A three-dimensional working model for a guide RNA from Trypanosoma brucei
A three-dimensional working model for a guide RNA from Trypanosoma brucei
Thomas Hermann1,+, Beate Schmid§, Hermann Heumann1, H. Ulrich Göringer*
Laboratorium für Molekulare Biologie, Genzentrum der Ludwig Maximillians Universität München, 82152 Martinsried, Germany and 1Max-Planck-Institut für Biochemie, 82152 Martinsried, Germany
Received March 19, 1997; Revised and Accepted April 23, 1997
ABSTRACT
RNA editing in protozoan parasites is a mitochondrial RNA processing reaction in which exclusively uridylate residues are inserted into, and less frequently deleted from, pre-mRNAs. Molecules central to the process are so-called guide RNAs (gRNAs) which function as templates in the reaction. For a detailed molecular understanding of the mechanism of the editing process knowledge of structural features of gRNAs will be essential. Here we report on a computer-assisted molecular modelling approach to construct the first three-dimensional gRNA model for gND7-506, a ND7-specific gRNA from Trypanosomabrucei. The modelling process relied on chemical modification and enzymatic probing data and was validated by in vitro mutagenesis experiments. The model predicts a reasonably compact structure, where two stem/loop secondary structure elements are brought into close proximity by a triple A tertiary interaction, forming a core element within the centre of the molecule. The model further suggests that the surface of the gRNA is primarily made up of the sugar-phoshate backbone. On the basis of the model, footprinting experiments of gND7-506 in a complex with the gRNA binding protein gBP21 could successfully be interpreted and provide a first picture for the assembly of gRNAs within a ribonucleoprotein complex.
INTRODUCTION
Guide RNAs (gRNAs) are small, metabolically stable RNA molecules only identified within the mitochondria of kinetoplastid protozoan parasites (1). The molecules are trans-acting molecular components and play a central role during the maturation of mitochondrial pre-mRNAs, a process known as kinetoplastid (k)RNA editing (reviewed in 2). During kRNA editing, uridine nucleotides are site-specifically inserted and deleted into otherwise non-translatable mRNA molecules. The primary sequences of gRNAs contain the information for the processing reaction and this information is transferred to the pre-edited mRNAs in a base pairing interaction. The process is likely catalysed by a series of enzymatic reactions, presumably acting within a large ribonucleoprotein (RNP) complex (3-5).
gRNAs have an average length of 50-70 nt with a strong A/U nucleotide bias. On average 15 uridines are post-transcriptionally added to their 3[prime]-ends. Our current understanding of the reaction steps of an editing cycle allows the assignment of three basic functions for gRNAs. First, by base pairing to the pre-edited mRNAs proximal to an editing site, they define the endonucleolytic cleavage sites within the mRNA molecules (6). Second, as mentioned above, they function as templates for U insertion and deletion. Third, they potentially prevent the diffusion and loss of the upstream mRNA portion by base pairing of the 3[prime] oligo(U) tail following endonucleolytic cleavage (5).
The guiding capacity of gRNAs is determined by the overall length of the molecules and, as a consequence, multiple gRNAs are generally required for complete editing of a specific pre-mRNA (7,8). These gRNAs, though differing in primary sequence, function in the same biochemical reaction and are assembled into an identical RNP complex (3-5, reviewed in 9). As a working hypothesis we propose that, due to the lack of similarity within the primary structure of gRNAs, it is higher order folding rather than a common sequence motif that is essential for recruiting gRNAs into the editing apparatus. This notion is supported on two accounts. First, different gRNAs have been shown to become cross-linked to the same set of mitochondrial proteins, which indicated common binding sites for the polypeptides (10-12). Second, a detailed structure probing analysis of four Trypanosoma brucei gRNA molecules demonstrated that these RNAs have similar secondary structures despite their variable primary sequences (13). The foldings are characterized by two imperfect stem/loop elements with both the 5[prime]- and 3[prime]-ends in a single-stranded conformation. Aside from these data, no other structural information exists for gRNAs, although >200 potential gRNA sequences from eight different kinetoplastid organisms have been published (14).
Here we describe the computer-assisted molecular modelling of the three-dimensional architecture of gRNA gND7-506 from T.brucei. As primary input data we used published information from accessibility probing experiments where base-specific modification reagents and structure-specific enzymatic probes were used (13). The reliability of the model was verified by examining two gND7-506 mutant RNAs where base interactions predicted from the model were either disrupted or altered. The proposed three-dimensional model is supported by footprinting data of gND7-506 in a complex with gBP21, a gRNA binding protein from T.brucei (15). The model is intended as a basis for further work on the structure/function correlation of gRNAs during the editing reaction.
