| Nucleic Acids Research | Pages |
Assignment of the L30-mRNA complex using selective isotopic labeling and RNA mutants
Introduction
Materials And Methods
RNA synthesis and purification
Complex formation
NMR spectroscopy
Sugar pucker and glycosidic conformations
Results
Sequence design of the L30 RNA and related mutants
Assignments of the RNA exchangeable resonances in the free form
Assignments of the RNA non-exchangeable resonances in the free form
Assignments of the RNA exchangeable resonances in the bound form
Assignments of the RNA non-exchangeable resonances in the bound form
Discussion
Loop dynamics in the free and bound RNAs
Unusual sugar carbon chemical shifts
Unusual NMR features of the internal loop
Conclusion
Acknowledgements
References
Assignment of the L30-mRNA complex using selective isotopic labeling and RNA mutants
Received June 21, 1999; Revised August 9, 1999; Accepted August 30, 1999
ABSTRACT The helix-loop-helix structure formed in the pre-mRNA and the mRNA of L30, a ribosomal protein from the yeast Saccharomyces cerevisiae, serves as an auto-regulatory binding site for the protein to suppress the L30 synthesis upon overproduction. Using a 33-nucleotide model RNA, the structures of the L30 binding site RNA in the presence and absence of the protein were investigated using nuclear magnetic resonance (NMR) spectroscopy. Homonuclear and 13C/15N-based resonance assignments and spectral comparisons indicated that the purine-rich internal loop is dynamic in the free RNA but becomes ordered in the presence of L30 protein. Although the resonances in the loop region are sharper and more disperse in the bound RNA, their assignment was extremely challenging, due to spectral complexity and broadened resonances caused by local dynamics. Two strategies, namely selective 13C/15N-labeling and NMR analyses of five complexes with RNA mutants, were used to overcome these difficulties. Only using these approaches could assignment of the internal loop resonances and identification of the unusual NOEs and nucleotide conformations within the internal loop be made. In the case of structural determination of the L30-mRNA complex, it was critical to be able to take advantage of the available biochemical information in order to complete the structure determination.
INTRODUCTION
RNA-mediated feedback regulation plays a crucial role in controlling the synthesis of ribosomal proteins (r-proteins) (1,2). Expression of r-proteins in prokaryotes is often regulated by binding of an r-protein to an mRNA site to provide translational repression (1), whereas synthesis of r-proteins in eukaryotes can be subject to feedback regulation at both the pre-mRNA processing and the mRNA translation levels (2). In the yeast Saccharomyces cerevisiae, overexpression of the r-protein L30 (formerly known as L32) (3,4) is negatively auto-regulated at two stages: its pre-mRNA splicing (5) and its mRNA translation (6). Specifically, the complete assembly of the spliceosome is disrupted when the L30 binds to the 5[prime]-region of its pre-mRNA, comprising the 5[prime]-exon I, the initiation codon AUG and the 5[prime]-splice site in intron (5,7). Similarly, the L30 translation is repressed when the protein binds to a comparable site in the matured mRNA (with a single nucleotide variation) (6).
The secondary structure of the L30 auto-regulatory RNA was first identified through a phylogenetic comparison (5). Subsequent biochemical analyses identified the minimal RNA binding site for the L30 protein as a helix-purine loop-helix structure, which is represented by the MiniL32 RNA (8) shown in Figure 1A. The purine-rich asymmetric internal loop appeared to be the important protein-binding region, as nucleotides critical for binding were found only within the loop and its adjacent regions. The bases of G11, A12, G13, A55, G56, G58 and A59 were protected from diethylpyrocarbonate (DEPC) modification or hydroxyl radical attack in the presence of L30, and N7-carboxethylation of the bases or ethylation of the backbone within the internal loop abolished the L30 binding (8). The loop size also matters, for RNAs with deletion of either G56 or A57 failed to bind the protein (9). In vitro selection experiments (10) further confirmed these chemical probing results. For example, the protein-binding RNA aptamers exhibited a consensus motif with two helices separated by a two-opposite-five internal loop. Five out of the seven bases in the internal loop showed a strong preference for a specific purine, and only A55 and A57 could be replaced by pyrimidines (Fig. 1B). These biochemical findings suggested that both the overall structure and the base identities of the internal loop are important for L30 binding.
Figure 1. Sequence and secondary structure of the L30 mRNA. Numbering is according to the L30 pre-mRNA and lower case letters represent the non-wild type nucleotides. (A) Secondary structure of the MiniL32 RNA used in biochemistry studies (8). DEPC modification of the nucleotides in red circles interferes with protein binding. (B) L30 RNA motif summarizing results from the in vitro selection (10). N, Y and N-N represent any nucleotide, pyrimidine and Watson-Crick base pair, respectively. Nucleotides in green are highly preferred nucleotides, and those in red are absolutely conserved nucleotides among RNA apatmers. (C) L30 mRNA constructs used in NMR studies. Wild type RNA is colored in black, and point mutations used for NMR studies are indicated with arrows and highlighted in cyan lower case letters.
Based on the sequence and secondary structure proposed by biochemical studies, we have carried out nuclear magnetic resonance (NMR) structural studies of the auto-regulatory RNA binding site for the L30 protein. Resonance assignments of both the free and protein-bound RNAs were performed using standard homonuclear and 13C/15N-edited heteronuclear NMR approaches. The assignment difficulties associated with the purine-rich nature of the internal loop and partial resonance broadening were overcome only by the usage of a 13C/15N G-only labeled sample and five active RNA point-mutants. Identification of unusual NOEs and sugar conformations in the RNA, in conjunction with information from the protein NMR studies allowed the final structural modeling of the complex (manuscript in preparation).
MATERIALS AND METHODS
RNA synthesis and purification
The RNAs were synthesized by in vitro T7 RNA polymerase transcription (11). More specifically, the N-terminal His-tag containing T7 RNA polymerase was overexpressed in Escherichia coli BL21 (DE3) and purified on a Ni-chelating sepharose column (Pharmacia). The 18-nucleotide DNA promoter sequence (`top-strand') is 5[prime]-CTAATACGACTCACTATA-3[prime]. The individual DNA template contains the 5[prime]-anti-sense coding sequence plus 5[prime]-TATAGTGAGTCGTATTAG-3[prime] (Operon, Inc., Alameda, CA). In addition, a 4-methylindole [beta]-deoxynucleotide was added at the 5[prime]-end of the DNA template for repressing the add-on nucleotide during transcription (12). The transcription buffer contains 80 mg/ml polyethylene glycol, 80 mM Na+-HEPES (pH 8.1), 10 mM DTT, 1 mM spermidine, 0.01% Triton X-100, 300 nM DNA template, 300 nM top-strand, and 0.5 U/ml inorganic pyrophosphatase (optional), with NTPs, MgCl2 and T7 RNA polymerase concentrations optimized for highest yields on each DNA template. For instance, the wild type RNA transcription reaction contained 3 mM GTP, 2 mM ATP, 2 mM CTP, 1 mM UTP, 12.8 mM MgCl2 and 26.4 µg/ml T7 RNA polymerase. After 3.5 h at 37°C, transcription reaction was subjected to phenol:chloroform extraction and ethanol precipitation, and product RNAs were separated on 20% (w/v) polyacrylamide (19:1) gels containing 8 M urea. The product band was excised from the gel, and the RNA was recovered by electro-elution in an Elutrap apparatus (Schleicher & Schuell) and ethanol precipitation. Unlabeled RNAs were prepared using the above protocol with NTPs purchased from Sigma, and the 13C/15N-labeled RNAs were prepared using the NTPs purified from the Methylophilus methylotrophus (13,14). The purified RNAs were completely suspended in ddH2O and dialyzed extensively against the NMR buffer (10 mM potassium phosphate, pH 6.0 or 6.5, 0.02% NaN3 and 0.1 mM EDTA).
