A new model for the three-dimensional folding of Escherichia coli 16 s ribosomal RNA. III. The...

J. Mol. Biol. (1997) 271, 566±587

A New Model for the Three-dimensional Foldingof Escherichia coli 16 S Ribosomal RNA.III{. The Topography of the Functional Centre

Florian Mueller1 and Holger Stark2Marin van Heel2Jutta Rinke-Appel1

and Richard Brimacombe1*

1Max-Planck-Institut fuÈ rMolekulare GenetikAG-Ribosomen, Ihnestrasse 7314195 Berlin, Germany2Imperial College of ScienceDepartment of BiochemistryLondon SW7 2AY, England

{ Paper II in this series is an accMueller & Brimacombe (1997b).

Abbreviation used: EM, electron

0022±2836/97/340566±22 $25.00/0/mb

We describe the locations of sites within the 3D model for the 16 S rRNA(described in two accompanying papers) that are implicated in ribosomalfunction. The relevant experimental data originate from many labora-tories and include sites of foot-printing, cross-linking or mutagenesis forvarious functional ligands. A number of the sites were themselves usedas constraints in building the 16 S model. (1) The foot-print sites for Asite tRNA are all clustered around the anticodon stem±loop of the tRNA;there is no ``allosteric'' site. (2) The foot-print sites for P site tRNA thatare essential for P site binding are similarly clustered around the P siteanticodon stem±loop. The foot-print sites in 16 S rRNA helices 23 and 24are, however, remote from the P site tRNA. (3) Cross-link sites fromspeci®c nucleotides within the anticodon loops of A or P site-boundtRNA are mostly in agreement with the model, whereas those fromnucleotides in the elbow region of the tRNA (which also exhibit extensivecross-linking to the 50 S subunit) are more widely spread. Again, cross-links to helix 23 are remote from the tRNAs. (4) The corresponding cross-links from E site tRNA are predominantly in helix 23, and these agreewith the model. Electron microscopy data are presented, suggestive ofsubstantial conformational changes in this region of the ribosome.(5) Foot-prints for IF-3 in helices 23 and 24 are at a position with closecontact to the 50 S subunit. (6) Foot-prints from IF-1 form a clusteraround the anticodon stem±loop of A site tRNA, as do also the sites on16 S rRNA that have been implicated in termination. (7) Foot-print sitesand mutations relating to streptomycin form a compact group on oneside of the A site anticodon loop, with the corresponding sites for specti-nomycin on the other side. (8) Site-speci®c cross-links from mRNA(which were instrumental in constructing the 16 S model) ®t well both inthe upstream and downstream regions of the mRNA, and indicate thatthe incoming mRNA passes through the well-de®ned ``hole'' at the head-body junction of the 30 S subunit.

# 1997 Academic Press Limited

Keywords: functional sites; site-directed cross-linking; foot-printing;mutagenesis; conformational changes
*Corresponding author
Introduction

Structural studies on macromolecular complexessuch as the ribosome are primarily of interest inso-far as they are able to shed some light on the waythat the complex concerned exerts its biological

ompanying paper,

microscopy.

971212

function. In the preceding two papers (Mueller &Brimacombe, 1997a,b) we have described a newmodel for the three-dimensional arrangement ofthe 16 S rRNA of Escherichia coli, which has been®tted to the 30 S moiety of an electron microscopic(EM) reconstruction of the 70 S ribosome at ca 20 AÊ

resolution (Stark et al., 1997a). We have demon-strated how this model accommodates purelystructural data such as intra-RNA cross-linkswithin the 16 S molecule (Mueller & Brimacombe,1997a), and how well it can be correlated with the

# 1997 Academic Press Limited

Functional Sites in the 16 S rRNA Model 567

known arrangement of the ribosomal proteins asdetermined by neutron scattering (Capel et al.,1988) or immuno electron microscopy (IEM;StoÈ f¯er & StoÈ f¯er-Meilicke, 1986; Oakes et al.,1986), via the RNA-protein interaction data fromcross-linking (Brimacombe, 1991) or foot-printing(Powers & Noller, 1995) studies (Mueller &Brimacombe, 1997b). These correlations form thephysical basis of the model, and the model itselfprovides a framework for examining the largebody of data that is available concerning sites offunctional interest within the 16 S rRNA molecule.

In the 70 S EM reconstruction by Stark et al.(1997a) that was used to ®t the 16 S rRNA struc-ture, the ribosomes were in the pre-translocationalstate, and the A and P site tRNA molecules weredirectly visualized. In the same publication, a simi-lar reconstruction of ribosomes in the post-translo-cational state correspondingly showed the tRNAmolecules at the P and E sites. Thus, the positionsof the anticodon stem±loops of the tRNAs at allthree sites could be located on the 30 S subunitwith some precision. Furthermore, the location ofthe mRNA in the decoding region is in turn dic-tated by the positions of the anticodon loops of atleast the A and P site tRNAs, via the codon-antico-don interactions at these two tRNA sites. Withregard to the arrangement of the 16 S rRNA withinthe 30 S subunit, there are a number of data sets inthe literature that directly relate speci®c regions ofthe 16 S molecule to the decoding region, as de-®ned by the functional complex of tRNAs andmRNA. The information includes site-directedcross-links from different positions on the mRNA(e.g. see Rinke-Appel et al., 1994), foot-prints fromtRNA at the A and P sites (Moazed & Noller,1990), and site-directed cross-links from variouspositions on tRNA located at the A, P or E site(e.g. see Prince et al., 1982; Rinke-Appel et al.,1995). These data sets were indeed used as primaryconstraints in the construction of the 16 S rRNAmodel, and in its ®tting to the EM contour(Mueller & Brimacombe, 1997a).

Here, we address the question of how well thismodel-building strategy has succeeded, and docu-ment the locations of all the sites just mentioned inthe completed 16 S rRNA model. In addition, weconsider the locations in the model of related setsof data within the 16 S rRNA, including foot-printsites of initiation factors (Moazed et al., 1995), foot-print sites or mutation sites causing resistance toantibiotics (e.g. see Moazed & Noller, 1987;Montandon et al., 1985; Sigmund et al., 1984), andmutation sites causing translational suppression(Murgola et al., 1988; Shen & Fox, 1989). It must beborne in mind that the various data sets were ob-tained, some with 30 S subunits and others with70 S ribosomes, under a wide variety of experimen-tal conditions. It is therefore not surprising that, aswas the case in the correlation with the RNA-pro-tein interaction data (Mueller & Brimacombe,1997b), there are some discrepancies. In particular,one helix of the 16 S rRNA, helix 23, contains a

number of functional sites including cross-linksand foot-prints to P site tRNA. However, in the16 S model this helix is forced by other modellingconstraints (such as the neutron map) into a lo-cation that is rather remote from that of the A andP site tRNA molecules in the EM reconstruction.This discrepancy raises the question of possibleconformational changes in the 30 S subunit underdifferent experimental conditions, and in this con-nection we describe a further EM reconstruction(cf. Stark et al., 1997a), which was made with 70 Sribosomes carrying a single uncharged tRNAmolecule at the ribosomal P site. The latter recon-struction, although at somewhat lower resolution,did indeed indicate a major conformational differ-ence (coupled with a possible shift in the positionof the P site tRNA) in the E site region of the ribo-some, where helix 23 is located. Similarly, thecross-linking data indicate that there is some move-ment of the mRNA in the downstream region closeto the A site codon, depending on whether a tRNAis present at the A site. Regardless of this, themRNA path in both the upstream and downstreamregions is rather clearly de®ned in terms of the16 S rRNA model. In particular, the incomingmRNA passes through the well-de®ned hole (cf.Stark et al., 1995; Frank et al., 1995) at the head-body junction of the 30 S subunit, giving rise to asituation that is reminiscent of the clamp or ringstructures that have been observed in other sys-tems, such as RNA or DNA polymerases (e.g.Jacobo-Molina et al., 1993; Polyakov et al., 1995;Herendeen & Kelly, 1996). In all these consider-ations, the fact that our 16 S rRNA model has beenrelated to the directly visualized positions of thetRNA molecules in the 70 S EM reconstructions,rather than to their inferred positions in theisolated 30 S subunit as in earlier models (cf.Brimacombe, et al., 1988; Stern et al., 1988;Malhotra & Harvey, 1994; Fink et al., 1996), has asigni®cant in¯uence on the interpretation of thedata.

Results and Discussion

The functional data

The data that we have considered in connectionwith the three-dimensional model of the 16 SrRNA are summarized in Tables 1 to 3, and theirlocations within the secondary structure of the 16 Smolecule are illustrated in Figure 1. Table 1 liststhe site-directed cross-links from mRNA to 16 SrRNA that have been analysed in our laboratory,in collaboration with the Moscow laboratory ofBogdanov (Dontsova et al., 1992), from variouspositions in the upstream and downstream regionsof mRNA close to the decoding region; the A andP site codons cover mRNA positions �1 to �6.With the exception of the single site within the Psite codon (position �2), all of these cross-linkswere induced by direct UV-irradiation of thionu-cleotides incorporated into the mRNA, and are

Table 1. Summary of mRNA-16 S rRNA cross-linkingdata

mRNA position Site in 16 S rRNA Reference

�2 h28 (926) (1)�4 h44 (1402) (2)�6 h34 (1052) (2)�7 h28 (1395) (2)�8,9 h34 (1196) (1)�11 h18 (532) (2)�12 h18 (530) (1)

ÿ1/ÿ8 h22 (665) (3)ÿ1/ÿ8 h43 (1360) (3)ÿ1/ÿ8 h45 (ca 1530) (3)

The mRNA downstream position �1 is the A-residue of theAUG start codon; upstream position ÿ1 is the nucleotideimmediately 50 to the latter residue. The upstream cross-linksÿ1/ÿ8 were observed in all positions between ÿ1 and ÿ8. Thelocations in 16 S rRNA show the number of the nearest helix inbold face (cf. Figure 1), with the precise cross-linked nucleotidein parentheses. Cross-links from Rinke-Appel et al. (1993, 1994)were zero-length cross-links from thioU residues in the mRNA;those from Sergiev et al. (1997) were zero-length cross-linksfrom thioG, with the exception of the cross-link at position �2,which was from a diazirine-modi®ed uridine residue with abridging distance of ca 10 AÊ . References: (1) Sergiev et al.(1997); (2) Rinke-Appel et al. (1993); (3) Rinke-Appel et al.(1994).

