Skip Navigation

This Article
Right arrow Abstract Freely available
Right arrow Print PDF (121K) Freely available
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrow Search for citing articles in:
ISI Web of Science (23)
Right arrowRequest Permissions
Right arrow Commercial Re-use Guidelines
for Open Access NAR Content
Google Scholar
Right arrow Articles by Peabody, D.
Right arrow Articles by Lim, F
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Peabody, D.
Right arrow Articles by Lim, F
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?

© 1996 Oxford University Press 2352-2360

Footnote

Complementation of RNA binding site mutations in MS2 coat protein heterodimers

Complementation of RNA binding site mutations in MS2 coat protein heterodimers David S. Peabody* and Francis Lim

Department of Cell Biology, University of New Mexico School of Medicine and Cancer Research and Treatment Center, Albuquerque , NM 87131, USA

Received February 20, 1996; Revised and Accepted April 19, 1996

ABSTRACT

The coat protein of bacteriophage MS2 functions as a symmetric dimer to bind an asymmetric RNA hairpin. This implies the existence of two equivalent RNA binding sites related to one another by a 2-fold symmetry axis. In this view the symmetric binding site defined by mutations conferring the repressor-defective phenotype is a composite picture of these two asymmetric sites. In order to determine whether the RNA ligand interacts with amino acid residues on both subunits of the dimer and in the hope of constructing a funtional map of the RNA binding site, we performed heterodimer complementation experiments. Taking advantage of the physical proximity of their N- and C-termini, the two subunits of the dimer were genetically fused, producing a duplicated coat protein which folds normally and allows the construction of the functional equivalent of obligatory heterodimers containing all possible pairwise combinations of the repressor-defective mutations. The restoration of repressor function in certain heterodimers shows that a single RNA molecule interacts with both subunits of the dimer and allows the construction of a functional map of the binding site.

INTRODUCTION

The coat protein of the RNA bacteriophage MS2 plays a dual role, functioning both as the major structural protein of the viral particle and as a translational repressor of the viral replicase gene ( 1 ). These functions rely on highly specific protein-protein and protein-RNA interactions. A coat protein dimer functions both as the translational repressor and as a precursor to viral capsids. A ribbon diagram of the dimer, derived from the X-ray structure of the viral particle ( 2 ), is shown in Figure 1 . The subunits are composed of two main elements: a five strand [beta]-sheet which, in the dimer, forms a continuous 10 strand [beta]-sheet and an [alpha]-helical segment which each monomer extends over one surface of the [beta]-sheet of its companion subunit. The two helices interdigitate in antiparallel fashion. Thus, the subunits are not independently folded, but their final conformations depend on their mutual association. Mutational analysis has identified residues on the solvent-exposed surface of the [beta]-sheet as components of the RNA binding site ( 3 ).


Figure 1 . The structure of the MS2 coat protein dimer (redrawn from 2).

We previously reported the isolation of coat mutants that produce repressor dimers but which are defective for assembly into capsids. The crystal structure of one such coat protein mutant confirms the 2-fold symmetry and indicates that the structure is fundamentally identical to that observed in the virus particle ( 2 , 4 ). This being the case, two symmetrically equivalent operator binding sites must exist before RNA binding. However, since the repressor binds only one molecule of operator RNA at saturation ( 5 ), occupation of one site somehow prevents occupation of the other site. This negative cooperativity could be accomplished by either of two mechanisms: (i) binding at one site induces a conformational change that reduces the affinity of binding at the second site; (ii) the two sites overlap so that simultaneous occupation of both is sterically prohibited.

Since the symmetric coat protein dimer binds only one molecule of an inherently asymmetric RNA, why is the repressor a dimer? One possibility is that amino acid residues on both subunits of the dimer form contacts with a single RNA molecule. According to this model the RNA binding site is formed through the participation of amino acid residues on both monomers. Dimerization is therefore necessary for RNA binding. Alternatively, an individual RNA binding site might be restricted to one monomer. The distinction between these two models is illustrated in Figure 2 . Consider each of the two symmetry-related binding sites to be composed of two non-identical half-sites. Call the half-sites of one binding site A and B and call the half-sites of the other binding site A' and B'. It is the conjunction of half-sites A and B or of A' and B' that results in the formation of one complete binding site. The half-sites cooperate by binding different portions of the operator. When considered in these terms we may then ask whether A and B, or A' and B', reside on the same or different subunits.


