The Structure of Ribosomal RNA and Its Organization Relative to Ribosomal Protein

The Structure of Ribosomal RNA and Its Organization Relative to Ribosomal Protein RICHARDBRIMACOMBE, PETERMALY,AND CHRISTIAN ZWIEB* Max-Planck-Institut f u r Molekulare Cenetik, Abteilung Wittmann, Berlin-Dahlem, Federal Republic of Germany I. Ribosomal RNA Sequences.. ................................. 11. Secpndary Structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. Derivation of Structures of Escherichia coli rRNA.. . . . . . . . . . . . B. Structures Proposed for rRNA from Other Sources.. . . . . . . . . . . . C. Evidence for Alternative Conformations (“Switches”). . . . . . . . . . 111. RNA-Protein Interactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. Protein Binding Sites on rRNA. . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. RNA-Protein Cross-Linking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . IV. Three-Dimensional Packing of E . coli rRNA. .................... A. Intra-RNA Cross-Linking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. Electron Microscopy of rRNA within Ribosomal Subunits . . . . . . V. Outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

2 5

8 12 21 24 24 26

37 37 39 41 43

The structure of the Escherichia coli ribosome was last reviewed by our laboratory in this series in 1976 (1).At that time, the emphasis was on the structure and function of the individual ribosomal proteins, and, although a considerable amount of information was already available concerning the ribosomal RNA, progress in this area was hampered by the lack of complete base sequence data for the ribosomal RNA molecules. Since then, as a direct result of the explosion in DNA sequencing technology ( 2 , 3 ) ,the situation has changed dramatically. Complete nucleotide sequences have now been established, corresponding not only to the ribosomal RNA molecules from E . coli, but also to those from a number of different organisms, covering a large portion of the evolutionary spectrum. * Present address: Department of Physiological Chemistry, Brown University, Providence, Rhode Island. 1 Progress in Nucleic Acid Research and Molecular Biology, Vol. 28

Copyright 8 1983 by Academic Press, Inc. All rights of reproduction in any form reserved. ISBN 0-12-540028-4

2

RICHARD BRIMACOMBE

et al.

We begin by reviewing briefly the current status of these sequences. Next, we describe how the sequence information has been used to derive convincing secondary structure models for the RNA from both subunits of the E . coli ribosome, and we compare the various models that have been proposed. We show how extrapolation of these data to ribosomal RNA molecules of widely differing size classes leads to the clear conclusion that the secondary structures, as well as significant regions in the primary sequences, have been conserved to a large extent throughout evolution. Section IV deals with the threedimensional organization of the ribosomal RNA and its arrangement with respect to the ribosomal proteins, concentrating once again on the E . coli ribosome. In particular, we include a review of the application of cross-linking techniques (bath RNA to protein and intra-RNA) to this problem. In general, rather than presenting an exhaustive survey of the literature, we have selected topics or examples to illustrate those problems or points of interest that we consider to be most relevant to the central objective in this field of research, namely, the elucidation of the three-dimensional structure of the ribosomal RNA in situ in the ribosome. I. Ribosomal RNA Sequences The organization and transcription of ribosomal RNA genes is a complex and fascinating subject, itself worthy of review. However, we confine ourselves here to a discussion of the mature ribosomal RNA (rRNA) species, as they occur in the completed ribosomal particles. The “standard” bacterial ribosome, as typified by that of E . coli, contains in its small subunit a single RNA molecule (16 S), which is about 1540 nucleotides in length ( 4 , 5 ) . The large subunit contains a 23 S RNA (ca. 2900 nucleotides) (6) and a 5 S RNA (120 nucleotides) (7). In other organisms, the size of these rRNA molecules varies considerably. The smallest so far reported are those from trypanosome mitochondria, which are only 640 and 1230 nucleotides in length, respectively, from the small and large subunits (8). Next come the mitochondria1 ribosomes from mammals, with RNA molecules of 12 S and 16 S (ca. 950 and 1550 nucleotides) (e.g., 9, lo), and these very small ribosomes contain no 5 S RNA. Other types of mitochondria (e.g., 1 1 ) , and also chloroplasts (e.g., 12, 13), have rRNA molecules corresponding in size to those of the bacterial ribosomes. The chloroplasts and mitochondria from higher plants both contain 5 S rRNA, and in addition the large subunit of higher plant chloroplast ribosomes contains a 4.5 S RNA species (14).

STRUCTURE OF RIBOSOMAL RNA

3

The largest ribosomes are those from the cytoplasm of eukaryotes, with RNA molecules of 18 S (ca. 1800 nucleotides) (e.g., 15, 16) and 26-28 S (up to 4000 nucleotides) (e.g., 17) in the small and large subunits, respectively. The large subunit contains a 5.8 S RNA molecule (e.g., 18) as well as 5 S RNA, and a 2 S RNA species is also observed in the large ribosomal subunit from Drosophila (19). Further, the large subunit rRNA genes for many eukaryotic ribosomes contain introns (e.g., 20, 21), and in Drosophila the final “28 S” transcription product appears in two distinct halves (reviewed in 22). As shown in Section II,B the small rRNA molecules (4.5 S , 5.8 S, and 2 S) all have clear counterparts within the 23 S rRNA from E . coli, and they can therefore better be regarded as products of posttranscriptional processing of the rRNA rather than as “extra” rRNA species. Many complete or partial sequences are now known for all these classes of rRNA, mostly obtained by determination of the corresponding rDNA sequences. A compilation of known 5 S and 5.8 S sequences has been made (23, 24),’ and this list continues to grow at a rate of about one sequence every two or three weeks. [One of the more interesting new sequences here is the “5 S” rRNA from Halococcus morrhuae, which contains a 108-base insertion (%).I The sequence of 2 S rRNA from Drosophila melanogaster is known (26),and also the sequences of 4.5 S rRNA from wheat (27),tobacco (28),and maize (29). More important for the purpose of this article are the sequences of the major rRNA molecules, and those sequences currently complete or nearly complete are listed in Table I (4-6,9-13,15-17,30-40). It can be seen from Table I that almost every common size class of large rRNA mentioned above is represented by sequences from two or more species. In addition, many partial sequences of the major rRNA molecules have been determined. In the case of the small subunit, sequences are available from a total of about 20 species for the 50-200 nucleotides at the 3’ terminus, and compilations of these have been published (41,42).In the case of the large-subunit rRNA, some partial sequences are available for the regions flanking introns in the rDNA of Chlamydomonas reinhardii chloroplast (43) and yeast mitochondrial ribosomes (20, 44), as well as of Drosophila virilis (45),D . melanogaster (46,47),Tetrahymena pigmentosa (21),and Physarum poZycephalum (48,49)cytoplasmic ribosomes. The positions in E . coli 23 S rRNA that correspond to the locations of these inserted sequences have been collated (36).Similar short sequences are known from the 5’ ends of the large-subunit rRNA of Aspergillus nidulans mitochonSee also Singhal and Shaw in this volume. [Ed.]

4

RICHARD BRIMACOMBE

et al.

TABLE I SEQUENCES OF RIBOSOMAL DNA (OR rRNA) MOLECULES" Organism

Small subunit rRNA Large subunit rRNA

Human mitochondrion Mouse mitochondrion Rat mitochondrion

12 s (9) 12 s (10)

Saccharomyces cerevisiae mitochondrion Aspergillus nidulans mitochondrion Paramecium primaurelia mitochondrion

15 S (11) 15 S (31)

Escherichia coli Proteus vulgaris Bacillus brevis Bacillus stearothernophilus Zea mays chloroplast Euglena gracilis chloroplast Saccharomyces cerevisiae cytoplasm Saccharomyces carlsbergensis cytoplasm Xenopus laevis cytoplasm

-

16 S (4, 5) 16 S (34) (16 S) (35)

-

16 S (12) 16 S (37) 18 S (15) 18 S (16)

16 S (9) 16 S (10) 16 S (30)

-

20 S (32) 23 S (6, 33)

-

23 S (36) 23 S (13)

-

26 S (17) 26 S (38) (28 S) (39,40)

0 Known sequences are listed, showing the size (S value in Svedberg units) of the corresponding rRNA molecule; parentheses denote that the sequence concerned is not yet fully complete; a dash indicates that the sequence is not available. All sequences were determined from ribosomal DNA, with the exception of E. coli 16 S rRNA (5) and 23 S rRNA (33), and P . vulgaris 16 S rRNA (34);these determinations were made directly from RNA. Numbers in parentheses indicate reference numbers.

dria (50)and Euglena gracilis chloroplasts (51, 52) and from the 3' ends of Aeromonas punctata and Proteus vulgaris (53).Last, a large amount of sequence data is contained in oligonucleotide catalogs made earlier from the rRNA of many species [summarized by Fox et al. (54)l. It has long been known that the sequence of 5 S rRNA is highly conserved (see, e.g., 55),and a corresponding pattern of conservation is seen among the major rRNA molecules. For example, P . vulgaris 16 S rRNA shows 93% homology to the 16 S rRNA from E . coli (34), whereas the homology between the 23 S rRNA molecules from B . stearothermophilus and E . coli is about 75% (36).Escherichia coli rRNA and Z. mays chloroplast rRNA also show about 75% sequence homology, in both subunits (12, 13). Xenopus 18 S and yeast 18 S rRNA are again about 75% homologous to one another (16).The two Saccharomyces 26 S rRNA sequences [ S . cerevisiae (17)and S . carlsbergensis ( I S ) ] are virtually identical, differing only in about 20 positions, and this order of difference is also observed between different

STRUCTURE OF RIBOSOMAL

RNA

5

ribosomal cistrons of the same organism; E . coli 16 S rRNA, for instance, contains 16 single-base cistron heterogeneities (56). When sequences from the different size classes of rRNA (Table I) are compared, it becomes apparent that a number of regions have been conserved in both the large- and small-subunit rRNA molecules of all species, despite the large differences in lengths of the polynucleotide chains. This type of homology becomes even more striking when the secondary structures of the various molecules are considered; it is discussed in more detail in Section I1,B. Since most of the sequence data (Table I) was determined from rDNA, information concerning modified nucleotides in the corresponding rRNA molecules is incomplete in many cases. Escherichia coli 16 S rRNA contains 9 methylated bases (56),and the 23 S rRNA contains 10 methylated bases and 3 pseudouridine residues (33).In P . vulgaris 16 S rRNA, 6 methylated bases are in homologous positions to their E . coli counterparts (34), and the two N6-dimethyladenine residues near the 3’ terminus of the 16 S rRNA seem to be a universal feature of all small-subunit RNA molecules (reviewed in 42). Some methylated bases have been localized in rodent mitochondria1 rRNA, and these also appear to be highly conserved (57).The pattern of nucleotide modification in eukaryotic cytoplasmic ribosomes is rather more complex. Xenopus 18 S rRNA contains 40 methylated bases, of which the majority are 2’-O-methyl groups (16).There is one hypermodified base as well as several base-methylated residues, and in addition there are 40 pseudouridine residues (58),which for the most part have not as yet been localized precisely. The corresponding yeast 18 S rRNA molecule has the same number of base methylations, but fewer 2‘-O-methyl groups (59). These have not yet been localized, whereas in yeast 26 S rRNA, 30 out of the 43 methyl groups have been placed (38).There are also some data on the corresponding sites of methylation in Xenopus 28 S rRNA (60). To conclude this section, it is clear that an enormous amount of sequence information has already been collected, and this provides the raw material for the next stage in the elucidation of the threedimensional organization of ribosomal RNA, namely the construction of models for the secondary structure.

II. Secondary Structures There are, broadly speaking, three different approaches by which a nucleotide sequence can be folded into a double-helical secondary structure: the theoretical, the experimental, and the comparative.

6

RICHARD BRIMACOMBE

et al.