568 Functional Sites in the 16 S rRNA Model

thus of ``zero length''. Corresponding site-directedcross-links to 16 S rRNA from speci®c positions intRNA located at the ribosomal A, P or E sites arelisted in Table 2. In contrast to the mRNA cross-

Table 2. Summary of tRNA-16 S rRNA cross-linking data

tRNA tposition Sites(s) on 16 S rRNA A

8 h29 (1338, 1339) ÿh45 (1531-1542) ÿ

20:1 h23 (694, 711) ��h23 (701) ÿh29 (1342) �h44 (1465-1500) �h45 (1500-1542) �

32 h23 (693) ÿh29 (936) �h29 (1338) ÿh31 (957, 966) ÿh43 (1376, 1378) ÿh43 (1378) ��

34 h44 (1400)

37 h45 (1510-1542)

47 h23 (686, 701, 717) ÿh29 (1345, 1348) ÿh43 (1350-1387) ÿ

The cross-link site locations in 16 S rRNA show the number of thelinked residue(s) in parentheses. The cross-links from tRNA positiontance of ca 10 AÊ , those from positions 20:1 and 47 were with a dithese locations; here the bridging distance, including the acp3U sidediazirine reagent, bridging distance ca 10 AÊ , and those from positioirradiation or with an azido-nucleotide, respectively. The latter two ccases the distribution of the cross-linking between the A, P and E sences: (1) Rinke-Appel et al. (1995); (2) DoÈring et al. (1994); (3) Prince

links, only those from tRNA positions 34 (Princeet al., 1982) and 37 (Wower et al., 1993a) are zero-length cross-links, the remainder, from our labora-tory, being the results of cross-linking with bifunc-tional reagents having a bridging distance of 10 to15 AÊ . Some of the cross-link sites, notably those inthe extreme 30-region of the 16 S rRNA, could notbe localized to sequences less than 30 to 50 nucleo-tides long; these are included in Table 2 for thesake of completeness, but will not be consideredfurther here.

Table 3 lists the relevant foot-printing and muta-tional data, including the foot-prints for A and Psite-bound tRNA (Moazed & Noller, 1990) and thefoot-prints for initiation factors IF-1 and IF-3(Muralikrishna & Wickstrom, 1989; Moazed et al.,1995), as well as foot-print and mutational sites re-lating to the antibiotics streptomycin, spectinomy-cin and tetracycline (see Table 3 for references).The corresponding foot-print data for a number ofother antibiotics (Woodcock et al., 1991) are not in-cluded in Table 3, as these foot-prints coincidealmost entirely with the A and P site tRNA foot-prints, and the antibiotic binding sites can thus beinferred from the positions of the respective tRNAfoot-print sites in the model. Mutational sites caus-ing translational suppression (Murgola et al., 1988;Shen & Fox, 1989) are also included in Table 3, asthe locations of such sites in relation to the decod-ing centre are of interest in the context of chaintermination.

RNA binding siteP E Reference

� ÿ (1)�� (1)

�� (1)ÿ � (1)� � (1)ÿ ÿ (1)� � (1)

�� (2)ÿ ÿ (2)� � (2)�� ÿ (2)ÿ �� (2)ÿ ÿ (2)

�� (3)

�� (4)

�� (5)�� ÿ (5)ÿ � (5)

nearest helix in bold face (cf. Figure 1), with the precise cross-8 were with an azido cross-linking reagent with a bridging dis-

azirine cross-linking reagent attached to the acp3U residues at-chain, is ca 15 AÊ . The cross-links from position 32 were with ans 34 and 37 are zero-length cross-links, induced by direct UV-ross-links were to the P and the E site, respectively. In the otherites is indicated by �� (strong), � (weak) or ÿ (absent). Refer-et al. (1982); (4) Wower et al. (1993a); (5) Osswald et al. (1995).

Table 3. Summary of footprint and mutational data on16 S rRNA for functional ligands, antibiotics, etc.

Ligand and datatype

Sites on 16 S rRNA Reference

A site tRNAfootprint

h18 (529-531);h44 (1408, 1492-1494)

(1)

P site tRNAfootprint

h18 (532); h23 (693); h24 (794-795); h28 (926, 1381); h29 (1338,1339); h31 (966); h44 (1399-1401)

(1)

IF-1 footprint h18 (530); h44 (1492-1493) (2)

IF-3 footprint h23 (700, 703); h24 (791) (2)h23 (701); h24 (793) (3)

Translational h18 (517) (4)suppression h34 (1054) (5)

Streptomycinfootprint

h2 (911-915) (6)

Streptomycin h2 (912) (7)resistance h18 (523) (8)

h1 (13); h2 (912-915); h18 (523) (9)h18 (507, 525) (10)

Spectinomycinfootprint

h34 (1063-1064) (6)

Spectinomycinresistance

h34 (1192) (11)

Tetracyclinefootprint

h27 (892) (6)

The sites on the 16 S rRNA are given by the number of thenearest helix (bold face; cf. Figure 1), with the exact nucleotidepositions in parentheses. See the text for further details.References: (1) Moazed & Noller (1990); (2) Moazed et al.(1995); (3) Muralikrishna & Wickstrom (1989); (4) Shen & Fox(1989); (5) Murgola et al. (1988); (6) Moazed & Noller (1987);(7) Montandon et al. (1985); (8) MelancËon et al. (1988); (9) Harriset al. (1989); (10) Powers & Noller (1991); (11) Sigmund et al.(1984).


It can be seen from Figure 1 that all of the siteslisted in Tables 1 to 3 are clustered in distinct areasof the 16 S rRNA. The areas concerned cover the``central core'' of the secondary structure, involvingall of the principal domain and sub-domain junc-tions. This is not surprising, since the function ofthe 30 S subunit is focussed on the ``processing'' ofthe mRNA-tRNA complex, and therefore the foot-print and cross-link sites would be expected tooccupy central locations; our strategy of construct-ing the 16 S rRNA model around the tRNA-mRNAcomplex merely re¯ects this basic principle. In-deed, the 16 S rRNA regions that are not includedin Figure 1 are all located towards the ``extremi-ties'' of the model (Mueller & Brimacombe, 1997a),and comprise helices 4 to 17 (in the subunit body),the upper part of helix 21 and the lower part of 44(also in the body), and helices 33, 35 to 40 and theloop-end proximal region of helix 41 (in the sub-unit head). The positioning of these more remoteregions within the model was accomplished bymaking use of the ®ne structure of the EM contour,

as well as of the RNA-protein interaction data etc.,as described in the accompanying papers (Mueller& Brimacombe, 1997a,b).

A closer look at Figure 1 reveals that the ma-jority of the functional sites are clustered into fourgroups. The ®rst is the group of sites in helix 18.The second is that comprising helices 22, 23 and24, and here it should be noted that the loop endsof helices 23 and 24 are in direct contact with oneanother, by virtue of the intra-RNA cross-link V(Mueller & Brimacombe, 1997a). The third groupof sites is at the main domain junction of the 16 SrRNA, formed by helices 1, 2, 28/29 and the upperpart of 44, and the ®nal group is that lying in helix34. It is furthermore noteworthy that in each ofthese four clusters there are at least two cross-linksites to either mRNA or tRNA (Tables 1 and 2),and therefore each group is directly, and notallosterically, linked to the functional centre. ThemRNA cross-linking data in helices 18 and 34 areof particular importance in this context. The cross-links from mRNA analogues carrying thiouridineresidues to the latter helices (Dontsova et al., 1992;Rinke-Appel et al., 1993) provided the ®rst indi-cation that the older generation of 16 S rRNAmodels (Brimacombe et al., 1988; Stern et al., 1988)were seriously in error. The newer, and entirely in-dependent, set of cross-links from mRNAs carryingthioguanosine residues (Sergiev et al., 1997) corro-borates this conclusion, because the new cross-links (Figure 1; Table 1) are also to neighbouringsites within helices 18 and 34. Yet another data setoffering direct evidence for the proximity of helices18 and 34 to the decoding region comes from thecleavage patterns in 16 S rRNA within these helicesproduced by phenanthroline-modi®ed tRNA atthe ribosomal P site in the presence of Cu2� (J. M.Bullard & W. E. Hill, personal communication).The question of the position of helix 18 in relationto proteins S7 and S12 is discussed in the preced-ing paper (Mueller & Brimacombe, 1997b), and onthe basis of the latter arguments, in combinationwith the unequivocal mRNA cross-linking andphenanthroline cleavage data, we discount theclaim (Powers & Noller, 1994) that the functionalsites in helix 18 (cf. Figure 1) are of allosteric origin(for further discussion of this point seeBrimacombe, 1992). A crucial constraint in buildingour 16 S rRNA model has therefore been to locateas many of the functional sites (Figure 1) aspossible in the direct neighbourhood of themRNA-tRNA complex.

The locations of the tRNA molecules in theEM reconstruction

Figure 2 shows the three-dimensional model forthe 16 S rRNA (Mueller & Brimacombe, 1997a),®tted to the 30 S moiety of the 70 S EM contour(Stark et al., 1997a). The neutron map of the masscentres of the ribosomal proteins (Capel et al.,1988) is included in the Figure (Mueller &Brimacombe, 1997b), as are the A and P site tRNA

Figure 1. Regions of the secondary structure of 16 S rRNA containing functional data. The helices are numbered asby Brimacombe (1995). The various symbols are: mX ÿ , mX � , upstream and downstream cross-links to mRNA, re-spectively; tX, approximate locations of cross-link sites to tRNA (see Table 2 for details of individual nucleotides andtRNA site involved); tFA, tFP, foot-print sites to A site and P site tRNA, respectively; IF1, IF3, foot-print sites toinitiation factors IF-1 and IF-3, respectively; TS, sites of mutation causing translational suppression; SmR, SmF,mutation sites causing resistance to streptomycin, and foot-print sites for streptomycin, respectively; SpcR, SpcF,resistance sites or foot-print sites for spectinomycin, respectively; TetF, foot-print site for tetracycline. See Tables 1 to3 for references.


Figure 2. Stereo views of the 30 S subunit, with helical regions of the 16 S rRNA model (cylinders), and the neutronmap positions (Capel et al., 1988) of the ribosomal proteins (spheres). The protein spheres are reduced in size, asby Mueller & Brimacombe (1997b). The rRNA domains are coloured dark blue (50-domain), red (central domain),light blue (30-domain) and yellow (30-minor domain) as by Mueller & Brimacombe (1997a). The proteins arecoloured blue-green (those in the subunit head), orange (side lobe) and darker blue (body). The A site tRNA, Psite tRNA and mRNA are included as pale blue, green and white backbone tube models, respectively. The modelis shown within the semi-transparent silhouette of the EM contour of the 30 S subunit (Stark et al., 1997a). (a) Viewfrom the interface side, with the A site tRNA towards the viewer. (b) View obtained by rotating that of (a) 60�about the vertical axis, with the side lobe towards the viewer. (The black areas are software artifacts, caused bypolygons in the rendering of the EM contour coinciding with those in the rendering of the spheres or cylinders.)


molecules, together with a short segment ofmRNA. The tRNA molecules lie within the EMdensity attributed to them (Stark et al., 1997a; andcf. Figure 3, later), with their CCA ends directedacross the 30 S:50 S interface cavity, towards thepeptidyl transferase area of the 50 S subunit. Theangle between the planes of the two tRNAs isapproximately 50�, which, as discussed by Stark

et al. (1997a), is in good agreement with bio-chemical data such as the ¯uorescence measure-ments by Paulsen et al. (1983) or the site-directedmutagenesis results reported by Smith & Yarus(1989). Relative to one another, the two tRNAs arein the S con®guration, as opposed to the Rcon®guration, which we previously preferred (Limet al., 1992). Cogent arguments in favour of the

Figure 3. Proteins that have been cross-linked to tRNA in the 30 S subunit. The proteins shown are S7 (blue), S9(orange), S11 (red), S13 (yellow) and S19 (purple). Positions of the A site (pale blue), P site (green) and E site (white)tRNAs are taken from Stark et al. (1997a). The view of the 30 S subunit is from the interface side. See the text fordetails.


