresis after Coomassie Brilliant Blue staining, using protein molecular weight markers as standard. Protein fractions were stored at -70°C until use.In vitro selection
A degenerate double-stranded DNA template was synthesized by PCR using the three oligonucleotides T7, Rev and N20 (template for PCR reaction). PCR was performed with 0.8 µg of T7 and Rev primers and 120 ng of N20, with native Pfu polymerase (Stratagene) for nine cycles.The resulting DNA pool consisted of about 3 × 1012 molecules (1012 unique sequences). An aliquot of 1 µg of the DNA pool was in vitro transcribed with T7 RNA polymerase (Gibco BRL) to give the 32P-labeled RNA pool C0. Selection steps were performed using Unr protein immobilized on Ni2+-NTA-agarose (five initial cycles) and by a gel mobility shift assay (five subsequent cycles). Selection of RNA on Unr-Ni2+-agarose beads was carried out in a batchwise fashion as follows: 5 µg of purified His-Unr protein were mixed with 25 µl of Ni2+-NTA-agarose equilibrated with RNA selection buffer [25 µg/ml BSA, 25 µg/ml tRNA, 5 U/ml RNAguard (Pharmacia), 1 mM PMSF, 1 mM [beta]-mercaptoethanol and 1 µg/ml of leupeptin, antipain and aprotinin, in TNG buffer] in a final volume of 100 µl and incubated for 45 min at 25°C with shaking. After three washes with 0.5 ml of selection buffer, Ni2+-agarose-bound His-Unr protein was mixed with 200 µl of the same buffer containing 0.25 µg of labeled RNA pool and 80 U of RNAguard. After a 30 min incubation at 30°C with occasional shaking, the resin was pelleted at 1000 g and free RNA was removed by washing three times with 0.5 ml of ice-cold binding buffer. Bound RNA was eluted by two incubations with 0.4 ml of elution buffer (0.1 M glycine-HCl, pH 4.0, 25 µg/ml tRNA, 0.025% SDS, 5 mM MgCl2), for 5 min at 35°C. The eluate was extracted with phenol/chloroform and chloroform and RNA was ethanol precipitated. Half of the selected RNA was reverse transcribed for 1 h at 42°C in 30 µl of RT buffer (Promega), with 10 U of avian myeloblastosis virus reverse transcriptase (Promega) and 100 ng of Rev primer. After phenol/chloroform and chloroform extractions, the cDNAs were ethanol precipitated with 1 µg of tRNA. The cDNAs were then diluted 2-fold and amplified by 15 cycles of PCR, with 2.5 U of native Pfu polymerase, 600 ng of T7 and Rev primers in 100 µl of PCR buffer. The PCR DNA product was then in vitro transcribed and utilized for the next round of selection. After the fifth cycle, the RNA pool was applied to Unr-less Ni2+-NTA-agarose to remove non-specific RNA. Five additional selection/amplification cycles were performed using a band shift assay to separate Unr-RNA complexes. An aliquot of 50 ng of labeled RNA pool was incubated in 40 µl of selection buffer with 250 ng (70 nM) of Unr protein. Following a 30 min incubation at 30°C, Unr-bound RNAs were electrophoresed in an 8% polyacrylamide gel in 1× Tris-borate-EDTA buffer at 4°C overnight, at 2.5 V/cm. After a -70°C gel exposure to X-ray hyperfilm (Amersham Life Science), the bands corresponding to complexed RNAs were located and excised. Elution was carried out in 0.5 M NH4Ac, 1 mM EDTA, overnight at 37°C. The eluate was extracted with phenol/chloroform and chloroform and ethanol precipited with 1 µg of tRNA. RT-PCR was carried out as described above. After the last PCR, the DNA was digested with EcoRI and BamHI and cloned into the plasmid pUC13 and 50 individual clones were sequenced using a primer located 60 nt upstream of the random sequence. DNA sequencing was performed with Sequenase v.2.0 (US Biochemical Corp.)