Figure 2 . Two alternative models of the RNA binding site. ( A ) In the model shown here each of the two equivalent binding sites is arbitrarily divided onto two non-identical half-sites that reside on different subunits. ( B ) In the alternative model the half-sites reside on the same subunit. In each case the consequences of producing heterodimers with mutational lesions (denoted by asterisks) in different half-sites are shown.

It is possible to distinguish these two models experimentally using a series of repressor-defective mutants with lesions in the RNA binding site. Such a collection of mutants has been described elsewhere ( 3 ). As illustrated in Figure 2 A, one model holds that a binding site for RNA requires contacts on both monomers. This model predicts that heterodimers formed from mutants defective in different half-sites will bring two normal half-sites together to form one intact binding site. In the alternative model (Fig. 2 B) the formation of heterodimers cannot restore a functional binding site. If the first model is correct, it should be possible to observe complementation with certain pairs of repressor-defective mutants. Furthermore, one might expect that a collection of such mutants would comprise two complementation groups, each corresponding to one of the two half-sites. Complementation tests might therefore permit the assignment of residues to different half-sites. Of course, some residues may contribute to both half-sites. Mutations that alter such residues will not be complemented by any other binding site mutation and, for this reason, should be regarded as constituting a third class of mutations, even though they are members of both functional half-sites. Alternatively, a mutation may not be complementable if the amino acid substitution in one half-site has effects on the structure of the other half-site.

A variety of approaches to the formation of heterodimers could be envisioned. However, inspection of the atomic model of the dimer ( 2 , 4 ) reveals that the C-terminus of each monomer closely approaches the N-terminus of the other monomer. The proximity of the N- and C-termini suggested the possibility of duplicating the coat sequence in such a way as to create a translational fusion between the identical subunits. The result would be a polypeptide of twice normal coat protein molecular weight in which the coat protein sequence appears twice. Assuming the duplicated coat protein sequence would fold properly when produced as a `two-domain' monomer rather than as the conventional dimer, its existence would permit the construction of recombinants in which different halves of the sequence contain different amino acid substitutions, thus creating the functional equivalent of obligatory heterodimers. This paper describes the construction of functional duplicated coat proteins and their utilization in the construction of mutant heterodimers. All possible pairwise combinations of repressor-defective mutants were introduced into heterodimers and tested for translational repression. Based on the resulting complementation pattern a model of the RNA binding site can be proposed. The X-ray structure of the MS2 virus-operator RNA complex has been reported recently ( 6 ). Here we compare the functional RNA binding site defined by our genetic studies with that identified by structural analysis.

MATERIALS AND METHODS

Recombinant plasmids were constructed by standard techniques as described briefly below. Oligonucleotides were synthesized by the University of New Mexico Protein Chemistry Facility. SDS-PAGE ( 7 ) and Western blot analysis ( 8 ) have been described before. The use of Sepharose CL-4B chromatography and agarose gel electrophoresis to identify assembly-defective MS2 coat mutants has also been reported elsewhere ( 3 , 9 ). Filter binding studies were conducted as described by Carey et al. ( 10 ) using 32 P-labeled RNA produced by run-off transcription of a cloned operator sequence using T7 RNA polymerase in vitro ( 11 ).

Plasmid pCT21, a coat duplication, was constructed from pCT119 ( 9 ) and pCT119dl-5N, which contains a version of the coat sequence into which a Nco I site has been introduced at the coat protein initiator AUG by single nucleotide substitution. A synthetic duplex oligonucleotide was used to join the two coat sequences together utilizing a Hpa II site near the 3'-end of the upstream copy (from pCT119) and the Nco I site at the 5'-end of the downstream copy (from pCT119dl-5N). The overall structure of the resulting plasmid is shown in Figure 3 A and the nucleotide sequence at the junction of the duplication is shown in Figure 3 B. The plasmid pCT22 was derived from pCT21 by the introduction of a single nucleotide substitution converting the initiator AUG of the downstream copy of the coat to UUG using site-directed mutagenesis ( 12 ).


Figure 3 . ( A ) The structure of pCT21 containing the duplicated coat sequence. ( B ) The duplication of the coat sequence was accomplished by joining the 3'-end of the upstream copy of coat by its Hpa II site to the 5'-end of the dowstream copy at its Nco I site using synthetic DNA as shown here.