MATERIALS AND METHODS
General modelling procedures
Graphical model building was performed using the SYBYL molecular modelling software package (Tripos, St Louis, MO) on a Silicon Graphics Indy workstation. Geometries of standard A-form ribonucleotides were taken from the monomer library implemented in SYBYL, which uses the coordinates from Arnott and Hukins (16). Bond geometries in graphically constructed models were corrected by local minimization under the AMBER forcefield (17), ignoring electrostatic contributions but taking into account hydrogen bonding. Coordinates of the model can be obtained from the authors upon request.
Biochemicals
Diethylpyrocarbonate (DEPC) was purchased from Serva and dimethylsulfate (DMS) from Merck. Ribonucleases T1 and T2, T4 polynucleotide kinase and T4 RNA ligase were from Gibco BRL. Cobra venom RNase V1 was purchased from USB and AMV reverse transcriptase from Stratagene. [32P]pCp (3000 Ci/mmol), [[alpha]-32P]ATP (3000 Ci/mmol) and [[gamma]-32P]ATP (5000 Ci/mmol) were purchased from NEN or Amersham. Oligodeoxynucleotides were synthesized by solid support chemistry using O-cyanoethyl-N,N-diisopropylphosphoramidites.
RNA synthesis and construction of mutant gND7-506
gRNAs were transcribed from linearized plasmid DNA templates with T7 polymerase following standard procedures. Transcripts were purified from non-incorporated NTPs by size exclusion chromatography, recovered by ethanol precipitation and dissolved in 50 mM potassium cacodylate, pH 7.2, 150 mM KCl, 2.6 mM MgCl2, 0.1 mM Na2EDTA. The RNAs were renatured after purification (95°C, 2 min, followed by cooling the samples to 25°C at a rate of 1°C/min). RNA concentrations were determined by UV absorbency measurements at 260 nm using extinction coefficients ([epsis]260 l/mol/cm) calculated from tabulated values for the di- and mononucleotides (18).
Synthetic genes for two mutated gND7-506 RNAs [single site mutations at either position 43 from A to C (A43C) or at position 14 from A to U (A14U)] were constructed by self-assembly of overlapping 5[prime]-phosphorylated oligodeoxynucleotides according to Reyes and Abelson (19). The annealed DNA fragments were cloned into plasmid pBS- (Stratagene) and transformed into competent Escherichia coli SURE cells (Stratagene). Positive clones were identified by restriction enzyme digestion of isolated plasmid DNA and sequenced to verify the desired base change.
Radioactive end-labelling
Oligodeoxynucleotides and gRNAs were 5[prime]-end-labelled using T4 polynucleotide kinase and [[gamma]-32P]ATP. gRNAs were 3[prime]-end-labelled using T4 RNA ligase and [32P]pCp according to standard procedures. All radiolabelled nucleic acids were purified by denaturing polyacrylamide gel electrophoresis.
RNA structure probing and primer extension analysis
Chemical modification reactions were performed as described by Christiansen et al. (20) using DMS and DEPC as modifying reagents. Reactions were performed with 0.2 [mu]g gRNA and 0.8 [mu]g bulk yeast tRNA as a carrier in a volume of 10 [mu]l. Incubation was for 5 and 10 min for DEPC modifications or 2 and 5 min for reactions with DMS at 27°C, which is the optimal growth temperature of insect stage trypanosomes.
Reverse transcription reactions were performed using the oligodeoxyribonucleotide primer AAAATTCACTATATA, complementary to positions 47-65 of gND7-506. Modified gRNAs (1.3 pmol) and 5[prime]-end-labelled primer (1.0 pmol, [sim]50 000 c.p.m.) were annealed in 10 mM Tris-HCl, pH 6.9, 40 mM KCl, 5 mM Na2EDTA at 75°C for 30 s, followed by incubation at room temperature for 30 min. Extension reactions were performed at 37°C for 30 min with 1 U AMV reverse transcriptase per 10 [mu]l reaction. Complementary DNA products were separated on denaturing 10% (w/v) polyacrylamide gels.
gND7-506/gBP21 complex formation and footprinting experiments
Recombinant (r) gBP21 protein (15) was mixed with 5[prime]- or 3[prime]-end-labelled gND7-506 RNA in 60 [mu]l 6 mM HEPES, pH 7.5, 50 mM KCl, 2.6 mM MgCl2, 0.1 mM Na2EDTA, 0.5 mM DTT, 6% (v/v) glycerol. To ensure complete binding of the gRNA, the protein was used in at least a 10-fold molar excess over RNA. Association of the two molecules was achieved at 27°C for 30 min. Complexes were digested with RNase T1 (125 mU/[mu]l), RNase T2 (100 mU/[mu]l) and RNase V1 from cobra venom (25 mU/[mu]l) at 27°C for 7.5 min in the presence of yeast tRNA (0.25 [mu]g/[mu]l) as a carrier. Control samples did not contain any enzyme but were otherwise treated identically throughout the experiment. Reactions were stopped by the addition of KOAc, pH 6 (final concentration 270 mM), followed by extraction with phenol and ethanol precipitation and were analysed on 10% (w/v) polyacrylamide gels containing 8 M urea. Experiments were performed with three different r-gBP21 isolates that had been tested for their binding activity in nitrocellulose filter binding assays (15,21).