RNA transcription
EcoRI-linearized clones or EcoRI-digested PCR products were in vitro transcribed using 50 U of T7 RNA polymerase, with 20 µCi of [[alpha]-32P]UTP (300 Ci/mmol) and 1 mM unlabeled nucleotides. Resulting labeled RNAs (62 nt long) were purified by phenol/chloroform and ethanol precipitation and their integrity was checked by electrophoresis on a urea-acrylamide gel. [[alpha]-32P]UTP incorporation was quantified to estimate RNA concentration.
Determination of RNA-binding affinities
Nitrocellulose filter-binding assays. Aliquots of 1-3 fmol of labeled RNA were incubated at 30°C for 20 min in 20 µl of RNA-binding buffer (25 µg/ml BSA, 10 ng of tRNA, 1 mM [beta]-mercaptoethanol and 1 mM PMSF, in TNG buffer), with increasing amounts of protein as indicated. After cooling on ice for 5 min, the reaction was filtered through a wet 0.45 µm pore nitrocellulose filter (Schleicher & Schuell) under gentle suction and dried. Retention of labeled RNA was analyzed by liquid scintillation. Each assay included a control for RNA retention in the absence of protein. Percentage of RNA bound to filters was corrected by substracting the retention of free RNA (1-2%) from data points. Affinity constants were derived by plotting the fraction of complexed RNA as a function of protein concentration (36). Efficiencies of retention for the RNAs studied here ranged from 50 to 80%. The fraction of complexed RNA was taken from the plateau of binding and the plateau was assigned as 1.0. The reported Kd values are an average of three binding curves. The variation in Kd values obtained from different experiments was ~20-30%.Gel retardation assays. RNA/protein incubations were performed as described for nitrocellulose assays, in a 40 µl final volume. After cooling on ice, the mixture was directly loaded on a 3.8% polyacrylamide gel (acrylamide:bisacrylamide, 40:1) containing 5% glycerol and 25 mM Tris-glycine, pH 8.0. The gel was electrophoresed at 4°C for 3-4 h at 20 V/cm and then dried. The amount of free and bound RNA was analyzed with a phosphor screen and a Storm 860 (Molecular Dynamics).RESULTS
In vitro selection of Unr-binding RNA
To investigate the sequence requirements for Unr-RNA interaction, we used the selection amplification procedure (SELEX; 33). The RNA population used consisted of 62 nt long molecules, with 20 nt of randomized sequence (35). Selection of RNA-protein complexes was performed with hexahistidine-tagged protein. Recombinant His-Unr protein was purified to homogeneity as shown in Figure
Figure 1. (A) Schematic representation of the Unr protein and of the mutants used in this study. The full-length human Unr protein and the CSD1, CSD5 and CSD1-2-3 derivatives contain a 36 amino acid histidine tag at their N-terminus. The CSDs are boxed and RNP1 and RNP2 motifs are indicated. (B) SDS-PAGE analysis of purified proteins. Aliquots of 2 µg of each protein were loaded on a 15% SDS-polyacrylamide gel that was stained with Coomassie Blue. The size of the molecular weight markers is indicated on the left. Figure 2. Nucleotide RNA sequences selected by Unr protein from a randomized pool. The names of the individual clones are given to the left of each selected sequence. Individual sequences are classified as the GU group (A) or the AC group (B) and aligned by shared sequence motif indicated by bold, underlined characters. The variable region is in capital letters and nucleotides belonging to the 5[prime] and 3[prime] flanking constant regions (two or more are shown) are in lower case. The frequencies of nucleotides selected at each position of the bolded underlined region and the two deduced consensus motifs are shown at the bottom of each group of selected sequences. Individual bases are specified when they appear with a frequency [ge]66%. Ten rounds of selection/amplification were performed. For the first five rounds of selection, His-Unr protein immobilized on a Ni2+-NTA matrix was used, as described for the identification of specific RNA ligands for the splicing factors ASF/SF2, SC35 and RBP1 (37,38). To prevent selection of RNAs that interact with the Ni2+ matrix (39), we used a band shift assay for the five subsequent rounds of selection to isolate Unr-RNA complexes. The first SELEX experiment, SELEX A, was performed in the absence of MgCl2. After a total of 10 cycles, 46 clones were sequenced.The sequences of 36 inserts of SELEX A are presented in Figure A second SELEX (SELEX B) was carried out under