Small deletions around the duplication junction were constructed as follows. Plasmid pCT21 was digested with Nco I and then treated with S1 nuclease. Ragged ends were repaired using the large fragment of DNA polymerase I and the plasmid was recirularized by treatment with T4 DNA ligase. The ligated DNA was again treated with Nco I and introduced by transformation into Escherichia coli . Individual isolates were tested for the ability to produce the duplicated coat protein by SDS-PAGE of cell lysates and Western blotting using anti-MS2 serum and 125 I-labeled protein A. Deletions were characterized by DNA sequence analysis.

The mutant `heterodimer' constructions were created using a 120 bp Eco RI- Sal I fragment containing the duplication junction from pCT2dl-13 (see Fig. 6 ) to join an upstream coat sequence by its Eco RI site to a downstream coat sequence at its Sal I site (see Fig. 3 A for the locations of these sites). The repressor-defective mutants used in this study were described previously ( 3 ).

RESULTS

Construction and expression of a duplicated coat sequence

The duplication of the coding sequence for bacteriophage MS2 coat protein was performed as described in Materials and Methods and as illustrated in Figure 3 . The construction utilized synthetic duplex oligonucleotides to join the two sequences at a Hpa I site near the 3'-end of the upstream gene and a Nco I site that had previously been introduced by site-directed mutagenesis at the initiator AUG of the downstream coat sequence. The result was the in-frame fusion of the two coat coding sequences with the insertion of a two amino acid linker peptide (Ala-Met) between the normal C-terminus of the upstream copy and the usual N-terminus of the downstream copy. The introduction of a linker peptide of this length was more for the purposes of convenient construction than the result of deliberate design based on protein engineering considerations. The plasmid containing the duplicated sequence is called pCT21. We will refer to its protein product as COAT21 to distinguish it from wild-type coat protein, which we call COAT.

Except for the duplication, pCT21 is similar to the previously described coat protein expression vector called pCT119 ( 3 , 9 ) and expresses the COAT21 sequence using the Lac promoter of E.coli . To examine the expression of COAT21, 1ysates of cells containing pCT21 were subjected to SDS-PAGE and proteins were visualized by Western blot analysis. These cells contain a major new coat protein species that is twice the size of COAT (Fig. 4 ) and is produced at least as abundantly as wild-type COAT. Notice also the production of small amounts of normal sized coat protein from pCT21. This might be the result of internal translation initiation at the second coat initiator AUG, even though the internal AUG is not preceded by any obvious Shine-Dalgarno homology. Alternatively, this product could result from proteolytic degradation of the COAT21 protein or from recombination between the duplicated sequences, occasionally resulting in the production of the unit length gene. In an attempt to eliminate synthesis of the shorter form the initiator AUG of the downstream copy of the coat sequence was converted by site-directed mutagenesis to UUG. The result is the substitution of Leu for Met. Although UUG occasionally serves as an initiation codon in E.coli , it is generally less efficient that AUG (for example see 13 ). Moreover, Leu is a relatively conservative substitution for Met, so that the functional properties of the protein should be preserved in this mutant. The plasmid containing this mutation is called pCT22 and its protein product is called COAT22. Figure 4 shows that pCT22 produces only slightly reduced quantities of the normal sized coat protein compared to pCT21. Deletion of nucleotides (including the AUG triplet) encoding two or three amino acids at the junction of the two sequences (described below) or use of a recA - host (results not shown) also fail to eliminate completely the synthesis of monomer sized coat protein. Thus the mechanism by which these small amounts of unit length protein are produced remains a mystery.


Figure 4 . Western blot analysis of the coat protein products of pCT119 (wild-type) and pCT21 and pCT22 (duplicated coat proteins).

Translational repression by the duplicated coat protein sequence

The construction of a two-plasmid system in which the synthesis of coat protein from one plasmid represses the translation of a hybrid replicase-[beta]-galactosidase protein encoded on the second plasmid has been described before ( 9 ). Translational repression results in the production of white colonies on plates containing the chromogenic substrate 5-bromo-4-chloro-3-indolyl-[beta]-D-galactoside (X-gal), while repressor-defective mutants give rise to blue colonies. In this colony color assay the presence of either pCT21 or pCT22 results in white colonies, indicating that both plasmids produce fully active translational repressors.