Monomer versus dimer states of RNA used were analyzed on a Bio-sil SEC 125-5 size-exclusion column (BioRad). Briefly, RNAs with concentration ranging from 10 to 1000 µM were heat denatured at 65°C for 1 min, and then flash cooled on ice for 5 min. Two microliters of RNA were injected into an SEC column equilibrated with the standard NMR buffer running with a flow rate of 0.5 ml/min, and the peaks were monitored by UV absorption at 260 nm. Retention time of a monomer was estimated by comparison to that of a previously known monomeric RNA (34 nucleotide), and it was also standardized by the retention times of protein markers with known molecular weights. Retention time of a dimer is expected to be less than that of a monomer, and the population of two states are determined by the ratio of the two peaks.
The procedure for testing the protein binding affinities of the mutant RNAs followed the method of Li et al. (8). A detailed description of the gel-shift binding assay is also presented elsewhere (manuscript in preparation).
Complex formation
After the concentration of the L30 protein ([epsiv]280nm = 8.98 mM-1 cm-1) and RNA ([epsiv]260nm = 363 mM-1 cm-1) were determined, the complex of 1:1 stoichiometric RNA: protein was prepared by mixing protein and RNA at ~50 µM concentration, then increasing the concentration slowly to 0.7 to 1.8 mM in a 2-ml centricon-3 concentrator (Amicon) by centrifugation at 6000 g. NMR samples of the RNA-protein complexes were then dialyzed against the NMR buffer (10 mM potassium phosphate, pH 6.0 or 6.5, 0.02% NaN3 and 0.1 mM EDTA) over 24 h with two changes of buffer.
NMR spectroscopy
NMR experiments were performed on Varian Inova 600, 750 MHz or FBML 500 MHz spectrometers equipped with z-axis shielded gradient triple resonance probes. The RNA samples include unlabeled 1.5 mM wild type, 1.3 mM U14C, 0.5 mM A55C, 0.7 mM A55U, 0.8 mM G56A, 0.5 mM A57C and 0.9 mM 13C/15N uniformly labeled wild type, and 0.5 mM 13C/15N G-only labeled wild type RNAs. The complex samples include: unlabeled L30 protein bound to non-labeled wild type or mutant A57C RNAs; 15N-labeled protein in complex with unlabeled wild type, mutant U14C, A55C, A55U and G56A RNAs; unlabeled protein bound to the 13C/15N uniformly labeled wild type or 13C/15N G-only labeled wild type RNAs; and 1.7 mM 13C/15N uniformly labeled protein bound to unlabeled wild type RNA. The experiments involving exchangeable protons were generally collected in 90% H2O/10% D2O at 15°C for the free RNAs, and 30°C for the complex. Whenever there were overlap ambiguities, experiments were also recorded at either lower or higher temperatures. Experiments involving only non-exchangeable resonances were recorded in 99.9996% D2O at 25°C for the free RNAs, and mostly at 30°C for the complex. Spectra were processed in NMRPipe (15), and analyzed in the PIPP and CAPP programs (16). The spectra were referenced to 3-(trimethylsilanly) propionate (TSP) for 1H and 13C resonances directly, and to external NH4Cl indirectly for 15N resonance.
A variety of homo- and heteronuclear experiments were used to make RNA resonance assignments and to generate structural constraints for the structure calculations. Homonuclear experiments included NOESY (17), DQF-COSY (18), TOCSY (19) and ROESY (20) experiments. Most of these experiments were recorded on the Varian Inova 600 MHz spectrometer. The 2D H2O NOESY (100 and 150 ms mixing time) spectra typically have 14 000 Hz sweep width in each dimension, and use the flip-back WATERGATE (21) for water suppression. The 2D D2O spectra for the free RNA typically included NOESY (300 ms), TOCSY (11 and 75 ms) and DQF-COSY spectra width of 6000 Hz in each dimension. 2D D2O spectra for the bound RNA included NOESY (50, 100, 150, 200 and 300 ms mixing time), TOCSY (11 and 75 ms), DQF-COSY and ROESY (50 ms), with typically spectra width is 6600 Hz in each dimension. Isotropic mixing period for TOCSY and ROESY experiments involved MLEV-17 (22)
Heteronuclear experiments were performed on uniformly labeled 13C/15N L30N RNA and 13C/15N guanine-only labeled L30N RNA in the free and bound form. These experiments included 2D (1H, 15N) HSQC (23), 2D (1H, 13C) HSQC (24), 2D (1H, 13C) CT-HSQC (25), 3D (1H, 1H, 15N/13C) NC-NOESY-HSQC (26), 3D (1H, 1H, 15N) NOESY-HSQC (27), 3D (1H, 1H, 13C) NOESY-HSQC (28) and 3D (1H, 13C, 1H) HCCH-TOCSY (29). Most of these experiments were acquired on the Varian 600 MHz spectrometer, except a few 3D experiments on the Varian 750 MHz spectrometer. Where appropriate, 13C or 15N WALTZ or GARP-1 (30) broadband decoupling was applied during the acquisition period. A DIPSI-3 isotropic mixing scheme (31) was applied during the 12 ms-spin-lock period in the HCCH-TOCSY experiments. 2D (1H, 13C) HSQC spectra were acquired by either centering 13C transmitter at 116 p.p.m. to maximize aromatic and H1[prime]/C1[prime] resonance 1H-13C correlation, or around 83 p.p.m. to obtain anomeric 1H/13C correlation. Constant time 13C-HSQC spectra centering 83 p.p.m. at 25 and 30°C were acquired with evolution periods optimized for JHC = 200 Hz, and the constant-time period set to 1/JCC(~27 ms). 2D (1H, 15N) HSQC spectra were recorded on FBML 500 MHz spectrometer with the 15N transmitter centering around 118 p.p.m. to obtain the full exchangeable 1H, 15N correlations, or around 150 p.p.m. to get the imino 1H, 15N correlations. 3D CN-NOESY-HSQC spectra were acquired centering the 15N transmitter around 120 p.p.m. to observe NOEs originated from all exchangeable proton, and 13C around 104 p.p.m. to observe NOEs from non-exchangeable aromatic and H1[prime] protons. 3D 13C-edited HSQC-NOESY (150 ms mixing time) and 3D HCCH-TOCSY (12 ms mixing time) spectra were recorded in D2O with the 13C transmitter centered around 83 p.p.m. to maximize the sugar proton assignment. Because of the relatively low concentration of the 13C/15N-guanosine labeled RNA sample, 3D NOESY and TOCSY experiments were acquired only on the uniformly 13C/15N-labeled sample to obtain RNA assignment in the free form and bound form. While the HCCH-TOCSY and CN-NOESY-HSQC spectra were acquired at 600 MHz, 13C HSQC-NOESY spectrum of the sugar resonances was performed on 750 MHz. Two additional experiments were also acquired to assist the assignment of the RNA in complex. 3D 13C-edited HSQC-NOESY spectrum centering 13C to 140 p.p.m. in the middle of C8/C6 and C2 region was acquired at 600 MHz to observe aromatic protein NOEs. A 3D 15N-edited NOESY-HMQC spectrum centering 15N to 150 p.p.m. was recorded at 750 MHz to observed imino-proton NOEs.