Circular dichroism measurements
CD spectra were measured at 27°C in 6 mM HEPES, pH 7.5, 50 mM KCl, 2.6 mM MgCl2, 0.1 mM Na2EDTA, 0.5 mM DTT, 6% (v/v) glycerol. Spectra were recorded from 300 to 195 nm with data acquisition every 0.1 nm. Scans were repeated 10 times and averaged. Mean molar residue ellipticities ([thetas]m) were calculated per mole of nucleotide monomer.
Temperature-dependent UV spectroscopy
Absorbance versus temperature profiles (melting curves) were recorded at 260 nm in 50 mM potassium cacodylate, pH 7.5, 50 mM KCl, 2.6 mM MgCl2, 0.1 mM Na2EDTA, 6% (v/v) glycerol using a thermoelectrically controlled Perkin Elmer lambda 16 spectrophotometer. The temperature was scanned at a heating rate of 1 or 2°C/min at temperatures between 10 and 95°C. Tm values were determined from derivative plots of absorbance versus temperature as 0.5 [Delta]A260.
RESULTS AND DISCUSSION
gRNA gND7-506
Trypanosomabrucei gRNA gND7-506 directs the editing of a sequence domain near the 5[prime]-end of the mRNA for subunit 7 of NADH dehydrogenase (ND7) (8). Within that sequence stretch 15 uridylate residues are inserted and six U nucleotides are deleted. gND7-506 has been identified in total RNA preparations from both procyclic and bloodform stage trypanosomes (8). A secondary structure model for gND7-506 was derived from surface probing experiments with base-specific modification reagents and structure-specific RNases in combination with temperature-dependent UV spectroscopy data (13). Figure 1A outlines the proposed structure, which consists of two hairpin elements, termed stem/loop I and stem/loop II, separated by seven single-stranded nucleotides. Both terminal ends of the molecule are likely in a single-stranded conformation, with evidence for a partly helical arrangement of the 3[prime] oligo(U) tail (13). Figure 1B shows a representative autoradiogram of a RNase accessibility experiment which emphazises the various domains of the gND7-506 RNA.
Figure 1 (A) Secondary structure model of T.brucei gRNA gND7-506, summarizing the chemical modification and enzyme accessibility data used as input in the modelling procedure (13). DMS, dimethylsulfate; DEPC, diethylpyrocarbonate; CMCT, 1-cyclohexyl-3(2-morpholinoethyl)carbodiimide; T1, RNase T1 (G-specific); T2, RNase T2 (single strand-specific); S1, nuclease S1 (single strand-specific); CV, cobra venom nuclease V1 (specific for double strands and stacked bases). The two stem/loop elements are marked as I and II. Bases near the 5[prime]-end which are involved in initial base pairing to the pre-mRNAs (`anchor interaction') are given as outline letters. (B) Representative autoradiogram of an enzyme accessibility experiment with cobra venom nuclease V1 (CV), RNases T1 and T2 and 32P 3[prime]-end-labelled gND7-506 RNA as described (13). The double-stranded (ds) domains of the RNA are shown as black rectangles, single-stranded (ss) regions are given as black bars and the 3[prime] oligo(U) tail as a grey rectangle. OH refers to an alkaline hydrolysis ladder.