the same conditions as SELEX A, except for the presence of 1 mM MgCl2. After a total of 10 cycles, 50 clones were sequenced. The purine enrichment was similar to that observed in SELEX A and 27 sequences could be aligned in the two groups defined by the results of SELEX A (not shown). As for SELEX A, the 23 other sequences presented a high purine content, but the presence of several pyrimidines made possible multiple alignments with the consensus which did not provide additional information. The presence of Mg2+ cation in the selection therefore did not reveal other sequence motifs or structures in the Unr RNA targets. In the subsequent analysis some sequences obtained in SELEX B were included for comparison and were denoted ‘b’. The high purine content of the selected sequences should prevent the formation of secondary structure by Watson-Crick base pairs. To further investigate the structural context of the purine clusters, we analyzed 15 sequences using the Mfold software (40). In 11 cases, and irrespective of the presence or absence of magnesium during the selection process, the selected sequence was clearly predicted to be located within a single-stranded domain of the molecule. These results thus agree with the previously observed lack of affinity of Unr for double-stranded RNA (31). In summary, the same consensus motifs were obtained in the two independent SELEX experiments. These consensus sequences are characterized by the motif AAGUA or AACG/A located downstream of a 5-8 nt long purine stretch.
Affinity of Unr for the selected RNAs
To further characterize the interaction of Unr with the selected sequences, we first used a gel mobility shift assay. Individual sequences were transcribed from the corresponding plasmids after linearization at their 3[prime]-end, yielding molecules of the same size as those used in the SELEX experiments. Figure
Figure 3. Gel mobility shift assay of His-Unr with several selected and unselected RNAs. Aliquots of 2 fmol of 32P-labeled RNAs were incubated with increasing protein concentrations and the products were resolved by native gel electrophoresis as described in Materials and Methods. The RNAs analyzed were from SELEX A (RNAs 76 and 77) or SELEX B (RNA 10b). A random RNA (Ctr) was used as a control. The RNAs are numbered on top and the concentrations of Unr are indicated above the lanes. Positions of free RNA (probe) and RNA-protein complexes (C) are indicated with arrows on the left. RNA insert sequences are presented in Figure 2 (RNAs 76 and 77) or in Table 1 (RNA 10b). In these experiments, one major band was observed which could be accounted for by the formation of 1:1 RNA-Unr complexes. However, at the highest Unr concentration one or two minor species could also be observed (Fig. These band shift experiments were suggestive of an apparent dissociation constant for the selected sequences in the 20 nM range. To refine and extend this analysis to a larger set of selected sequences, we used a nitrocellulose filter-binding assay (Fig. Figure 4. His-Unr binding curves on several selected RNAs. Graphs show some results of filter-binding experiments, in which the percentage of complexed labeled RNA retained on the filter is plotted against the amount of protein added to the binding reaction. The protein concentration is varied, whereas RNA (~104 c.p.m.) is held at a constant concentration of 0.15 nM. Binding reaction and filtration are described in Materials and Methods. (A) Incubation mixture does not contain MgCl2; (B) incubation mixture contains 1 mM MgCl2. Symbols used to represent selected RNAs and Ctr RNA (control) are indicated at the top. Table 1. Figure The dissociation constants of a total of 13 sequences are compiled in Table 1. Ten sequences had dissociation constants of 10 ± 2 nM and three had slightly higher Kd values (16, 18 and 20, respectively). Thus a majority of the isolated sequences are high affinity binding sites with indistinguishable affinities for Unr, a few sequences having slightly lower affinities. As the SELEX procedure is known to converge very slowly in the presence of multiple targets with comparable affinities these results clearly indicate the efficiency of the selection. More specifically, data presented in Table In summary, all of the aligned selected RNA sequences, whether of the GU or AC consensus group, are bound with a high affinity by Unr. Moreover, single base changes (Pu->Py) have very limited effect on Unr-RNA recognition.