S con®guration have been put forward byEasterwood et al. (1994) and, although Lim (1997)still adheres to the R con®guration, the morerecent ribosomal topographical data also stronglysupport the S con®guration (for discussion, seeBrimacombe, 1995). The arrangement of the A andP site tRNA molecules within the ribosome asdepicted in Figure 2 is thus similar to the modelsproposed by Wower et al. (1989) or by Noller et al.(1990).

Further support for this tRNA arrangement hascome from an even more recent EM reconstructionof kirromycin-stalled ribosomes (Stark et al.,1997b), in which density corresponding to theEF-Tu/tRNA complex was directly visualized. Thecrystallographically derived atomic coordinates ofthe EF-Tu-GMPPNP-Phe-tRNAPhe complex fromThermus aquaticus (Nissen et al., 1995) could be®tted precisely into this EM density, with the pos-ition of the A site tRNA corresponding closely tothat reported by Stark et al. (1997a) as illustrated inFigure 2. It should, however, be noted at this junc-ture that a signi®cantly different arrangement forthe A, P and E site tRNA molecules has been re-ported in an EM reconstruction by Agrawal et al.(1996). In the latter study, ribosomes were loadedwith three uncharged tRNA molecules, and theEM reconstruction showed a much larger angle (ca160�) between the planes of the A and P sitetRNAs. The differences between the two modelswere pointed out by Stark et al. (1997a), and arediscussed further below.

We have already noted in the Introduction thatthe position of helix 23 in our rRNA model poses aproblem in connection with the locations of thefunctional sites (Figure 1), and this is immediatelyapparent from Figure 2(b). Helix 23 (subhelices

23.1 and 23.2) interacts with protein S11 (seeMueller & Brimacombe, 1997b), and both S11(orange) and helices 23.1 and 23.2 (red) are visiblein Figure 2(b) as the respective protein and rRNAelements in the side lobe of the 30 S subunit on theextreme right-hand side of the Figure (cf.Figure 4(c) of Mueller & Brimacombe, 1997a, andFigure 3(c) of Mueller & Brimacombe, 1997b); theseelements are clearly not close to the P site tRNA,although helix 23 contains both cross-link sites(Table 2) and foot-print sites (Table 3) to the P sitemolecule (which will be discussed in more detailbelow, see Figures 4 and 5). The S11 area is, on theother hand, very close to the E site tRNA, and thisis illustrated in Figure 3, which shows all threetRNAs (A, P and E), together with the proteinsthat have been cross-linked to tRNA; here, proteinS11 (in red) can be seen lying directly behind thewhite E site tRNA molecule.

The E site tRNA was located by Stark et al.(1997a) in an EM reconstruction of disomes in thepost-translocational state. This reconstruction wasessentially identical in its overall morphology withthat of the ribosomes in the pre-translocationalstate (containing A and P site tRNAs), and accord-ingly the position of the E site tRNA from the dis-ome reconstruction is superimposed in Figure 3 onthe 30 S moiety of the pre-translocational EM re-construction. The A and P site tRNAs are alsoshown (cf. Figure 2), as well as the neutronlocations (Capel et al., 1988) of proteins S7, S9, S11,S13 and S19 (Mueller & Brimacombe, 1997b), all ofwhich have been cross-linked from tRNAs at oneor other of the A, P and E sites in a number of stu-dies (summarized by Wower et al., 1993b; for morerecent data, see Osswald et al., 1995; Rinke-Appelet al., 1995). Position 37 of E site-bound tRNA


showed cross-links to the extreme 30-terminal re-gion of the 16 S rRNA (Wower et al., 1993a), whichis located on the side lobe of the 30 S subunit (cf.Mueller & Brimacombe, 1997a), and also to proteinS11. Similarly, position 8 of E site-bound tRNAwas cross-linked to S11 (Rinke-Appel et al., 1995),and these data ®t very well to the location of the Esite tRNA, as illustrated in Figure 3. Cross-linksfrom a number of positions in the E site-boundtRNA have been observed to helix 23 (Table 2, andsee Figure 7(a)), and these also ®t the location ofthe E site tRNA in Figure 3 (cf. Figure 2(b)).However, as noted above, cross-links and foot-prints to this same helix were concomitantly foundfrom the P site tRNA. Similarly, cross-links to pro-tein S11 have been observed from P site-boundtRNA (Rinke-Appel et al., 1995) and, conversely,cross-links to S7 (cf. Figure 3) from E site tRNA(Osswald et al., 1995). Obviously there is a discre-pancy here, which is inherent in the positioning ofthe 16 S rRNA helices in relation to the neutronmap and the dimensions of the EM reconstructionon the one hand, as opposed to the directly visual-ized locations of the P and E site tRNA moleculeswithin the 70 S ribosome on the other hand. Thepossibility of conformational changes in this areaof the 30 S subunit under differing experimentalconditions will be discussed later (see Figures 8and 9).

The other four proteins shown in Figure 3 (S7,S9, S13 and S19) have been cross-linked fromvarious positions in A or P site-bound tRNA(Ofengand et al., 1986; Wower et al., 1993b;Osswald et al., 1995). While the neutron positionsof S7 (dark blue) and S9 (orange) ®t well to theposition of the P site tRNA (from which most ofthe observed cross-links occurred), the positions ofS13 (yellow) and S19 (purple) clearly do not. Thepositions of these proteins are discussed in detailby Mueller & Brimacombe (1997b), where wenoted that the IEM location of S19 (Olson et al.,1988) is close to the position of S7, in good agree-ment with the cross-linking to S19 from both P andA site tRNAs (e.g. see Ofengand et al., 1986). It wasalso noted that the neutron position of S13 is out ofcontact with the rRNA, both in our model (Mueller& Brimacombe, 1997b) and in others (Stern et al.,1988; Malhotra & Harvey, 1994; Fink et al., 1996).In the context of tRNA cross-linking to protein S13,it is noteworthy that the cross-link site from S13 tothe 16 S rRNA (nucleotides 1338-1339 in helix 29,see Table 1 of Mueller & Brimacombe, 1997b) co-incides with sites of both cross-linking and foot-printing on the 16 S rRNA from P site tRNA(Tables 2 and 3).

In the following sections, the locations of thefunctional sites (Figure 1) in our 16 S rRNA modelare documented. As in the accompanying papers(Mueller & Brimacombe, 1997a,b), we use an all-atom presentation to illustrate the data, but onceagain we emphasize that this is for display pur-poses only and should not be over-interpreted atthe current level of resolution of the model. Also as

before, the diagrams are intended to be viewedwith a standard 6 cm magnifying stereoscope. Sub-divisions of the 16 S rRNA helices are referred tousing the nomenclature (e.g. 18.1, 18.2 or h18d1,h18d2) of Figure 1 of Mueller & Brimacombe(1997a). In each Figure, the relevant elements ofthe rRNA are shown within the semi-transparentcontour of the 30 S subunit.

Foot-prints and cross-links for tRNA at the Aand P sites

The foot-print sites for A and P site-bound tRNA(Moazed & Noller, 1990) are illustrated inFigure 4(a) and (b), respectively. In Figure 4(a), thepale blue ball-and-stick nucleotides depicting the Asite foot-prints can be seen to surround the antico-don stem±loop of the tRNA molecule, with thoseat nucleotides 1492 to 1494 in helix 44.3-44.4 in thecentre, that at nucleotide 1408 (also in helix 44.4)below on the right, and those at nucleotides 529 to531 in the loop end of helix 18 on the left. Only thesub-helices 18.2 and the pseudo-knot helix 18tfrom the helix 18 group are displayed inFigure 4(a); the lower part of this helix (18.1, notshown in the Figure) extends to the left towards toposition of protein S4 (cf. Figure 4(a) and (c) ofMueller & Brimacombe, 1997a). All of the A sitefoot-print positions are thus located in closeproximity to the tRNA, with no allosteric sites, asmentioned above. It was noted by Mueller &Brimacombe (1997a) that there is a sharp bend inthe upper region of helix 44, between sub-helices44.3 and 44.4. The A site foot-prints at nucleotides1492-1494 (Figure 4(a)) form the outer curve of thisbend in the helix, which, again as noted byMueller & Brimacombe (1997a), correlates with theNMR structure derived for this region of helix 44by Fourmy et al. (1996).

The corresponding P site foot-prints (Figure 4(b)),represented by the green ball-and-stick nucleotides,show a cluster of sites (those at nucleotides 1399 to1401 in helix 44.1-44.2, together with that atnucleotide 926 in helix 28) just below the anticodonstem±loop of the tRNA. The site at nucleotide 966in the loop end of helix 31 lies above this lattercluster, penetrating the anticodon stem±loop on itsupper side, with the sites at nucleotides 1338-1339(adjacent to helix 29) just above and to the right,and the site at nucleotide 1381 in the single strandjoining helices 28 and 43 just to the right of thetRNA anticodon loop. Not surprisingly, the foot-print site at position 532 in the loop-end of helix 18is closer to the anticodon loop of the A site tRNA(cf. Figure 4(a)) than it is to the P site moleculeand, as expected from the discussion in the pre-vious section (Figures 2(b) and 3), the remainingfoot-print sites in helices 23 and 24 (at nucleotides693 and 794-795) are remote from the P site tRNA(Figure 4(b), bottom). It is noteworthy that the sitesthat have been found to be essential for binding oftRNA to the P site (Von Ahsen & Noller, 1995),nucleotides 926, 966, 1401 and (sometimes) 1338,

Figure 4. Foot-print sites to A and P site tRNA. The A and P site tRNAs are shown as pale blue and green backbonemodels, respectively, with the mRNA codons at the A and P sites as a white backbone model. (a) The 16 S rRNAhelices 18.2 and 18t, and 44.1 to 44.5, with the A site foot-prints (Table 3) high-lighted as pale blue ball-and-sticknucleotides. (b) The 16 S rRNA helices 18.2 and 18t, 23, 24.3, 28, 29, 31, 43 and 44.1 to 44.5, with the P site foot-printsas green ball-and-stick nucleotides.


are among those that are clustered closely aroundthe anticodon stem±loop of the P site tRNA inFigure 4(b).