Construction of the three-dimensional model
Starting from the secondary structure, construction of the three-dimensional model was performed in a stepwise fashion, beginning with independent modelling of the two helices. As initial building blocks we used coordinates from sequence-identical domains of either known RNA crystal structures or NMR-derived RNA solution structures stored in the PDB database. For stem I, coordinates from the PDB files 4TRA (tRNAPhe from E.coli; 22), 1SCL (sarcin/ricin loop of rat 28S rRNA; 23) and 1RNA [synthetic poly(UA) helix; 24] were used. Stem II was constructed from sub-structures of PDB files 1ELH (helix I of E.coli 5S rRNA; 25) and 1RNK (synthetic pseudoknot; 26), in addition to the aforementioned files 1SCL, 1RNA and 4TRA. Coordinates for the three G/U base pairs within gND7-506 (U31/G55, G32/U54, U33/G53) were derived by superimposition of several G/U pairs obtained from four crystal structures of tRNAPhe (PDB files 2TRA, 3TRA, 4TRA and 5TRA) (22) and from G/U base pairs in 1ELH and 1RNK. The A/A bulge within stem II (positions A34 and A52) was modelled according to a similar motif in the sarcin/ricin loop (PDB file 1SCL). Linking of the individual base pairs was done taking into account standard A-form double strand RNA geometry, resulting in a base stacking pattern in line with the experimental CV probing data (13). For an additional verification, we performed circular dichroism (CD) measurements and confirmed all typical A-form features (27) for gND7-506: a large negative elipticity at 210 nm, a moderate negative elipticity at 240 nm and a positive band around 265 nm (data not shown).
The bases of several single-stranded nucleotides immediately flanking stem I were modelled under the assumption of optimal stacking, again, as indicated by the CV probing data (13). G18, the first loop nucleotide of stem/loop I was built to stack on top of base C17 and bases A24, A25 and U26 were arranged to staple onto A23, thus extending the A-type conformation of helix I.
Restraining G18 as described reduced the possible conformational space for modelling of the 3 nt loop of stem I (bases G18-A19-U20). The final structure of the loop was assessed using the following data: A19 was strongly modified by DMS and DEPC and thus was arranged to be fully solvent accessible by pointing outward from the loop. In contrast, U20 was only weakly reactive towards 1-cyclohexyl-3(2-morpholinoethyl)carbodiimide (CMCT) and, as a consequence, was arranged to point to the inside of the loop.
For the relative orientation of the two helices to each other we examined the experimental probing data for inaccessible base positions and identified several adenosine residues which, based on the secondary structure model, showed unexpected reactivities: A14 at the 5[prime] base of stem/loop I was unreactive towards DEPC and any of the single strand- and double strand-specific enzymatic probes. It reacted only very weakly with DMS. Similarly, A43 within the loop on top of stem II was unreactive towards DMS and DEPC, but reactive towards the single strand-specific RNase T2. Finally, the two adenosines at positions 24 and 25 were DMS and DEPC reactive but also displayed a sensitivity towards the double strand-specific cobra venom enzyme. An inspection of the possible orientations of stem I and stem II to accommodate the modification data of A14 and A43 revealed a possible triple A interaction involving in addition A24 at the 3[prime]-end of stem I. Precedence for tertiary interactions of three adenylate residues has been reported for tRNAGln (28) and thus we further investigated this possibility in more detail. Due to their proximity to stem I, the positions of A14 andA24 were more or less fixed and the resulting A/A interaction caused a slight widening of the helix, probably explaining the observed fraying effect in the chemical and enzymatic probing studies (13). For the docking of A43 to A14/A24 at the base of stem I, two mutually exclusive orientations were conceivable. While the two possibilities lead to a different packing of stems I and II, considering the given length of the single-stranded region between the two stems only one orientation remained as a realistic alternative. Experimental evidence for the existence of the triple A interaction in gND7-506 will be shown below. After fixing the orientations of stems I and II, the best arrangement of the stems was obtained by a manual docking procedure which continuously monitored the distances of the RNA fragment termini that had to be connected by single-stranded regions (Distance Monitoring option, SYBYL software).
Lastly, the single-stranded but very likely stacked 3[prime] oligo(U) extension was modelled from the poly(U) half of a poly(A/U) RNA helix and further optimized by forcefield minimization in vacuo (17). The tail was joined with stem/loop II and the resulting final model is outlined in Figure 2. It is characterized by a tight parallel packing of the two stem structures whose major grooves face each other. The 3[prime] oligo(U) tail is dangling from that core structure, showing a high degree of flexibility. The majority of the surface of the molecule is made up by the sugar-phosphate backbone of the RNA, since almost all single-stranded bases are either pointing inwards from the loop structures or are buried at the interface of the two helices. Bases protruding freely into the solvent are exceptions, such as A19, which was strongly modified in the chemical probing experiments. The insert in Figure 2 depicts the exact geometry and hydrogen bonding pattern of the triple A interaction.
Figure 2 Ribbon model of the helix arrangement of the gND7-506 gRNA. Stems I and II are in green, loop regions are in orange and the oligo(U) tail is in blue. The bases A14, A24 and A43 are annotated in red. Bases A34 and A52, which form a bulge within helix II, are in green. Stems I and II are parallel packed facing their major grooves. The 7 nt loop on top of stem II is folded in between the helical grooves, forming a triple A tertiary interaction with the basis of stem I. The geometry of the triple A interaction of bases A14/A24/A43 is shown as an insert (bottom, right). The drawing was made on a Silicon Graphics Indy workstation using SYBYL 6.1 (Tripos, St Louis).