RNA sequence requirements for binding
To assess the functional significance of the upstream purine stretch and of the adjacent conserved motif, mutations were introduced in two selected sequences, 85 and 78 (Table 2). Introduction of a single pyrimidine within the upstream purine stretch (mutant 85M2) decreased the affinity slightly, while the introduction of a pyrimidine at position +1 (mutant 85MT4) had no significant effect. These results therefore confirm the limited effect on Unr binding of point mutations within the consensus. In mutants M3 and M5, pyrimidine-rich sequences were introduced to preserve only the conserved core motif. These mutations reduced the interaction by 2.5- to 3-fold, revealing that the core motif surrounding the pyrimidine is not sufficient for a high affinity interaction. Eliminating the conserved core motif while preserving the upstream purine stretch (mutant MT3, AGUA->UUCU) decreased the affinity by 5-fold, demonstrating the importance of the core motif for Unr-RNA interaction.
Table 2.
These results underscore the importance of the observed organization of the consensus sequences and suggest that, by itself, a high purine content should not be sufficient to create an optimal binding site for Unr. To confirm this we generated an RNA target with a 70% purine content but containing no stretch of more than three purines (Sc Pu). Despite its high purine content, this sequence displayed a 10-fold lower affinity for Unr.
In summary, both the upstream purine stretch and the conserved environment of the pyrimidine contribute to the high affinity for Unr. The presence of only one of these elements reduces affinity by a factor of three and, in their absence, the affinity decreases by a factor of 10.
RNA binding properties of Unr derivatives
Since Unr contains five CSDs and two consensus sequences were identified in the SELEX experiments, it was possible that different CSDs had selected these two sets of sequences. To assess the role of individual CSDs we constructed three proteins, which are schematically presented in Figure Figure 5. RNA binding specificity of CSD1. The binding of CSD1 to Unr-selected RNAs, mutant RNAs and random control RNAs was analyzed using a nitrocellulose filter-binding assay as in Figure 4. Each binding curve represents the results of three independent measurements. Symbols used to represent tested RNAs are indicated at the top. Table 3. These results indicate that a single CSD (CSD1) can account for a specific interaction with the two groups of winner sequences. This observation is also true for the C-terminal CSD (CSD5) although in this case, the overall affinity for RNA is reduced by 100-fold. The presence of three CSDs as in CSD1-2-3 is sufficient to confer the same affinity for the winner sequences as the full-length protein indicating that not all the CSDs are required. Intriguingly, the presence of a functional CSD1 is not required to achieve either affinity or specificity, suggesting the existence of redundancy between the CSDs of Unr.
DISCUSSION
In this study, in vitro selection experiments were performed to determine the RNA-binding specificity of the human Unr protein. Two SELEX experiments were carried out, differing only by the presence or absence of MgCl2. In both experiments, long purine-rich sequences (11-14 purines and a single pyrimidine) were recovered after 10 rounds of selection. Alignment of the sequences of 36 of 46 clones (SELEX -Mg2+) and of 27 of 50 clones (SELEX +Mg2+) identified two related consensus motifs: 5[prime]-(Pu)5AAGUA(Pu)-3[prime] and 5[prime]-(Pu)8AACG/A(Pu)2-3[prime] characterized by an upstream purine tract and an adjacent conserved motif surrounding the pyrimidine. Selected RNA sequences from either consensus group bound Unr with a high affinity (Kd values in the range of 10 nM), as determined by nitrocellulose and mobility shift assays. Stable secondary structures that rely on canonical base pairs do not exist for the two consensus sequences, suggesting that Unr discriminates between RNA via their primary sequences rather than their secondary structure. Interestingly, the affinity of Unr for the selected sequences shows little sensitivity to the presence of MgCl2 in contrast to our previous observation with longer RNA substrates. One possibility, which we are currently investigating, is that on long RNA molecules multimeric Unr complexes are formed which are sensitive to the presence of Mg2+. Alternatively, magnesium could act via an effect on RNA conformation, by stabilizing secondary structures, which are frequently extensive on long RNA molecules.