The cross-linking data from speci®c positions inthe anticodon loops and elbow regions of A and Psite-bound tRNA are shown in Figure 5. Themajority of these cross-links (summarized byRinke-Appel et al., 1995) were formed from tRNAmolecules carrying a photoreactive group speci®-cally attached to a naturally modi®ed base. The re-sults showed that when the photo-reactive moietywas in the anticodon loop of the tRNA, cross-link-ing occurred exclusively to the 30 S subunit (eitherto protein or to 16 S rRNA) from all three (A, P or

E) sites, and the corresponding cross-linking fromthe CCA end of the molecule was exclusively tothe 50 S subunit. However, with photoreactivelabels in the elbow region, concomitant cross-link-ing to both subunits was observed and, further-more, overlaps were found with the respectivecross-linking data from the anticodon loop andCCA end of the tRNA. Since the bridging distanceof the cross-linking reagents was only 10 or 15 AÊ

(including the side-chains of the naturally modi®ednucleotides to which the reagents were attached),such an overlap of cross-linking sites to the 16 SrRNA from the elbow and anticodon regionsshould not be possible. It therefore seems likely

Figure 5. Site-directed cross-links from A and P site tRNA. In each case the tRNA molecule is shown as a black tube,with the positions of site-directed cross-linking (cf. Table 2) in different colours. The ball-and-stick nucleotides in the16 S rRNA model indicate the cross-linked targets, coloured correspondingly; tRNA position 8 (red), 20:1 (yellow), 32(pale blue), 34 (purple) and 47 (dark blue). (a) The A site tRNA, together with 16 S rRNA helices 23, 28, 29 and 43.(b) The P site tRNA, together with 16 S rRNA helices 22a, 23, 28, 29, 30, 31, 43 and 44.1 to 44.3.


that the tRNA molecules are bound to the ribo-some, at least under the conditions of the cross-linking experiments, in a somewhat ¯exiblemanner, which is re¯ected in the spread of thecross-links from the elbow region of the tRNA inthe 16 S rRNA model (Figure 5). Moreover, in theEM reconstruction used to ®t the 16 S rRNA theareas of density corresponding to the elbow re-gions of both the A and P site tRNAs are not veryclose to any density corresponding to parts of the30 S subunit itself (Figure 5). Again, this suggestssome ¯exibility in the tRNA binding, and has theconsequence that a really satisfactory ®t (withinthe bridging distance of the cross-linking reagents)

can a priori not be realized between the 16 S rRNAmodel and the cross-link data from the elbowregion of the tRNA in this particular con®gurationof the EM reconstruction. Nevertheless, it is note-worthy that the cross-link sites involve the sameregions of the 16 S rRNA (see Figure 1) as thosefound in the foot-printing studies (cf. Figure 4).

For the A site tRNA (Figure 5(a)), cross-linksfrom the anticodon loop (position 32, light blue inFigure 5(a)) were found to nucleotides 936 in thesingle strand connecting helices 28 and 29, and1378 in the corresponding strand connecting he-lices 28 and 43; the two sites lie very close togetherin the model (Figure 5(a), upper right), although it


should be noted that the latter single strand(connecting helices 28 and 43) is 11 bases long andthus potentially very ¯exible. The only other cross-links were from position 20:1 of tRNA from lupinseeds (yellow) to helix 23 (Figure 5(a), lower right;Rinke-Appel et al., 1995). In the case of the P sitetRNA (Figure 5(b)), the cross-links from position32 (light blue) are to nucleotides 693 (helix 23),1338 (adjacent to helix 29), and to 957 and 966 inhelix 31, and these sites correspond very closely tothe foot-print data (Figure 4(b)). The cross-linksfrom the elbow region, positions 8 (red), 20:1(yellow) and 47 (dark blue), are concentrated intwo areas of the model, the one (helices 28, 29, 43)lying above the anticodon loop of the tRNA andthe other (in helix 23) lying beneath (Figure 5(b),lower centre). Thus, again it is the data involvinghelix 23 (particularly the cross-links from positions32 of the P site tRNA and from position 20:1 of theA site tRNA) that are problematic.

The remaining cross-link, namely the well-known direct UV-induced cross-link from pos-ition 34 of the P site tRNA (Prince et al., 1982) isshown in purple, and here the cross-linked targetnucleotide (C-1400) can be seen lying directlybeneath the anticodon loop, close to position 34of the tRNA.

Functional sites close to the A site tRNA

A number of data sets in Table 3 involve foot-print or mutation sites at locations in the vicinityof the A site tRNA, and these are illustrated inFigure 6. The ®rst and most obvious is the set offoot-print sites for initiation factor IF-1, which co-incide exactly with the A site foot-prints (Moazedet al., 1995). The IF-1 foot-prints are coloured pur-ple in Figure 6(a) (cf. Figure 4(a)), and the sameFigure shows the two mutation sites causing trans-lational suppression at nucleotides 517 (Shen &Fox, 1989) and 1054 (Murgola et al., 1988). The for-mer site (in helix 18, yellow) lies very close to theA site anticodon stem±loop and to the IF-1 sites atnucleotides 1492-1493, whereas the latter (in helix34, orange) lies to the left of the anticodon loop,close to the IF-1 foot-print site at nucleotide 530.Thus, sites related both to initiation and termin-ation appear to occupy very similar locations nearthe A site tRNA.

Figure 6(b) shows the mutational and foot-printdata relating to the antibiotics spectinomycin,streptomycin and tetracycline. The spectinomycinfoot-print and resistance sites are on oppositestrands of helix 34 (Figure 1), and these can beseen (as purple ball-and-stick nucleotides) on theupper left in Figure 6(b). The spectinomycin sitesare in helix 34.3, and are therefore somewhatfurther away from the A site tRNA than thesuppressor site at nucleotide 1054 (Figure 6(a)), thelatter being between helices 34.1 and 34.2. Incontrast to spectinomycin, the corresponding foot-print and mutational sites for streptomycin arelocated in two distinct regions of the 16 S rRNA

(Table 3; Figure 1); namely, in helix 18 and in he-lices 1 and 2. These two regions are, however,close together in the 16 S rRNA model, with theresult that the streptomycin sites form a compactcluster (yellow in Figure 6(b)); the resistance sitesin helix 18 are just to the left of the anticodon loopof the A site tRNA, with the resistance and foot-print sites in helices 1 and 2 lying beneath them.Sites of resistance to both spectinomycin and strep-tomycin have been identi®ed in ribosomal proteinS5 (Piepersberg et al., 1975), and in the crystalstructure of the protein (Ramakrishnan & White,1992) the respective sites lie on opposite sides ofthe molecule. In the neutron map (Capel et al.,1988) of the 30 S proteins, ®tted to the EM recon-struction (Mueller & Brimacombe, 1997b), proteinS5 occupies a position just to the left of the clusterof streptomycin sites (in the orientation ofFigure 6(b)), and the atomic structure of the proteincan be oriented so as to bring the spectinomycinand streptomycin sites into the neighbourhood oftheir counterparts in the 16 S rRNA (cf. Figure 6(b));this modelling study will be published elsewhere(cf. also Brimacombe, 1992, 1995).

The sites for streptomycin in Figure 6(b) extenddownwards towards helix 27, in the single strandconnecting helix 2 to the latter (Figure 1; Table 3),and the foot-print for tetracycline (red inFigure 6(b), bottom centre) is located in the corre-sponding single strand connecting helix 19 to helix27. Helix 27 occupies this position in the lowerpart of the body of the 16 S rRNA model by virtueof an intra-RNA cross-link to helix 11 (cross-linkIV, Mueller & Brimacombe, 1997a), and in conse-quence the tetracycline foot-print site is forced intoa location remote from the decoding region. How-ever, in the absence of this cross-linking constraint,the single strands leading from helices 2 and 19 tohelix 27 (Figure 1) would be long enough to enablehelix 27 to approach the decoding area, withoutdisturbing other more rigid areas of the 16 Smodel. Finally in this context, we have notedabove that there are a number of other antibioticswith foot-prints that essentially overlap those ofthe A and P site tRNA molecules (Woodcock et al.,1991).

Cross-links to tRNA at the E site andfoot-prints for IF-3

The site-directed cross-links for E site-boundtRNA are shown in Figure 7(a). It can be seen thatin helix 23 (Figure 7(a), lower centre) there arecross-links from tRNA positions 32 (light blue),20:1 (yellow) and 47 (dark blue), and the proximityof the E site tRNA to helix 23 (and to protein S11)has been discussed above (Figures 2(b) and 3). Theposition of the E site tRNA was determined by itslocation in the EM reconstruction of disomes in thepost-translocational state (Stark et al., 1997a),whereas the location of helix 23 in the 16 S rRNAmodel is determined by the ®ne structure of theEM contour, the position of protein S11, and the

Figure 6. (a) Foot-print sites to IF-1, and mutation sites causing translational suppression (cf. Table 3). The 16 S rRNAhelices 18.2, 18t, 34 and 44.1 to 44.5 are displayed, with coloured ball-and-stick nucleotides denoting IF-1 foot-printsites (purple) and suppression sites in helices 18 and 34 (yellow and orange, respectively). (b) Mutation and foot-printsites for antibiotics. The 16 S rRNA helices 1, 2, 3, 18, 19, 27 and 34 are shown, with ball-and-stick nucleotides denot-ing streptomycin sites (yellow), tetracycline (red) and spectinomycin (purple). In (a) and (b) the A site tRNA isshown as a pale blue tube model.


intra-RNA cross-link between the loop end of helix23 and that of helix 24 (Mueller & Brimacombe,1997a,b). Nevertheless it is clear that a 45� rotationeither of the E site tRNA anticodon stem±loop in aclockwise direction, or of helix 23 in an anticlock-wise direction (as viewed in Figure 7(a)), wouldbring the respective cross-link sites on the tRNAand the 16 S rRNA into close correspondence; thisis another example where there appears to be some¯exibility in the positioning of the tRNA moleculeand/or a conformational heterogeneity in the lo-cation of the helix 23 region. The same applies tothe remaining E site tRNA cross-links from pos-ition 32 to the single strand connecting helices 28and 43 (nucleotides 1376 and 1378; Figure 7(a), topright). These cross-links were speci®c for the E site

(DoÈring et al., 1994), whereas the corresponding Asite cross-link (Table 2; Figure 5(a)) was to position1378 only. As noted above, the single strand con-necting helices 28 and 43 is long and therefore ¯ex-ible in the model, but it is nonetheless obvious thatnucleotide 1378 cannot be simultaneously juxta-posed to both the A and the E site tRNA in a singlestructure. There are no foot-print data for the E sitetRNA on 16 S rRNA.