Comparison of the base accessibilities in the model with experimental probing data
As outlined above, construction of the model relied in part on experimental data of base accessibilities derived from chemical probing experiments with kethoxal, DEPC, CMCT and DMS. To allow for a quantitative comparison of the model with the probing results, we calculated theoretical accessibilities for relevant base atoms using the GEPOL algorithm (29) and the atomic radii of Rose et al. (30). The values were normalized with accessibilities of corresponding atoms in isolated, unpaired model nucleotides and finally scaled to match the 0-4 scale that was used for classification of the experimental data (13). Figure 3 depicts a graphical representation of these results, showing the differences in base accessibilities (model data minus experimental data) for the four chemical probes. From the 25 nt that had been identified to be reactive to the different probes, the model predicted the accessibility of 20 base positions correctly (80%). For 3 nt (12%) the model was in line with the modification for one probe but deviated for a second modification reagent. Only for two nucleotides (U15 and G35) did the calculated accessibilities notably underestimate the experimental values (8%).
Figure 3 Quantitative comparison of the base accessibilities in the model with the experimental probing data (13). The differences of base accessibilities towards DMS, DEPC, kethoxal and CMCT (model minus experimental) are shown using a relative reactivity scale ranging from 0 (not accessible) to 4 (very accessible). Adenosines are listed as two values, accounting for modification of the N7 position by DEPC or the N1 position by DMS. The values for guanosines were averaged to account for modification of both the N1 and N2 nitrogens by kethoxal. Hydrolysis positions in untreated control samples are annotated with an x and were not considered in the calculation. Considering an error margin of ±1, the model is in line with 80% of the experimental data.
Site-directed mutagenesis
To collect experimental evidence for the predicted triple A interaction of A14, A24 and A43 within the core of the gND7-506 structure, two single site mutant gRNAs were constructed. A base change at position 43 from A to C (mutant A43C) was predicted to disrupt the interaction (Fig. 4A), whereas an A[rarr]U transversion at position 14 (mutant A14U) was thought to potentially stabilize the tertiary interaction (Fig. 4B), similar to the A9/A23/U12 triple in E.coli tRNAPhe (29).
Figure 4 Evaluation of single site mutant gND7-506 gRNAs to test the presence of the triple A tertiary interactionwithin the core of the gRNA structure (dotted line connecting bases A14, A24 and A43). (A) Secondary structure of gND7-506 emphasizing the A43[rarr]C (A43C) base change, which was predicted to disrupt the tertiary interaction. (B) Mutant A14[rarr]U (A14U) was anticipated to stabilize stem I by the formation of a U14·A24 base pair and not interfere with the tertiary interaction. (C and D) Results of the chemical probing experiments for both gND7-506 mutant RNAs. The two graphs list the accessibility differences (mutant minus wild-type) of bases 10-39 for DMS and DEPC. (C) mutant A43C; (D) mutant A14U. The data are derived from densitometer scans of non-saturated autoradiograms scaling the accessibility values from 0 to 4 (see legend to Fig. 3). Positive values correspond to an increase in the accessibility and negative values convey a shielding effect when compared with the wild-type.
The two mutant gRNAs were transcribed in vitro from DNA templates which had been verified for the presence of the desired base changes. As an initial test of their structures we examined the UV melting profiles of the two mutant RNAs. Wild-type gND7-506 RNA, as published previously (13), is characterized by a main melting transition with a Tm of 39°C. This transition was interpreted to reflect primarily the melting of stem/loop II with the melting of stem/loop I superimposed on it. In addition, a small low temperature transition at [sim]20°C and a minor high temperature transition at 70°C have been described. They were assigned to either temperature-induced changes of the secondary structure or unidentified higher order transitions. Interestingly, both mutant gRNAs exhibited melting profiles indistinguishable from wild-type gND7-506 RNA (data not shown). This was not too surprising, since a disruption of the predicted H bonding pattern of the triple A interaction, which is certainly of low thermodynamic stability (see insert in Fig. 2), need not necessarily create a distinct change in the melting profile. Therefore, we decided to rely on chemical modification experiments with base-specific reagents (DMS and DEPC) to test the structures of the two gND7-506 mutant RNAs. Figures 4C and D summarize the results of the chemical probing experiments. For the A43C mutant RNA an enhanced accessibility predominantly at nucleotide positions at the base of stem I was found (Fig. 4C). In contrast, the A14U mutant RNA showed a small but significant reduction in the accessibilites of bases in that region (Fig. 4D). These data can be interpreted in the first instance as a disruption of the triple A interaction as a consequence of the A43C mutation and in the second case as an enhanced core interaction by the formation of a U14/A24/A43 triple.