Unr binds specifically to the selected RNAs; its affinity for selected sequences is at least 10-fold higher than for non-selected control sequences. Independently of the nature of the pyrimidine, the interaction of Unr with selected RNAs is characterized by a high affinity (Kd = 8-20 nM) which varies little between individual sequences. These data indicate that Unr-RNA recognition is not rigidly sequence dependent. To better understand the basis of Unr-RNA binding specificity, mutations were introduced in two of the selected RNAs. While a single base change interrupting the purine stretch decreases the affinity by less than 2-fold, eliminating one of the elements of the consensus, i.e. the upstream purine stretch or the pyrimidine-containing core motif, decreases the affinity by 3- to 4-fold. Finally, scrambling of a selected sequence leads to a 10-fold lower affinity. Together, these data eliminate the possibility that Unr affinity for RNA depends only on the purine content of the molecules; they support the notion that the two parts of the consensus sequences, the upstream purine stretch and the adjacent pyrimidine-containing core motif, are necessary to constitute an optimal substrate for Unr.
Two broad classes of RNA-binding proteins can be distinguished with respect to their RNA target structures. In the first class, RBPs interact with stem-loop structures. These proteins include U1A (43), U2B[prime][prime] (44), nucleolin (45), iron regulatory factor IRF (46) and eIF-4B (47). Proteins from this class can bind to their RNA targets with high affinity (Kd = 10-9-10-11 M) and specificity. In the second class, RBPs interact specifically with single-stranded RNA sequences. These proteins include poly(A)-binding protein (48), the splicing factors ASF/SF2, SC35 and Sxl (37,49), the polypyrimidine tract-binding protein (49,50), and the polyadeny-lation factor CstF (51,52). SELEX analysis of the binding specificity of these proteins usually leads to degenerate consensus sequences reflecting the existence of several high affinity binding sites. Point mutations of the conserved nucleotides of these consensus sequences are often associated with limited variations in affinity in vitro (49). Unr, with its degenerate consensus and limited sensitivity to point mutations, clearly belongs to this class of proteins. In this class, it is interesting to note the analogy with PTB, a protein involved in regulation of splicing (49,50) and internal initiation of translation (53) for which the SELEX consensus is similarly organized with a core motif embedded in a pyrimidine-rich sequence PTB (50).
As Unr contains five CSDs, the pool of selected sequences could be the resultant of the sequences selected by individual CSDs. Specifically, the two consensus sequences, which differ by the nature of the pyrimidine, could have been selected by distinct CSDs. However, a protein which contains only the N-terminal CSD (CSD1) exhibits the same sequence specificity as Unr, albeit with a 5-fold lower affinity. Moreover, although a protein which contains only the C-terminal CSD (CSD5) has a much lower affinity for RNA than Unr (Kd values in the micromolar range), it retained a similar sequence specificity. Thus, these studies indicate that a single CSD has the ability to interact with the two consensus sequences obtained in the SELEX experiments. In this respect, Unr differs from many of the RBPs with multiple RRM, as in several instances it has been shown that the observed sequence specificity requires the contribution of several RRMs: ASF/SF2, (37); PABP, (54,55); nucleolin, (56); Sxl, (57).
Within Unr several CSDs are required to achieve a high affinity for the selected sequences indicating that several CSDs can contribute to the binding energy. The properties of CSD1 suggest that the N-terminal CSD could be the major contributor to the binding energy, other CSDs, as in CSD1-2-3 and Unr, playing the role of auxiliary domains. However, analysis of CSD1*2-3-4-5 reveals that CSD1 is not absolutely required for high affinity binding, indicating that other CSDs can substitute for it. Taken together these results suggest that within Unr several CSDs can contribute to the binding energy, but that several combinations of CSDs can lead to similar RNA-protein interaction. This and the idea that a single CSD is sufficient to achieve the observed sequence specificity lead to the proposition that there is some redundancy between the CSDs of Unr. Further studies will be required to determine how many CSDs can simultaneously interact with a substrate.