The theoretical approach relies on the use of computer programs to select the thermodynamically most favorable structure for the sequence, using thermodynamic parameters derived from the melting properties of model oligonucleotides or RNA fragments (e.g., 61-64). A large number of computer algorithms generated for this purpose have been reported (e.g., 65-69), and the newer versions (e.g., 68) have been considerably improved, in that they can select structures for an entire long sequence rather than for just short sections. However, the approach suffers from the obvious disadvantage that the computer program can be only as good as the thermodynamic data put into it, and these data are by no means comprehensive. The effects of imperfections such as “bulges” and “loops” in the helices cannot be computed very accurately, and-more important-the thermodynamic effects of tertiary structure or interactions with protein can hardly be assessed at all. As a result, this approach can lead to erroneous structure predictions, and it has, for example, been shown (70) that the computer may predict quite different structures for two phylogenetically closely related sequences. Nevertheless, when combined with experimental or comparative data, the computer approach is a powerful tool for screening potential secondary structures. The experimental approach to secondary structure determination has made use of a number of different methods; these include chemical modification (e.g., 71-76), analysis of enzyme cutting points (e.g., 56, 77), oligonucleotide binding (e.g., 78, 79), isolation of base-paired RNA fragments (80,81), and intra-RNA cross-linking (e.g., 82, 83). In the case of chemical modification, the RNA is treated with base-specific reagents to test the accessibility (or single-strandedness) of individual residues. The reagents that have been used include kethoxal ( 71)and glyoxal ( 72) (G-specific);monoperphthalic acid ( 73),m-chloroperoxybenzoic acid (72), and diethyl pyrocarbonate (74) (A-specific); and methoxyamine (75) and bisulfite (72)(C-specific). Dimethyl sulfate (74)(C- and G-specific) and soluble carbodiimides (76)(U- and G-specific) have also been applied. The analysis of enzyme cutting points is, in principle, a very similar method, in which the RNA is subjected to a mild digestion by a single-strand-specific nuclease (such as nuclease S1, or ribonucleases A and Tl), and the resulting fragments are analyzed to pinpoint those residues at which the polynucleotide chain has been cut (56).The converse approach, using the double-strand-specific nuclease from cobra venom, has also been applied with success (77).Oligonucleotide binding is a probe for singlestranded regions, in which the putatively exposed sequences can be tested for their ability to bind a short complementary oligonucleotide

STRUCTURE OF RIBOSOMAL

RNA

7

(78,79).Data from all these methods become progressively more difficult to interpret as the sequence under consideration becomes longer, a disadvantage that is not shared by the “base-paired-fragment” approach (80, 81). Here, the RNA is subjected to a mild digestion with nuclease, and the fragments are isolated with base-pairing intact by gel electrophoresis under nondenaturing conditions. This is followed by a second electrophoretic dimension in denaturing conditions, and analysis of the products gives direct information as to which regions of the sequence are base-paired with one another. The last method mentioned above, namely intra-RNA cross-linking (82, 83), also gives direct data on neighborhoods between various parts of the RNA chain; this is discussed in more detail in Section IV,A. The third general approach to secondary structure determination, namely the comparative approach, is without doubt the most powerful and is in essence very simple (84).If similar sequences from different organisms are compared, then a base change in one strand of a putative helical region must be compensated by a complementary base change in the other strand. Thus an A * U pair in one species could become a G C pair at the corresponding positions in the second species, and so on. If the base changes do not compensate each other in this way, then the implication is that the proposed element of secondary structure is incorrect. The approach can also be inverted since, once a secondary structure has been established, similar structures can be sought in more distantly related species by screening for stretches of identical base sequence rather than for differences between the two species concerned. It is clear from the preceding section that a Iarge number of suitable rRNA sequences are available for the application of this approach, and its great advantage is that it generates a “positive feedback” situation in which each new sequence studied not only gives support to the concept that secondary structure in rRNA has been highly conserved among molecules of the same size class, but also serves to confirm and extend the original secondary structure model. In practice, the models derived for the secondary structure of the E . coli rRNA molecules all make use of combinations of several of the methods just outlined, and these models are described in the next section. It should be noted that in general the percentage of residues proposed to be involved in base-pairing is rather less than that predicted by physicochemical studies [e.g., hyperchromicity measurements (85, 8611. This is quite reasonable, since the models on the whole describe minimum secondary structures and take no account, for instance, of tertiary interactions or complex stacking of helical

8

RICHARD BRIMACOMBE

et al.

elements, both of which would contribute to the hypochromicity of the overall structure. A. Derivation of Structures of Escherichiu coli rRNA 1. 5 S RNA Among the large number of secondary-structure models proposed over the years for 5 S rRNA (see, e.g., 55), that of Fox and Woese (84) soon became established as the most reasonable “minimum” structure for prokaryotic 5 S rRNA, the model being based on comparative evidence (see above)’. This model for E . coli 5 S rRNA is illustrated schematically in Fig. 1. More recently, extensions to the base-pairing scheme have been proposed, and two examples of these are also included in Fig. 1. The model of Studnicka et al. (87) was derived by a “filtering” method that combines the theoretical and comparative approaches; minimum energy structures were accepted only if they could be drawn for all the prokaryotic 5 S sequences under consideration. This model also predicts a tertiary interaction between residues 24 and 39. The model of Pieler and Erdmann (88),on the other hand, is based on a combination of comparative evidence and S l-nuclease cleavage data from E . coli 5 S rRNA; it predicts a “tertiary” basepairing between residues 41-44 and 74-77, as well as additional base-pairs in the stem region of the molecule. A similar tertiary base40

,110

1

1 12

40

FIG.1. Three models for the secondary structure of Escherichia coli 5 S rRNA. The sequence is numbered from the 5’ end, and the bars denote base-pairing. (A) The model of Fox and Woese (84), (B) that of Studnicka et al. (87), and (C) that of Pieler and Erdmann (88), indicating the additional tertiary base-pairing proposed between residues 41-44 and 74-77.

STRUCTURE OF RIBOSOMAL

RNA

9

pairing has also been proposed by Hancock and Wagner (89) between residues 37-40 and 73-76, on the basis of an intra-RNA cross-link generated between E . coli residues 41 and 72. It should be noted that short tertiary base-pairings of this nature can be accommodated in the structure, without generating “knots” in the RNA chain; clearly, however, a longer base-pairing of this type (extending over more than half a helix turn) does generate a potential “knot” situation (cf. Sections II,C and IV,A). A model for two eukaryotic 5 S rRNA species, proposed by Troutt et al. (70),is essentially very similar to that of Studnicka et al. (87),and the latest model from Woese and his group (90)offers a general structure for all 5 S rRNA molecules, both prokaryotic and eukaryotic. In this model, additional base-pairing is proposed on the basis of a comparison with 5 S rRNA from the archaebacterium Sulfolobus acidocaldarius, in which the whole region between bases 69 and 111 forms a continuously base-paired hairpin loop. In order to generate such a helix in the case of E . coli 5 S rRNA, a number of mismatched basepairs must be postulated, including three A * G pairs. While it has been shown that A G pairs are common features at the ends of helices in tRNA (91),there is as yet no evidence that they can exist within a continuing helix, and as the authors themselves point out (go), such a helix would certainly be very irregular and readily denatured. The important inference to be drawn here is that it may be possible to replace several Watson-Crick base-pairs by mismatched or nonpaired bases, without altering the overall three-dimensional appearance of the molecule.

2. 16 S AND 23 S rRNA Several research groups have been independently involved in the construction of secondary-structure models for the major rRNA molecules, each using different sets of experimental data derived from E . coli rRNA, and each using sequence comparisons to refine the models. Noller et al. in California, in collaboration with Woese et al. in Illinois (72, 35, 36) have exploited primarily the analysis of sites of chemical modification by kethoxal, glyoxal,bisulfite, and m-chloroperoxybenzoic acid in order to obtain their experimental data. The sequences used for comparative analysis were those of B . breuis 16 S rRNA (35)and B . stearothermophilus 23 S rRNA (36) (see Table I), as well as the extensive oligonucleotide catalogs of Woese et al. (54).The groups in Strasbourg (33,34,92) have made use of single- and double-strand-specific nucleases to generate specific cuts in the RNA, and have used the sequences of P . uulgaris 16 S rRNA (34) and 2. mays chloroplast 16 S

10

RICHARD BFUMACOMBE

et al.

and 23 S rRNA (12,13) for comparison. Our approach has been based and intra-RNA crosson the analysis of base-paired fragments (80,81) links (83) (see also Section IV,A). These methods proved to be very valuable in establishing the “long-range interactions” in the rRNA molecules (Le,,base-pairings between segments of RNA that are remote in the primary sequences), particularly in the case of 23 S rRNA (93), and the models derived from these data (81, 93) were subsequently refined in collaboration with Kossel et al. in Freiburg (94-96) by comparative analysis using the Z. mays chloroplast 16 S and 23 S rRNA sequences, The degree of homology between these sequences and those of E . coli (ca. 75%; see Section I) makes them ideal for comparative analysis, since the two sets of sequences are thus similar enough to be compared directly with one another in detail, and at the same time the degree of divergence is sufficient to allow a large number of “compensating base changes” to be found in double-helical regions of the structures. A comparison of the structure derived for E . coli 23 S rRNA with the corresponding structure for 2. mays chloroplast rRNA shows, for example, a total of over 450 such compensating base changes (95). The latest versions of these models for 16 S rRNA from the three research groups can be found in references 35, 92, 96, and the first complete versions for 23 S rRNA in references 33, 36, 95. It is very difficult, however, to compare these structures with one another, since they have been derived independently and are presented in different formats (particularly the 23 S rRNA structures). For this reason, in Figs. 2 and 3 we have transposed the “California” and Strasbourg” models into our format, in order to allow easy comparison of all the versions for each sequence.2 In those cases where two alternative conformations have been suggested for particular regions of the RNA (35,92),we have chosen the alternative that gives the best agreement among all three models. In the case of the 16 S rRNA (Fig. 2), it can be seen that the molecule is organized into “domains” clearly defined by the longrange interactions already mentioned. These long-range interactions are common to all three models, at least as alternatives, although it should be noted that one of them (between bases 564-570 and 880-886 in the central region) is based only on an interaction obFor the sake of simplicity, the models in Figs. 2 and 3 are referred to as “Berlin,” “California,” and “Strasbourg” models, respectively. In fact, as mentioned above, the “California” model for both 16 S and 23 S rRNA was a collaborative venture between the Californiaand Illinois groups, and the Freiburg group was involved in the construction of both the “Berlin” and “Strasbourg” models for 23 S rRNA.

”

BERLIN

‘I

b

P

m-

“

C A L I F 0 R N I A“ b

P

M

“STRASBOURG“

Ly

w

FIG.2. Comparison of secondary structure models for Escherichia coli 16 S rRNA. The sequence is numbered from the 5‘ end (every 50 bases, with a stroke at every tenth base), and is divided into three domains (a to c) as in reference 96, the bars denoting base-pairing. The “California” (35)and “Strasbourg” (92)models have been rearranged to the “Berlin” (96)format, so that identical structural elements from each model have identical orientations, to facilitate comparison.

12

RICHARD BRIMACOMBE

et al.

served between two rather long fragments of E. coti rRNA encompassing the “binding site” for ribosomal protein S4 (97). This particular interaction is obviously crucial in maintaining the topography of the whole central part of the molecule, and, in contrast to the other longrange interactions, is so far not very well supported by comparative evidence. Figure 2 also shows that the differences between the models are largely differences in the detailed base-pairing arrangements within individual helical elements, but there are a few more serious discrepancies. On the whole, however, the degree of agreement is very satisfactory, in view of the fact that the models are derived from entirely independent sets of data. It should be noted that the interaction between bases 17-20 and 915-918 in the “California” model (domains a and b, Fig. 2) is not incompatible with the hairpin loop (bases 9-25, domain a) in the “Berlin” model; rather, it represents a “tertiary” interaction situation comparable to that proposed for 5 S RNA (see Fig. 1C and 88). The 23 S rRNA (Fig. 3) is also arranged in clear domains, and the whole molecule appears to be a giant loop, closed by an interaction between its extreme 5‘ and 3’ ends (53).Again, there is good agreement between the models with regard to the principal long-range interactions, with the notable exception that bases 578-584 are paired with 1255-1261 in the “California” model (36),as opposed to bases 805-811 in the other two models (33, 95). There are some further significant discrepancies in various parts of the models, in particular the regions between bases 450-510, 1340-1380, 1415-1585, 1900-1975, 2500-2585, and 2765-2785. However, as with the 16 S rRNA, the majority of the remaining differences are again local details of the base-pairing arrangements within well-defined domains; although it will obviously take some time before all these differences are ironed out, it is nonetheless clear that in both 16 S and 23 S rRNA the broad basis of the secondary structures can be regarded as firmly established.

B. Structures Proposed for rRNA from Other Sources The structures just described for E. coli 16 S and 23 S rRNA are, by

virtue of the fact that they are based to a large extent on sequence comparisons, simultaneously general models for ribosomal RNA molecules of this size class. The slightly smaller 15 S rRNA sequences from S. cerevisiae mitochondria ( 1 1 ) and A. nidulans mitochondria (31)fit precisely into the same structure as the 16 S molecules (98,31), and this raises the question as to how far the sequence comparison approach can be extended to other ribosomal RNA molecules, outside

STRUCTURE OF RIBOSOMAL

RNA

13

the 16 S and 23 S size class. For this purpose, both the smaller mitochondrial rRNA and the larger eukaryotic cytoplasmic rRNA species (Table I) can be used for comparative analysis. In both cases the overall degree of primary sequence homology with respect to E . coli is at first sight not very high, but nevertheless there are significant regions of conserved sequence, and secondary structures can be built up around such conserved regions, using the E . coli secondary structure models as a basis. The results of these comparisons are very striking indeed. The small mitochondrial 12 S and 16 S rRNA species, although they are only about half the size of the corresponding E . coli molecules, can be arranged in secondary structures for which many elements are precisely equivalent to their 16 S and 23 S counterparts (e.g., 95,96). The reduction in size is achieved by simple amputation of secondary structural loops or by erosion of whole domains. If the E . coli structures are compared with the larger eukaryotic molecules, a precisely analogous situation is observed (e.g., 96), except that here there are extra sequences instead of missing ones. Furthermore, if the sequences within these secondary structure models are compared, it becomes immediately apparent that the degree of sequence homology between the various species is in fact a good deal higher than was at first supposed. Many short stretches of four- to five-base homology, which would not be regarded as significant in a simple comparison of the primary sequences, appear at identical positions in the various secondary structures, and seem to constitute a “core” of conservation running right through the molecules (95,96).The longer stretches of homology tend to be in single-stranded regions (e.g., 72), but there are also significant sequence homologies in double-helical regions. Interestingly, in the case ofA. nidulans 15 S rRNA there appears even to be more conservation of sequence in the double-stranded regions than in the single-stranded areas (31).It should also be noted in this context that the very highly conserved areas of sequence (e.g., bases 1390-1408 and 1492-1502 in the E . coli 16 S rRNA; see reference 42 for review) cannot be placed in secondary structures by the sequence comparison approach, simply because no compensating base changes are to be found in such regions, and this may have led to distortion of the relative importance of sequence Conservation in single-stranded as opposed to double-stranded regions. Some examples of structures deduced for rRNA molecules from different size classes are illustrated in Figs. 4 and 5. Figure 4 shows the models that we have proposed (96) for mitochondrial 12 S rRNA and eukaryotic cytoplasmic 18 S rRNA, compared with the E . coli 16 S

“BERLIN”

’.. ”

.-

CALIFORNIA ”

a

*

w

‘.

