Structural evidence for specific S8–RNA and S8–proteininteractions within the 30S ribosomal subunit: ribosomal protein S8 from Bacillus stearothermophilus at 1.9 Å resolutionChristopher Davies1, V Ramakrishnan2* and Stephen W White1*
Background: Prokaryotic ribosomal protein S8 is an important RNA-bindingprotein that occupies a central position within the small ribosomal subunit. Itinteracts extensively with 16S rRNA and is crucial for the correct folding of thecentral domain of the rRNA. S8 also controls the synthesis of several ribosomalproteins by binding to mRNA. It binds specifically to very similar sites in the twoRNA molecules.
Results: S8 is divided into two tightly associated domains and contains threeregions that are proposed to interact with other ribosomal components: twopotential RNA-binding sites, and a hydrophobic patch that may interact with acomplementary hydrophobic region of S5. The N-terminal domain fold is found inseveral proteins including two that bind double-stranded DNA.
Conclusions: These multiple RNA-binding sites are consistent with the role ofS8 in organizing the central domain and agree with the latest models of the 16SRNA which show that the S8 location coincides with a region of complicatednucleic-acid structure. The presence in a wide variety of proteins of a regionhomologous to the N-terminal domain supports the idea that ribosomal proteinsmust represent some of the earliest protein molecules.
IntroductionDuring the past five years, new insights into the architec-ture, mechanism and evolution of the ribosome have beenprovided by detailed structures of individual ribosomalproteins [1]. Of some 50 proteins in the prokaryotic ribo-some, the structures of nine have now been determined,six from the large 50S subunit [2–7] and three from thesmall 30S subunit [8–10]. The structures provide high-resolution landmarks within the ribosome and they are alsonow being used as detailed probes of local ribosomal(r)RNA structure [11]. The eventual aim of this work is thereconstruction of the whole ribosome structure using theimproving three-dimensional (3D) models of the rRNA[12] and the increasingly detailed views of the particleafforded by electron microscopy [12,13] and X-ray crystal-lography [14]. This task has been facilitated by the abilityto clone and overexpress the genes for ribosomal proteins[15] which has allowed the more important proteins to betargeted for structural analysis.
S8 is a medium-sized ribosomal protein, and its role as an important primary RNA-binding protein in the 30Ssubunit has been known for some time [16,17]. It is crucialfor the correct folding of the central domain of 16S rRNA[18,19], and mutations within the protein have beenshown to result in defective ribosome assembly [20].Neutron diffraction studies [21] and protein–proteincrosslinking data [22] have shown that the neighbours of
S8 in the 30S subunit are S2, S4, S5, S12, S15 and S17. InEscherichia coli, the S8-binding site within 16S rRNA hasbeen independently investigated by several techniquesincluding nuclease protection [18,23], RNA modification[24], RNA–protein crosslinking [25–27], chemical probing[19] and hydroxyl-radical footprinting [28]. All agree thatS8 binds to an extended stem-loop structure comprisingnucleotides 583–653, which is highly conserved [29]. Thissite has been extensively analyzed by a number of groups,and S8 appears to recognize a small internal loop atnucleotides 596–597/641–643 [30–36]. Fragments of 16SrRNA containing this feature also bind S8 [34,35].
S8 is also one of the principal regulatory elements thatcontrols ribosomal protein synthesis by the translationalfeedback inhibition mechanism discovered by Nomuraand colleagues [37]. S8 regulates the expression of the spcoperon that encodes, in order, the ten ribosomal proteinsL14, L24, L5, S14, S8, L6, L18, S5, L30 and L15. The S8protein binds specifically to a region of the polycistronicspc transcript near the start of the L5 gene [35,39]. Thisregion of the mRNA is very similar to the S8-binding siteon 16S rRNA [40,41], but subtle differences have beenshown to account for the fivefold greater affinity for the16S rRNA site [35].