Figure 7(b) shows the foot-print data for in-itiation factor IF-3 (Table 3), which are located inhelices 23 and 24; the sites in helix 23 lie just abovethose in helix 24 in the Figure. The loop end ofhelix 24 is believed to lie at the 30 S:50 S subunitinterface (Herr et al., 1979), and there is indeed aclose contact to the 50 S subunit in the EM recon-

Figure 7. (a) Site-directed cross-links from E site tRNA. The E site tRNA is shown as a black tube, with positions ofsite-directed cross-links marked yellow (position 20:1), pale blue (32) and dark blue (47). Cross-link sites within the16 S rRNA helices 22a, 23, 28 and 43 (Table 2) are marked by ball-and-stick nucleotides in the corresponding colours.(b) Foot-print sites to IF-3. The 16 S rRNA helices 23 and 24 are displayed, with the IF-3 foot-print sites (Table 3) asred ball-and-stick nucleotides. The P site tRNA is shown as a green tube model for reference.


struction (Stark et al., 1997a) at the position wherethe loop end of helix 24 is located in the 16 S rRNAmodel (discussed by Mueller & Brimacombe,1997a; see also Lata et al., 1996). The foot-print datafor IF-3 in this area would thus be consistent withits known function in dissociating 70 S ribosomes(Subramanian & Davies, 1970). The two partiallylocalized cross-links between IF-3 and 16 S rRNA(to nucleotides 819 to 859 in helices 25 and 26, andto nucleotides 1506 to 1529 in helix 45; Ehresmannet al., 1986) are consistent with a location of IF-3 inthis area (cf. Mueller & Brimacombe, 1997a). Onthe other hand, IF-3 is also important for selectionof the correct initiator codon at the P site (Hartz

et al., 1989), and the P site tRNA is included inFigure 7(b) for reference; as will be evident fromthe data already presented in Figures 2 to 5, the Psite tRNA is remote from the IF-3 foot-print pos-itions.

In contrast to the situation with the ``530 loop''(helix 18) discussed above, we have been unable to®nd a location for helix 23 in the 16 S rRNA modelthat satis®es all of the data sets; there are too manyconsiderations ``pulling in opposite directions'', in-cluding, as already noted, the ®ne structure of theEM reconstruction, the locations of the P and E sitetRNA molecules, and the positions of the riboso-mal proteins, as well as the biochemical data just


described (Figures 2(b), 3, 4(b), 5(b), 7(a) and 7(b)).We believe that this is a situation that cannot beexplained in terms of a static model of the ribo-some, and two dynamic solutions to the problemcan be considered.

The ®rst possibility is that there are multiplecon®gurations for the tRNA at both the P and Esites. Two con®gurations for the E site tRNA havebeen proposed (Robertson & Wintermeyer, 1987),and intermediate states in the movement of thetRNA molecules across the ribosome have been de-scribed by Moazed & Noller (1989). In our owncross-linking studies (DoÈring et al., 1994), slightlydifferent cross-linking patterns were observed for Psite tRNA when the ribosomal complexes wereprepared under different conditions. Moreover, thedata reported by Rheinberger et al. (1986), whichindicate that there is codon-anticodon interactionat the E site, must also be accounted for; as pointedout by Stark et al. (1997a), the location of the E sitetRNA in the EM reconstruction (cf. Figure 3) is toofar away from the P site for codon-anticodon inter-action to occur. It could thus be postulated thatthere are intermediate locations for the P and E sitetRNAs, lying between those de®ned by the EM re-construction. The second possibility is that majorconformational changes occur within the ribosomeitself at different stages of the protein biosyntheticcycle, and such changes have been reported(OÈ fverstedt et al., 1994). In the next section we de-scribe an EM reconstruction, which, although atlower resolution than the reconstructions reportedby Stark et al. (1997a), strongly suggests that bothof the above possibilities contribute to the ``helix 23problem''.

Two conformational states of the 70 S ribosome

The EM reconstructions by Stark et al. (1997a)showed that there was essentially no difference inthe overall morphology of ribosomes in the pre-translocational state (carrying tRNAs at the A andP sites) and those in the post-translocational state(disomes, carrying tRNAs at the P and E sites).However, another reconstruction, made from ribo-somes programmed only with tRNAMet

f at the Psite in the presence of a suitable mRNA analogue(see Materials and Methods) shows a signi®cantlychanged morphology. The EM reconstruction ofthe disomes (Stark et al., 1997a) is compared withthis latter reconstruction, which we refer to as themonosome-P reconstruction, in Figures 8 and 9, re-spectively. In each case, three characteristic pro-teins from the neutron map (Capel et al., 1988) areshown for reference within the 30 S moiety of the70 S reconstruction (cf. Mueller & Brimacombe,1997b); namely, S4 (dark blue), S7 (light blue) andS11 (red). The P site tRNA molecule is included asa green backbone tube model. Two views of eachreconstruction are shown, the ®rst from the ``L1side'' of the 70 S particle (Figures 8(a) and 9(a)),and the second from ``above'' the ribosome(Figures 8(b) and 9(b)). Because of the non-quanti-

tative occupancy of the P site (ca 70%; seeMaterials and Methods), the ribosomal populationis conformationally heterogeneous here, and as aresult the resolution of the monosome-P recon-struction is lower (25 AÊ ) than that of the disomes(21 AÊ ). Nevertheless some signi®cant differencesare immediately apparent. First, the connection be-tween the head of the 30 S subunit and the centralprotuberance of the 50 S subunit in the disomes (tothe left of protein S7 and covering the tRNA mol-ecule in Figure 8(b)) is broken in the monosome-Pribosomes (Figure 9(b)), and here the con®gur-ations of both the 30 S head and the 50 S centralprotuberance are substantially altered. Second, theL1 protuberance (which is the prominent mush-room-shaped object protruding towards the viewerjust to the left of the tRNA molecule in Figures 8(a)and 9(a)) occupies a different position in the tworeconstructions. In the disomes (Figure 8(a)) thereis a clear gap between the L1 protuberance and thehead of the 30 S subunit (also visible in Figure 8(b)),but in the monosome-P reconstruction the L1 pro-tuberance has moved in towards the centre of theribosome, and is now in contact with the head ofthe 30 S subunit (Figure 9(a) and (b)). Furthermore,the whole side lobe of the 30 S subunit (the areasurrounding protein S11 in Figure 8(a)) has alsomoved inwards towards the centre of the ribosomein the monosome-P structure (Figure 9(a)) conco-mitantly with the L1 protuberance, and it is the30 S side lobe that carries the 16 S rRNA helix 23.

Another difference between Figures 8 and 9 isthat the ``knob'' of density in the disome EM recon-struction (Figure 8) corresponding to the elbow re-gion of the P site tRNA (cf. Figures 5(b) and 7(b))has disappeared in Figure 9, where instead a corre-sponding knob (not present in unprogrammed ri-bosomes; data not shown) can be discerned justbelow the L1 protuberance in Figure 9(a), and justabove and behind the latter in Figure 9(b). As dis-cussed by Stark et al. (1997a), the fact that in thethree-dimensional reconstructions density corre-sponding to the P site tRNA is observed only inthe central region of the molecule suggests thatthere is some ¯exibility in the orientation of thistRNA (cf. the foregoing description of the P sitetRNA cross-linking data). In Figure 9 we have ten-tatively modelled the P site tRNA onto this knobof density, and this gives a con®guration for the Psite tRNA that is reminiscent of that reported byAgrawal et al. (1996) in their reconstruction of ribo-somes carrying three tRNA molecules at the A, Pand E sites. Notably, as in the monosome-Preconstruction here, Agrawal et al. (1996) useduncharged tRNA to programme their ribosomes.

The monosome-P reconstruction was made withribosomes from the same batch that was used tomake our original reconstruction of ``empty'' ribo-somes (Stark et al., 1995). The latter reconstructionwas morphologically very similar to the recon-structions of the pre and post-translocationalribosomes by Stark et al. (1997a), and therefore it isthe programming with a single tRNA molecule at

Figure 8. The 70 S ribosome, taken from the EM reconstruction of disomes made by Stark et al. (1997a). The P sitetRNA is shown as a green tube model, together with the neutron positions of proteins S4 (dark blue), S7 (light blue)and S11 (red). (a) View from the solvent side of the 30 S subunit, with the 50 S subunit to the left. (The L1 protuber-ance is the mushroom-shaped object in front and to the left of the elbow of the tRNA). (b) View from ``above" theribosome, with the 30 S subunit on the right (i.e. the view in (a) has been rotated 90� towards the viewer about thehorizontal axis).


the P site that is responsible for the conformationalchanges just noted. Furthermore, the monosome-Preconstruction was notably unstable or ``fuzzy'' injust those regions, the L1 protuberance and the30 S side lobe, where the changes occur. It remainsto be seen whether these changes have anyphysiological signi®cance, but the importance ofthe result is that the P site tRNA cross-linkingexperiments (summarized by Rinke-Appel et al.,1995), as well as numerous other relevant bio-chemical studies, were made with ribosomes underconditions similar to those used for the monosome-P reconstruction. The net result in the monosome-Preconstruction appears to be that the side lobe ofthe 30 S subunit has moved inwards and at thesame time the P site tRNA has moved outwards(Figure 9), relative to the corresponding situationin the disome reconstruction (Figure 8) or in thatused to ®t the 16 S rRNA model (Mueller &Brimacombe, 1997a,b). These movements wouldobviously offer a solution to the helix 23 problem(cf. Figure 3), and could well be relevant to the a-emodel for the translocation reaction proposed byNierhaus et al. (1995). Although this conclusion is

tentative, it underscores the point that, as the resol-ution of the rRNA model improves, differences inthe conditions used for collecting the experimentaldata, or for preparing the ribosomes for EM recon-struction, will increasingly need to be taken intoaccount.

Figures 8(b) and 9(b) show an unaccustomedview of the ribosome, which also serves to demon-strate the importance of building the model for the16 S rRNA in the context of the whole 70 S ribo-some as opposed to the isolated 30 S subunit. Thethree ``marker'' proteins in these Figures (S4, S7and S11) have been chosen because of their charac-teristic positions in the neutron map (Capel et al.,1988) relative to the 30 S moiety of the EM recon-struction, and because the ®t to the RNA-proteininteraction data (Mueller & Brimacombe, 1997b) isparticularly good in all three cases relative to the16 S rRNA model. In the corresponding modelsproposed by Stern et al. (1988) and Fink et al.(1996), the location of helix 23 in the side lobe ofthe 30 S subunit, close to protein S11, is essentiallysimilar to that in our model, apart from its orien-tation in the ``vertical'' direction (discussed by

Figure 9. An EM reconstruction of 70 S monosomes carrying uncharged tRNAMetf , at a resolution of ca 25 AÊ , ``mono-

some-P'' (cf. Stark et al., 1997a). The P site tRNA and ribosomal proteins S4, S7 and S11 are shown as in Figure 8.The two views, (a) and (b), correspond precisely to those in Figure 8(a) and (b). See the text for further explanation.


Mueller & Brimacombe, 1997a). However, Sternet al. (1988) and Fink et al. (1996) locate the P sitetRNA with its anticodon loop at the decodingcentre (as in our model), but with the central folddirected towards protein S11 and helix 23, in orderto accommodate the P site foot-prints in the latterhelix. It is immediately clear from the location ofS11 in Figures 8(b) and 9(b) that this is not a viablearrangement in the context of the 70 S ribosome; itplaces the plane of the P site tRNA at an angle ofmore than 90� relative to its location in Figures 8(b)or 9(b), so that the CCA end of the molecule is di-rected towards the solvent below protein S11 inthese Figures, rather than towards the peptidyltransferase centre in the 50 S subunit.