”

STRAS B OUR G

‘I

b

FIG.3. Comparison of‘secondary structure models for Escherichia coli 23 S rRNA. The structure is divided into six domains (a to f) as in reference 95, and the “Berlin” (95), “California” (36),and “Strasbourg” (33) models are compared as in Fig. 2.

STRUCTURE OF RIBOSOMAL

RNA

15

"BERLIN" f

"CALIFORNIA" f

"

ST RA S B 0 UR G " f

e

FIG.3 (continued).

IHUMAN MITOCHONDRION 125 rRNA

E.COLI 165 rRNA

5. CEREVISIA E I 0 5 r RNA

b

LIP

FIG.4. Secondary structure comparisons between human mitochondria1 12 S rRNA, Escherichia coli 16 S rRNA and Saccharomyces cereoisiae 18 S rRNA. The diagrams are sketches from the structures proposed in reference 96, arranged as in Fig. 2; corresponding secondary structural elements in the three molecules lie in the same orientation.

16

HUMAN MITOCHONDRION 165 r R N A

a

b

d

f

U E.COLI 23s r R N A

FIG.5. Secondary structure comparison between Escherichia coli 23 S rRNA and human mitochondria1 16 S rRNA. The diagrams are sketches of the structures proposed in reference 95, arranged as in Fig. 3 (cf. Fig. 4). 17

18

RICHARD BRIMACOMBE

et al.

structure; in Fig. 5, there is a corresponding comparison of mitochondrial 16 S rRNA with E . coli 23 S rRNA (95). These models were derived using the mitochondrial 12 S and 16 S sequences from mouse (10) and man (9) and the 18 S sequences from S. cerevisiae (15) and X. laevis (16). It can be seen from Fig. 4 that the extra sequences in the 18 S rRNA occur largely in the region from bases 640 to 850 (equivalent to 590 to 650 in the E . coli sequence), and that precisely this region is entirely missing in the mitochondrial 12 S rRNA. Other regions of the 18 S structure (e.g., domain c) are virtually superimposable on the corresponding E . coli structure, whereas several parts of the mitochondrial 12 S structure in domains a and c have been seriously eroded. Interestingly, in the human (but not the mouse) mitochondrial rRNA, there are a number of instances where a missing hairpin loop is replaced by a stretch of four consecutive C residues (96). Similarly, it can be seen in Fig. 5 that in domains b, d, and e of the large subunit rRNA there is strong conservation of the secondary structure, with a certain number of loops being cleanly amputated in the mitochondrial rRNA. In domains a, c, and f, on the other hand, the mitochondrial domains are considerably reduced in size. [The original publications (95, 96) should be consulted for descriptions of the primary sequence homologies in the various structures.] In a model proposed for yeast 26 S rRNA (38),an analogous situation to that of the 18 S rRNA is observed, with the “extra” sequences relative to 23 S rRNA occurring in large clusters rather than being uniformly distributed throughout the molecule. Further similar structural derivations have been made by a number of authors, proposing models for yeast 18 S rRNA (99, IOO), general structures for the small subunit rRNA (98, lo]), a model for 12 S rRNA (IOZ), and comparisons of bacterial 23 S and mitochondrial 16 S rRNA (33).While there is in general a very high degree of agreement concerning the most strongly conserved features of the structures, there is, not surprisingly, some divergence in areas where the conservation patterns are not so clear. In some cases this arises from ambiguity in the possibIe alignments of the primary sequences under comparison, and one interesting example of this is shown in Fig. 6, where two apparently significant stretches of primary sequence homology have “leap-frogged” each other in the sequence of mitochondrial 16 S rRNA (cf. 33,95),as compared to that of E . coEi 23 S rRNA. As a result, different secondary structures were derived for this part of the mitochondrial molecule (33,95),depending on which stretch of homology

STRUCTURE OF RIBOSOMAL

f

/

/

-\

\

MOUSE MITOCHONDRIAL

I I

451-U

19

RNA

,

I

,/-165

RNA ( 5 ‘ )

fA-110 G---G A---A A ---A

I I

c---c c u

A---A G---G

---

i

MOUSE MITOCHONDRIAL

u---u 460-A---A c -- -c -120

1 6 5 RNA

(3’)

\

u

G

\ c G A- --A A- - -A

c A

21 0G---G G---G 470 -A---A A---A-1 A---A G---G G A

c

480-A

c

30

u

U

A

A I A---A A---A A---A G---G A 1

u---u

\

B

180

t

\ \---C

u---u c- - - - cu - --u-200 r- - - - -r

G---G A---A A---A A---A G-139

I

I I I

\

c---c

G---G

\ \

l

I

560

I

I

G U A

,

I

I

C C U

I

!

I

200-c---c A---A A---AH190 U G\ C I A

190

I

I

I

I

I

i

A I

I

I

U U U G U A U A A U G

E. C O L I

I

5 70

I I

2 3 s RNA ( 5 ’ )

t E. C O L I 2 3 5 RNA ( 3 ’ )

FIG.6. “Leap-frogging” sequence homologies. Homologous regions of Escherichia coli 23 S rRNA and mouse mitochondrial 16 S rRNA are compared. The sequences are numbered from the 5’ ends, and the thin dashed lines indicate the homologous residues. The homology in region C occurs at similar positions (around residue 200) in the two sequences, but is less strong than the homologies in regions A and B, which occur 3’ relative to region C in the E. coli sequence, but 5’ relative to region C in the mitochondrial sequence. Homology regions A and B were used in deriving the secondary structure model for the mitochondrial rRNA in reference 95, whereas regions A and C were used for the model of reference 33; not surprisingly, the two proposed mitochondria] models are very different in this part of the molecule.

20

RICHARD BRIMACOMBE

et al.

was taken into consideration (see legend to Fig. 6). Another situation of this type is discussed in Section I1,C. A further consequence of the secondary structure comparisons is that the small rRNA species (5.8S, 4.5 S, and 2 S rRNA; see Section I) can now be accounted for with reasonable certainty. It has already been observed (103) that 5.8 S rRNA shows a significant degree of sequence homology with the 5‘-end region of E . coli 23 S rRNA, and 2 S rRNA in Drosophila corresponds to the 3‘ region of other 5.8 S rRNA species (26).If the stem region of the secondary structure proposed for 5.8 S rRNA (104,105) is opened up (cf. 106), then the molecule becomes very similar to the 5’ region of the E . coli 23 S rRNA structure (95,36,107).This is illustrated in Fig. 7, and a comparison of Figs. 7B and 7C also suggests how the 5.8 S rRNA, which is known to be strongly hydrogen-bonded to the large subunit rRNA (e.g., 108, 109), is attached to the latter (cf. 106,110).However, it should be noted that in the secondary structure proposed for yeast 26 S and 5.8 S rRNA (38) there is no hydrogen-bonding between the extreme 5’ end of 5.8 S rRNA and the extreme 3’ end of the 26 S rRNA [cf. the interaction between the extreme 5’- and 3’ ends of E . coli 23 S rRNA (53)in Fig.

-

Y-

,_---. .

FIG.7. The equivalence of eukaryotic 5.8 S rRNA to the 5‘ terminus of Escherichia coli 23 S rRNA. (A) The “burp gun” model for trout 5.8 S rRNA (105).(B) The same model with the stem region opened up, oriented for comparison with C. (C) The 5’ region of E. coli 23 S rRNA (cf. Fig. 3, domain a). The dashed lines alongside the structure indicate sequence homologies between the 5.8 S and 23 S molecules (103,95).

STRUCTURE OF RIBOSOMAL

RNA

21

31. There is therefore a disagreement with the results of Kelly and Cox ( I l l ) ,who observed an interaction between 5.8 S rRNA and fragments isolated from the 3’ end of Neurospora crassa 28 S rRNA ( 1 1 1 ) . This point remains to be resolved. In a similar manner the 4.5 S rRNA from plant chloroplasts corresponds very clearly to the 3’ terminus of E . coli 23 S rRNA. This was

proposed on the basis of sequence homology (113),and the secondary structure of the 4.5 S molecule [which in contrast to 5.8 S rRNA does not appear to be strongly hydrogen-bonded to the chloroplast 23 S rRNA (114)l corresponds precisely to the 3‘-terminal 100 bases of the E. coli 23 S structure (Fig. 3) (53,29). A slightly different secondary structure for 4.5 S rRNA has been proposed (107),but this structure is, in our opinion, both energetically and phylogenetically less favorable.

C. Evidence for Alternative Conformations (“Switches”)

The studies just described demonstrate that, despite differences in size, base composition, and base sequence, all ribosomal RNA molecules so far investigated fall into the same overall pattern of structure. This raises the question of why the high level of conservation has been necessary, and here it is difficult to avoid the conclusion that the ribosomal RNA must be initimately involved in the protein biosynthetic function. This in turn leads to the question of whether the secondary structures are more or less rigid, or whether gross changes in conformation take place in ribosome assembly or during the ribosomal cycle (cf. 84, 115, 116). There is some strong, but unfortunately still not conclusive, evidence that the latter possibility is the correct one; this evidence comes from the analysis of base-paired fragments (80,81,95), where a few RNA sequences from both E. coli 16 S and 23 S rRNA are paired with different partners in the various experiments. In particular, a fragment of 5 S rRNA was identified in association with the 23 S sequence region 1760-1770, which is base-paired to the 1980-1990 region in the secondary structure model of Fig. 3 (domain d) (95). In the 16 S rRNA, the sequence 1050-1070, which is normally associated with the sequence 1190-1210 (domain c, Fig. 2), was also found in association with the sequence 385-400 (domain a, Fig. 2) (81).In such “switch” situations, the sequence comparison approach can again be invoked, to test whether the results are experimental artifacts, or whether they represent genuine multiple conformations of the RNA. In the 5 S-23 S rRNA interaction, a similar but not entirely satisfactory structure can be drawn for the corresponding sequences in 2. mays chloroplasts (95),and also for the equivalent regions (117) in

22

RICHARD BRIMACOMBE

et al.

yeast 26 S rRNA (38) and 5 S rRNA (118). In the case of the 16 S “switch,” the base-pairing concerned is not only conserved from E . coli to 2. mays chloroplast rRNA (96), but a precisely similar switch structure can also be drawn in both S . cerevisiae and X . laevis 18 S rRNA (94,96),i.e., between sequences 1250-1280 (domain c in yeast 18 S RNA, Fig. 4) and sequences 390-410 (domain a). I n every case, this structure is a strong helix composed of 13-15 base-pairs. Similar, although less strong, complementarities can also be drawn (117) for the corresponding regions from A. nidulans (31)and S. cerevisiae (11) mitochondrial 15 S rRNA; all these structures are illustrated in Fig. 8. Additional evidence for an interaction of some kind between these two regions of the E . coli 16 S rRNA comes from a ribonucleoprotein fragment isolated by Spitnik-Elson et al. (119); the RNA from this fragment contained sequences from the regions concerned (among others) and migrated as a single complex after removal of the protein. On the other hand, there is not a trace of a corresponding switch interaction in the 12 S mitochondrial rRNA. This is not too serious, since the small mitochondrial rRNA molecules must inevitably have lost some of the properties of their larger counterparts, and this switch could be an example, since the mitochondrial rRNA is considerably eroded in the region of the 3’-proximal component of the switch (cf. Fig. 4). It should also be noted that the switch structure generates a potential “knot” in the RNA (cf. Section II,A), and it has been suggested (81)that this could be compensated by appropriate twisting of other parts of the structure, leading to a chain reaction of cooperative conformational changes in the RNA. Evidence against the existence of the switch is the fact that the secondary structure of the 18 S rRNA in the region between bases 200 and 500 (encompassing the switch sequence, Fig. 4) is currently in dispute (96,98-100). The controversy arises because two distinct sets of primary sequence homology between yeast and E . coli can be discerned in this region. One set of homology leads to the structure shown for yeast 18 S rRNA in Fig. 4 (domain a) (96),whereas the other set leads to a different structure (98-loo), in which the switch sequence is no longer in an equivalent position to the corresponding E . coli switch sequence. The former structure shows the greatest similarity to the E . coli secondary structure in the region between bases 400 and 500, whereas the latter model shows better primary sequence homology, but less secondary structural similarity. We have already mentioned (Fig. 6) a similar case where two different sets of apparently significant sequence homology led to the deduction of different secondary structures. It may well be that in such cases both sets of

23

RNA

STRUCTURE OF RIBOSOMAL

S.CEREVISIAE

A“’

‘1069

!-:

1185

ro69/C’G-F’C\185

. .

. .

1060-U-A

--

G-C

4

1300-c-G

. .

5-G

C-?

1211

I050

1262\A.

c-c.

u’

. . . . C-$

U-G C-G

-400

U-A

G-C

G-C

C-C

c’

.. . . .

. .

140

A.c-G<4

C’ ‘G. U-G.

c-c a-u

“-! 1052,

. .

. .

.. . . .

U-A 9SC-G-u

:-I:

G-C

U-?

. . G--u .

1270-U-A

. .

k.u-cc

. .

. . . .

U-A

G-c

. . c.u-a,

C-G‘

C-!

F-G

A-U

C-G/

G-u-390

“:C-G G-C

p ’ C - G

c.A.A-V-1060

. .