gRNA-protein interaction
Finally, support for the model was gained from footprinting experiments of gND7-506 in a complex with gBP21, a gRNA binding protein from T.brucei (15). The arginine-rich polypeptide has been shown to bind to gRNA molecules with an equilibrium dissociation constant (Kd) in the nanomolar range. Protein motifs known to mediate RNA binding (32) are absent in gBP21. Based on the sensitivity of the gRNA/gBP21 complexes to elevated monovalent cation concentrations, the current model for the interaction of the two molecules involves, at least in part, ionic contacts. The association with gBP21 stabilizes the gRNA structure as judged from hyperchromicity experiments of the RNP complex in comparison with naked gRNA (15).
The footprinting experiments were performed with the single strand-specific RNases T1 and T2 and the double strand-specific cobra venom nuclease V1. Protected nucleotides were found exclusively within and at the base of stem/loop II (Fig. 5A and B). Nucleotide A37, loop nucleotide G44 and the sequence stretch G53-A56 at the 3[prime]-end of helix II were strongly protected. Several other positions were shielded to a weaker or only to a very weak extent (see colour code in Fig. 6A). One nucleotide position (U42) exhibited an enhanced reactivity in the RNP complex when compared with naked gRNA. U42 is located in the apical loop of hairpin II. Its sugar-phosphate residue is exposed on the surface of the model and thus it is very accessible, even in free gRNA. A graphical representation of these data on the three-dimensional model revealed a clustering of protected nucleotides on the solvent-accessible side of helix II (Fig. 6B) across its minor groove, which is indicative of a defined gRNA binding site for gBP21.
Figure 5 Determination of the binding site of T.brucei gBP21 protein on gND7-506 RNA. (A) Representative autoradiogram of a nuclease footprint experiment with 32P 3[prime]-end-labelled gND7-506 (0.6 [mu]M) digested with cobra venom nuclease V1 (CV, 25 mU/[mu]l) and RNase T1 (T1, 125 mU/[mu]l) in the presence (+) and absence (-) of recombinant gBP21 (6 [mu]M). Protected areas are marked with a vertical bar. OH refers to an alkaline hydrolysis ladder. (B) Bar graph summarizing the relative accessibility differences (+/- r-gBP21 protein) of nucleotide positions for cobra venom nuclease (CV, left panel), RNase T1 (right panel, black bars) and RNase T2 (right panel, grey bars) derived from three independent experiments. Nucleotide positions without a bar remained unchanged in their accessibility to the three enzymes when in a complex with gBP21. Bars pointing to the left are equivalent to protection of the nucleotide position in the presence of the protein, whereas bars pointing to the right represent enhanced accessibility. The data stem from densitometer scans of non-saturated autoradiographs using a normalized scale of 2 to -4 (0 no, 1 weak, 2 medium, 3 strong and 4 very strong difference in the nucleotide accessibility; + and - designate enhancement and protection respectively).
Figure 6. (A) Secondary structure and (B) space filling model of gND7-506 annotating protected backbone positions upon gBP21 binding in red > orange > yellow. Positions with no change in the accessibility pattern are shown in green, bases are given in blue. The sugar-phosphate backbone at nucleotide U42, which exhibited enhanced reactivity in the RNP complex, is coloured light blue. The two orientations of the model differ by a 90° rotation around the vertical axis. The footprinting pattern suggests binding of gBP21 protein to gND7-506 RNA at one side of stem II.
CONCLUSION
The computer-assisted modelling of RNA structures has proven a valuable tool in providing an understanding of the relationship between structure and function of a variety of RNAs (33-35). Although gRNAs have been identified as key components in the mitochondrial RNA editing reaction, a molecular understanding of their precise function is still lacking. This deficit, at least in part, is a consequence of the scarcity of structural data for these RNAs.
In the present study we have used chemical and enzymatic probing data to model the three-dimensional architecture of gND7-506, an ND7-specific gRNA from T.brucei. The main features and conclusions from the model are as follows. The folding of gND7-506 presents a remarkably compact structure. Two hairpins, which are the basic secondary structure elements of gND7-506, are closely packed together and likely connected by a tertiary interaction of three adenosine residues. The triple A interaction is formed between two adenosines at the base of stem I and a third A residue which is part of the apical loop of hairpin II. As a consequence, the two helices are lined up in an almost parallel fashion, facing each other with their major grooves.