The path of the mRNA through the 30 S subunit

The site-directed cross-linking data relating tomRNA (Table 1) are illustrated in Figure 10. Thecross-links concerned, with one exception (seeTable 1), were all induced by direct UV-irradiationof thionucleotides incorporated into the mRNA,and are therefore ``zero-length'' cross-links. Inconsequence, the data ®t much more closely to the16 S rRNA model than the corresponding site-di-rected cross-links from tRNA (Figures 5 and 7(a)),which predominantly involved the use of bifunc-tional reagents with an effective bridging distance

of up to 15 AÊ . The cross-links in the upstream re-gion of the mRNA were observed from a range ofmRNA positions between the Shine-Dalgarno se-quence and the AUG initiator codon (Rinke-Appelet al., 1994), and the cross-link formation was onlypartially dependent on the presence of tRNAMet

f .These cross-links are shown in Figure 10(a). Theupstream region of the mRNA (positions ÿ1 to ÿ8;cf. Table 1) is marked in yellow, and the threecross-link sites to the 16 S rRNA at nucleotides 665(in helix 22), 1360 (helix 43) and 1530 (close to helix45) are shown as yellow ball-and-stick nucleotides.The cross-link at nucleotide 1360 is above themRNA in Figure 10(a), and the other two arebelow, with that at nucleotide 1530 being on theleft.

The ®nding that the upstream cross-links canoccur from a range of mRNA positions suggeststhat this part of the mRNA is quite ¯exible. In con-trast, the downstream cross-links were highlyspeci®c with regard to the mRNA position, and allof the cross-links are to universally conservednucleotides in the 16 S rRNA (Rinke-Appel et al.,1993; Sergiev et al., 1997). Furthermore, all of thedownstream cross-links were entirely dependenton the presence of tRNAMet

f . Three of the cross-linksites concerned, those to nucleotides 532, 1196 and1395 (Table 1), have been observed by Bhangu et al.(1994), although these authors were not able to

Figure 10. Site-directed cross-links from mRNA (cf. Table 1). The P site tRNA is shown as a green tube model. ThemRNA is shown as a white tube with the positions of site-directed cross-linking marked in different colours; the tar-gets of cross-linking in the 16 S rRNA are displayed as ball-and-stick nucleotides, coloured correspondingly.(a) Cross-links from the upstream region of mRNA. mRNA positions ÿ1 to ÿ8 are coloured yellow. The 16 S rRNAhelices 22, 43 and 45 are shown, with the cross-link sites as yellow ball-and-stick nucleotides. (b) Cross-links from thedownstream region of the mRNA, without the semi-transparent EM contour. The mRNA positions marked are �2(purple), �4 (light blue), � 6 (dark blue), �7 (green), �8 and �9 (yellow), �11 (orange) and �12 (red). The 16 SrRNA helices 18.2, 18t, 28, 34 and 44.1 to 44.5 are shown, with the cross-link sites as coloured ball-and-stick nucleo-tides. (c) The same as (b), rotated ca 50� about the vertical axis, and with the EM contour of the 30 S subunit added.



identify the precise mRNA positions involved inthe cross-links. The positions of the downstreammRNA cross-links in the 16 S rRNA model areshown in Figure 10(b) and (c). In Figure 10(b) theEM contour is omitted for clarity, and the cross-link sites can be seen (from left to right) as follows.The red cross-link is from mRNA position �12 tonucleotide 530 (in helix 18), orange is position �11to nucleotide 532 (helix 18), yellow �8 or �9 to1196 (helix 34), green �7 to 1395 (close to helix 28),dark blue �6 to 1052 (helix 34), light blue �4 to1402 (helix 44), and purple �2 to 926 (helix 28).The cross-link from position �2 was formed usinga 10 AÊ bifunctional reagent attached to a modi®eduridine residue in the AUG codon (Sergiev et al.,1997).

At the current resolution of the model, the cross-link sites from mRNA (Figure 10(b)) ®t well totheir corresponding targets on the 16 S rRNA, withthe exception of the cross-links (yellow and darkblue) in helix 34 (from mRNA positions �6, �8,�9). Helix 34 lies in the head of the 30 S subunit,and from the foregoing descriptions (e.g. Figures4(a) and 6) it is clear that in the functional complexof tRNA and mRNA the anticodon stem±loop ofthe A site tRNA is interposed between the mRNAand the helix 34 region (Figure 6). In the presenceof a tRNA at the A site, mRNA contacts to helix 34would therefore not be expected (for furtherdiscussion of this point, see Brimacombe, 1995).Indeed, we have reported (Rinke-Appel et al., 1993)that the cross-linking from mRNA positions �4,�6 and �7 (in the region of the A site codon, pos-itions �4 to �6) is abolished by the presence of atRNA at the A site. In Figure 10(b) the mRNA pos-ition is modelled as if an A site tRNA was present(i.e. as in Figure 4(a)), and the cross-linking datathus indicate that, in the absence of an A sitetRNA, the mRNA in the �6 to �9 region must be¯exible enough to loop up towards helix 34(Figure 10(b)). (Alternatively, a conformationalchange in the 16 S rRNA could allow a comp-lementary movement of helix 34 downwardstowards the mRNA.) In contrast, the cross-linksfurther out along the mRNA at positions �11 and�12 were not affected by A site tRNA binding (e.g.see Rinke-Appel et al., 1993). It is noteworthy thatboth the upstream and downstream regions of themRNA as depicted in Figure 10(a) and (b) appearto be in an almost fully extrended con®guration.

It has been proposed (Stark et al., 1995; Franket al., 1995) that the incoming (i.e. downstream)mRNA passes through the ``hole'' in the 30 S sub-unit at the head-body junction. The hole can beseen in Figure 3, lying behind and between the an-ticodon stem±loops of the A and P site tRNAs,and the details of the mRNA cross-links in this re-gion are shown in Figure 10(c), which illustratesthe same data as Figure 10(b) (rotated ca 50� aboutthe vertical axis) but with the EM contour added.It can be seen that our 16 S rRNA model does in-deed make the very clear prediction that themRNA passes through the hole in the 30 S subunit.

Furthermore, the fact that the mRNA cross-linkingdata extend out to position �12 is in excellentqualitative agreement with the toe-printing resultsreported by Hartz et al. (1989), which indicate thatthe ribosome has a sharp ``boundary'' correspond-ing to mRNA position �15. However, at the sametime this mRNA arrangement immediately raises aphilosophical problem. Namely, the initial contactbetween an mRNA molecule and the ribosome isthe formation of the Shine-Dalgarno interaction,which involves the upstream region of the mRNA(cf. Figure 10(a)). It therefore seems inconceivablethat, prior to forming the Shine-Dalgarno inter-action, the mRNA should have to thread its waythrough a hole in the 30 S subunit. We propose asolution to this apparent dilemma (cf. Sergiev et al.,1997), which is essentially the same as that recentlysuggested by Lata et al. (1996), as follows.

The hole in the 30 S subunit, through which themRNA passes, arises from the presence of twohead-body contact regions. According to the 16 SrRNA model (Mueller & Brimacombe, 1997a) themain head-body contact (on the left in Figure 10(c),or to the right of the hole in Figure 3) representsthe covalent connection within the rRNA (helix 28)joining the head and body regions of the 16 S mol-ecule. On the other hand, the minor head-bodycontact (on the right in Figure 10(c), or to the leftin Figure 3) is not covalent and must be formedeither by tertiary interactions within the 16 SrRNA, or by RNA-protein interactions. The thick-ness of the latter contact is seen to vary betweendifferent EM reconstructions (cf. Stark et al., 1995,1997a), and we postulate that it can transientlyopen. Thus, after formation of the Shine-Dalgarnointeraction, the downstream region of the mRNAwould wrap around the 30 S subunit so as to ®ndits contacts with the 16 S rRNA (Figure 10(b) and(c)), and the right-hand head-body junction wouldthen close, clamping the mRNA into position. Sucha mechanism would ensure that the incomingmRNA is precisely positioned as it approaches thedecoding area, and is consistent with the conceptof a ``stand-by'' site for the mRNA (Canonaco et al.,1989; La Teana et al., 1995), whereby the mRNAbecomes partially relocated under the in¯uence ofthe initiation factors. The proposed mechanism isfurthermore closely analogous to the opening andclosing ring or clamp structures that have been ob-served in DNA polymerase III (Herendeen &Kelly, 1996), E. coli core RNA polymerase(Polyakov et al., 1995) and reverse transcriptase(Jacobo-Molina et al., 1993); this may thus representa general principle for the ``reading'' of nucleicacid sequences.

Conclusions

The primary reason for attempting to buildthree-dimensional models for the ribosomal RNAis to show to what extent the available data, bothstructural and functional, can be accommodated in


a single structure, and to identify experimentalfacts or interpretations that are in con¯ict with oneanother. The advent of the EM reconstructions at20 AÊ resolution (Stark et al., 1997a) has added aseries of rigorous constraints on the folding of therRNA molecules, and has for the ®rst timeprovided a real physical basis on which theserRNA models can be built. As we have alreadyemphasized (Mueller & Brimacombe, 1997a), at thecurrent level of resolution the 16 S rRNA modelhas not been ®nalized, and many improvementsand modi®cations will surely need to be made asmore biochemical data are accumulated and theresolution of the EM structures improves. Never-theless, it is encouraging that the adaptation of ourprevious ``cartoon'' model of the 16 S rRNA (cf.Brimacombe, 1995) so as to satisfy the EM con-straints has, in general, led to an equally good oreven improved ®t to the biochemical data, ratherthan to a more strained situation. There are ofcourse exceptions, among which the helix 23 pro-blem described here (Figures 3, 4 and 5) is themost prominent. In this case, the location of thehelix concerned did not appear to present anydif®culty in models of the isolated 30 S subunit(Brimacombe et al., 1988; Stern et al., 1988;Malhotra & Harvey, 1994; Fink et al., 1996), wherethe binding sites for the tRNA molecules were in-ferred as opposed to being directly visualized; theproblem becomes apparent only when the 16 SrRNA model is considered in the light of the directlocalization of the P site tRNA in the EM recon-struction of the complete 70 S ribosome (Stark et al.,1997a).

The helix 23 problem has raised the question ofmajor conformational changes during the riboso-mal cycle (see Figures 8 and 9), and in fact the EMimage reconstruction approach is particularly wellsuited to the investigation of the ribosome inspeci®c functional states. This has already beenexploited in the reconstruction of pre and post-translocational ribosomes (Stark et al., 1997a), andin the visualization of the EF-Tu/tRNA complex inkirromycin-stalled ribosomes (Stark et al., 1997b).Thus, the positions of the ribosomal ligands, aswell as conformational changes or heterogeneitieswithin the ribosome itself, can in principle beobserved at each stage of the process. As the meth-odology improves, it should become possible to di-rectly test concepts such as that proposed here forthe opening and closing of the mRNA track at theinitiation of protein synthesis. At the same time,these developments have the obvious consequencethat it is no longer possible to consider the ribo-some as a static object, and in future it will becomenecessary to correlate speci®c sets of biochemicaldata with individual EM reconstructions of theribosome in a particular state, and with corre-sponding con®gurations of the 16 S rRNA model.In this context it is important to note that there areno apparent theoretical limits to the resolution thatmight be achieved by the EM image reconstructiontechnique (Henderson, 1995) and, as has been

pointed out by Moore (1995), the application of theEM methodology is therefore likely to be morefruitful than the X-ray crystallographic approachfor some time to come.