Here we present the crystal structure of S8 at 1.9 Å resolu-tion. The protein contains two tandemly arranged domains
Addresses: 1Department of Microbiology, DukeUniversity Medical Center, Durham, NC 27710,USA, 2Department of Biochemistry, University ofUtah School of Medicine, Salt Lake City, UT84132, USA.
of approximately equal size which are tightly associated toform a globular molecule. A number of potential RNA-binding sites have been identified, and the C-terminaldomain contains a distinctive concave hydrophobic patch atits surface for potential protein–protein interactions withribosomal protein S5. The fold of the N-terminal domain ispresent as a module in a diverse array of other proteinsincluding two that bind double-stranded DNA.
ResultsCrystallizationCrystals of Bacillus stearothermophilus S8 were obtainedunder two conditions, using ammonium phosphate orsodium chloride as precipitants. Those grown from sodiumchloride were large, grew reproducibly and gave excellentdiffraction to 1.9 Å resolution, and were therefore used forthe crystallographic analysis. The space group was foundto be orthorhombic, P21212, and the cell dimensions area=80.2 Å, b=85.9 Å and c=39.6 Å. Only by assuming thatthere are two molecules in the asymmetric unit does thecrystal density lie within the accepted range for typicalprotein crystals [42].
Structure determinationFollowing an extensive search for heavy-atom derivativesfor use in the multiple isomorphous replacement (MIR)method, four suitable candidates were obtained and usedto phase the diffraction data. Three were found by con-ventional soaking techniques, and the fourth involved themetabolic incorporation of selenium. The principal deriv-ative was obtained with platinum, and two sites wereobserved in the difference Patterson map which were cor-rectly assumed to correspond to one site per molecule inthe two-molecule asymmetric unit. Initial phases fromthis derivative allowed the three other derivatives to be identified by difference Fourier techniques. Theseincluded a two-site gadolinium, a single-site lead and afour-site selenium (as selenomethionine). All sites wereconfirmed by difference Pattersons and cross differenceFouriers. The anomalous data from the platinum and
gadolinium derivatives were measurable and providedadditional phase information. The overall figure of meritwas 0.741 and this improved to 0.853 after solvent flatten-ing. The data collection and final phasing statistics areshown in Tables 1 and 2, respectively.
The electron-density map was of high quality and alloweda complete and unambiguous fitting of the B. stearother-mophilus S8 amino-acid sequence [43] (Fig. 1a). As anindependent confirmation of the model, the locations ofthe methionine sulphurs were consistent with the posi-tions of the selenium atoms determined from the differ-ence Fourier. Only the N-terminal valine was not visiblein the initial electron-density map, but excellent densitywas eventually observed around this residue after refine-ment. The model was subjected to several rounds ofmodel building and refinement with simulated annealing,using all data between 8 Å and 1.9 Å. The final structurehas an R factor of 22.2%, an R free [44] of 28.8% and goodgeometry (Tables 3,4). A total of 124 water moleculeswere also modelled into the structure. A portion of thefinal 2|Fo-Fc| electron-density map is shown in Figure 1b.
Description of the structureOverall, the S8 molecule is relatively compact with dimen-sions 45Å ×40Å ×35Å, and it contains no regions of signif-icant flexibility. Topologically, the protein is divided intoan N-terminal and a C-terminal domain that are imme-diately adjacent within the sequence. The domains areindependently folded but interact closely through anextended interface. Stereoviews of the a-carbon backboneand the ribbon diagram are shown in Figure 2, and thelocations of the secondary structural elements within thesequence are shown in Figure 3.
The N-terminal domain comprises two a-helices and athree-stranded b-pleated sheet with the connectivitya1–b1–a2–b2–b3. These secondary structural elementsare approximately parallel to each other and criss-crossalong the length of the domain. The two a helices pack
In each case, all data were collected from a single crystal. *Derivativeabbreviations: KCNPT, K2Pt(CN)4; GDCL, GdCl2; SEMET, selenome-thionine; PBAC, Pb(CH3COOH)2. †Rsym = S|Ii–Im|/ SIm where Ii is the
intensity of the measured reflection and Im is the mean intensity of allsymmetry-related observations.
onto one surface of the b sheet, and the opposite surface iscompletely exposed. All of the loops within the N-termi-nal domain are short apart from loops 5 and 6 which arepoorly conserved and disposed at the N and C termini ofstrand b3. Loop 6 contains the region of primary structurewith the highest sequence and length variability (Fig. 3). Astriking feature of this domain is the number of chargedresidues; these cluster at the back of the molecule as
viewed in Figure 2, on the exposed surface of the b sheetand at the interface between a2 and b2.