U-?

U-?

1190-G--F A-y

. . u-! . .

. . . .

C-G

C-G

U-G

Mitochondria1 15s

1277/U‘G--f’C\391 A-U

. . . .

A-U

. .

A-U

N I OULhNS

A.

18s

. .

940-C-G

U-A.

c4,08

. .

A A.

. .

. .

C-C

401

917,.

G’

C-G.

A’

148

u

C-G

‘

\u.u-*.c/

I045

Z.IIAYS

X.LAEVIS

Chloroplast 165

185

1107

. . . . . . . .

/u.c-c.c\

C-G

U-G

U-A

U-A-410

G-C-360

C-G. . U-A‘

. . . . 1100-

. .

. .

C-G

C-C

C-G

. .

A

?-A

U-A

G-u

. .

U-G

. .

. .

?

. . . .

A-U

. . . . . .

G-c-790 A

. . . . . .

U-

. . . .

U-A

G-C

C-G

U-A

406

A-U

. .

4

5.CEREVISXAE

Mitochondria1 15s

G-C

. .

G-U 129J,A.U-A.

c/421

C-G

FIG.8. Possible “switch” structures in small-subunit rRNA molecules. The two structures drawn for Escherichia coli 16 S rRNA were observed experimentally (81),the “normal” structure (cf. Fig. 2) being shown 011 the left and the “switch” structure on the right. The corresponding “switch” structures shown for the other five sequences indicate the possible alternative base-pairings from the equivalent or nearly equivalent regions of the secondary structures in domains a and c (Figs. 2 and 4; and cf. refs. 96,11, 31).See text for further explanation.

homology have significance in the three-dimensional structure of the RNA, but one is also left with the uncomfortable possibility that one or both sets of homology are (despite the extremely low probability) statistical artifacts. Clearly this type of discrepancy needs to be re-

24

RICHARD BRIMACOMBE

et al.

solved before the all-important question can be settled of whether or not major conformational switches do indeed take place during the ribosomal cycle.

I11. RNA-Protein Interactions

The secondary structures described in the preceding sections are in effect “two-dimensional” maps of the RNA, which, in the ribosome itself, undergo further folding to fit into the compact ribosomal particles, together with the ribosomal proteins. The interaction between RNA and protein and the three-dimensional packing of the RNA are the subjects of this section and Section IV, respectively. Unfortunately, in relation to the foregoing sections, the state of our knowledge in both these areas is still extremely fragmentary. A. Protein Binding Sites on rRNA The “classical” approach to the study of RNA-protein interactions in the ribosome has been to make use of the fact that a number of ribosomal proteins can bind singly and specifically to their cognate rRNA molecules. Such complexes can be submitted to mild nuclease digestion, and the RNA fragments remaining bound to the protein can be analyzed. Alternatively, whole ribosomal subunits can similarly be digested and separated into RNP fragments whose protein and RNA content can then be determined. The data obtained from such experiments have been reviewed often and in detail (e.g., 35,120,121).To summarize very briefly, “binding sites” on E. coli 16 S rRNA have been localized at least partially for proteins S4, S7, S8, S15, S17, and S20 (see review reference 121, and references 97 and 122 for original literature). Protein S1 has been implicated in binding to the 3’ terminus of the 16 S rRNA (123,124),and the association of specific regions of the RNA with various groups of proteins has also been reported (120). In addition, some protein attachment sites on 16 S rRNA have been located by electron microscopy (125).In the case of 23 S rRNA, binding sites have been localized for proteins L l , L20, L23, L24 (121) and (more recently) for L11 (126),and other single proteins or groups of proteins have been isolated in association with large subfragments of the 23 S rRNA molecule (120).It should be noted here that in some cases the binding sites described in the reviews (120,121)are based on older partial sequence data for the 23 S rRNA, and precise placings within the complete 23 S sequence ( 6 )have only recently been made (33).In the case of protein L20, the exact location of the binding site is

STRUCTURE OF RIBOSOMAL

RNA

25

still unclear. The more detailed results from all these experiments are included in the summary diagrams (see Figs. 9 and 10). On 5 S rRNA, binding sites for proteins L5, L18, and L25 have been determined (121, 127-129), but there is still disagreement among the various results. To some extent this reflects the fact that the intact 5 S molecule is already smaller than most of the RNA segments isolated as binding sites for proteins on the 16 S and 23 S rRNA; as a result, attempts to localize binding sites within the 5 S rRNA are a priori focused at a more detailed level. An interesting finding emerging from these studies is that all 5 S rRNA molecules contain a bulged base at the same position (A-66 in E . coli; cf. Fig. 1); this residue is implicated in the binding of protein L18 (130).On the basis of similar features in other RNA-protein complexes, it is proposed (130) that such bulged nucleotides may represent primary recognition sites for protein-nucleic acid interactions. Some RNA-protein complexes have been described with 5 S and 5.8 S rRNA from eukaryotes (e.g., 131,132),and it has also been reported that E . coli proteins L18 and L25 can bind specifically to 5.8 S RNA (133).This led to the suggestion that eukaryotic 5.8 S rRNA is the equivalent of prokaryotic 5 S RNA, a proposition in contradiction to the role of 5.8 S RNA indicated by the comparative sequencing studies (Section 11,B). Heterologous complexes can be formed between rRNA from a variety of sources with E. coli ribosomal proteins, including proteins S4, S7, S8, S15, S17, and S20 (e.g., 134,135) and, most dramatically, protein L1(136,137),which binds specifically even to the eukaryotic 26 S rRNA from Dictyostelium discoideum (136).Such observations provide powerful support for the concept presented in the preceding section that secondary structure in the rRNA (and presumably also the three-dimensional structure) has been strongly conserved throughout evolution. The binding sites of E . coli proteins S8 and S15 on several heterologous 16 S rRNA molecules are indeed strikingly conserved (138),and in the case of protein L1, the binding site on D . discoideum 26 S rRNA was shown to correspond precisely to the equivalent binding site isolated from the cognate E . coli 23 S rRNA (136).A similar binding site could be constructed in the 28 S rRNA from X . Zaeuis

(136).

However, the application of the binding site approach has limitations. In the first place, the size of the binding site varies considerably from one protein to another. The smallest site, that of L11, is 61 nucleotides long (126),whereas the binding site of protein S4 corresponds to almost one-third of the 16 S molecule (97).The binding sites corre-

26

RICHARD BRIMACOMBE

et al.

spond very closely to secondary structural domains in the rRNA (see Figs. 9 and lo), and sometimes [e.g., in the cases of S4 and L24 (reviewed in 121)] they contain noncontiguous sequences of RNA that migrate as a single complex held together by long-range interactions (see Section I1,A) even when the protein has been removed. Further, although quite a large number of proteins bind singly and specifically to the 16 S or 23 S rRNA [with a strong dependence on the method used for the isolation of both RNA and protein moieties (reviewed in 120 and 139)],it has not yet been possible to localize binding sites for more than a few of these proteins. In many cases, as already mentioned, the proteins (e.g., S7) have been isolated only in association with very large subfragments of the RNA (120). This suggests that binding sites may have been found only in those cases where the protein concerned is indeed associated with a single secondary structural domain of the RNA; in other cases (e.g., if a protein were to be supported between two such domains) then the complex would very likely disintegrate upon mild nuclease digestion, and no binding site would be detected. Regardless as to whether this surmise is correct, it is clear that additional experimental approaches must be applied in order to obtain more detailed information concerning the organization of protein and RNA within the ribosome, and for this reason a number of research groups have turned their attention to RNA-protein crosslinking techniques.

B. RNA-Protein Cross-Linking The cross-linking approach is very simple in concept, but in practice it poses a number of problems, and we therefore discuss the crosslinking of ribosomal proteins to their cognate rRNA molecules within the ribosome at some length. First of all, it is important to remember that cross-linking is a purely topographical probe; that is to say, it offers the possibility of determining neighborhoods or contacts between the various components of a system, but these neighborhoods may or may not reflect strong physical associations between the components concerned. Thus a cross-linking experiment gives a different type of information, compared to the binding site approach described above. We begin by listing the various methods and reagents used to induce RNA-protein cross-links in the E . coli ribosome and go on to discuss the analysis of the cross-linked products. Since the ultimate objective here is the construction of a detailed topographical map of the ribosome at the amino acid-nucleotide level, we pay particular attention to those cases where a precise analysis of the cross-link sites has been possible.

STRUCTURE OF RIBOSOMAL

RNA

27

1. CROSS-LINKING METHODSAND REAGENTS The most direct type of cross-linking is where an amino acid or nucleotide is activated (either chemically or photochemically) and is then able to react directly with a neighboring residue within the ribosome, generating a covalent link. For such a reaction to occur, the two residues concerned must obviously be very close to, if not actually in contact with, one another. The simplest method of this type is irradiation with ultraviolet light (see also Section IV), which generates a number of RNA-protein cross-links in the ribosome (140-144).Crosslinks can be induced in a similar manner in the presence of photoactivatable dyes such as methylene blue (145).Chemical activation leading to cross-linking can be effected by the use of N-ethyl-N’-(dimethylaminopropy1)carbodiimide (146), which presumably functions by activating carboxyl groups of aspartic or glutamic acid, with a subsequent nucleophilic attack by the amino group of a nucleotide residue. A further rather specialized example is the use of periodate to oxidize the 3’-terminal cis-hydroxyl groups of the RNA (147,148); the dialdehyde thus formed can react with protein amino groups to form Schiff bases, which can then be stabilized by borohydride reduction (cf., however, reference 149,and see below). The more usual type of cross-linking involves the use of bifunctional reagents, which generate a bridge between two nucleotide or amino acid residues. Here the distance between the cross-linked residues is obviously dependent on the distance between the two functional groups of the cross-linking reagent, but, since these reagents are seldom rigid molecules, it should be remembered that this distance will be less than or equal to the fully extended length of the reagent. Such reagents are of two kinds, symmetrical or heterobifunctional. A symmetrical reagent must contain functional groups able to react with both nucleotides and amino acids, and this type of reagent is a priori therefore rather nonspecific. Formaldehyde has been used to generate RNA-protein cross-links in the ribosome (150), but with only limited success, since the cross-linking reaction is spontaneously reversible, and therefore a detailed analysis of the cross-linked products could not be made. The chemical reactions involved here are not clear, but presumably lead to methylene bridge formation. More useful are the “nitrogen mustards” (see also Section IV), in particular bis(2-chloroethylfamine (151)and its N-methyl derivative (252,153), which react on the one side with the N-7 atom of guanine residues in the RNA, and on the other side with various amino-acid side chains, thus leading to formation of stable RNA-protein cross-links. Very similar in

28

RICHARD BRIMACOMBE

et al.

reactivity is the reagent diepoxybutane ( 154, 155), which generates cross-linked products containing a cis-diol group that can, if desired, be cleaved subsequently by periodate oxidation. Another example is trichloro-2,4,6-triazine (sym-triazine trichloride) (156),a rather rigid molecule, in which the chlorine atoms can b e substituted one at a time at different temperatures, resulting in the formation of RNA-protein cross-links. The use of a symmetrical reagent obviously leads also to the formation of intra-RNA and protein-protein cross-links, and the inherent lack of specificity can cause problems such as denaturation or aggregation (see below). Considerable attention has therefore been given to the development of heterobifunctional reagents, in which each step in the cross-linking reaction can be independently controlled. One of the first reagents of this type to be applied to the ribosome was 4-nitrophenyl 3-(2-bromo-3-oxobutane-l-sulfonyl)propionate, which contains at one end a bromoketone group (to react with a nucleotidic amino group at p H 6), and at the other end an activated ester (to react with a protein amino group at pH 8) (157). However, this reagent yielded only very low levels of RNA-protein cross-linking (151), partly because of solubility problems, but also very probably because the reagent was “too specific”; that is to say, the type of nucleotideamino acid neighborhood that could be cross-linked by the reagent may have been rather rare in the ribosome. This leads to the concept of a reagent in which one reactive group causes a specific reaction, whereas the second group is designed to react as unspecifically as possible, in order to have a high chance of forming a cross-link with any functional group in its vicinity. Photoactivatable aromatic azides have proved to be very useful for this latter purpose and have been applied in a number of systems. I n the case of the ribosome, methyl azidophenylacetimidate (158,159)and ethyl 4azidobenzamidoacetimidate (160) (both of which carry an imido ester function for specific reaction with lysine residues) have been used successfully, the photoactivatable azide group showing a preference for reaction with the nucleic acid moiety. Azido derivatives of halogenated pyridine (161)belong to this same class of reagents, but probably react with protein in a less specific manner. In another compound of similar type, (4-azidopheny1)glyoxal (162), the glyoxal group is able to react specifically either with unpaired guanine residues in the RNA or with arginine residues in the protein. With all these compounds, the specific reaction is allowed to take place first, excess reagent is then removed, and the nonspecific azide reaction, which proceeds via nitrene formation, is then induced by mild ultraviolet irradiation, The