Up to now, a characterization of RNA binding specificity for CSD proteins using the selection/amplification procedure has only been performed for Unr (this work) and for the two homologous Y-box proteins FRGY1 and FRGY2 (58). Unr and FRGY1/FRGY2 have the capacity to bind RNA with sequence specificity, but distinct consensus sequences were obtained. The consensus for FRGY1 and FRGY2 (AACAUC) clearly differs from the two related purine-rich consensus sequences for Unr. FRGY1 and FRGY2 contain only one CSD (almost identical in the two proteins) and one additional tail-binding domain. As shown by Bouvet et al. (58), it is the CSD that determines RNA-binding specificity of FRGY1/FRGY2, the tail domain contributing to the overall affinity. The bacterial cold shock proteins CspA and CspB, which consist of only a single CSD, bind RNA with a low affinity (Kd = 3 × 10-5-5 × 10-6 M) and a broad specificity (5,59). It thus appears that the CSD can mediate specific RNA interactions, as observed for Unr and FRGY1/FRGY2, or more relaxed ones, as reported for CspA and CspB. In this respect, the CSD behaves similarly to the RRM, which can bind to a variety of RNA sequences with various levels of specificity (60,61). According to the analysis of the human U1A protein (62,63), the high level of amino acid variation between CSDs (up to 70%) should allow a large range of sequence specificity.
Because of the sequence flexibility of the consensus, the results of the present study do not allow a direct identification of the mRNA targets for Unr. However, an ongoing study of cellular factors involved in internal initiation of translation of the HRV-2 rhinovirus has identified Unr as one of them (32). This raises the possibility that Unr could be involved in internal initiation of translation of cellular mRNAs. Indeed, an increasing number of cellular mRNAs, including those of some growth factors and developmental regulators, have been shown to be translatable by internal initiation (64-66). The consensus identified in this work will foster studies on the role of Unr in internal initiation of translation both in vitro and in vivo.
ACKNOWLEDGEMENTS
We thank Dr Jackson for communication of experimental data prior to publication and for giving us the mutant protein CSD1*2-3-4-5. We are grateful to D. Karlin for help in the computer analysis of RNA secondary structures and to P. Gondran, N. Modjtahedi, G. Pierron and D. Weil for their critical comments of the manuscript. This work was supported by the association pour la Recherche contre le Cancer by a research grant (6307).
REFERENCES
This article has been cited by other articles:
This page is run by Oxford University Press, Great Clarendon Street, Oxford OX2 6DP, as part of the OUP Journals
Comments and feedback: www-admin{at}oup.co.uk
Last modification: 26 Mar 1999
Copyright©Oxford University Press, 1999.
CiteULike 
Connotea 
Del.icio.us What's this?
E. C. Anderson, S. L. Hunt, and R. J. Jackson
Internal initiation of translation from the human rhinovirus-2 internal ribosome entry site requires the binding of Unr to two distinct sites on the 5' untranslated region
J. Gen. Virol.,
November 1, 2007;
88(11):
3043 - 3052.
[Abstract]
[Full Text]
[PDF]
C.-Y. A. Chen, Y. Yamashita, T.-C. Chang, A. Yamashita, W. Zhu, Z. Zhong, and A.-B. Shyu
Versatile applications of transcriptional pulsing to study mRNA turnover in mammalian cells
RNA,
October 1, 2007;
13(10):
1775 - 1786.
[Abstract]
[Full Text]
[PDF]
M. Dinur, R. Kilav, A. Sela-Brown, H. Jacquemin-Sablon, and T. Naveh-Many
In Vitro Evidence that Upstream of N-ras Participates in the Regulation of Parathyroid Hormone Messenger Ribonucleic Acid Stability
Mol. Endocrinol.,
July 1, 2006;
20(7):
1652 - 1660.
[Abstract]
[Full Text]
[PDF]
I. Abaza, O. Coll, S. Patalano, and F. Gebauer
Drosophila UNR is required for translational repression of male-specific lethal 2 mRNA during regulation of X-chromosome dosage compensation
Genes & Dev.,
February 1, 2006;
20(3):
380 - 389.