The model is supported by a very good fit with the experimental probing data and was further strengthened by analysis of single site mutant gND7-506 RNAs, created by in vitro mutagenesis techniques. A base transversion, introduced to disrupt the triple A tertiary interaction, was confirmed to lead to a more open gRNA structure, with an increased accessibility of bases at the interface of the two helices. Conversely, a mutation with the potential to strengthen the tertiary interaction indeed resulted in a gRNA that showed characteristics of a more compact folding.
Lastly, footprinting experiments on gND7-506 in a stable complex with the gRNA binding protein gBP21 were used to support the model. Binding of the polypeptide resulted in the protection of a well-defined set of nucleotide positions, identifying a substantial part of stem/loop II as the interaction site for the protein. Very likely, this binding domain can be defined more precisely if the footprinting experiments are performed with chemical modification reagents instead of the rather bulky enzymatic probes. Interestingly, the gRNA structure remained largely unchanged when in a complex with gBP21. Only one nucleotide (U42) became more accessible to RNase T2. This result is in line with CD measurements of gND7-506 in the presence and absence of gBP21 (15). The data indicated no gross structural rearrangement of the RNA molecule upon protein binding. Since the functional groups of the bases in minor grooves are rather inaccessible and do not allow good discrimination of different base pairs (36), we feel that it is unlikely that the interaction relies on some form of base specificity. The indiscriminate binding behaviour of gBP21 to different gRNA molecules is further support for this assumption (15). As previously suggested (37), it is more likely that the RNA molecule adopts a three-dimensional folding that results in presentation of defined patches of negative charges on the surface of the molecule which correspond to a complementary array of positive charges within the binding domain of the protein. This hypothesis rationalizes experimental data for the cation sensitivity of the gRNA/gBP21 interaction (15). Whether ionic interaction is the general principle for the binding between gRNAs and proteins and whether it also applies for the recruitment of gRNAs into the editing machinery cannot be assessed at the moment. However, the phenomenon might in part be responsible for the observed salt sensitivity of the RNA editing activity in vitro (38).
The obvious question that arises from the presented model for gND7-506 is whether all gRNAs can be folded into a similar three-dimensional architecture. For only three additional gRNA molecules has secondary structure information been collected. These RNAs also consist of two imperfect stem/loop elements (13). Thus, similar foldings seem to be conceivable, although they clearly cannot rely on the same triple A tertiary interaction, simply because their primary sequences differ in the relevant sequence domains. However, many other forms of tertiary interactions can easily be envisaged (28). Even for gND7-506, we predict that other interactions contribute to the overall stability of the molecule. Due to the low resolution of the model, only the triple A interaction could be resolved, but additional contacts are required to mediate a stable interaction of the two hairpin elements. Cross-linking studies of gND7-506 as well as of other gRNAs must be performed to ultimately solve this question.
Finally, we would like to emphasize that we view the model only as an initial step towards a molecular understanding of the structure of gRNAs. Hopefully it will stimulate further work on the structure/function correlation of gRNAs to ultimately gain a molecular understanding of the role of gRNAs in the RNA editing process.
ACKNOWLEDGEMENTS
We wish to thank A.Souza, U.Müller and A.S.Paul for critical reading of the manuscript. J.Köller is thanked for r-gBP21 protein preparations and E.Weyher-Stingl and L.Moroder for their help performing the CD measurements. Work in the H.U.G. laboratory is supported by the German Ministry for Education and Research (BMBF) and the German Research Foundation (DFG). T.H. acknowledges support from the Studienstiftung des deutschen Volkes and H.H. from the DFG.
REFERENCES
1. Blum,B., Bakalara,N. and Simpson,L. (1990) Cell, 60, 189-198.MEDLINE Abstract
2. Hajduk,S.L. and Sabatini,R.S. (1996) In Smith,D.F. and Parsons,M. (eds), Molecular Biology of Parasitic Protozoa. IRL Press, Oxford, UK pp.134-158.
17. Weiner,S.J, Kollman,P.A., Case,D.A., Singh,U.C., Ghio,C., Alagona,G., Profeta,S. and Weiner,P. (1984) J. Am. Chem. Soc., 106, 765-784.
18. Puglisi,J. and Tinoco,I. (1989) Methods Enzymol., 180, 304-325.MEDLINE Abstract
19. Reyes,V.M. and Abelson,J.N. (1989) Methods Enzymol., 180, 63-69.MEDLINE Abstract
20. Christiansen,J., Egebjerg,J., Larsen,N. and Garrett,R.A. (1990) Ribosomes and Protein Synthesis. A Practical Approach. IRL Press, Oxford, UK, pp. 229-252.