Unlike the N-terminal domain, the C-terminal half has avery unusual structure. It comprises six b strands and ashort a-helix with the following connectivity, b4–b5–b6–b7–a3–b8–b9. The principal feature is an anti-parallelfour-stranded b-sheet structure which is bifurcated and
Research Article Ribosomal protein S8 Davies et al. 1095
Table 2
Final phasing data and statistics.
Derivative* KCNPT GDCL SEMET PBAC
Rmerge† 6.6 9.2 7.7 3.9
Number of sites 2 2 4 1
Type of data# Iso Ano Iso Ano Iso IsoResolution (Å) 2.5 2.5 2.5 2.5 2.5 2.5RCullis
*Derivative abbreviations: KCNPT, K2Pt(CN)4; GDCL, GdCl2; SEMET,selenomethionine; PBAC, Pb(CH3COOH)2. †Rmerge=S|FPH–FP|/S|FPH |.#Iso is isomorphous and Ano is anomalous. ‡RCullis =S|(FPH±FP)–FH(calc)|/S |FPH–FP|. §RKraut = S|FPH(obs)–FPH
(calc)| /SFPH(obs). ††Phasing power = FH/ERMS. FP , FPH and FH are theprotein, derivative and heavy-atom structure factors respectively, andERMS is the residual lack of closure.
Figure 1
Stereoviews of the electron-density maps ofribosomal protein S8 (displayed using the Oprogram [75]). The area shown is theinterdomain hydrophobic core around residueIle126. Both maps are contoured at 1.5s. Thecoordinates shown are those of the finalmodel. (a) The solvent flattened MIR mapcalculated at 2.5 Å that was used to trace thepolypeptide chain. (b) The 2|Fo-Fc| calculatedphased map generated from the refinedcoordinates at 1.9 Å.
bent to form two sides of a triangular pyramid. Going fromthe base of the pyramid to the apex, one side is formed bystrands b4, b9, b6 and b7 and the other by strands b5, b8,b6 and b7. Strands b4 and b5 are connected by the highlyconserved and structured loop 7, and strands b8 and b9are delineated by a kink at Ile125. Loop 8 betweenstrands b5 and b6 is in an extended conformation, and itforms the third side of the pyramid along with helix a3.The conformations of three residues (89–91) within loop 8conform to a 310 helix.
The interface between the two halves of S8 is formed byhelices a1 and a2 of the N-terminal domain and onesurface of the b4, b9, b6 and b7 b sheet from the C-termi-nal domain. Therefore, a1 and in particular a2 are sand-wiched between two b sheets (Fig. 2). The interfacecontains conserved hydrophobic residues and representsan interdomain hydrophobic core. There also appears tobe several interdomain salt bridges that are common toboth molecules in the asymmetric unit and are apparentlynot dictated by the crystal environment.
The two S8 molecules within the asymmetric unit areextremely similar and can be overlapped with a rms
deviation on a carbons of 0.32 Å. Small differences occurin loop 5, loop 7 and at Lys76 within strand b4, and in theorientation of some surface side chains. The arrangementof the eight S8 molecules within the unit cell conforms topseudo I222 symmetry, and the crystals have beenobserved to change to this space group in some heavy-atom soaks. This movement of molecules within thecrystal is probably facilitated by the relatively small area ofcontact between the two molecules within the asymmetricunit which is centered on His90 at the tip of the C-termi-nal domain. S8 exists as a monomer within the ribosome,and the dimer probably has no biological significance.