STRUCTURE OF RIBOSOMAL,

RNA

29

reagent methyl 4-(6-formyl-3-azidophenoxy)butyrimidate (163) is also worthy of mention in this context, although it has up to now been used only as a protein-protein cross-linking agent. Reagents containing sulfhydryl groups can also be used in RNAprotein cross-linking reactions. N-Acetyl-N'-( p-glyoxyloylbenzoy1)cystamine ( 1 6 4 ) reacts with RNA and can be used to generate either intra-RNA cross-links (by coupling adjacent SH groups), or, in combination with a second photoactivatable derivative, RNA-protein crosslinks. More recently, the well-known protein-protein cross-linking reagent 2-iminothiolane (165, 166) (or methyl 4-mercaptobutyrimidate) has been used as an RNA-protein cross-linker (167)by allowing the imidoester function to react first with protein and then giving a mild ultraviolet irradiation treatment, which causes cross-linking to the RNA, presumably by reaction of the sulfhydryl function with excited pyrimidine residues. As already mentioned, the advantage of heterobifunctional reagents is that each step of the reaction can be independently controlled. Such compounds also tend to give a plateau level of reaction as the concentration of reagent is increased, but the yields of the photochemical reactions are usually very low, particularly in the case of the aromatic azides; the nitrene intermediates are so reactive that many of the activated species are lost by reaction with solvent, and the resulting maximum yield of RNA-protein cross-linked product usually corresponds to about 5% of the total ribosomal protein (e.g., 159). In contrast, the symmetrical bifunctional reagents such as diepoxybutane (154) or nitrogen mustard (mechlorethamine) (151) show an accelerating reaction with increasing reagent concentration, which implies that at higher reagent concentrations the reaction is accompanied by a randomization of the ribosomal conformation; this in turn allows more cross-links to form. This point was underscored by experiments with bisulfite, a potentially very useful cross-linking reagent that can substitute the 4-amino group of cytosine with the Eamino group of a lysine residue (168).This reagent showed no RNAprotein cross-linking at all in ribosomal subunits in the presence of magnesium, but when magnesium was removed a large yield of RNAprotein cross-links was obtained (151). In this context it should also be noted that many cross-linking agents are not water-soluble and must be dissolved in organic media, which may have an adverse effect on the ribosomal conformation, even at low concentrations. Ultraviolet irradiation causes destruction of the ribosomal conformation at high doses (142),and, as a general principle with all the cross-linking systems described above, it is clear

30

RICHARD BRIMACOMBE

et al.

that the most specific cross-links are formed under minimal conditions. Loss of ribosomal activity (e.g., in polyphenylalanine synthesizing ability) is usually rather rapid during the cross-linking reactions (e.g., 1/59),but this can be due either to the relatively large number of monovalently reacted cross-linker molecules that become attached to the ribosomal subunits or to a “freezing” of an important part of the ribosome as a direct result of a cross-link, and it does not necessarily indicate a destruction of the ribosomal conformation. Finally, the question of cleavable versus noncleavable reagents must be considered here. Of the reagents described above, only diepoxybutane (154) can be cleaved after the reaction, by virtue of the cis-diol group generated by the reaction. Another cleavable crosslinking system (169) was applied successfully to a synthetic complex between protein L24 and 23 S rRNA, but the system showed virtually no RNA-protein cross-linking in ribosomal subunits (170). In any event, it must be remembered that, after cleavage of the cross-linker, a cross-linked site becomes in general indistinguishable from a site where the reagent has formed a monovalent adduct. In other words, if the ultimate objective of the cross-linking experiment is the analysis of the sites of cross-linking (see below), then the use of a cleavable cross-linking agent may not offer any advantage.

2. ANALYSESOF

CROSS-LINKED PRODUCTS

The first stage in any analysis of the products of an RNA-protein cross-linking reaction with the ribosome is the identification of the proteins involved, and, insofar as they are known, the cross-linked proteins are listed in Table I1 for all of the reagents decribed in the foregoing section. In the earlier experiments the identities of the proteins cross-linked to rRNA were inferred in a negative manner by their disappearance from the pattern of total ribosomal proteins on polyacrylamide gels (140, 141, 147). Since most of the cross-linking reactions are accompanied by a number of side reactions, the disappearance of a protein is obviously a very unsatisfactory criterion, which can in any case be observed only if the level of reaction is very high; as noted in the preceding section, this is not usually the case. Identification with the heIp of antibodies to the individual proteins has been used (148,142), but in most of the recent analyses (e.g., 149151, 159-162), the cross-linked proteins have been identified by removing non-cross-linked protein on dodecyl sulfate sucrose gradients, isolating the RNA plus cross-linked protein, digesting this with nucleases, and analyzing the resulting protein-oligonucleotide complexes on two-dimensional polyacrylamide gels. The existence of a covalent

TABLE I1 RNA-PROTEIN CROSS-LINKING AGENTSO

Reagent or method of cross-linking Ultraviolet irradiation Methylene blue activation N-Ethyl-”-(dimethy laminopropy 1)carbodiimide Periodate oxidation Formaldehyde

Bridge length

(A)

Principal proteins identified as cross-linked to RNAb

30 S subunit proteins

<1 <1 <1

s7c s3, s4, s5, s 7 S4, S5, S7, S9/11, S13, S15-18

<1 2

Bis(2-chloroethy1)amine N-Methylbis(2-chloroethy1)amine

5 5

Diepoxybutane sym-Triazine trichloride (trichloro-2,4,6hiazine) 3-(2-Bromo-3-oxobutane-l-sulfonyl)propionate Methyl 4-azidophenylacetimidate Ethyl 4-azidobenzamidoacetimidate

4 3

s1, S21d S3, S4, S5, S9/11, S13, S21 (S7, S12, S18) S3, S4, S5, S9/11, S13 S2, S3, S4, S5, S6, S7, S9/11, S13 Many s3, s 4

8 6 8

s3, s4, s5, s7, s9

4-Azido-3,5-dichloro-2,6-difluoropyridine 4-Azidopheny lgly oxal N-Acetyl-N’-(4-g1yoxyloylbenzoyl)cystamine 2-Iminothiolane

3 5 >10 5

50 S subunit proteins L2, L4 L2, L3

-

L13, L27, L32 (L2, L6, L16) L1, L2, L3, L5 L1, L2, L3, L5, L20

Reference 142,143 145 146 148 150 151 153

Many L2

154, 155 156

S3, S4, S9/11, S13

L2

157

s3, s4, s5, s 7

L1, L2, L4, L25/29

-

(S12, S13/14, S16, S17, S18) s4, s7, s9 s2, s3, s4, s5, s7, (S12)

160

161 162

-

S3, S4, S5, S7, S8, S9/11, S10

159

164 L2, L4, L6, L21, L23, L27, L29

167, 171

a The cross-linking reagents or methods are listed in the order in which they appear in the text. The bridge length gives the approximate maximum distance between the active functional groups of the cross-linker. b Only positively identified proteins are listed. In many cases other proteins either were not resolved or may have been present as aggregates (see text). Proteins in parentheses are minor products; dashes indicate that no data are available. c Other authors (140, 144) found many more proteins cross-linked (see text). Other authors (149)found many more proteins cross-linked (see text).

32

RICHARD BRIMACOMBE

et al.

cross-link can be demonstrated by using 32P-labeled RNA, in which case the isolated complexes carry a measurable 32P label. In this way, quite low levels of cross-linking of individual proteins can be observed, but the analyses are not unequivocal. In the first place, the cross-linked proteins have altered mobilities in the gels, and this can lead to ambiguities in identification, especially of the smaller proteins. Even if a cleavable cross-linker is used (e.g., 154), the presence of monovalently reacted or cross-linked reagent molecules on the protein can alter its mobility. Further, in all cases the gels of cross-linked proteins show relatively large amounts of protein aggregates; it cannot be excluded that some proteins are particularly prone to aggregation after treatment with the cross-linking reagent, and therefore do not appear on the gels. For this reason the lists of proteins given in Table I1 should be regarded as minimum descriptions of the cross-linked products, and estimates of the yields of individual cross-linked proteins on gels (e.g., by radioactivity measurements) may be quite misleading. A further problem is that the low levels of cross-linking generally observed raise the uncomfortable possibility that only a small, possibly inactive, subpopulation of the ribosomal particles are involved in a particular cross-link, but this uncertainty is common to any study based on chemical modification of the ribosome. Three points from Table I1 deserve special mention. First, in the case of ultraviolet-induced cross-linking, only proteins S7, L2, and L4 become cross-linked at low doses of irradiation (142,143).Turchinsky et al. (144)have described the cross-linking of some additional 30 S proteins, and Gorelic (140, 141) has reported that virtually all the proteins can be cross-linked. At least in the latter instance, the crosslinking was carried out under such vigorous conditions that the ribosomal conformation was certainly destroyed (142).Second, in the case of the periodate oxidation method, Czernilofsky et al. (148) found cross-linking only to proteins S1 and S21, whereas we found a large number of proteins to be involved (149).There are several possible explanations (149)for this discrepancy, which has yet to be resolved. Third, no proteins have so far been reported cross-linked to 5 S rRNA. This is almost certainly a simple consequence of the methods used for the isolation of the RNA-protein cross-linked complexes, which are in general designed only to separate proteins cross-linked to the large 16 S or 23 S rRNA molecules. It should also be noted in this context that some RNA-protein cross-links across the subunit interface (i.e., 30 S proteins cross-linked to 23 S rRNA, or 50 S proteins cross-linked to 16 S rRNA) have been reported using diepoxybutane (154, 1 5 9 ,

STRUCTURE OF RIBOSOMAL

RNA

33

whereas in our experiments with methyl azidophenylacetimidate (159) no such cross-links could be detected in significant amounts. These latter experiments (159) were conducted in such a way as to exclude any possible confusion resulting from low-level cross-contaminations between the 16 S and 23 S rRNA during the analyses of the cross-linked products. Table I1 shows that most if not all of the ribosomal proteins can be cross-linked to their cognate rRNA molecules by one method or another, which reflects an obviously very large number of RNA-protein neighborhoods in the ribosome. It is clear that useful information can be extracted from these data only if the precise sites involved in the cross-links can be identified on both protein and RNA, but so far this has been achieved in only a few instances. In the periodate oxidation cross-linking method (148), the crosslink should a priori involve the 3' terminus of the rRNA, but this has not been positively established (148).The first example of a positive identification of an RNA-protein cross-link site was that of the ultraviolet-induced cross-link between protein S7 and 16 S RNA within the E . coli 30 S subunit (172,173).Here, the site on protein S7 was established by isolating a cross-linked 32P-Iabeled S7-oligonucleotide complex, digesting this with trypsin, and then searching for peptides containing the 32P label. This led to the identification of a short labeled peptide containing five amino acids, in which one amino acid (Met-114) was absent from the sequence ( 1 74) and was presumed to be the site of the cross-link (172).The corresponding site on the RNA was established ( 1 73) by first isolating partially digested 32P-labeled cross-linked S7-RNA fragments containing 30-40 nucleotides, and determining the positions of these fragments in the 16 S rRNA sequence ( 4 ) .This showed unequivocally that the sequence region concerned spanned residues 1215-1255, and that one octanucleotide (CUACAAUG, 1234-1241) was absent. Second, an S7-oligonucleotide cross-linked complex was isolated from a total ribonuclease T1 digest of the S7-RNA complex, and further digestion of this complex with ribonuclease A released C, U, AC, and G. Finally, a third digestion with ribonuclease T2 released two A-residues from the protein-oligonucleotide product after ribonuclease A digestion, which clearly implicates U-1240 as the site of the cross-link. A subsequent report (175), in which only the second of these three digestions was conducted, also led to the same release of C, U, AC, and G from the S7-oligonucleotide complex, but the authors claimed that the crosslink site was within a different oligonucleotide (ACCUCG, positions 1261- 1266).

a

5 20

FIG.9. The secondary structure of Escherichia coli 16 S rRNA (cf. Fig. 2), showing sites of interaction with protein and sites of intra-RNA cross-linking.Protein binding sites are indicated by the regions boxed in with dashed lines, and in the cases of proteins S4 and S8, S15, the smallest of the various published binding sites (97,122)are shown. RNA-protein cross-link sites are indicated by the arrows (e.g., protein S7 at residue 1240),and intra-RNA cross-link sites by the arrowed loops (e.g., at positions 600-650). The site of cross-linking to tRNA is also shown. Sites of psoralen-induced cross-links in the RNA have been localized between approximate positions 930 and 1540, 510 and 1540, 450 and 1540,O and 1540,950 and 1400, and 550 and 870. See text for references.

35

f

L6

5SRNA. L5. L18. L25

FIG.10. The secondary structure of Escherichia coli 23 S rRNA (cf. Fig. 3), showing protein binding sites, RNA-protein cross-link sites, and intra-RNA cross-link sites, as in Fig. 9.

36

RICHARD BRIMACOMBE et

al.