Failure of COAT21 and COAT22 to assemble into virus capsids

We previously reported that coat protein produced from a plasmid co-migrated with authentic virus during native agarose gel electrophoresis and that assembly-defective mutants behave very differently ( 3 ). Using this method COAT21 and COAT22 proteins were found to be assembly-defective. Figure 5 A shows that they fail to produce a discrete species with the mobility of normal capsids. Instead they produce a smeared pattern of lesser mobility. This is characteristic of most assembly-defective mutants. However, the COAT21 and COAT22 proteins do not produce the banding pattern typical of previously characterized mutants that form only dimers ( 2 , 14 ). Since any gross folding defect would result in translational repressor defects, these results apparently reflect a reluctance of the properly folded polypeptide to efficiently assemble into stable capsids.


Figure 5 . Agarose gel electrophoresis/Western blot analysis ( A ) of the wild-type, COAT21 and COAT22 proteins and ( B ) of the various duplicated coat proteins.


Figure 6 . Sepharose CL-4B elution profiles of coat proteins produced from ( A ) pCT119, ( B ) pCT21 and ( C ) pCT21dl-13.

The elution behavior of coat protein during gel exclusion chromatography on columns of Sepharose CL-4B is also an indicator of assembly. As we have described elsewhere, plasmid-produced coat protein co-elutes with authentic virus ( 9 ). When analyzed by this means about half of the duplicated coat protein is present in capsid-sized material (Fig. 6 ). By this measure COAT21 is partially defective for capsid assembly.

Construction of deletions of the linker peptide

The failure of COAT21 and COAT22 to assemble into stable virus-like particles might be the result of the presence of the two amino acid linker peptide at the junction of the duplication. Inspection of the structure of the MS2 virus particle suggests that extra sequences in the vicinity of the N- and C-termini could indeed be expected to interfere with interdimer contacts required for capsid assembly. Therefore, random, short deletion mutations around the Nco I site of pCT21 were introduced by treatment of the Nco I-digested plasmid with S1 nuclease followed by ligation (see Materials and Methods). The resulting deletion mutants were introduced by transformation into the colony color indicator strain and were screened for translational repressor activity by production of white colonies on X-gal plates. Twenty four white or pale blue colonies were picked for analysis. Plasmids were isolated and each was found to lack the Nco I site. SDS-PAGE and Western blotting identified 14 that continued to produce a coat protein twice normal size and native agarose gel electrophoresis indicated that each of these assembled into virus-like particles. Six of these were subjected to DNA sequence analysis. Figure 7 shows that three different types of deletions were identified. The plasmids containing these deletions are called pCT2dl-1, pCT2dl-6 and pCT2dl-13. In each case two or three amino acids have been removed from the COAT21 sequence. Figures 5 B and 6 C show that these deletions restore the ability to assemble into a virus-like particle as assessed both by agarose gel electrophoresis and by chromatography on Sepharose CL-4B.


Figure 7 . The nucleotide sequence at the duplication junction of pCT21 and three deletion mutations that restore capsid assembly activity.

Comparison of in vitro RNA binding activities of wild-type and duplicated coat proteins

The experiments described above show that the duplicated coat protein produced by pCT2dl-13 behaves like the wild-type protein with respect to its capacity for formation of virus-like particles and in its ability to repress translation of the replicase-[beta]-galactosidase fusion protein. To determine whether the repressor activity of COAT2dl-13 was the result of a normal capacity for RNA binding, proteins were purified by methods previously described ( 9 ) and subjected to protein-excess filter binding analyses. In the experiments shown in Figure 8 the abilities of increasing concentrations of the wild-type and duplicated proteins to cause the retention of 32 P-labeled operator RNA on nitrocellulose filters were measured. They produced indistinguishable binding profiles, indicating equivalent RNA binding affinities. In each case the binding curves closely fit a theoretical curve calculated assuming the previously reported value for K d of 3 nM ( 10 ).


Figure 8 . Filter binding analysis of the wild-type (circles) and the COAT21dl-13 (triangles) coat proteins. The curve was calculated assuming K d = 3 nM.