Sugar pucker and glycosidic conformations
The C3[prime]-endo or C2[prime]-endo sugar pucker conformations were estimated qualitatively from the 3JH1[prime]-H2[prime] coupling constants measured from homonuclear 2D COSY and TOCSY (11 ms) spectra. In the free RNA, seven distinct H1[prime]-H2[prime] cross-peaks have been observed in both spectra COSY and TOCSY spectra. In both the free and bound RNAs, G5 at the 5[prime]-end, C19 and A20 in the tetraloop have coupling constants ~5-8 Hz, suggestive of C2[prime]- and C3[prime]-endo sugar pucker inter conversion; G11, G56, A57 and G58 within the internal loop appeared as C2[prime]-endo pucker conformation. All other nucleotides in helices I and II appeared to have 3[prime]-endo sugar conformation in both the free and bound RNA.
Anti or syn glycosidic conformations were estimated from the NOEs between base H8/H6 protons to their sugar H1[prime] proton-a strong NOE is associated with the syn conformation (~2.5 Å distance), whereas a weaker one is associated with anti conformation (~3.7 Å distance) (32). Only G56 has a syn glycosidic conformation, other nucleotides have anti glycosidic conformations in both the free and bound L30 RNAs.
The 1H, 13C and 15N chemical shifts of the free and bound L30 RNAs have been deposited into the BioMagResBank, and are available under accession codes 4346 and 4345 respectively.
RESULTS
Sequence design of the L30 RNA and related mutants
The 33-nucleotide RNA used in the NMR structural studies (Fig. 1C) was adapted from the MiniL32 RNA (Fig. 1A) with two modifications. The single-stranded regions at the 5[prime]- and 3[prime]-end were replaced with a G-C pair, and the non-wild type loop capping the helix II was substituted with a C-G pair and a GCAA tetraloop. These modifications appeared to stabilize the monomeric conformation of the L30 RNA, for <10% of the free RNA was detected as a dimer at millimolar concentration using a size-exclusion chromatography. According to the biochemical studies (8,10), these modifications should not affect the RNA-protein binding, because they are far away from the important internal loop. Gel-shift assays using this construct confirmed that the RNA indeed binds to the MBP-L30 fusion protein with a wild-type binding affinity (Kd = 10 ± 3 nM) (8).
Besides the wild type RNA described above, five active point-mutants, U14C, A55C, A55U, G56A and A57C (cyan letters in Fig. 1C) were also prepared to resolve the assignment ambiguities. The RNA with the U14C mutation (creating a C14-G53 Watson-Crick pair) allowed differentiation of the U14·G53 wobble pair in helix II from the G10·U60 pair in helix I. The internal loop assignment was particularly challenging, due to the high density of purines and partial line broadening in this region. We were able to take advantage of the sequence variations that were observed in the in vitro selection experiments, such as C or U replacement of A55, A57C or G56A mutations (10), to assist the assignment of the internal loop. All five point-mutants were subjected to gel-shift binding assays to verify that these mutations did not significantly disrupt the L30 binding. Mutants U14C, A55C, A55U and A57C have wild-type affinity (Kd = 12 ± 4 nM) for the MBP-L30 fusion, while G56A has 7-10-fold reduced affinity. The 2D (1H, 15N) HSQC spectra of the L30 protein in complexes with the mutant RNAs (i.e. U14C, A55C, A55U and G56A) showed very similar correlation peaks to those in the wild type complex (Fig. 2). The largest 1H differences are ~0.2 p.p.m., which were observed in the (1H, 15N) HSQC spectrum of the G56A-protein complex. The similarity of the HSQC spectra suggests that the structure of the complex is comparable for the wild type and mutant RNAs, and that the point-mutants are suitable for assisting the wild type RNA resonance assignments.
Figure 2. Superposition of the L30 protein (1H, 15N) HSQC spectra in complexes with the wild type and mutant RNAs, pH 6.5 at 30°C. The protein correlation peaks are colored according to the complexes: in black with the wild type RNA, in red with A55C, in green with A55U, in blue with A56A (in blue) and in magenta with U14C. The chemical shifts of L30 protein amides share close similarities among complexes with the wild type and mutant RNAs.
Assignments of the RNA exchangeable resonances in the free form
A 2D (1H, 15N) HSQC spectrum of the free RNA acquired at 15°C showed 16 imino correlation peaks: 12 from guanines (out of 13) and four from uracils (Fig. 3A). Among them, only 13 imino resonances from helices I and II and the tetraloop were identified using the H2O NOESY spectra. The assignments were primarily based on NOEs between imino protons, between imino protons and cytosine amino protons, and between imino protons and neighboring H1[prime] protons (33). Resonance differentiation of the wobble base pairs U14·G53 from G10·U60 was also based on spectral comparison of the wild type RNA with the mutant U14C-resonances at 11.93 p.p.m. (for U14) and 11.14 p.p.m. (for G53) were absent in the mutant imino proton spectrum. Analysis of the NOE patterns indicated that G6, G10, G13, U14, G15, U16, G50, G53, U60, G61, G62 and U63 in helices I and II are indeed involved in base pairs outlined in Figure 1C, and G18 in the GCAA tetraloop is involved in a G18·A21 shear base pair (34). However, imino resonances of G5 at the 5[prime]-end and G11, G56 and G58 within internal loop appeared to undergo either fast solvent exchange or µs to ms time scale conformational exchange, and were therefore not assigned.