So far we have concentrated our attention on themodelling of the 16 S rRNA within the 30 S sub-unit. However, the 50 S subunit shows a verystable and detailed ®ne structure in the EM recon-structions, and since the publication of our ®rstcartoon model of the 23 S rRNA (Mitchell et al.,1990) a considerable amount of new biochemicalinformation has been collected (for a review, seeBrimacombe, 1995). A new model for the 23 SrRNA, and hence for the complete rRNA contentof the bacterial ribosome, is the next challenge.

Materials and Methods

Preparation of 70 S ribosomes carrying tRNA at theP site

tRNAMetf (from Sigma) was labelled with 32P at the 50

terminus by the procedure of Gnirke et al. (1989) andwas bound to E. coli 70 S tight-couple salt-washed ribo-somes in the presence of the synthetic mRNA analoguemRNA 3 (DoÈring et al., 1994), which contains both aShine-Dalgarno sequence and an AUG initiator codon.The reaction mixtures contained 120 pmol of 70 S ribo-somes together with a fourfold molar excess of mRNAand a threefold molar excess of tRNA in 160 ml of 20 mMHepes-KOH (pH 7.5), 7.5 mM MgCl2 and 60 mM NH4Cl.Incubation was for 15 minutes at 37�C. Aliquots (20 ml)were assayed by nitrocellulose ®ltration, which indicateda tRNA binding level of 65 to 70%.

Electron microscopy and image processing

All of the relevant procedures have been described byStark et al. (1997a,b).

Computer graphics and rRNA modelling

All of the relevant procedures have been described byMueller & Brimacombe (1997a).

Acknowledgements

We thank all members, past and present, of our re-search groups, whose work in collecting the biochemicaldata and developing the electron microscope method-ology have contributed to this study. Our thanks are dueto the laboratory of Drs A. Bogdanov and O. Dontsova(Moscow) and to Dr D. Bochkariov (California) for thecollaborations that have played an important role in thecross-linking studies. We are grateful to the Max-Planck-Gesellschaft for providing the computer graphics hard-ware. Parts of the work have been supported by theDeutsche Forschungsgemeinschaft (grant nos HE 2162/1-1 and BR 632/3-2). We are grateful to Frau RenateSpann for her patient help in preparing the manuscripts.


References

Agrawal, R. K., Penczek, P., Grassucci, R. A., Li, Y.,Leith, A., Nierhaus, K. H. & Frank, J. (1996). Directvisualization of A-, P-, and E-site transfer RNAs inthe E. coli ribosome. Science, 271, 1000±1002.

Bhangu, R., Juzumiene, D. & Wollenzien, P. (1994).Arrangement of mRNA on E. coli ribosomes withrespect to 10 16 S rRNA cross-linking sites. Biochem-istry, 33, 3063±3070.

Brimacombe, R. (1991). RNA-protein interactions in theE. coli ribosome. Biochimie, 73, 927±936.

Brimacombe, R. (1992). Structure-function correlations(and discrepancies) in the 16 S ribosomal RNA fromE. coli. Biochimie, 74, 319±326.

Brimacombe, R. (1995). The structure of ribosomal RNA;a three-dimensional jigsaw puzzle. Eur. J. Biochem.230, 365±383.

Brimacombe, R., Atmadja, J., Stiege, W. & SchuÈ ler, D.(1988). A detailed model of the three-dimensionalstructure of E. coli 16 S ribosomal RNA in situ inthe 30 S subunit. J. Mol. Biol. 199, 115±136.

Canonaco, M. A., Gualerzi, C. O. & Pon, C. L. (1989).Alternative occupancy of a dual ribosomal bindingsite by mRNA affected by translational initiationfactors. Eur. J. Biochem. 182, 501±506.

Capel, M. S., Kjeldgaard, M., Engelman, D. M. & Moore,P. B. (1988). Positions of S2, S13, S16, S17, S19 andS21 in the 30 S ribosomal subunit of E. coli. J. Mol.Biol. 200, 65±87.

Dontsova, O., Dokudovskaya, S., Kopylov, A.,Bogdanov, A., Rinke-Appel, J., JuÈ nke, N. &Brimacombe, R. (1992). Three widely separated pos-itions in the 16 S RNA lie in or close to the riboso-mal decoding region; a site-directed cross-linkingstudy with mRNA analogues. EMBO J. 11, 3105±3116.

DoÈring, T., Mitchell, P., Osswald, M., Bochkariov, D. &Brimacombe, R. (1994). The decoding region of 16 SRNA; a cross-linking study of the ribosomal A, Pand E sites using tRNA derivatized at position 32in the anticodon loop. EMBO J. 13, 2677±2685.

Easterwood, T. R., Major, F., Malhotra, A. & Harvey,S. C. (1994). Orientations of transfer RNA in theribosomal A and P sites. Nucl. Acids Res. 22, 3779±3786.

Ehresmann, C., Moine, H., Mougel, M., Dondon, J.,Grunberg-Manago, M., Ebel, J. P. & Ehresmann, B.(1986). Cross-linking of initiation factor IF-3 to E. coli30 S ribosomal subunit by trans-diamminedichloro-platinum (II); characterization of two cross-link sitesin 16 S rRNA; a possible way of functioning forIF-3. Nucl. Acids Res. 14, 4803±4821.

Fink, D. L., Chen, R. O., Noller, H. F. & Altman, R. B.(1996). Computational methods for de®ning theallowed conformational space of 16 S rRNA basedon chemical foot-printing data. RNA, 2, 851±866.

Fourmy, D., Recht, M. I., Blanchard, S. C. & Puglisi, J. D.(1996). Structure of the A site of E. coli 16 S riboso-mal RNA complexed with an aminoglycosideantibiotic. Science, 274, 1367±1371.

Frank, J., Zhu, J., Penczek, P., Li, Y., Srivastava, S.,Verschoor, A., Radermacher, M., Grassucci, R.,Lata, R. K. & Agrawal, R. K. (1995). A model ofprotein synthesis based on cryo-electron microscopyof the E. coli ribosome. Nature, 376, 441±444.

Gnirke, A., GeigenmuÈ ller, U., Rheinberger, H. J. &Nierhaus, K. H. (1989). The allosteric three-site

model for the ribosomal elongation cycle. J. Biol.Chem. 264, 7291±7301.

Harris, E. H., Burkhart, B. D., Gillham, N. W. &Boynton, J. E. (1989). Antibiotic resistance mutationsin the chloroplast 16 S and 23S rRNA genes of C.reinhardtii; correlation of genetic and physical mapsof the chloroplast genome. Genetics, 123, 281±292.

Hartz, D., McPheeters, D. S. & Gold, L. (1989). Selectionof the initiator tRNA by E. coli initiation factors.Genes Dev. 3, 1899±1912.

Henderson, R. (1995). The potential and limitations ofneutrons, electrons and X-rays for atomic resolutionmicroscopy of unstained biological molecules.Quart. Rev. Biophys. 28, 171±193.

Herendeen, D. R. & Kelly, T. J. (1996). DNA polymeraseIII; running rings around the fork. Cell, 84, 5±8.

Herr, W., Chapman, N. M. & Noller, H. F. (1979).Mechanism of ribosomal subunit association; dis-crimination of speci®c sites in 16 S RNA essentialfor association activity. J. Mol. Biol. 130, 433±449.

Jacobo-Molina, A., Ding, J. P., Nanni, R. G., Clark, A. D.,Lu, X., Tantillo, C., Williams, R. L., Kamer, G.,Ferris, A. L., Clark, P., Hizi, A., Hughes, S. H. &Arnold, E. (1993). Crystal structure of humanimmunode®ciency virus type I reverse transcriptasecomplexed with double-stranded DNA at 3.0 AÊ res-olution shows bent DNA. Proc. Natl Acad. Sci. USA,90, 6320±6324.

Lata, K. R., Agrawal, R. K., Penczek, P., Grassucci, R.,Zhu, J. & Frank, J. (1996). Three-dimensional recon-struction of the E. coli 30 S ribosomal subunit in ice.J. Mol. Biol. 262, 43±52.

La Teana, A., Gualerzi, C. O. & Brimacombe, R. (1995).From stand-by to decoding site. Adjustment of themRNA on the 30 S ribosomal subunit under the in-¯uence of the initiation factors. RNA, 1, 772±782.

Lim, V. I. (1997). Analysis of interactions between thecodon-anticodon duplexes within the ribosome;their role in translation. J. Mol. Biol. 266, 877±890.

Lim, V., Venclovas, C., Spirin, A., Brimacombe, R.,Mitchell, P. & Mueller, F. (1992). How are tRNAsand mRNA arranged in the ribosome? An attemptto correlate the stereochemistry of the tRNA-mRNAinteraction with constraints imposed by the riboso-mal topography. Nucl. Acids Res. 20, 2627±2637.

Malhotra, A. & Harvey, S. C. (1994). A quantitativemodel of the E. coli 16 S RNA in the 30 S ribosomalsubunit. J. Mol. Biol. 240, 308±340.

MelancËon, P., Lemieux, C. & Brakier-Gingras, L. (1988).A mutation in the 530 loop of E. coli 16 S ribosomalRNA causes resistance to streptomycin. Nucl. AcidsRes. 16, 9631±9639.

Mitchell, P., Osswald, M., SchuÈ ler, D. & Brimacombe, R.(1990). Selective isolation and detailed analysis ofintra-RNA cross-links induced in the large riboso-mal subunit of E. coli; a model for the tertiary struc-ture of the tRNA binding domain in 23S RNA.Nucl. Acids Res. 18, 4325±4333.

Moazed, D. & Noller, H. F. (1987). Interaction of anti-biotics with functional sites in 16 S ribosomal RNA.Nature, 327, 389±394.

Moazed, D. & Noller, H. F. (1989). Intermediate states inthe movement of transfer RNA in the ribosome.Nature, 342, 142±148.

Moazed, D. & Noller, H. F. (1990). Binding of tRNA tothe ribosomal A and P sites protects two distinctsets of nucleotides in 16 S rRNA. J. Mol. Biol. 211,135±145.


Moazed, D., Samaha, R. R., Gualerzi, C. & Noller, H. F.(1995). Speci®c protection of 16 S rRNA by transla-tional initiation factors. J. Mol. Biol. 248, 207±210.

Montandon, P. E., Nicolas, P., SchuÈ rmann, P. & Stutz, E.(1985). Streptomycin resistance of E. gracilis chloro-plasts; identi®cation of a point mutation in the 16 SrRNA gene in an invariant position. Nucl. Acids Res.13, 4299±4210.

Moore, P. B. (1995). Ribosomes seen through a glass lessdarkly. Structure, 3, 851±852.

Mueller, F. & Brimacombe, R. (1997a). A new model forthe three-dimensional folding of E. coli 16 S riboso-mal RNA: I. Fitting the RNA to a 3D electronmicroscopic map at 20 AÊ . J. Mol. Biol. 271, 524±544.