The important point in the analysis of the S7-cross-link site just described is that the presence of the cross-linked product could be positively demonstrated, right u p until the last stage of the analyses. In a number of early experiments, such positive demonstrations were lacking. For example, in our own experiments with formaldehyde (ISO), the spontaneously reversible nature of the cross-linking reaction made it impossible to establish the presence of cross-linked protein on small RNA fragments, and the existence of cross-links on such fragments could only be inferred. Similarly, in attempts (176)to identify sites of cross-linking within synthetic complexes of S20 or S4 with 16 S RNA, the absence of certain RNA fragment bands from digests of the cross-linked complexes was taken as evidence that these bands contained the cross-linked sites, and peptides missing from a corresponding tryptic digest of an S7-16 S RNA complex (177) were assumed to be involved in the cross-linking. In all these cases, the negative nature of the analysis leaves considerable doubt as to the validity of the conclusions. More recently, the site of ultraviolet-induced cross-linking between protein L4 and 23 S rRNA within the E . coli 50 S subunit was determined (178) [Tyr-35 in protein L4 (179)was linked to U-615 in the 23 S RNA (6)l.In this case the analysis of the cross-linking site on the RNA was made using a two-dimensional gel electrophoresis system (178)that allows a number of cross-link sites to be analyzed simultaneously. The system has since been applied to 50 S subunits crosslinked with 2-iminothiolane, and cross-link sites on the 23 S RNA could be positively established for proteins L4, L6, L21, L23, L27, and L29 (167).In the 30 S subunit, a cross-link site for protein S8 was identified (171).On the other hand, analysis of the corresponding sites of cross-linking on the proteins still poses problems, very likely as a result of the monovalently reacted cross-linker molecules present on each protein in addition to the reagent molecule actually involved in the cross-link (171). Finally, the cross-linking of protein S1 to the 16 S RNA in 30 S subunits, using ethyl 4-azidobenzamidoacetimidate, has been described (180). In this experiment, the protein was first treated with reagent, then added to the ribosome and irradiated to generate the cross-link. Protein S1 was therefore the only protein to be involved in any cross-linking, and this is an elegant example of how the independent control of the functional groups in a heterobifunctional reagent can be exploited. It should be noted, however, that the authors were unable to demonstrate positively the presence of the cross-linked protein on the RNA fragment (residues 861-889) they isolated. This

STRUCTURE OF RIBOSOMAL

RNA

37

leaves open the possibility that the protein was in fact cross-linked to a different RNA fragment, which in turn was bound to residues 861889 by hydrogen-bonding. The results of all the cross-linking experiments described above are included in the summary diagrams (Figs. 9 and 10).

IV. Three-Dimensional Packing of €. coli rRNA A. Intra- RNA Cross- Linking Many of the problems discussed in the preceding section are also relevant to the question of intra-RNA cross-linking, and several of the cross-linking agents (Table 11) also lead to formation of intra-RNA cross-links. Compounds such as sulfur mustard (181),nitrogen mustard (152),or diepoxybutane (182)have long been known to generate cross-links in nucleic acids, and all have been shown to cause intraRNA cross-linking in E . coli ribosomal subunits (183,184,185,respectively), as has also simple ultraviolet irradiation (83,95).Another symmetrical bifunctional reagent that has been useful is 1,4-phenyldiglyoxal(1,4-phenylenebisglyoxal)(82,89),and photoactivatable derivatives of psoralens [which intercalate into double-helical regions of the nucleic acid (e.g., 186)]have been successfully applied (187,188). As is the case with the RNA-protein studies, analysis of the sites of cross-linking is obviously essential here, but until recently this has proved to be difficult. Wagner and Garrett (82)were able to isolate a short cross-linked oligonucleotide from the stem region of 5 S rRNA after reaction with 1,4-phenylenebisglyoxal and ribonuclease T1 digestion. By exploiting the G-specificity and ready reversibility of this cross-linking reagent, they could pinpoint the cross-link to residues G-2 and G-112 in the 5 S rRNA sequence (cf. Fig. 1).Similarly, Rabin and Crothers (189) identified a psoralen cross-link in the stem region of 5 S rRNA after partial digestion, making use of the photoreversibility of the psoralen reaction. Monovalently reacted cross-linker molecules cause considerable difficulties in this type of analysis, and recently the problem was overcome for 1,4-phenyldiglyoxal by coupling such monoaddition products to a solid support (89).With this sytem, a new cross-link was identified between residues G-41 and G-72 in the E . coli 5 S RNA (89).The significance of this cross-link for the threedimensional structure of 5 S molecule is described in Section II,A (cf. Fig. l),and these examples serve to illustrate the general principle in intra-RNA cross-linking studies, uiz., the cross-link may be “within”

38

RICHARD BRIMACOMBE

et al.

the secondary structure (as for the G-2 to G-112 cross-link), or it may give an indication of a neighborhood or tertiary structural interaction “outside” the secondary structure (as for residues G-41 and G-72). The analytical methods used in these intra-RNA cross-linking experiments with 5 S rRNA have not so far been applicable to 16 S or 23 S rRNA, since the nuclease digestion products of the larger molecules are too complex. Here, some intra-RNA cross-links have been localized by other methods, involving either two-dimensional gel electrophoretic techniques (83) or electron microscopy (187,188).The first of these methods relies on the fact that RNA fragments from partial digests of rRNA are strongly retarded in denaturing gel systems if they contain an intra-RNA cross-link. The cross-linked fragments can therefore be readily distinguished from the large numbers of noncross-linked fragments on a suitable two-dimensional gel system and can be isolated and subjected to oligonucleotide analysis. Provided that the cross-link lies in a “characteristic” region of the sequence with respect to such an oligonucleotide analysis, the cross-link site can be located, and this has been achieved for a number of ultravioletor diepoxybutane-induced cross-links in both 16 S (83,185)and 23 S rRNA (95).However, so far all the cross-links found by this method are “within” the secondary structure (see above), which almost certainly reflects the selective nature of the partial digestion conditions used. The locations of these cross-links are included in the summary diagrams (Figs. 9 and 10). In contrast, all the intra-RNA cross-links localized by electron microscopy are “outside” the secondary structure. Here Cantor and his colleagues (187,188)used psoralen derivatives to generate cross-links in 16 S rRNA or 30 S subunits, and the cross-link sites were localized to a first approximation by measuring the dimensions of the resulting loop structures that appeared in electron micrographs of the crosslinked RNA. Some of the loops observed are very large, and were originally interpreted as evidence that there are “knots” in the 16 S rRNA structure (190) (cf. Sections II,A and C), since psoralen is an intercalating molecule and long double-helices could be constructed around the cross-link sites. More recently, however, these authors (191, 192) suggested that their cross-links reflect tertiary structural interactions [cf. the tertiary interaction described above for 5 S rRNA (88, 89) (see Fig. l)].In the latest series of experiments, Wollenzien and Cantor used a retarding denaturing gel system to separate the various different cross-linked species of 16 S RNA according to the size of the cross-linked loop (191).In addition, the problem of determining the polarity of the cross-linked RNA molecules in the electron

STRUCTURE OF RIBOSOMAL

RNA

39

microscope has been solved by hybridization and covalent cross-linkage of specific DNA fragments to one end of the rRNA molecule (192). The positions of these psoralen cross-links are indicated in the legend to the summary diagram (Fig. 9) and will obviously be of considerable importance in helping to “fold” the 16 S rRNA secondary structure into three dimensions. A final example of intra-RNA cross-linking is the localization of a cross-link between 16 S rRNA and tRNA. This involves the hypermodified uridine at the 5‘-anticodon position of tRNAVa’or tRNASer, which can be cross-linked very specifically by mild ultraviolet irradiation to 16 S RNA when the tRNA is bound at the ribosomal P-site (193). The cross-link site on 16 S RNA was pinpointed by a series of partial digestions (193-195); it is residue C-1400 in the 16 S rRNA sequence, and a precisely analogous cross-link has since been found in yeast 18 S rRNA (196). The cross-linked nucleotide occurs in a sequence region that is very highly conserved in all small subunit rRNA molecules so far sequenced (see reference 42). The position of this crosslink is also indicated in Fig. 9.

B. Electron Microscopy of rRNA within Ribosomal Subunits

There is now a reasonabIe consensus of agreement among the various models proposed for the E . coli ribosomal subunits on the basis of electron microscopy, and a large amount of data has been collected concerning the distribution of the individual ribosomal proteins on these models (reviewed in 197,198).However, the ribosomal subunits maintain their gross morphological features as seen under the electron microscope even when a large proportion of the proteins has been removed (199). This suggests that it is the rRNA that is largely responsible for determining the overall architecture of the subunits, and some specific regions of the rRNA molecules have been localized by immune electron microscopy. The 3’ ends of isolated rRNA molecules can be oxidized with periodate and allowed to react with a suitable hapten. The modified RNA is subsequently reconstituted into ribosomal subunits and treated with hapten-specific antibody. Using this approach, several groups have made electron microscopic localizations of the 3‘ ends of E . coli 5 S (200,201),16 S (202,203),and 23 S rRNA (201,204);the positions of the 3’ ends in the 30 S and 50 S subunits are illustrated in Fig. 11. A haptenization method has been developed for the 5’-terminal phosphate group of an RNA molecule, and this has been used to locate the 5’ end of 16 S rRNA, by a similar immunological procedure

40

RICHARD BRIMACOMBE

et al.

BASES 925-1395 I PROTllNS 57, 59 510, 511.51?1

BASES 2090-2200

165 RNA Il’-fNOl

165 RNA IS-INOI

N6-OIMETHYL ADENOSINE

30s SUBUNIT 50s SUBUNIT FIG.11. Location of regions of 16 S and 23 S rRNA in Escherichia coli 30 S and 50 S subunits, derived from electron microscopic studies. The sketches crudely summarize the data from several models and several laboratories (see text for details and references).

(205). The 5‘ end of 23 S rRNA has not yet been localized, but it should be noted that, if the secondary structure model (Fig. 3) is correct, then the 5’ and 3’ termini of 23 S rRNA should be virtually coincident, as is also the case with 5 S rRNA (Fig. 1). The use of antibodies against modified bases in the rRNA to pinpoint specific RNA regions by immune electron microscopy has been successful so far only in the case of N6-dimethyladenosine (206,207). Two such residues occur adjacent to one another at a position 23 nucleotides from the 3’ end of the 16 S rRNA ( 4 , 5 ) ,and their location in the 30 S subunit is also indicated in Fig. 11. It is noteworthy that antibodies to a haptenated tRNA molecule crosslinked to the ribosomal P-site (cf. Section IV,A) also bind predominantly to the same region of the 30 S subunit (208), and, taken together with the established involvement of the 3‘ terminus of 16 S rRNA in messenger RNA recognition (209, 210),this region of the 30 S subunit is now rather clearly implicated as being the messenger decoding site. The location of some regions of the 16 S and 23 S rRNA molecules can be inferred from the data on RNA-protein interactions (see Section 111,A). Thus, nucleotides 2090-2200 in 23 S rRNA, which constitute the binding site for protein L1 (136,137),should lie in the vicinity of the antibody attachment site for protein L1, which occupies a very characteristic position on the 50 S subunit (211)(Fig. 11). Similarly, it is highly probable that nucleotides 925-1395 in the 16 S rRNA constitute the “head” of the 30 S subunit, by virtue of the association of this section of the RNA with proteins S7, S9, S10, S14, and S19 (120), all of which have antibody binding sites in the “head” region

STRUCTURE OF RIBOSOMAL

41

RNA

(197,198). Other regions, such as the 16 S rRNA sequences binding to protein 54 or to proteins S6, S8, S15, and S18, can also be tentatively placed (see, e.g., 35). The location of the 5 S rRNA relative to 23 S rRNA within the 50 S subunit is not yet clear; a possible “switch” involving sequences in 5 S and 23 S rRNA has already been discussed (Section II,C), and a ribonucleoprotein fragment containing proteins L5, L18, and L25 together with 5 S rRNA and a short section of 23 S rRNA has also been described (212) (see Fig. 10). These both offer possible contact sites for 5 S rRNA (cf. Fig. ll),but, on the other hand, a strong base-paired interaction previously proposed between 5 S rRNA and residues 143154 of the 23 S rRNA (213) has since been discounted on the basis of comparative evidence (36). Finally, it should be noted in this context that there is a considerable body of evidence concerning the accessibility (i-e.,surface topography) of specific regions of the rRNA within the subunits or at the subunit interface. The evidence comes from modification with kethoxal as well as from nuclease digestion studies, both described in Section I1,A. The identities of oligonucleotides that occur at the subunit interface have been inferred from their differential reactivity toward kethoxal in 70 S ribosomes as opposed to isolated subunits (214, 215). In addition, the identities of oligonucleotides released from the 30 S subunit by digestion with nuclease have been analyzed (216), and digestion with ribonuclease H has been used to probe the sites of binding to the RNA of complementary oligodeoxynucleotides (217, 218). This type of information cannot as yet be interpreted in terms of the precise three-dimensional folding of the rRNA, but will obviously have to be taken into account in any detailed models of the ribosomal subunits in the future.

V. Outlook We show in this article how the sequence information now availa-

ble for ribosomal RNA from many organisms has been used to help

construct secondary structure models for these molecules. It is to be expected that these models will become progressively refined during the next few years, as further sequences are determined and more experimental data are collected, but, as we have stressed (Section II,A), the principal features of the secondary structures for both 16 S and 23 S rRNA are already well established. As a result, the emphasis in working out the three-dimensional topography of the ribosomal RNA in situ in the subunits must now focus on the questions of the

42

RICHARD BRIMACOMBE

et al.

tertiary folding of the RNA, its interaction with ribosomal protein, and the organization of the interface between the subunits. Here, again as we have stressed (Section 111), our knowledge is still very fragmentary. It is not yet clear how far the comparative sequencing approach, so useful in establishing the secondary structures, will be of help in these areas, and the main burden will be the collection of suitably direct experimental information. Cross-linking techniques, despite their drawbacks and the difficulties involved in their application, still offer the best hope for delivering the type of detailed information that is needed, and for this reason we have devoted a considerable amount of space to a description of the current status of these techniques (Sections I11 and IV). As a preliminary picture of the folding of the rRNA takes shape, it will gradually become possible to correlate the structure with data from other sources, in particular from electron microscopy. In this field, progress is being made in the three-dimensional reconstruction and averaging of electron micrographs of ribosomal subunits, both in ribosome microcrystals (219) and as individual subunits (220, 221), and the shape of the ribosomal RNA within the subunits is beginning to emerge. Coupled with improved data from immune electron microscopy (cf. Section IV,B), and with other data on the three-dimensional arrangement of the proteins [e.g., protein-protein cross-linking data (222), or neutron scattering studies of the protein distribution (223)], this should lead to an increasingly coherent picture of the topographical organization of RNA and protein in the subunits. However, here it must be added that establishing the principles that govern the actual physical interaction between proteins and RNA remains one of the most difficult problems to approach in this field. As far as function of the ribosomal RNA is concerned, one emerging technique is particularly worthy of mention, namely the site-directed mutagenesis of rRNA. This involves the construction via rDNA of specific ribosomal RNA mutants, and it offers the possibility of examining the importance of selected areas of the RNA in a manner that has hitherto been unthinkable. Some preliminary results have already been reported (224), and this new technology could well become a vital link in correlating the ribosomal RNA function with the type of structural studies that we have outlined in this article.