Figure 3. Imino regions of the (1H, 15N) HSQC spectra of (A), the free and (B), the bound wild-type L30 RNA. The spectrum of (A) was recorded on the uniformly 13C/15N-labeled RNA in 10 mM potassium phosphate buffer, pH 6.0, containing 0.02% NaN3 and 0.1 mM EDTA at 15°C. The spectrum of (B) was recorded on the uniformly 13C/15N-labeled RNA in complex with unlabeled L30 protein in 10 mM potassium phosphate buffer, pH 6.0, containing 0.02% NaN3 and 0.1 mM EDTA at 30°C. The concentration of the free and bound RNA is ~0.7 and 0.9 mM, respectively.
Six pairs of cytosine amino protons (except for C19 in the tetraloop and C65 at the 3[prime]-end) were identified in a (1H, 15N) HSQC spectrum and assigned from the strong NOEs to their paired G imino protons. Only a few purine amino protons, however, were observable in HSQC or NOESY spectra. In the helical regions, the fast and freely rotating amino protons of G10 (6.45 p.p.m.) and G53 (6.43 p.p.m.) in the G·U pairs displayed strong NOEs to their imino proton in the H2O NOESY spectrum. In addition, the amino protons of G13 in the G13-C54 Watson-Crick pair exhibited two distinct NOEs (at 8.63 and 7.00 p.p.m.) to its imino proton. The observable purine amino proton resonances within a Watson-Crick pair were previously reported in the AMP-aptamer (35), and could be explained by a more frequent breathing that occurred at the terminal of base pair. It is likely that G13 in the L30 RNA undergoes similar motion, but it is also possible that the G13 amino protons involve in a two-hydrogen-bonding conformation (i.e. base triple) that further slow down the intermediate C-N rotation occurred in the normal Watson-Crick base pair. The absence of distinct purine amino proton resonances in the L30 RNA internal loop region is most likely due to conformational exchange broadening and not a result of hydrogen bonding.
Assignments of the RNA non-exchangeable resonances in the free form
Analyses of homonuclear 2D NOESY, COSY and TOCSY spectra, in combination with heteronuclear 2D (1H, 13C) CT-HSQC, 3D (1H, 1H, 13C) NOESY-HMQC and (1H, 13C, 1H) HCCH-TOCSY spectra recorded on a 13C/15N uniformly labeled sample provided most of the non-exchangeable proton assignments of the free RNA. First, the assignments of base and H1[prime] protons were carried out following the H1[prime] to H8/H6 sequential NOE connectivity (32) identified from a 3D (1H, 1H, 13C) NOESY-HMQC spectrum acquired in D2O. The H1[prime] assignments were further verified from sequential NOEs between H1[prime] and 5[prime]-side G imino protons in a H2O NOESY spectrum (33). Next the H1[prime] protons were correlated with the remaining sugar protons (i.e. H2[prime], H3[prime], H4[prime], H5[prime] and H5[prime][prime]). Only a few, namely the H2[prime] and H3[prime] resonances of G5, G11, A20 and A21 that have large H1[prime]-H2[prime] coupling constants were assigned using homonuclear COSY or TOCSY (75 ms mixing time) spectra. Most sugar proton assignments were based on a 3D (1H, 13C, 1H) HCCH-TOCSY (12 ms mixing time), and confirmed with a 3D (1H, 1H, 13C) NOESY-HMQC (150 ms mixing time). Almost complete assignments were made for the nucleotides in the two helices and the tetraloop, and limited assignments were obtained in the internal loop.
Analysis of the non-exchangeable resonance spectra of the free RNA generally supported the conclusion obtained from the exchangeable proton NMR spectra. Overall, resonances in helices I and II showed features typical of an A-form helix, such as moderate intra- and inter-nucleotide H1[prime] to H8/H6 NOEs (Fig. 4). In addition, sequential NOEs within the GCAA tetraloop were consistent with the reported structure (34). By contrast, the internal loop displayed significant broadening, and consequently, there were insufficient NOEs to define the exact conformation in this region. Nonetheless, NMR data did reveal some interesting observations of the internal loop. For instance, G56 has an interesting upfield shift of H8 (7.24 p.p.m.) and very unusual down field shift of C8 (140.8 p.p.m.). Moreover, G56 also appears to have a syn glycosidic conformation, revealed by the relatively strong NOE for the H8 to its own H1[prime] proton (at 4.89 p.p.m.). Other unusual features include the C2[prime]-sugar puckers observed for four purines (i.e. G11, G56, A57 and G58) in the internal loop. These observations indicated that the internal loop likely has an unusual structure in the free RNA; however, it was not possible to determine the detailed structure.
Figure 4. (A) Aromatic proton (7.1-8.8 p.p.m.) to ribose H1[prime] (4.8-7.0 p.p.m.) region of the D2O NOESY (300 ms mixing time) spectrum for the free wild type RNA, pH 6.0 at 25°C. The labeled peaks indicate intra-nucleotide base H8/H6 to sugar H1[prime] proton NOEs. The solid lines trace sequential NOE connectivities between base H8/H6 and their own and 5[prime]-flanking sugar H1[prime] protons from G5 to G10 and from G13 to A21, and dashed lines trace the connections from G50 to A55 and from G61 to C65. (B) Schematic of NMR data for the free RNA. White, black and gray pentagons represent sugar with respective C3[prime]-, C2[prime]- and C2[prime]/C3[prime] mixture sugar puckers. Dashed boxes indicate unassigned and/or unobserved nucleotides. Thick dashes indicate base pairs and thin dashes indicate observed sequential imino-imino, sequential base to sugar proton NOEs.
Assignments of the RNA exchangeable resonances in the bound form
Upon complex formation, imino resonances within the two helices remained largely unperturbed (<0.27 p.p.m. changes) (Fig. 3B), and retained their characteristic sequential and intra-base NOEs (Fig. 5) observed in the free RNA. By contrast, the internal loop region exhibited several significant changes. For instance, the imino protons of G10 and U60 adjacent to the internal loop became broader, and two new G-imino proton resonances in the internal loop appeared as strong correlation peaks and gave several inter-residue NOEs (Figs 3B and 5). The two new resonances, however, do not have any sequential imino-imino NOEs, and their assignments were facilitated with mutant studies. The presence of a medium-strong NOE between the G11 imino (at 11.18 p.p.m.) and the G56 H8 proton (at 7.38 p.p.m.) and absence of G11 amino proton resonances strongly suggested that G11 be involved in an unusual G·G base pair. The imino proton resonance at 11.18 p.p.m. was assigned to G11 based on sequential NOEs, and was also confirmed by the spectral comparison between the wild type and mutants A55U and A55C. The G11 imino resonance shifted upfield for ~0.1 p.p.m. in both mutants as a result of pyrimidine substitution. The other slowly exchanging imino resonance observed at 8.98 p.p.m. was assigned to G56 based on disappearance of this peak in the spectrum of the G56A mutant. In a 15N-edited NOESY-HSQC spectrum, the G56 imino resonance clearly showed several NOEs to Phe85 on the L30 protein (Fig. 5B). Thus, the unusually high upfield shift of the G56 imino resonance is most likely caused by a direct aromatic shielding effect from Phe85 in the protein.