Incorporation of mutant and wild-type sequences into the duplicated coat gene

Elsewhere we have described the isolation of a series of mutant coat sequences defective for translational repression and RNA binding, but competent for assembly of virus shells, suggesting that the mutations affect RNA binding site residues ( 3 ). Mutations causing amino acid substitutions at 10 different sites per monomer were isolated. All of them affect residues whose side chains project from the solvent-exposed surface of the [beta]-sheet on the interior of the viral particle. Mutants with substitutions at these 10 sites formed the basis of an experiment designed to test whether some pairs of mutations would restore RNA binding and translational repression when present in heterodimers. Such recombinants were constructed using the wild-type sequence and mutants with lesions at each of the 10 sites previously identified as important for translational repression ( 3 ). In order to test all possible pairwise combinations of the wild-type and mutant sequences a total of 121 pCT2dl-13 recombinants were produced. These recombinants are listed in a matrix in Figure 9 , where the mutation present in the 3'-half of a given heterodimer is listed across the top and the mutation present in the 5'-half is listed down the left side. Each combination of mutations occurs twice, since any given mutation appears both in the 5'- and 3'-halves of heterodimers. Each recombinant was tested for its ability to produce blue or white colonies on X-gal plates. In most cases the distinction between complementation and non-complementation was obvious, since the colonies were clearly blue or white. Such cases are indicated with an x symbol in Figure 9 . In other cases a definite, but less dramatic, improvement in repressor activity was observed. These instances are denoted by a / symbol in Figure 9 . Complementation was achieved in a number of cases and the pattern of complementation was mostly symmetric about the diagonal. Most mutations seem to segregate into two groups. One group (call it group A) is made up of Y85H, N87S and E89K. The other group (call it group B) contains S47R, R49H, N55D, K57E, T59S and T91I. There are a few anomalies in the complementation behavior. Note, for example, that S47R behaves in all respects as though it belongs to group B except for its complementation in both dimer halves by T59S and in one half by K57E. K61R was incapable of being complemented by the wild-type or any mutant sequence. This is the behavior expected of a residue belonging to both half-sites. E89K seems to fit a pattern similar to Y85H and N87S in generally failing to complement the same set of mutations. However, E89K weakly complements Y85H and N87S.


Figure 9 . A matrix showing all the pairwise combinations of heterodimers containing the wild-type and 10 mutant coat sequences. Amino acid substitutions present in the 3'-half of the fused dimer are listed across the top, while those present in the 5'-half are listed down the left side. In ( A ) the relative intensities of blueness of colonies on X-gal plates are compared. ( B ) The complementation pattern. The blueness of each heterodimer pair was compared to the blueness of the least blue (i.e. least defective) of its homodimer parents (found on the diagonal of the matrix). Only those mutant combinations that gave rise to a - or +- phenotype were considered to be candidates for complementation. Those improving from ++ or +++ to +-, or those improving from +, ++ or +++ to - were defined as strongly complementing and assigned an x symbol. Those that showed improvement from + to +- complemented less strongly and were assigned a / symbol.

Some may wonder whether these results are influenced by the low level production of the monomer sized species shown in Figure 4 , because it could presumably self-associate to form homodimers. We argue that this protein is produced in such small quantities that it should make a negligible contribution to translational repression. Moreover, any homodimers produced in this fashion would be repressor-defective in all cases where both parents are repressor-defective and thus could not account for complementation between two mutants. We cannot, however, rule out categorically a contribution to translational repression in cases where the wild-type sequence is paired with a mutant.

DISCUSSION

One purpose of this work was to determine whether a duplicated coat protein was capable of folding properly into the structure typical of the dimer. Figure 1 shows that the two halves of the normal coat protein dimer are not independently folded, but are intertwined so that the final conformation of each monomer is determined by association with its identical partner. The results reported here indicate that a duplicated coat sequence is able to fold correctly, since its translational repressor, RNA binding and capsid assembly functions are essentially indistinguishable from the those of the wild-type protein. As originally constructed in pCT21 the duplicated coat sequence contained two extra amino acids linking the halves. One is an alanine encoded in the synthetic DNA sequence which links the two coat coding sequences. The other is the methionine which would normally be cleaved from the N-terminus, but which now resides at an internal position in the duplicated sequence. The presence of the two extra amino acids in the original duplication construction (pCT21) resulted in a partial defect in capsid assembly or stability, but apparently did not affect the formation of dimer-like molecules, since COAT21 is perfectly capable of translational repression. Inspection of the structure of the MS2 viral particle ( 4 ) reveals that the extra amino acids reside in a location where they may interfere with contacts between dimers at the quasi-3-fold axis ( 15 ). Consistent with this idea, our results show that removal of two or three amino acids at the junction of the duplication restores the capacity to form stable capsids.