ACKNOWLEDGMENTS The authors are very grateful to Dr. H. G. Wittmann for his critical reading of the manuscript, and to many colleagues who have sent us their results prior to publication.

STRUCTURE OF RIBOSOMAL

43

RNA

REFERENCES I. R. Brimacombe, K. H. Nierhaus, R. A. Garrett, and H. G. Wittmann, This Series 18,

1 (1976). F. Sanger and A. R. Coulson, J M B 94,441 (1975). A. M . Maxam and W. Gilbert, PNAS 74, 560 (1977). J. Brosius, M. L. Palmer, P. J. Kennedy, and H. F. Noller, PNAS 75, 4801 (1978). P. Carbon, C. Ehresmann, B. Ehresmann, and J. P. Ebel, FEBS Lett. 94, 152 ( 1978). 6. J. Brosius, T. J. Dull, and H. F. Noller, PNAS 77, 201 (1980). 7 . G. G Brownlee, F. Sanger, and B. G. Barrell, Nature (London)215,735 (1967). 8. P. Borst and L. A. Grivell, Nature (London)290, 443 (1981). 9. I. C. Eperon, S. Anderson, and D. P. Nierlich, Nature (London)286, 460 (1980). 10. R.A. van Etten, M. W. Walberg, and D. A. Clayton, Cell 22, 157 (1980). 1 1 . F. Sor and H. Fukuhara, C . R . Hebd. Seances Acad. Sci. Ser. D 291,933 (1980). 12. Z. Schwarz and H. Kossel, Nature (London)283, 739 (1980). 13. K. Edwards and H. Kossel, NARes 9, 2853 (1981). 14. C. M. Bowman and T. A. Dyer, BJ 183,605 (1979). 15. P. M. Rubtsov, M. M. Musakhanov, V. M. Zakharyev, A. S. Krayev, K. G. Skryabin, and A. A. Bayev, NARes 8,5779 (1980). 16. M. Salim and B. E. H. Maden, Nature (London)291,205 (1981). 17. 0. I. Georgiev, N. Nikolaev, A. A. Hajiolov, K. G. Skryabin, W. M. Zakharyev, and A. A. Bayev, NARes 9,6953 (1981). 18. G. M. Rubin, JBC 248,3860 (1973). 19. B. R. Jordan, FEBS Lett. 44, 39 (1974). 20. B. Dujon, Cell 20, 185 (1980). 21. M. A. Wild and R. Sommer, Nature (London)283,693 (1980). 22. D. M. Glover, Cell 26, 297 (1981). 23. V. A. Erdmann, NARes 8, r31 (1980). 24. V. A. Erdmann, NARes 9, r25 (1981). 25. K. R. Luehrsen, D. E. Nicholson, D. C. Eubanks, and G. E. Fox, Nature (London) 293, 755 (1981). 26. G. N. Pavlakis, B. R. Jordan, R. M. Wurst, and J. N. Vournakis, NARes 7, 2213 (1979). 27. A. G . Wildeman and R. N. Nazar,JBC 255, 11896 (1980). 28. F. Takaiwa and M. Sugiura, Mol. Cen. Genet. 180, 1, (1980). 29. K. Edwards, J. Bedbrook, T. Dyer, and H. Kossel, Biochem. Int. 2,533 (1981). 30. C. Saccone, P. Cantatore, G. Gadaleta, R.Gallerani, C. Lanave, G. Pepe, and A. M. Kroon, NARes 9,4139 (1981). 31. H. G . Kochel and H. Kiintzel, NARes 9,5689 (1981). 32. J. J. Seilhammer and D. J. Cummings, NARes 9, 6391 (1981). 33. C. Branlant, A. Krol, M. A. Machatt, J. Pouyet, J. P. Ebel, K. Edwards, and H. Kossel, NARes 9,4303 (1981). 34. P. Carbon, J. P. Ebel, and C. Ehresmann, NARes 9, 2325 (1981). 35. H. F. Noller and C. R. Woese, Science 212, 403 (1981). 36. H. F. Noller, J. Kop, V. Wheaton, J. Brosius, R. R. Gutell, A. M. Kopylov, F. Dohme, W. Herr, D. A. Stahl, R. Gupta, and C. R. Woese, NARes 9,6167 (1981). 37. L. Graf, E. Roux, E. Stutz and H. Kossel, NARes 10,6369 (1982). 38. G. M. Veldmann, J. Klootwijk, V. C. H. F. de Regt, R. J. Planta, C. Branlant, A. Krol, and J. P. Ebel, NARes 9, 6935 (1981). 2. 3. 4. 5.

44

RICHARD BRIMACOMBE

et al.

R. Course and S. Gerbi, NARes 8, 3623 (1980). S. Gerbi, personal communication. A. A. Azad and N. J. Deacon, NARes 8,4365 (1980). R. van Charldorp and P. H. van Knippenberg, NARes 10, 1149 (1982). B. Allet and J. Rochaix, Cell 18, 55 (1979). J. L. Bos, K. A. Osinga, G . van der Horst, N. B. Hecht, H. F. Tabak, G. B. van Ommen, and P. Borst, Cell 20, 207 (1980). 45. P. M. M. Rae, B. D. Kohorn, and R. P. Wade, NARes 8,3491 (1980). 46. I. B. Dawid and M. L. Rebbert, NARes 9,5011 (1981). 47. H. Roiha and D. M. Glover, NARes 9,5521 (1981). 48. H. Nomiyama, Y. Sakaki, and Y. Takagi, PNAS 78, 1376 (1981). 49. H. Nomiyama, S. Kuhara, T. Kukita, T. Otsuka, and Y. Sakaki, NARes 9, 5507 (1981). 50. H. Kochel, C. Lazarus, N. Basak, and H. Kuntzel, Cell 23, 625 (1981). 51. L. Graf, H. Kossel, and E. Stutz, Nature (London) 286, 908 (1980). 52. E. M. Orozco, K. E. Rushlow, J. R. Dodd, and R. B. HallickJBC 255,10997 (1980). 53. A. Machatt, J. P. Ebel, and C. Branlant, NARes 9, 1533 (1981). 54. G. E. Fox, E. Stackebrandt, R. B. Hespell, J. Gibson, J. Maniloff, T. A. Dyer, R. S. Wolfe, W. E. Balch, R. S. Tanner, L. J. Magrum, L. B. Zablen, R. Blakemore, R. Gupta, L. Bonen, B. J . Lewis, D. A. Stahl, K. R. Luehrsen, K. N. Chen, and C. R. Woese, Science 209,457 (1980). 55. V. A. Erdmann, This Series 18,45 (1976). 56. P. Carbon, C. Ehresmann, B. Ehresmann, and J. P. Ebel, EJB 100,399 (1979). 57. R. J. Baer and D. T. Dubin, NARes 9,323 (1981). 58. D. G . Hughes and B. E. H. Maden, BJ 171,781 (1978). 59. J . Klootwijk and R. J. Planta, EJB 39,325 (1973). 60. B. E. H. Maden, Nature (London) 288,293 (1980). 61. I. Tinoco, P. N. Borer, B. Dengler, M. D. Levine, 0. C. Uhlenbeck, D. M. Crothers, and J. Gralla, Nature NB 246, 40 (1973). 62. J. Gralla and D. M. Crothers, J M B 78, 301 (1973). 63. P. N. Borer, B. Dengler, I. Tinoco, and 0. C. Uhlenbeck,JMB 86, 843 (1974). 64. D. Lightfoot, NARes 5,3565 (1978). 65. G. M. Studnicka, G . M. Rahn, I. W. Cummings, and W. A. Salser, NARes 5,3365 (1978). 66. R. Nussinov and A. B. Jacobson, PNAS 77,6309 (1980). 67. M. Zuker and P. Stiegler, NARes 9, 133 (1981). 68. R. Nussinov, I. Tinoco, and A. B. Jacobson, NARes 10,351 (1982). 69. P. E. Auron, W. P. Rindone, C. P. H. Vary, J. J . Celentano, and J. N. Vournakis, NARes 10,403 (1982). 70. A. Troutt, T. J. Savin, W. C. Curtiss, J. Celentano, and J. N. Vournakis, NARes 10, 653 (1982). 71. H. F. Noller, Bchem 13,4694 (1974). 72. C. R. Woese, L. J. Magrum, R. Gupta, R. B. Siege], D. A. Stahl, J. Kop, N. Crawford, J. Brosius, R. Gutell, J. J. Hogan, and H. F. Noller, NARes 8, 2275 (1980). 73. F. van der Haar, E. Schlimme, V. A. Erdmann, and F. Cramer, Bioorg. Chem, 7, 282 (1971). 74. D. A. Peattie and W. Gilbert, PNAS 77,4679 (1980). 75. C . Bellemare, B. R. Jordan, J. Rocca-Serra, and R. Monier, Biochimie 54, 1453 (1972). 76. J. C. Lee and V. M. Ingram,JMB 41,431 (1969). 39. 40. 41. 42. 43. 44.

STRUCTURE OF RIBOSOMAL

RNA

45

77. S. K. Vassilenko, P. Carbon, J. P. Ebel, and C. Ehresmann,JMB 152,699 (1981). 78. J. B. Lewis and P. Doty, Nature (London) 225,510 (1970). 79. P. Wrede, 0 . Pongs, and V. A. Erdmann, JMB 120,83 (1978). 80. A. Ross and R. Brimacombe, Nature (London) 281,271 (1979). 81. C. Glotz and R. Brimacombe, NARes 8, 2377 (1980). 82. R. Wagner and R. A. Garrett, NARes 5,4065 (1978). 83. C. Zwieb and R. Brimacombe, NARes 8,2397 (1980). 84. G. Fox and C. R. Woese, Nature (London) 256,505 (1975). 85. R. A. Cox, BJ 98,841 (1966). 86. A. Araco, M. Belli, and G. Onori, NARes 2, 373 (1975). 87. G. M. Studnicka, F. A. Eiserling, and J. A. Lake, NARes 9, 1885 (1981). 88. T. Pieler and V. A. Erdmann, PNAS 79,4599 (1982). 89. J. Hancock and R. A. Wagner, NARes 10, 1257 (1982). 90. D. A. Stahl, K. R. Luehrsen, C. R. Woese, and N. R. Pace, NARes 9,6129 (1981). 91. A. Rich and U. L. RajBhandary, ARB 45,805 (1976). 92. P. Stiegler, P. Carbon, M. Zuker, J. P. Ebel, and C. Ehresmann, NARes 9, 2153 (1981). 93. R. Brimacombe, in “Biological Implications of Protein-Nucleic Acid Interactions” (J. Augustyniak, ed.), p. 44. ElsevieriNorth-Holland, Amsterdam, 1980. 94. R. Brimacombe, Biochem. Int. 1, 162 (1980). 95. C. Glotz, C. Zwieb, R. Brimacombe, K. Edwards, and H. Kossel, NARes 9, 3287 (1981). 96. C. Zwieb, C. Clotz, and R. Brimacombe, NARes 9,3621 (1981). 97. C. Ehresmann, P. Stiegler, P. Carbon, E.Ungewickel1, and R. A. Garrett, EJB 103, 439 (1980). 98. P. Stiegler, P. Carbon, J. P. Ebel, and C. Ehresmann, EJB 120,487 (1981). 99. A. S. Mankin, A. M. Kopylov, P. M. Rubtsov, and K. G. Skryabin, Dokl. Acad. Nauk. USSR 256, 1006 (1981). 100. A. S. Mankin, A. M. Kopylov, and A. A. Bogdanov, FEBS Lett. 134, 11 (1981). 101. H. Kiintzel and H. G. Kochel, Nature (London) 293,751 (1981). 102. A. S. Mankin and A. M. Kopylov, Biochem. Int. 3,587 (1981). 103. R. N. Nazar, FEBS Lett. 119, 212 (1980). 104. R. N. Nazar, T. 0. Sitz, and H. Busch, JBC 250,8591 (1975). 105. R. N. Nazar and K. L. Roy, JBC 253,395 (1978). 106. N. R. Pace, T. A. Walker, and E. Schroeder, Biochemistm 16,5321 (1977). 107. C. G. Clark and S. A. Gerbi,]. Mol. Euol. 18,329 (1982). 108. J. J. PBne, E. Knight and J. E. Darnell, JMB 33, 609 (1968). 109. R. A. Weinberg and S. Penman,JMB 38,289 (1968). 110. R. N. Nazar and T. 0. Sitz, FEBS Lett. 115, 71 (1980). 1 1 1 . J. M . Kelly and R. A. Cox, NARes 9, 1111 (1981). 113. R. M. MacKay, FEBS Lett. 123, 17 (1981). 114. P. R. Whitfield, C. J. Leaver, W. Bottomley, and B. A. Atchison, BJ 175, 1103 (1978). 115. H. F. Noller, in “Ribosomes” (G.Chambliss, G. R. Craven, J. Davies, K. Davis, L. Kahan, and M. Nomura, eds.), p. 3. Univ. Park Press, Baltimore, Maryland, 1980. 116. H. Weidner, R. Yuan, and D. M. Crothers, Nature (London) 266, 193 (1977). 117. R. Brimacombe, unpublished results. 118. M. Miyazaki, NARes Spec. Publ. 3, s153 (1977). 119. P. Spitnik-Elson, D. Elson, S. Avital, and R. Abramowitz, NARes 10,4483 (1982). 120. R. Brimacombe, G . Stoffler, and H. G. Wittmann, ARB 47,217 (1978).