Figure 5. NOEs observed for the imino protons in the wild-type L30 RNA-protein complex at 30°C. (A) Imino proton (10.5-15.0 p.p.m.) region to the base, amino and sugar protons, and the protein (3.8-9.0 p.p.m.) of the H2O NOESY (100 ms mixing time) spectrum of the L30 RNA-protein complex. The imino peaks are labeled with their assignments above the spectrum. The G11 imino proton to G56 H8 cross-peak and the G13 imino proton to its two amino proton cross-peaks are illustrated. (B) G56 imino proton stripe of the 3D (1H, 15N) NOESY-HSQC (100 ms mixing time) spectrum of the 13C/15N-labeled wild type L30 RNA in complex with unlabeled protein. NOEs of the G56 imino proton to the protein Phe85 aromatic protons are illustrated.
Assignments of the RNA non-exchangeable resonances in the bound form
In the RNA-protein complex, resonances of the two helices were similar to those in the free RNA, and their assignments were straightforward using the homonuclear and heteronuclear strategies described for the free RNA. In contrast to the broad and poorly resolved resonances in the free RNA, resonances within the internal loop displayed significant improvements upon the L30 protein binding. The chemical shift dispersion in this region is significantly increased, and there are a number of unusual chemical shifts such as 8.68 p.p.m. for an H8, 6.78 p.p.m. for an H1[prime] and 5.22 p.p.m. for an H4[prime], 3.43 p.p.m. for an H2[prime] and 3.44 p.p.m. for an H4[prime] resonance. In addition, there is a general sharpening of line widths in the internal loop, and significantly more NOEs were also observed. These improvements in spectral properties have made possible a detailed structural study of the internal loop in the bound form. Despite these improvements, assignments in this region were nearly impossible solely on the basis of a single 13C/15N uniformly labeled RNA, because there are four adenine and five guanine residues within the short spans of G10 to G13 and C54 to U60. The presence of unusual non-sequential NOEs made it even more difficult to unambiguously assign residues on the basis of NOE connectivities. Triple-resonance experiments such as HCCNH-TOCCY (36) and HCN (37) that can provide unambiguous through-bond connectivities were not successful, presumably due to the short transverse relaxation time (T2) for the 22 kDa L30-mRNA complex.
The use of the 13C/15N guanosine-specific labeled wild type RNA and four unlabeled point-mutants proved to be critical in clarifying the assignment ambiguities in the internal loop. In the larger strand of the loop (A55 to A59), NOEs between the sugar H1[prime] to aromatic H8 were absent from A55 to G56 and from G56 to A57, and were very weak from A57 to G58, and it was impossible to differentiate H8 resonances of the adenosines (i.e. A55, A57 and A59) from those of the guanosines (i.e. G56 and G58). The G-specific labeling proved to be very powerful in this case, allowing the guanosine resonances to be identified in a (1H, 13C) HSQC spectrum. For example, an unusual guanine H8/C8 correlation peak at 7.38 p.p.m./142.3 p.p.m. was easily identified in Figure 6B, but could not have been readily identified in the spectrum of the uniformly labeled RNA shown in Figure 6A. Based on the comparison of the (1H, 13C) HSQC spectra between the fully labeled and G-only labeled RNAs, it was possible to identify the H8 proton shifts at 7.38 and 8.40 p.p.m. to two guanines (G56 and G58), and identifying the resonances at 8.18, 8.25 and 8.68 p.p.m. to the three adenines (A55, A57 and A59). Similarly, the sugar proton and carbon shifts of adenosine and guanosine residues were also differentiated.
Figure 6. Comparison of the aromatic (H8/H6 and C8/C6) regions in (1H, 13C) HSQC spectra of (A), 13C/15N uniformly labeled and (B), 13C/15N G-only labeled L30 RNA in complex. The guanine residues can be unambiguously identified in the G-only labeled spectrum. The peaks are labeled with their assignment. Note that G56 has an unusual downfield 13C chemical shift.
Although the G-only labeling proved to be critical in identifying the guanine resonances in the internal loop, the differentiation among the A55, A57 and A59 resonances still remained ambiguous. The usage of the four point-mutants (A55C, A55U, G56A and A57C) proved to be the key for clarifying this ambiguity. In some cases, simple comparisons of the homonuclear 2D NOESY, COSY and TOCSY spectra between the wild type and a mutant complex were sufficient to reveal important assignments. No large structural changes were anticipated because the observed chemical shift perturbations observed in the mutant spectra are very local, but subtle shift perturbations can reveal peaks that are overlapped in the wild type. The disappearance of certain aromatic (H8 and/or H2) to sugar peaks in a NOESY spectrum due to a specific mutation help to secure the assignments of that nucleotide. The appearance of base H6 resonances for three pyrimidine point-mutants (A55C, A55U and A57C) could be easily identified from the new H6/H5 cross-peaks in the COSY spectra of the mutants.
We initiated the mutant-based assignment by first introducing the A57C mutation in the middle of the larger loop. A comparison of NOESY H8/H1[prime] regions between the wild type (Fig. 7A) and A57C (Fig. 7B) indicated that the H8 and H2 resonances of A57 at 8.68 and 8.32 p.p.m., respectively, were absent in the mutant and were replaced by the H6 resonance of C57 at 8.22 p.p.m. The comparison not only revealed the assignments of the A57 base protons and left the two adenosines A55 and A59 to be differentiated, but perturbation of the H8 resonance from the wild type 8.40 p.p.m. to the mutant 8.35 p.p.m. also hinted at the identity of the G58 H8. The ultimate differentiation of the G56 and G58 resonances was made in the G56A mutant, where the H8 and H1[prime] resonances of G58 remain unchanged, but those of mutant A56 were both shifted downfield (~0.2 p.p.m.) relative to the resonances of G56 (Fig. 7C). Lastly, the differentiation of A59 and A55 was made clear by the A55U (Fig. 7D) and A55C mutants (note, the A55U mutant was also used for probing base pairing in A12 and A55). The combination of the 13C/15N labeled samples and mutations therefore enabled the tracing of H8/H6 to H1[prime] NOEs within the internal loop and its flanking nucleotides-sequential connectivities were weak at the G10-G11, A12-G13, A59-U60-G61 segments and were discontinuous at the A55-G56-A57 segment (Fig. 7A).