The major purpose of this work was to test whether a single operator RNA molecule forms interactions with both subunits of the coat protein dimer. By the arguments elaborated in the Introduction and summarized in Figure 2 , if the RNA binding site is formed from residues on both subunits, certain pairs of repressor-defective mutants should reconstitute functional repressors when present in different halves of the dimer. This condition was satisfied in a number of cases. We also hoped that by creating all possible heterodimers of the repressor-defective mutants a complementation pattern might emerge, allowing us to assign specific amino acid residues to different half-sites. Accordingly, we used mutants with amino acid substitutions at 10 different positions within the RNA binding site to create a series of pCT2dl-13 `heterodimer' recombinants. Figure 9 shows all the combinatorial possibilities and summarizes the results. The following points should be noted.

(i) When the wild-type sequence is present in either half of the molecule, repressor activity is restored in all cases except with K61R.

(ii) K61R is not complemented by the wild-type nor by any mutant sequence. We interpret this to mean that residue 61 is a constituent of both half-sites.

(iii) Although the complementation behavior is not ideal in all respects, it does seem that two groups of complementable mutations can be discerned. One group contains Y85H, N87S and E89K and the other contains S47R, R49H, N55D, K57E, T59S and T91I.

(iv) Although the mutations seem to segregate roughly into two complementation groups, there are some anomalies that are not easily explained within the framework of the two half-sites model. For example, S47 behaves as though it belongs to half-site B, except for its unexpected complementation in either dimer half by K57E and in one dimer half by T59S. E89K seems to group with Y85H and N87S, but weakly complements them both. We don't know how to explain these behaviors. However, it is an implicit assumption of our two half-sites model that any particular amino acid substitution in one half-site does not affect the function of amino acid residues in the other half-site. This is probably an over-simplification. For these reasons we view this model of the binding site as tentative.

(v) Note that the pattern of complementation is essentially symmetric about the diagonal, indicating that the two halves of the dimer are equivalent with respect to their RNA binding functions.

An alternative interpretation of our results is that the introduction of certain pairs of amino acid substitutions into the fused dimer results in protein folding or stability defects. In this view the complementation pattern reflects the correction of defects in protein structure, not in interaction with RNA. We cannot categorically exclude this possibility, but suggest it is a highly unlikely one. Each of the repressor-defective mutants was able to fold properly in the conventional homodimer ( 3 ). Moreover, work currently in progress shows that subunit fusion substantially stabilizes the protein against denaturation by urea and reverses the destabilizing effects of certain amino acid substitutions and peptide insertions. For example, we have recently shown by site-directed mutagenesis that T45 is exceptionally sensitive to substitution. Even conservative replacements at this site result in failure to produce the properly folded protein, as judged by the inability to form capsids. In the fused dimer, however, the residue 45 substitutions do not prevent folding (unpublished observations). Thus, even if a given pair of substitutions were destabilizing in the conventional, non-covalent heterodimer, subunit fusion itself would be expected to compensate for the defect.

We suggest that half-site B resides on one subunit and is made up of residues 47, 49, 55, 57, 59, 61 and 91. Half-site A is comprised of residues 61, 85, 87 and 89 and resides on the other subunit of the dimer. Figure 10 B schematically illustrates the composition of one full binding site based on these genetic considerations alone. It is a subset of the sites shown in Figure 10 A and is related to an identical second site by two-fold symmetry. Note that T45, substitutions of which were absent from our original collection of repressor-defective mutants, is included in this illustration of the binding site. Recently we introduced several substitutions at this site and find that they are not complemented when paired with the wild-type sequence in heterodimers, indicating that residue 45 is a component of both half-sites (unpublished observations).


Figure 10 . Schematic views of the RNA binding site of the coat protein dimer as viewed from the underside of the molecule shown in Figure 1. The two subunits of the dimer (and their associated half-sites) are labeled A and B in brackets. ( A ) The filled circles indicate the positions of amino acids whose substitution may result in an RNA binding defect (3). This represents a composite picture of two equivalent binding sites. ( B ) Amino acid residues (filled circles) making up a single RNA binding site identified by the heterodimer complementation studies described in the text. ( C ) A model of the RNA binding site derived from inspection of the X-ray structure of the protein-RNA complex. Here the filled circles represent amino acid residues that appear to contact RNA.