46

RICHARD BRIMACOMBE

et al.

121. R. A. Zimmermann, in “Ribososmes” (G. Chambliss, G. R. Craven, J. Davies, K. Davis, L. Kahan, and M. Nomura, eds.), p. 135. Univ. Park Press, Baltimore, Maryland, 1980. 122. R. Miiller, R. A. Garrett, and H. F. Noller,JBC 254,3873 (1979). 123. A. E. Dahlberg and J. E. Dahlberg, PNAS 72,2940 (1975). 124. R. C. Yuan, J. A. Steitz, P. B. Moore, and D. M. Crothers, NARes 7,2399 (1979). 125. M. D. Cole, M. Beer, T. Koller, W. A. Strycharz, and M. Nomura, PNAS 75, 270 (1978). 126. F. J. Schmidt, J. Thompson, K. Lee, J. Dijk, and E. Cundliffe,JBC 256, 12301 (1981). 127. J. Zimmermann and V. A. Erdmann, M o l . Gen. Genet. 160,247 (1978). 128. S. Douthwaite, R. A. Garrett, R. Wagner, and J. Feunteun, NARes 6,2453 (1979). 129. M. Speek and A. Lind, NARes 10, 947 (1982). 130. D. A. Peattie, S. Douthwaite, R. A. Garrett, and H. F. Noller, PNAS 78,7331 (1981). 131. N. Ulbrich, K. Todokoro, E. J. Ackermann, and I. G. Woo1,JBC 255,7712 (1980). 132. K. Todokoro, N. Ulbrich, Y. L. Chan, and I. G. Wool, JBC 256, 7207 (1981). 133. P. Wrede and V. A. Erdmann, PNAS 74,2706 (1977). 134. M. Geisser and G. A. Mackie, EJB 70, 159 (1976). 135. D. L. Thurlow and R. A. Zimmermann, PNAS 75,2859 (1978). 136. R. L. Gourse, D. L. Thurlow, S. A. Gerbi, and R. A. Zimmermann, PNAS 78,2722 (1981). 137. C. Branlant, A. Krol, A. Machatt, and J. P. Ebel, NARes 9, 293 (1981). 138. D. Thurlow and R. A. Zimmermann, in preparation. 139. K. H. Nierhaus, Cum. Top. Microbiol. Immunol. 97, 81 (1982). 140. L. Gorelic, Bchem 14,4627 (1975). 141. L. Gorelic, BBA 390, 209 (1975). 142. K. Moller and R. Brimacombe, MoZ. Gen. Genet. 141,343 (1975). 143. 0. G. Baca and J. W. Bodley, BBRC 70, 1091 (1976). 144. M. F. Turchinsky, N. E. Bronde, K. S. Kussova, G. G. Abduraschidova, E. V. Muchamedganova, J. N. Schatsky, T. F. Bystrova, and E. J. Budowsky, EjB 90,83 (1978). 145. D. E. Zook and S . R. Fahnestock, BBA 517,400 (1978). 146. C. Chiarrutini and A. Expert-Bezancon, FEBS Lett. 119, 145 (1980). 147. R. A. Kenner, BBRC 51,932 (1973). 148. A. P. Czemilofsky, C. G. Kurland, and G. Stoffler, FEBS Lett. 58,281 (1975). 149. J. Rinke and R. Brimacombe, MoZ. Biol. Rep. 4, 153 (1978). 150. K. Moller, J. Rinke, A. Ross, G. Buddle, and R. Brimacombe, EJB 76, 175 (1977). 151. E. Ulmer, M. Meinke, A. Ross, G. Fink, and R. Brimacombe, M o l . Gen. Genet. 160, 183 (1978). 152. P. E. Geiduschek, PNAS 47,950 (1961). 153. P. Maly, unpublished results. 154. H. G. Baumert, S. E. Skold, and C. G. Kurland, EJB 89,353 (1978). 155. S. E. Skold, Biochimie 63, 53 (1981). 156. C. Oste and R. Brimacombe, Mol. Gen. Genet. 168, 81 (1979). 157. G. Fink and R. Brimacombe, Biochem. SOC. Trans. 3,1014 (1975). 158. G. Fink, H. Fasold, W. Rommel, and R. Brimacombe, Anal. Biochem. 108, 394 (1980). 159. J. Rinke, M. Meinke, R. Brimacombe, G. Fink, W. Rommel, and H. Fasold,JMB 137, 301 (1980). 160. R. Millon, M. Olomucki, J. Y. LeGall, B. Golinska, J. P. Ebel, and B. Ehresmann, EJB 110,485 (1980).

STRUCTURE OF RIBOSOMAL

RNA

47

161. R. Millon, J. P. Ebel, F. le Goffic, and B. Ehresmann, BBRC 101, 784 (1981). 162. S. M. Politz, H. F. Noller, and P. D. McWhirter, Bchem 20, 372 (1981). 163. J. A. Maassen, E. N. Schop, and W. Moller, Bchem 20, 1020 (1981). 164. A. Expert-Bezanqon and D. Hayes, EJB 103, 365 (1980). 165. R. R. Traut, A. Bollen, T. T. Sun, J. W. B. Hershey, J. Sundberg, and L. R. Pierce, Bchem 12,3266 (1973). 166. A. Sommer and R. R. Traut,JMB 106,995 (1976). 167. I. Wower, J. Wower, M. Meinke, and R. Brimacombe, NARes 9, 4285 (1981). 168. I. Boni and E. I. Budowsky,]. Biochem. (Tokyo) 73,821 (1973). 169. C. Oste, R. Parfait, A. Bollen, and R. R. Crichton, M d Gen. Genet. 152,253 (1977). 170. C. Oste and R. Brimacombe, unpublished results. 171. J. Wower, I. Wower, and R. Brimacombe, unpublished results. 172. K. Moller, C. Zwieb, and R. Brimacombe, JMB 126, 489 (1978). 173. C. Zwieb and R. Brimacombe, NARes 6, 1775 (1979). 174. J. Reinbolt, D. Tritsch, and B. Wittmann-Liebold, FEBS Lett. 91, 297 (1978). 175. B. Ehresmann, C. Backendorf, C. Ehresmann, R. Millon, and J. P. Ebel, EJB 104, 255 (1980). 176. B. Ehresmann, C. Backendorf, C. Ehresmann, and J. P. Ebel, FEBS Lett. 78,261 (1977). 177. B. Ehresmann, J. Reinbolt, C. Backendorf, D. Tritsch, and J. P. Ebel, FEBS Lett. 67, 316 (1976). 178. P. Maly, J. Rinke, E. Ulmer, C. Zwieb, and R. Brimacombe, Bchem 19, 4179 (1980). 179. M. Kimura and B. Wittmann-Liebold, FEBS Lett. 121, 317 (1980). 180. B. Golinska, R. Millon, C. Backendorf, M. Olomucki, J. P. Ebel, and B. Ehresmann, EJB 115,479 (1981). 181. P. Brookes and P. D. Lawley, BJ 77, 478 (1960). 182. P. D. Lawley and P. Brookes,JMB 25, 143 (1967). 183. R. M. Malbon and J. H. Parish, BBA 246,542 (1971). 184. C. Zwieb, A. Ross, J. Rinke, M. Meinke, and R. Brimacombe, NARes 5, 2705 (1978). 185. C. Zwieb, unpublished results. 186. L. Musajo and G. Rodighiero, Photochem. Photobiol. 11,27 (1970). 187. P. Wollenzien, J. E. Hearst, P. Thammana, and C. R. Cantor,JMB 135,255 (1979). 188. P. Thammana, C. R. Cantor, P. Wollenzien, and J. E. HearstJMB 135,271 (1979). 189. D. Rabin and D. M. Crothers, NARes 7,689 (1979). 190. C. R. Cantor, in “Ribosomes” (G. Chambliss, G. R. Craven, J. Davies, K. Davis, L. Kahan, and M. Nomura, eds.), p. 23, Univ. Park Press, Baltimore, Maryland, 1980. 191. P. L. Wollenzien and C. R. Cantor,JMB 159, 151 (1982). 192. P. L. Wollenzien and C. R. Cantor, PNAS 79, 3940 (1982). 193. R. A. Zimmermann, S. M. Gates, I. Schwartz, and J. Ofengand, Bchem 18, 4333 (1979). 194. B. H. Taylor, J. B. Prince, J. Ofengand, and R. A. Zimmermann, Bchem 20, 7581 (1981). 195. J. B. Prince, B. H. Taylor, D. L. Thurlow, J. Ofengand, and R. A. Zimmermann, PNAS 79,5450 (1982). 196. J. Ofengand, P. Gornicki, K. Chakraburtty, and K. Nurse, PNAS 79,2817 (1982). 197. G. StofRer, R. Bald, B. Kastner, R. Liihrmann, M. Stoffler-Meilicke, and G. Tischendorf, in “Ribosomes” (G. Chambliss, G. R. Craven, J. Davies, K. Davis, L. Kahan, and M. Normura, eds.), p. 171. Univ. Park Press, Baltimore, Maryland, 1980.

48

RICHARD BRIMACOMBE

et al.

198. J. A. Lake, in “Ribosomes” (G. Chambliss, G. R. Craven, J. Davies, K. Davis, L. Kahan, and M. Nomura, eds.), p. 207. Univ. Park Press, Baltimore, Maryland, 1980. 199. V. D. Vasiliev and V. E. Koteliansky, FEES Lett. 76, 125 (1977). 200. I. N. Shatsky, A. G. Evstafieva, T. F. Bystrova, A. A. Bogdanov, and V. D. Vasiliev, FEBS Lett. 121,97 (1980). 201. M. Stoffler-Meilicke, G. Stoffler, 0. W. Odom, A. Zinn, G. Gamer, and B. Hardesty, PNAS 78,5538 (1981). 202. I. N. Shatsky, L. V. Mochalova, M. S. Kojouharova, A. A. Bogdanov, and V. D. Vasiliev,JMB 133, 501 (1979). 203. R. Luhrmann, M. IStoffler-Meilicke, and G. Stoffler, Mol. Gen. Genet. 182, 369 (1981). I 204. I. N. Shatsky, A. G!.Evstafieva, T. F. Bystrova, A. A. Bogdanov, and V. D. Vasiliev, FEBS Lett. 122, 251 (1980). 205. L. V. Mochalova, I. N. Shatsky, A. A. Bogdanov, andV. D. VasiIiev,/MB 159,637 (1982). 206. S. M. PolitL and D. G. Glitz, PNAS 74, 1468 (1977). 207. P. Thammana and C. R. Cantor, NARes 5,805 (1978). 208. M . Keren-Zur, M. Boublik, and J. Ofengand, PNAS 76, 1054 (1979). 209. J. Shine and L. Dalgarno, PNAS 71, 1342 (1974). 210. J. A. Steitz and K. Jakes, PNAS 72,4734 (1975). 211. E. R. Dabbs, R. Ehrlich, R. Hasenbank, B. H. Shroeter, M. Stoffler-Meilicke, and G. Stoffler, JME 149,553 (1981). 212. C. Branlant, A. Krol, J. Sriwidada, and R. Brimacombe, EJB 70,483 (1976). 213. W. Herr and H. F. Noller, FEBS Lett. 53,248 (1975). 214. W. Herr and H. F. Noller,JME 130,421 (1979). 215. W. Herr, N. M. Chapman, and H. F. Noller, J M B 130,433 (1979). 216. N. L. Teterina, A. M. Kopylov, and A. A. Bogdanov, FEBS Lett. 116,265 (1980). 21 7. A. S. Mankin, E. A. Skripkin, N. V. Chichkova, A. M. Kopylov, and A. A. Bogdanov, FEBS Lett. 131, 253 (1981). 218. C. Backendorf, C. J. C. Ravensbergen, J. Van der Plas, J. H. Van Boom, G. Veeneman, and J. Van Duin, NARes 9, 1425 (1981). 219. A. E. Yonath, J. Mussig, B. Tesche, S. Lorenz, V. A. Erdmann, and H. G . Wittmann, Biochem. Znt. 1,428 (1980). 220. V. Knauer, R. Hegerl, and W. Hoppe,JMB, in press. 221. H. Oettl, R. Hegerl, and W. Hoppe,JME, in press. 222. R. R. Traut, J. M. Lambert, G. Boileau, and J. W. Kenny, in “Ribosomes” (G. Chambliss, G. R. Craven, J. Davies, K. Davis, L. Kahan, and M. Nomura, eds.), p. 89. Univ. Park Press, Baltimore, Maryland, 1980. 223. P. B. Moore in “Ribosomes” (G. Chambliss, G. R. Craven, J. Davies, K. Davis, L. Kahan, and M. Nomura eds.), p. 11 1. Univ. Park Press, Baltimore, Maryland, 1980. 224. R. L. Course, M. J. R. Stark, and A. E. Dahlberg,JMB 159,397 (1982).