A
 |
B
 |
C
 |
D
 |
Figure 7. The ribose H1[prime] (4.55-7.00 p.p.m.) and aromatic H8/H6 (7.15-8.75 p.p.m.) regions of D2O NOESY (150 ms mixing time) spectra of (A), L30 wild type; (B), A57C; (C), G56A; and (D), A55U RNAs in complex. The solid lines trace the sequential NOE connectivities (H1[prime] to H8/H6) from G5 to A21, and dashed lines trace the connection from G50 to A55, and A57 to C65. The aromatic A H2 proton chemical shifts are marked on the left side of (A). The labeled peaks are intra-nucleotide H1[prime] to H8/H6 NOEs. The cyan labels in (B), (C) and (D) each indicates the intra-nucleotide H1[prime] to H8/H6 NOE of the point-mutant. The green labels in (B), (C) and (D) indicate the chemical shifts of the cross-peaks affected by mutation.
The most unusual structural features identified from the NOESY spectra is the existence of three sets of long-range H1[prime]-H1[prime] contacts between A12 and G13, between G13 and A59, and between A55 and A59. These medium to medium-strong NOE cross-peaks which can be seen clearly in the 50 ms NOESY spectrum (Fig. 8), were therefore not due to spin-diffusion. To rule out the possibility of conformational exchange peaks, a ROESY spectrum was recorded and confirmed these cross-peaks are true NOEs-having opposite sign to the diagonal peaks [note, exchange peaks should be the same sign to the diagonal (38,39)]. In addition, these NOE patterns were also present in all five RNA mutants (U14C, A55C, A55U, G56A and A57C). The (1H, 13C) HSQC spectrum recorded on the 13C/15N G-only labeled sample indicated that there was only one guanosine H1[prime] (6.78 p.p.m.) among the set of four H1[prime] resonances, and the other three were adenosine H1[prime] resonances. The resonance at 6.78 p.p.m. in the wild type was assigned to G13 according to normal sequential G13 H8-G13 H1[prime]-U14 H6 NOEs, and further confirmed by a direct chemical shift change from wild type 6.78 to 6.65 p.p.m. in the U14C mutant. The resonance at 6.05 p.p.m. was assigned to A55 according to the sequential C54 H1[prime]-A55 H8-A55 H1[prime] NOEs. And lastly, the assignment of 6.30 p.p.m. to A59 was according to sequential G58 H1[prime]-A59 H8-A59 H1[prime] NOEs. All three independent pieces of data (i.e. ROEs, mutant studies and 13C/15N G-labeling) have confirmed the existence of these unusual H1[prime]-H1[prime] contacts and indicated long-range proximity between the backbone of the two loops.
Figure 8. Three sets of medium to medium strong H1[prime]-H1[prime] sugar NOEs observed in NOESY (50 ms mixing time) spectrum, between G13 and A12, between G13 and A59, and between A59 and A55.
DISCUSSION
Loop dynamics in the free and bound RNAs
The internal loop of L30 RNA displayed very different dynamic properties between the free and bound forms. In the absence of L30 protein, regions from G11 to G13 and from G56 to U60 exhibited severe line broadening, which is a clear indication of a µs to ms time-scale conformational exchange. Such an intermediate exchange has been observed in the free form of various internal-loop containing RNAs, including RRE (39), lead-dependent ribozyme (40), ATP-aptamer (35) and the iron responsive element (41). For these RNAs, internal loops critical for binding are often disordered in their free forms, but become well ordered in the complexes. Likewise, ordering of the internal loop structure accompanied the L30 protein binding. The slowing of conformational exchange and the improvement of resonance line widths are particularly apparent for nucleotides G11, G56, A57, G58 and A59, some evidence for intermediate exchange still persists in nucleotides A12 and U60. Nevertheless, it was still possible to obtain assignments and a wealth of observable intra- and inter-residues NOE that has led a detailed structural interpretation of the internal loop in the bound form.
Unusual sugar carbon chemical shifts
13C chemical shift, which is highly sensitive to local conformation, has been successfully applied in protein structure prediction (42). Albeit the chemical shift based RNA structure prediction is still in its early stage, recent studies have noted that the 13C chemical shifts of sugar C1[prime], C3[prime] and C4[prime] are highly dependent on backbone torsion angles [gamma], [alpha] and [delta] as well as sugar pucker conformation (i.e. C2[prime]-endo or C3[prime]-endo) (35,43-45). For example, the C1[prime] carbons resonate at ~93 p.p.m. in C3[prime]-endo sugars, but they shift upfield by ~7 p.p.m. on average in C2[prime]-endo sugars, whereas different trends are seen for the C3[prime] or C4[prime] carbons. The C3[prime] carbons resonate at ~73 p.p.m. and C4[prime] carbons at ~82 p.p.m. in C3[prime]-endo sugars, but both shift downfield by ~2-6 p.p.m. in C2[prime]-endo sugars (43).
Conformation-dependent 13C chemical shifts have also been observed in the free and bound L30 RNA. In particularly, the 13C chemical shifts of four internal loop nucleotides in the bound form have distinctive shifts characteristic of a C2[prime]-endo sugar conformation. The C1[prime], C3[prime] and C4[prime] shifts for these nucleotides are G11 (89.1, 80.0 and 87.4 p.p.m.), G56 (90.8, 79.2 and 85.5 p.p.m.), A57 (89.7, 78.3 and 85.4 p.p.m.) and G58 (86.1, 82.1 and 87.2 p.p.m.). The relatively downfield shift for the C1[prime] carbon of G56 (90.7 p.p.m.) is due to the syn glycosidic torsion angle for this residue in the complex, and similar shifts have been seen in other structures (35,46). Conclusions drawn from the chemical shift trends are consistent with J-coupling constants and NOE measurements.
Unusual NMR features of the internal loop
The internal loop, which is undoubtedly the most important segment of the RNA, undergoes structural and dynamical changes upon complex formation. The 13C/15N based resonance assignments, which were greatly facilitated by the usage of selective labeling and mutants, allowed the identification of unusual sequential and long-range NOEs within this region (Table 1 and Fig. 9). The L30-mRNA complex structure has been solved recently, and the RNA average structure in an energy-minimized complex is shown in Figure 10. The overall structure of the RNA is composed of two helical stems that are interrupted by a highly distorted internal loop with extensive purine stacking (detailed manuscript in preparation). Describing below are some of the NMR observations of the two loop segments (G10 to G13 and C54 to U60) that define the structure and conformation of the internal loop region.
Figure 9. Schematic representation of the inter-residue NOEs in the internal loop region. The open pentagons and magenta rectangles represent sugars and bases, respectively. The filled red and black circles and the open black circles represent phosphate atoms, base H8/H6 protons, and adenine H2 protons, respectively. For clarity, the backbone from G10 to G13 is in green, and from C54 to U60 is in cyan. The dash lines with arrows indicate the observed inter-residue NOEs.