The publication by Valegard et al. ( 6 ) of the crystal structure of a complex of MS2 coat protein bound in capsids to operator RNA allows us to compare the RNA binding site defined by our genetic studies (described above and summarized in Fig. 10 B) with that derived from the X-ray structure (summarized in Fig. 10 C). The reader is referred to their paper for the details of the proposed nucleotide-amino acid contacts. First we consider half-site A. In spite of the tentative nature of our binding site model, there are strong similarities with the results of X-ray analysis. In the structure of the protein-RNA complex six amino acid residues form contacts with RNA in half-site A. Three of these (residues 61, 85 and 87) were predicted by our genetic analyses. We also note that our work with specificity mutants specifically predicted an interaction of Asn87 with the pyrimidine at position -5 in the translational operator ( 16 ), a contact which is confirmed by the structure. We did not find mutants with substitutions at residues 29 or 45 in our original collection of repressor-defective mutants. We have discussed above the reason for the absence of residue 45 mutants. Being defective for capsid formation they failed the screens employed in isolation of the repressor-defective mutants. Regarding residue 29, we previously reported the isolation of a mutant (V29I) that confers the ability to bind operator RNA substantially more tightly than wild-type ( 14 , 17 ). The effects of this substitution suggest a possible role for V29, but do not indicate whether it is a member of one or both half-sites. Another potential discrepancy involves E89, which our analysis assigns to half-site A, but which does not contact RNA in the X-ray structure of the complex. Although the mutant E89K is repressor-defective, our recent codon-directed mutagenesis experiments show that this position will tolerate a wide range of other substitutions without affecting repressor activity, suggesting that E89 is not an important RNA contact in the wild-type complex (M.Spingola and D.Peabody, manuscript in preparation). The only remaining obvious discrepancy in half-site A is the prediction from the structure that residue 47 is a constitutent of both half-sites. Some heterodimers are good repressors when they contain a functional residue 47 in only one half of the dimer.

The genetic data are also in substantial agreement with the structural assignment of residues to half-site B. Our studies demonstrate the importance of residues 45, 47, 49, 57, 59 and 61 and suggest a potential role for residue 29. In the complex, residue 51 is observed to contact RNA. However, we have not yet found a mutation affecting this amino acid and cannot comment on its importance. On the other hand, we find that substitutions of residue 55 can result in either the repressor-defective (N55D) or the the super-repressor (N55K) phenotypes ( 3 , 16 ), suggesting a role for this residue that is not apparent from inspection of the structure. We also assigned residue 91 to half-site B. Since no RNA contact with this residue is obvious from the structure, we suggest that substitutions here probably inactivate half-site B by steric interference with RNA binding and not by the disruption of a favorable contact.

It is important to recognize that genetic and structural methods represent complementary technical approaches to mapping a binding site and that each may provide information that is not necessarily available to the other. It is not surprising to discover that, in spite of general agreement, some aspects of the genetically defined binding site are different from those defined by inspection of the structure of the RNA-protein complex. In particular we point out the following.

(i) The binding site defined by the repressor-defective mutations apparently includes some amino acid residues that are not sites of protein-RNA contact, but whose substitution may interfere with RNA binding by steric occlusion or by altering the conformations of true contacting residues.

(ii) On the other hand, our genetic approach will have failed to identify any constituent of the binding site also playing a role in protein structure, since mutants affected at such sites would not survive the screen for capsid formation we employed in the isolation of repressor-defective mutants ( 3 ). T45 is apparently in this category.

(iii) The solution conditions of in vitro experiments, including those used in crystallization, sometimes fail to mimic faithfully those found in vivo . Some aspects of the RNA-protein interaction may differ in the two environments. We previously reported an example of this phenomenon with the coat protein of the RNA phage GA, which shows marked differences in its RNA binding specificity when the results of in vivo and in vitro experiments are compared ( 16 ).

(iv) We wonder whether the RNA binding process involves intermediate states requiring residues not implicated in the final RNA-protein interaction.

(v) The actual contributions of individual residues to the free energy of binding may not be obvious from simple inspection of the X-ray structure. Functional studies are required to assess and confirm the roles of individual amino acid residues in RNA binding.

ACKNOWLEDGEMENT

This work was supported by a grant from the National Institutes of Health.