Mueller, F. & Brimacombe, R. (1997b). A new model forthe three-dimensional folding of Escherichia coli 16 Sribosomal RNA. II. The RNA protein interactiondata. J. Mol. Biol. 271, 545±565.

Muralikrishna, P. & Wickstrom, E. (1989). E. coli in-itiation factor 3 protein binding to 30 S ribosomalsubunits alters the accessibility of nucleotideswithin the conserved central region of 16 S rRNA.Biochemistry, 28, 7505±7510.

Murgola, E. J., Hijazi, K. A., GoÈringer, H. U. &Dahlberg, A. E. (1988). Mutant 16 S ribosomalRNA: a codon-speci®c translational suppressor.Proc. Natl Acad. Sci. USA, 85, 4162±4165.

Nierhaus, K. H., Beyer, D., Dabrowski, M., SchaÈ fer,M. A., Spahn, C. M. T., Wadzack, J., Bittner, J. U.,Burkhardt, N., Diedrich, G., JuÈ nemann, R., Kamp,D., Voss, H. & StuÈ hrmann, H. (1995). The elonga-ting ribosome; structural and functional aspects.Biochem. Cell. Biol. 73, 1011±1021.

Nissen, P., Kjeldgaard, M., Thirup, S., Polekhina, G.,Reshetnikova, L., Clark, B. F. C. & Nyborg, J.(1995). Crystal structure of the ternary complex ofPhe-tRNAPhe, EF-Tu and a GTP analog. Science, 270,1464±1472.

Noller, H. F., Moazed, D., Stern, S., Powers, T., Allen,P. N., Robertson, J. M., Weiser, B. & Triman, K.(1990). Structure of rRNA and its functional inter-actions in translation. In The Ribosome; Structure,Function and Evolution (Hill, W. E., Dahlberg, A. E.,Garrett, R. A., Moore, P. B., Schlessinger, D. &Warner, J. R., eds), pp. 73±92, ASM Press, Washing-ton, DC.

Oakes, M., Henderson, E., Scheinman, A., Clark, M. &Lake, J. A. (1986). Ribosome structure, function andevolution. mapping ribosomal RNA, proteins andfunctional sites in three-dimensions. In Structure,Function and Genetics of Ribosomes (Hardesty, B. &Kramer, G., eds), pp. 47±67, Springer-Verlag, NewYork.

Ofengand, J., Ciesiolka, J., Denman, R. & Nurse, K.(1986). Structural and functional interactions of thetRNA-ribosome complex. In Structure, Function andGenetics of Ribosomes (Hardesty, B. & Kramer, G.,eds), pp. 473±494, Springer-Verlag, New York.

OÈ fverstedt, L. G., Zhang, K., Tapio, S., Skoglund, U. &Isaaksson, L. A. (1994). Starvation in vivo foraminoacyl-tRNA increases the spatial separationbetween the two ribosomal subunits. Cell, 79, 629±638.

Olson, H. M., Olah, T. V., Cooperman, B. S. & Glitz,D. G. (1988). Immune electron microscopic localiz-ation of dinitrophenol-modi®ed ribosomal proteinS19 in reconstituted E. coli 30 S subunits using anti-bodies to dinitrophenol. J. Biol. Chem. 263, 4801±4806.

Osswald, M., DoÈring, T. & Brimacombe, R. (1995). Theribosomal neighbourhood of the central fold oftRNA; cross-links from position 47 of tRNA locatedat the A, P or E site. Nucl. Acids Res. 23, 4635±4641.

Paulsen, H., Robertson, H. J. & Wintermeyer, W. (1983).Topological arrangement of two transfer RNAs onthe ribosome. Fluorescence energy transfer measure-ments between A and P site-bound tRNAPhe. J. Mol.Biol. 167, 411±426.

Piepersberg, W., BoÈck, A., Yaguchi, M. & Wittmann,H. G. (1975). Genetic position and amino acid repla-cement of several mutations in ribosomal protein S5from E. coli. Mol. Gen. Genet. 143, 43±52.

Polyakov, A., Severinova, E. & Darst, S. A. (1995).Three-dimensional structure of E. coli core polymer-ase; promoter binding and elongation confor-mations of the enzyme. Cell, 83, 365±373.

Powers, T. & Noller, H. F. (1991). A functional pseudo-knot in 16 S ribosomal RNA. EMBO J. 10, 2203±2214.

Powers, T. & Noller, H. F (1994). Selective perturbationof G530 of 16 S rRNA by translational miscodingagents and a streptomycin-dependence mutation inprotein S12. J. Mol. Biol. 235, 156±172.

Powers, T. & Noller, H. F (1995). Hydroxyl radical foot-printing of ribosomal proteins on 16 S rRNA. RNA,1, 194±209.

Prince, J. B., Taylor, B. H., Thurlow, D. L., Ofengand,J. & Zimmermann, R. A (1982). Covalent cross-link-ing of tRNAVal

l to 16 S RNA at the ribosomal P site;identi®cation of cross-linked residues. Proc. NatlAcad. Sci. USA, 79, 5450±5454.

Ramakrishnan, V. & White, S. W (1992). The structureof ribosomal protein S5 reveals sites of interactionwith 16 S rRNA. Nature, 358, 768±771.

Rheinberger, H. J., Sternbach, H. & Nierhaus, K. H.(1986). Codon-anticodon interaction at the riboso-mal E site. J. Biol. Chem. 261, 9140±9143.

Rinke-Appel, J., JuÈ nke, N., Brimacombe, R.,Dokudovskaya, S., Dontsova, O. & Bogdanov, A.(1993). Site-directed cross-linking of mRNA ana-logues to 16 S ribosomal RNA; a complete scan ofcross-links from all positions between �1 and �16on the mRNA, downstream from the decoding site.Nucl. Acids Res. 21, 2853±2859.

Rinke-Appel, J., JuÈ nke, N., Brimacombe, R., Lavrik, I.,Dokudovskaya, S., Dontsova, O. & Bogdanov, A.(1994). Contacts between 16 S ribosomal RNA andmRNA, within the spacer region separating theAUG initiator codon and the Shine-Dalgarnosequence; a site-directed cross-linking study. Nucl.Acids Res. 22, 3018±3025.

Rinke-Appel, J., JuÈ nke, N., Osswald, M. & Brimacombe,R. (1995). The ribosomal environment of tRNA;cross-links to rRNA from positions 8 and 20:1 inthe central fold of tRNA located at the A, P or Esite. RNA, 1, 1018±1028.

Robertson, J. M. & Wintermeyer, W. (1987). Mechanismof ribosomal translocation; tRNA binds transientlyto an exit site before leaving the ribosome duringtranslocation. J. Mol. Biol. 196, 525±540.

Sergiev, P., Lavrik, I., Wlasoff, V., Dokudovskaya, S.,Dontsova, O., Bogdanov, A. & Brimacombe, R.(1997). The path of mRNA through the bacterialribosome; a site-directed cross-linking study withnew photoreactive derivatives of guanosine anduridine. RNA, 3, 464±475.

Shen, Z. & Fox, T. D. (1989). Substitution of an invariantnucleotide at the base of the highly conserved ``530-


loop'' of 15S rRNA causes suppression of yeastmitochondrial ochre mutations. Nucl. Acids Res. 17,4535±4539.

Sigmund, C. D., Ettayebi, M. & Morgan, E. A. (1984).Antibiotic resistance mutations in 16 S and 23S ribo-somal RNA genes of E. coli. Nucl. Acids Res. 12,4653±4663.

Smith, D. & Yarus, M. (1989). tRNA-tRNA interactionswithin cellular ribosomes. Proc. Natl Acad. Sci. USA,86, 4397±4401.

Stark, H., Mueller, F., Orlova, E. V., Schatz, M., Dube,P., Erdemir, T., Zemlin, F., Brimacombe, R. & vanHeel, M. (1995). The 70 S E. coli ribosome at 23 AÊ

resolution; ®tting the ribosomal RNA. Structure, 3,815±821.

Stark, H., Orlova, E. V., Rinke-Appel, J., JuÈ nke, N.,Mueller, F., Rodnina, M., Wintermeyer, W.,Brimacombe, R. & van Heel, M. (1997a). Arrange-ment of tRNAs in pre- and post-translocationalribosomes revealed by electron cryomicroscopy.Cell, 88, 19±28.

Stark, H., Rodnina, M., Rinke-Appel, J., Brimacombe, R.,Wintermeyer, W. & Van Heel, M. (1997b). The tern-ary complex elucidated in the ribosomal A site byelectron cryomicroscopy. Nature, in press.

Stern, S., Weiser, B. & Noller, H. F. (1988). Model forthe three-dimensional folding of 16 S ribosomalRNA. J. Mol. Biol. 204, 447±481.

StoÈ f¯er, G. & StoÈ f¯er-Meilicke, M. (1986). Immuno elec-tron microscopy on E. coli ribosomes. In Structure,

Function and Genetics of Ribosomes (Hardesty, B. &Kramer, G., eds), pp. 28±46, Springer-Verlag, NewYork.

Subramanian, A. R. & Davis, B. D. (1970). Activity of in-itiation factor 3 in dissociating E. coli ribosomes.Nature, 228, 1273±1275.

Von Ahsen, U. & Noller, H. F. (1995). Identi®cation ofbases in 16 S rRNA essential for tRNA binding atthe 30 S ribosomal P site. Science, 267, 234±237.

Woodcock, J., Moazed, D., Cannon, M., Davies, J. &Noller, H. F. (1991). Interaction of antibiotics withA- and P-site speci®c bases in 16 S ribosomal RNA.EMBO J. 10, 3099±3103.

Wower, J., Hixson, S. S. & Zimmermann, R. A. (1989).Labelling the peptidyl transferase center of theE. coli ribosome with photoreactive tRNAPhe deriva-tives containing azidoadenosine at the 30-end of theacceptor arm; a new model of the tRNA-ribosomecomplex. Proc. Natl Acad. Sci. USA, 86, 5232±5236.

Wower, J., Scheffer, P., Sylvers, L. A., Wintermeyer,W. & Zimmermann, R. A. (1993a). Topography ofthe E site on the E. coli ribosome. EMBO J. 12, 617±623.

Wower, J., Sylvers, L. A., Rosen, K. V., Hixson, S. S. &Zimmermann, R. A. (1993b). A model of the tRNAbinding sites on the E. coli ribosome. In The Transla-tional Apparauts; Structure, Function, Regulation, Evol-ution (Nierhaus, K. H., Franceschi, F., Subramanian,A. P., Erdmann, V. A. & Wittmann-Liebold, B.,eds), pp. 455±464, Plenum Press, New York.

Edited by D. E. Draper

(Received 20 March 1997; received in revised form 18 June 1997; accepted 18 June 1997)

A new model for the three-dimensional folding of Escherichia coli 16 s ribosomal RNA. III. The...

Documents

Transcript of A new model for the three-dimensional folding of Escherichia coli 16 s ribosomal RNA. III. The...