Figure 10. The average RNA structure in the L30 RNA-protein complex. Nucleotides from G5 to C9 and G61 to C65 in helix I, and from U14 to G53 in helix II are in blue. The backbone from G10 to G13 and from C54 to U60 is in green and cyan, respectively. The bases from G10 to C13, and C54 to U60 are in magenta. Residues are labeled next to the phosphate atoms highlighted in red ball (except for G5 at the 5[prime]-end).
Table 1. Unusual sequential and long-range (between non-adjacent residues) NOEs in L30 RNA
| Proton | NOEs |
| G11 | |
| NH-1 | A12 (H2, H2[prime]), G56 (H1[prime], H8) |
| H1[prime] | A12 (H8) |
| H4[prime] | A12 (H8) |
| A12 | |
| H2 | G11 (NH-1), G13 (H1[prime]), A55 (H1[prime], H2), A59 (H1[prime]) |
| H8 | G11 (H1[prime], H4[prime]) |
| H1[prime] | G13 (H1[prime]) |
| G13 | |
| H1 | A55 (H2) |
| NH2-2 | A55 (H1[prime]), A59 (H1[prime]) |
| H8 | A55 (H2) |
| H1[prime] | A12 (H1[prime]), U14 (H5), A55 (H2), A59 (H1[prime], H2) |
| U14 | |
| H5 | G13 H1[prime] |
| H1[prime] | A59 (H1[prime], H2) |
| A55 | |
| H2 | A12 (H2), G13 (H1, H1[prime], H8), A59 (H1[prime]) |
| H1[prime] | A12 (H2), G13 (NH2-2), A59 (H1[prime], H8) |
| H2[prime] | A59 (H1[prime]) |
| G56 | |
| H1, NH2-2 | A12 (H2) |
| H8 | G11 (NH-1) |
| H1[prime] | G11 (NH-1) |
| H4[prime] | A57 (H8, H2[prime], H3[prime], H4[prime]), G58 (H1[prime], H2[prime], H4[prime] H8) |
| A57 | |
| H8, H2[prime], H3[prime] | G56 (H4[prime]) |
| H4[prime] | G56 (H4[prime]), G58 (H8) |
| G58 | |
| H8 | G56 (H4[prime]), A57 (H4[prime]) |
| H1[prime], H2[prime], H4[prime] | G56 (H4[prime]) |
| A59 | |
| H2 | G13 (H1[prime],H2[prime]), U14 (H1[prime]), A55 (H2) |
| H8 | A55 (H1[prime]), G58 (H1[prime]) |
| H1[prime] | A12 (H8), G13 (NH2-2, H1[prime]), U14 (H1[prime]), A55 (H1[prime], H2[prime]) |
Backbone extension from G10 to G11. The structural transition point from helix I to the shorter loop (green backbone in Figs 9 and 10) is at nucleotide G10, which forms a wobble base pair with U60. Although the helical conformation is maintained between C9 and G10 with characteristic NOEs between the C9 sugar protons and the G10 H8 proton, it is altered between G10 and G11. The following observations are indicative of backbone extension: the weak sequential NOEs between both the H2[prime] and H3[prime] of G10 and the H8 of G11, the nearly absent NOE between the H1[prime] of G10 and the H8 of G11, and the C2[prime]-endo conformation of the G11 sugar pucker. The sugar protons (H1[prime], H2[prime] and H3[prime]) of G11 show NOEs almost exclusively to their own base protons, with the exception of a medium-strong NOE between the H1[prime] of G11 and the H8 of A12. The observations of weak sequential NOE connectivities and the C2[prime]-sugar conformation of G11 suggested that the base stacking might be loosened and the phosphate backbone be extended in this segment.
Backbone reversal from G11 to A12. Loop nucleotide A12 is adjacent to the G13-C54 base pair of helix II, however, no NOE was observed from the G13 imino proton to either the A12 H2 proton or to the A12 H8 proton. The observable intra-nucleotide sugar to base NOEs of A12 are fairly weak, partially due to resonance broadening. However, there are medium-strong NOEs between the H1[prime] of G11 and the H8 of A12, between the H1[prime] of A12 and the H1[prime] of G13, and weak to medium sequential NOEs between both the H1[prime] and H2[prime] protons of A12 and the H8 proton of G13. In addition, the H2 proton of A12 shows NOEs to the H1[prime] protons of G13, A55 and A59. The above observations imply a reversal of the orientation of the A12 ribose and an unusual turn between A12 and G13.
Stacking between A55 and the G13-C54 pair. The bridging residue of helix II and the larger loop (cyan backbone in Figs 9 and 10) is A55. The complete stacking of A55 above the G13-C54 pair is based on a handful of sequential NOEs that are typical for an A-form helix: between the sugar H1[prime], H2[prime] and H3[prime] of C54 and the H8 of A55, between the H6 of C54 and the H8 of A55, and between the H2 of A55 and the imino proton of G13. These sequential sugar-to-base and base-to-base NOEs were also observed in both A55U and A55C mutant RNA-protein complexes, and clearly supported the stacked arrangement between A55 and the G13-C54 Watson-Crick pair.
Hairpin turn at G56-A57-G58. While G56 forms a reverse-Hoogsteen base pair with G11 and stacks above A55 residue, A57 and G58 are bulged out the RNA helix. The sequential sugar H1[prime] to aromatic H8 NOEs are absent between A55 and G56, and between G56 and A57, and very weak between A57 and G58. Moreover, base-to-base NOEs were absent between the G56 H8 and the A57 H8 protons, and weak between the A57 H8 and the G58 H8 protons. The absence of sequential NOEs to either the H8 or H2 of A57 indicates that base is bulged out from the RNA helical region. Thus, it is likely that G56, A57 and G58 are not involved in sequential stacking interactions. The backbone of this segment is stretched and elongated because all three nucleotides have unusual C2[prime]-endo sugar pucker conformation. Several unusual NOEs between G56, A57 and G58 indicate that the ribose segments are close in space in a hairpin-like arrangement. For instance, the G56 H4[prime] proton not only displayed weak inter-nucleotide NOEs to the base H8 protons of both A57 and G58, but also to the sugar protons of these two nucleotides. Moreover, the A57 H4[prime] show weak NOE to the G58 base proton. An unusual large upfield shifts for G56 H4[prime] (3.44 p.p.m.) indicates a possible ring current effect by the surrounding purines. Furthermore, A57 and G58 give many NOEs to the L30 protein, thus the hairpin structure is important for position these bases for protein recognition.