REFERENCES

1 Bernardi,A. and Spahr,P.F. (1972) Proc. Natl. Acad. Sci. USA, 69, 3033-3037. MEDLINE Abstract

2 Valegard,K., Liljas,L., Fridborg,K. and Unge,T. (1990) Nature, 345, 36-41. MEDLINE Abstract

3 Peabody,D.S. (1993) EMBO J., 12, 595-600.

4 Ni,C.-Z., Syed,R., Kodandapani,R., Wickersham,J., Peabody,D.S. and Ely,K.R. (1995) Structure, 3, 255-263.

5 Beckett,D. and Uhlenbeck,O.C. (1988) J. Mol. Biol., 204, 927-938. MEDLINE Abstract

6 Valegard,K., Murray,J.B., Stockley,P.G., Stonehouse,N.J. and Liljas,L. (1994) Nature, 371, 623-626. MEDLINE Abstract

7 Laemmli,U.K. (1970) Nature, 227, 680-685. MEDLINE Abstract

8 Burnette,W.N. (1981) Anal. Biochem., 112, 195-203. MEDLINE Abstract

9 Peabody,D.S. (1990) J. Biol. Chem., 265, 5684-5689. MEDLINE Abstract

10 Carey,J., Lowary,P.T. and Uhlenbeck,O.C. (1983) Biochemistry, 22, 4723-4730. MEDLINE Abstract

11 Draper,D.E., White,S.A. and Kean,J.M. (1988) Methods Enzymol., 164, 221-237. MEDLINE Abstract

12 Kunkel,T.A., Roberts,J.D. and Zakour,R.A. (1987) Methods Enzymol., 154, 367-3381. MEDLINE Abstract

13 Reddy,P., Peterkofsky,A. and McKenney,K. (1985) Proc. Natl. Acad. Sci. USA, 82, 5656-5660. MEDLINE Abstract

14 Peabody,D.S. and Ely,K.R. (1992) Nucleic Acids Res., 20, 1649-1655.

15 Golmohammadi,R., Valegard,K., Fridborg,K. and Liljas,L. (1993) J. Mol. Biol., 234, 620-639. MEDLINE Abstract

16 Lim,F., Spingola,M. and Peabody,D.S. (1994) J. Biol. Chem., 269, 9006-9010. MEDLINE Abstract

17 Lim,F. and Peabody,D.S. (1994) Nucleic Acids Res., 22, 3748-3752. MEDLINE Abstract

Return

* To whom correspondence should be addressed
Add to CiteULike CiteULike   Add to Connotea Connotea   Add to Del.icio.us Del.icio.us    What's this?


This article has been cited by other articles:


Home page
 RNAHome page
B. HOOK, D. BERNSTEIN, B. ZHANG, and M. WICKENS
RNA-protein interactions in the yeast three-hybrid system: Affinity, sensitivity, and enhanced library screening
RNA, February 1, 2005; 11(2): 227 - 233.
[Abstract] [Full Text] [PDF]

Home page
 Proc. Natl. Acad. Sci. USAHome page
I. Golding and E. C. Cox
RNA dynamics in live Escherichia coli cells
PNAS, August 3, 2004; 101(31): 11310 - 11315.
[Abstract] [Full Text] [PDF]

Home page
 Appl. Environ. Microbiol.Home page
S. Caballero, F. X. Abad, F. Loisy, F. S. Le Guyader, J. Cohen, R. M. Pinto, and A. Bosch
Rotavirus Virus-Like Particles as Surrogates in Environmental Persistence and Inactivation Studies
Appl. Envir. Microbiol., July 1, 2004; 70(7): 3904 - 3909.
[Abstract] [Full Text] [PDF]

Home page
 J. Biol. Chem.Home page
A. A. Leonov, P. V. Sergiev, A. A. Bogdanov, R. Brimacombe, and O. A. Dontsova
Affinity Purification of Ribosomes with a Lethal G2655C Mutation in 23 S rRNA That Affects the Translocation
J. Biol. Chem., July 3, 2003; 278(28): 25664 - 25670.
[Abstract] [Full Text] [PDF]

Home page
 J. Biol. Chem.Home page
D. S. Peabody and A. Chakerian
Asymmetric Contributions to RNA Binding by the Thr45 Residues of the MS2 Coat Protein Dimer
J. Biol. Chem., September 3, 1999; 274(36): 25403 - 25410.
[Abstract] [Full Text] [PDF]

Home page
 J. Virol.Home page
Y. Zhang, H. Qian, Z. Love, and E. Barklis
Analysis of the Assembly Function of the Human Immunodeficiency Virus Type 1 Gag Protein Nucleocapsid Domain
J. Virol., March 1, 1998; 72(3): 1782 - 1789.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Print PDF (121K) Freely available
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrow Search for citing articles in:
ISI Web of Science (23)
Right arrowRequest Permissions
Right arrow Commercial Re-use Guidelines
for Open Access NAR Content
Google Scholar
Right arrow Articles by Peabody, D.
Right arrow Articles by Lim, F
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Peabody, D.
Right arrow Articles by Lim, F
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?