[Abstract]
[Full Text]
[PDF]
G. P. Patel, S. Ma, and J. Bag
The autoregulatory translational control element of poly(A)-binding protein mRNA forms a heteromeric ribonucleoprotein complex
Nucleic Acids Res.,
December 14, 2005;
33(22):
7074 - 7089.
[Abstract]
[Full Text]
[PDF]
E. A. S. Lang and M. V. Marques
Identification and Transcriptional Control of Caulobacter crescentus Genes Encoding Proteins Containing a Cold Shock Domain
J. Bacteriol.,
September 1, 2004;
186(17):
5603 - 5613.
[Abstract]
[Full Text]
[PDF]
Y. Martineau, C. Le Bec, L. Monbrun, V. Allo, I.-M. Chiu, O. Danos, H. Moine, H. Prats, and A.-C. Prats
Internal Ribosome Entry Site Structural Motifs Conserved among Mammalian Fibroblast Growth Factor 1 Alternatively Spliced mRNAs
Mol. Cell. Biol.,
September 1, 2004;
24(17):
7622 - 7635.
[Abstract]
[Full Text]
[PDF]
T.-C. Chang, A. Yamashita, C.-Y. A. Chen, Y. Yamashita, W. Zhu, S. Durdan, A. Kahvejian, N. Sonenberg, and A.-B. Shyu
UNR, a new partner of poly(A)-binding protein, plays a key role in translationally coupled mRNA turnover mediated by the c-fos major coding-region determinant
Genes & Dev.,
August 15, 2004;
18(16):
2010 - 2023.
[Abstract]
[Full Text]
[PDF]
E. C. Brown and R. J. Jackson
All five cold-shock domains of unr (upstream of N-ras) are required for stimulation of human rhinovirus RNA translation
J. Gen. Virol.,
August 1, 2004;
85(8):
2279 - 2287.
[Abstract]
[Full Text]
[PDF]
H. L. Wilson and H. C. O'Neill
Identification of differentially expressed genes representing dendritic cell precursors and their progeny
Blood,
September 1, 2003;
102(5):
1661 - 1669.
[Abstract]
[Full Text]
[PDF]
O. Boussadia, M. Niepmann, L. Creancier, A.-C. Prats, F. Dautry, and H. Jacquemin-Sablon
Unr Is Required In Vivo for Efficient Initiation of Translation from the Internal Ribosome Entry Sites of both Rhinovirus and Poliovirus
J. Virol.,
March 15, 2003;
77(6):
3353 - 3359.
[Abstract]
[Full Text]
[PDF]
O. I. Stenina, K. M. Shaneyfelt, and P. E. DiCorleto
Thrombin induces the release of the Y-box protein dbpB from mRNA: A mechanism of transcriptional activation
PNAS,
May 30, 2001;
(2001)
121592298.
[Abstract]
[Full Text]
[PDF]
M. Pelletier, M. M. Miller, and L. K. Read
RNA-binding properties of the mitochondrial Y-box protein RBP16
Nucleic Acids Res.,
March 1, 2000;
28(5):
1266 - 1275.
[Abstract]
[Full Text]
[PDF]
O. I. Stenina, K. M. Shaneyfelt, and P. E. DiCorleto
Thrombin induces the release of the Y-box protein dbpB from mRNA: A mechanism of transcriptional activation
PNAS,
June 19, 2001;
98(13):
7277 - 7282.
[Abstract]
[Full Text]
[PDF]
This Article 
Abstract
Print PDF (287K)
Alert me when this article is cited
Alert me if a correction is posted
Services
Email this article to a friend
Similar articles in this journal
Similar articles in ISI Web of Science
Similar articles in PubMed
Alert me to new issues of the journal
Add to My Personal Archive
Download to citation manager
Search for citing articles in:
ISI Web of Science (25)
Request Permissions
Commercial Re-use Guidelines
for Open Access NAR Content
Google Scholar
Articles by Triqueneaux, G.
Articles by Jacquemin-Sablon, H.
Search for Related Content
PubMed
PubMed Citation
Articles by Triqueneaux, G.
Articles by Jacquemin-Sablon, H.
Social Bookmarking
What's this?