A59·G13-C54 base triple (layer 1 in Fig. 9). The position of A59 in the L30 RNA secondary structure depicted in Figure 1B is near helix I and far away from helix II. However, A59 in the modeled structure is unusually close in space to the G13-C54 Watson-Crick base pair in helix II, which leads to a highly distorted phosphate backbone. The direct NMR observations defining this structural arrangement are the long range NOEs from the A59 H1[prime] to the A55 H1[prime], to the G13 H1[prime], and to the H2 protons of both A12 and A55. A59 appears to approach the G13-C54 base pair from the minor groove side, as the H2 base proton of A59 displayed long-range NOEs to the H1[prime] proton of both G13 and U14, and to the H2[prime] proton of G13. Similar NOE patterns was observed in the NMR study of the base-triple domain of the Tetrahymena group I intron, and leaded to structural characterization of a A·G-C minor groove base triple (47). Additional NMR observations in L30 RNA also supported the base triple proposal. For instance, the exocyclic amino protons of G13 showed two distinct correlation peaks, presumably by formation of a hydrogen bond with C54, and another one with A59. The chemical shifts of the A59 H2 and the G13 H1[prime] protons are shifted unusually downfield to 8.42 and 6.78 p.p.m., respectively, which is indicative of an in-plan orientation of A59 and G13. These unusual NOE patterns indicate that the phosphate backbone on the larger loop swings back close to G13-C54 and U14-G53 base pairs, and A59 forms a possible base triple with nucleotides G13 and C54 (layer 1 in Fig. 9).
A12 and A55 (layer 2 in Fig. 9). In the RNA secondary sequence, A12 and A55 are both located adjacent to the G13-C54 base pair, thus formation of an A12·A55 base pair might be anticipated. However, it is unlikely that the bases of A12 and A55 orient side to side, because the A55U mutant showed no evidence of a Watson-Crick base pair between A12 and U55 (i.e. no observable NOEs between the U55 imino proton to the A12 base protons). Furthermore, the in vitro selection results did not suggest any potential isosteric base pairs [e.g. G·G mismatch or C·A mismatch (48)] between the two bases. Among the two nucleotides, A12 is absolutely conserved in the selection, whereas A55 can be replaced with a C or a U without affecting the L30 protein-binding affinity. Nonetheless, the bases of A12 and A55 are close in space, judging from the medium-strong NOE between the H2 protons of the two adenines. Based on the sequence conservation and inward orientation of A12, this base could be more structurally important than A55 in organizing the purine stack.
G11·G56 mismatch (layer 3 in Fig. 9). Two NMR observations indicate formation of the G11·G56 mismatch base pair. First, the imino proton of G11 gives medium-strong NOE to G56 H8 proton, and through spin diffusion to G56 H1[prime]. Second, the exocyclic amino proton resonances of G11 are absent, most likely due to a hydrogen bond conformation that slows down the rotation about the C2-N2 bond. Two hydrogen bonds are thus anticipated: between the imino proton of G11 and the N7 of G56, and between an amino proton of G11 and the O6 of G56. The imino proton of G11 resonates at 11.18 p.p.m., which has more upfield shift relative to an average imino resonance hydrogen bonded with a nitrogen (35), however, this chemical shift may be affected by the presence of the L30 protein.
Normally, G·G mismatches using the side-to-head base pairing have two possible configurations (Fig. 11). In the first arrangement, which shows hydrogen bonds between the imino proton of one guanine and the O6 of another guanine and between a exocyclic amino proton of the first guanine and the N7 of the second guanine, the glycosidic angles for both guanosines are anti in a regular continuous helix (Fig. 11A). In contrast, in the second arrangement, which reverse above the hydrogen bonding pattern, the glycosidic angles require a syn conformation for one guanosine (Fig. 11B). The latter one (i.e. Ganti·Gsyn) is compatible with the NOE data of the L30 RNA. First, G56 shows a strong intra-residue NOE between the H8 and the H1[prime] (<2.5 Å), which is characteristic of a syn glycosidic conformation. In the syn glycosidic conformation there is a large ring current effect of the G56 base on its own H1[prime], which accounts for the unusual upfield chemical shift (4.60 p.p.m.) for this resonance. Secondly, the imino of G11 also gives a medium-strong NOE to the G56 H8 (<4.0 Å), which further supports the Ganti·Gsyn mismatch conformation. If the two bases adopted the first conformation, there would be a much weaker NOE between the G11 imino proton and the G56 H8 proton. Moreover, at millimolar NMR concentration, the G56A mutant forms a stable complex with the L30 protein, and maintains the syn glycosidic conformation at the 56 position. However, biochemistry showed that substitution of G56 with an adenine reduces the protein binding 7-10-fold. It is clear from the arrangement of G11anti·G56syn base pair that produces two hydrogen bonds, whereas G11anti·A56syn can only maintain one hydrogen bond using this configuration.
Figure 11. Schematic of two G·G side-to-head mismatches: (A), Ganti·Ganti and (B), Ganti·Gsyn.
CONCLUSION
The assignments of the two helices in the auto-regulatory mRNA of the yeast S.cerevisiae ribosomal protein L30 were mostly obtained by analyses of homonuclear spectra and heteronuclear spectra acquired on the uniformly 13C/15N-labeled sample. However, the assignment of purine-rich internal loop in the bound form was only made possible with the assistance of the selective 13C/15N-labeling, and the usage of five RNA mutants designed according to the available biochemical information. NMR studies revealed that the two helical stems were present in both the free and the protein-bound forms but that the internal loop was stable only in the presence of the protein. Thus sufficient identifiable intra- and inter-nucleotide NOEs of the internal loop in the L30-mRNA complex allowed further detailed structural modeling. The purine-rich internal loop in the complex displays an unusual conformation, and shows three major characteristics: the highly distorted backbone on both sides of the internal loop, the bulged A57 and G58, and the three-layer base stack. In combination with the NMR studies carried on the L30 protein, the complex structure provides detailed views of the base pairing and spatial orientation of the nucleotides in the internal loop, and reveals insights to the structural role of internal loop in mediating the L30 RNA-protein recognition.
ACKNOWLEDGEMENTS
We would like to thank Professor Susan White at Bryn Mawr College for sharing valuable biochemical information on this project. We thank Dr Kwaku Dayie and Dr John Chung at TSRI and Dr Christopher Turner at MIT for their assistance with NMR spectrometers. This project is supported by a grant from the National Institute of Health (GM-53320) to J.R.W.
REFERENCES
*To whom correspondence should be addressed. Tel: +1 858 784 8740; Fax: +1 858 784 2199; Email: jrwill{at}scripps.edu
This page is run by Oxford University Press, Great Clarendon Street, Oxford OX2 6DP, as part of the OUP Journals
Comments and feedback: jnl.info{at}oup.co.uk
Last modification:
Copyright© Oxford University Press, 1999.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  S. A. WHITE, M. HOEGER, J. J. SCHWEPPE, A. SHILLINGFORD, V. SHIPILOV, and J. ZARUTSKIE Internal loop mutations in the ribosomal protein L30 binding site of the yeast L30 RNA transcript RNA, March 1, 2004; 10(3): 369 - 377. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||