21. Witherell,G.W. and Uhlenbeck,O.C. (1989) Biochemistry, 28, 71-76.MEDLINE Abstract
22. Westhof,E., Dumas,P. and Moras,D. (1988) Acta Crystallogr., A44, 112-123.
23. Szewczak,A.A. and Moore,P.B. (1995) J. Mol. Biol., 247, 81-98.MEDLINE Abstract
31. Quigley,G.J and Rich,A. (1976) Science, 194, 796-806.MEDLINE Abstract
32. Burd,C.G. and Dreyfuss,G. (1994) Science, 265, 615-620.MEDLINE Abstract
33. Westhof,E., Romby,P., Ehresmann,C. and Ehresmann,B. (1990) In Beveridge,D.L. and Lavery,R. (eds), Theoretical Biochemistry and Molecular Biophysics. Adenine Press, New York, NY, pp. 399-409.
34. Gautheret,D. and Cedergren,R. (1993) FASEB J., 7, 97-105.MEDLINE Abstract
35. Westhof,E. (1993) J. Mol. Struct.Theochem., 286, 203-211.
36. Weeks,K.M. and Crothers,D.M. (1993) Science, 261, 1574-1577.MEDLINE Abstract
37. Göringer,H.U., Koslowsky,D.J., Morales,T.H. and Stuart,K. (1994) Proc. Natl. Acad. Sci. USA, 91, 1776-1780.
38. Seiwert,D.S. and Stuart,K. (1994) Science, 266, 114-117.
Present addresses: +Institut de Biologie Moléculaire et Cellulaire, 67084 Strasbourg, France and §Memorial Sloan-Kettering Cancer Center, New York, NY 10021, USA
CiteULike Connotea Del.icio.usWhat's this?
This article has been cited by other articles:
M. J. Madej, J. D. Alfonzo, and A. Huttenhofer Small ncRNA transcriptome analysis from kinetoplast mitochondria of Leishmania tarentolae
Nucleic Acids Res.,
March 12, 2007;
35(5):
1544 - 1554.
[Abstract][Full Text][PDF]
D. E. GOLDEN and S. L. HAJDUK The 3'-untranslated region of cytochrome oxidase II mRNA functions in RNA editing of African trypanosomes exclusively as a cis guide RNA
RNA,
January 1, 2005;
11(1):
29 - 37.
[Abstract][Full Text][PDF]
L. SIMPSON, R. APHASIZHEV, G. GAO, and X. KANG Mitochondrial proteins and complexes in Leishmania and Trypanosoma involved in U-insertion/deletion RNA editing
RNA,
February 1, 2004;
10(2):
159 - 170.
[Abstract][Full Text][PDF]
L. SIMPSON, S. SBICEGO, and R. APHASIZHEV Uridine insertion/deletion RNA editing in trypanosome mitochondria: A complex business
RNA,
March 1, 2003;
9(3):
265 - 276.
[Abstract][Full Text][PDF]
M. Lorger, M. Engstler, M. Homann, and H. U. Goringer Targeting the Variable Surface of African Trypanosomes with Variant Surface Glycoprotein-Specific, Serum-Stable RNA Aptamers
Eukaryot. Cell,
February 1, 2003;
2(1):
84 - 94.
[Abstract][Full Text][PDF]
U. F. Muller and H. U. Goringer Mechanism of the gBP21-mediated RNA/RNA annealing reaction: matchmaking and charge reduction
Nucleic Acids Res.,
January 15, 2002;
30(2):
447 - 455.
[Abstract][Full Text][PDF]
J. Cruz-Reyes, A. Zhelonkina, L. Rusche, and B. Sollner-Webb Trypanosome RNA Editing: Simple Guide RNA Features Enhance U Deletion 100-Fold
Mol. Cell. Biol.,
February 1, 2001;
21(3):
884 - 892.
[Abstract][Full Text]
S. S. Leung and D. J. Koslowsky RNA editing in Trypanosoma brucei: characterization of gRNA U-tail interactions with partially edited mRNA substrates
Nucleic Acids Res.,
February 1, 2001;
29(3):
703 - 709.
[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]
M. L. Hayman and L. K. Read Trypanosoma brucei RBP16 Is a Mitochondrial Y-box Family Protein with Guide RNA Binding Activity
J. Biol. Chem.,
April 23, 1999;
274(17):
12067 - 12074.
[Abstract][Full Text][PDF]
T. E. Allen, S. Heidmann, R. Reed, P. J. Myler, H. U. Göringer, and K. D. Stuart Association of Guide RNA Binding Protein gBP21 with Active RNA Editing Complexes in Trypanosoma brucei
Mol. Cell. Biol.,
October 1, 1998;
18(10):
6014 - 6022.
[Abstract][Full Text]
CiteULike 
Connotea 
Del.icio.us What's this?