Possible functional sitesWe have developed a strategy for locating potential RNA-binding sites on ribosomal proteins that involves identify-ing patches of basic and aromatic residues that are highlyconserved in their many known primary structures[4–6,8,10]. Although these predictions have yet to be sub-stantiated by the direct visualization of a protein–rRNAcomplex, they have recently been supported by a numberof different studies. These include protein–RNA crosslink-ing experiments [45], site-specific cleavage of rRNA usingFe(II) tethered to ribosomal proteins ([11], HF Noller, per-sonal communication, and mutational analyses [46]. Severalribosomal proteins also have conserved exposed hydropho-bic residues that are candidates for either protein–protein orprotein–RNA interactions within the ribosome [4,6].
S8 has a number of potential functional regions distributedover the front surface of the protein, as viewed in Figure 2.These can be grouped into three sites. The overall loca-tions of these sites are shown in Figures 3 and 4, and eachof the sites is shown in more detail in Figure 5. Also shownin Figure 5 is the molecule’s electrostatic surface poten-tial, which has been useful for evaluating possible func-tional regions in these proteins [6,10].
Site 1 is located at the top of the N-terminal domain andincludes loop 2 and the C-terminal half of a1 (Fig. 5a). Itis bordered to the top and bottom by regions of positivepotential (Fig. 5d). The top region is composed of Arg20and Arg70, and the bottom region contains Arg14 as wellas Lys76, Lys80 and Arg77 which also border site 3 and areshown in Figure 5c. The sequence associated with site 1 isgenerally highly conserved, and the adjacent Arg14 andAsn15 are completely conserved (Fig. 3).
1096 Structure 1996, Vol 4 No 9
Table 3
Crystallographic parameters of the refined structure.
Resolution (Å) 1.9–8.0s cutoff 0.0Number of reflections 21545Completeness of outer shell (1.9 Å–1.99 Å) (%) 95.08Number of non-hydrogen protein atoms* 2058Number of waters (total, mol A, mol B)* 124, 69, 55Overall R factor 0.222Overall R free [44] 0.288Overall G factor [72] 0.19Mean B factor
Site 2 (Fig. 5b) is restricted mainly to the C-terminaldomain and extends across its lower surface as viewed inthe normal orientation (Fig. 2). Several of the residues inthis site are completely conserved including Arg84, Tyr86,Ser105, Ser107, Gly122, Gly123 and Glu124. The centrallocation of the two glycine residues adjacent to Tyr86 isparticularly interesting because they create a gap within thesite that may allow for the binding of an RNA molecule.Similar features have been noted in ribosomal proteins S5[8] and L9 [5].
Site 3 is also in the C-terminal domain and encompasses a large area including the concave inner surface of thepyramid-like substructure (Fig. 5c). It contains the moststriking feature of S8 which is a large conserved hydropho-bic area flanked by two basic regions (Fig. 5d). The smallerbasic region to the left contains the conserved residuesArg94 and Arg117. The larger basic region to the right rep-resents the most extensive area of positive potential on S8and includes the totally conserved Lys76, Arg77 andLys80. This patch also borders site 1, as described above.Site 3 is complicated as it has features that are typical ofboth protein–protein and protein–RNA interactions. Thelatter include the adjacent residues 128 and 130 which arearomatic amino acids (tyrosine and trytophan, respectively)in the majority of S8 sequences (Fig. 3). Site 3 may repre-sent a docking site for another protein–RNA complex.
Structurally homologous proteinsMost of the known ribosomal protein structures are homol-ogous with other families of proteins [47–50], and a subset
are homologous with each other [1]. These homologies arenot only interesting from an evolutionary standpoint, butthey have also helped in understanding the functions ofthese proteins in the ribosome. To search for protein struc-tures that are homologous with S8, the coordinates weresubmitted to the program ‘Dali’ [51].
The structure of S8 is not homologous with any of theknown ribosomal protein structures, but segments of fourother proteins were found to share a close similarity with itsN-terminal domain. Each segment contains two a helicesand three b strands with identical topology to S8 and whichcan be overlapped with Ca rms deviations of 2.5Å to 3.5Å(Fig. 6). The loop regions are somewhat variable and werenot included in the rms calculation. In order of decreasingsimilarity, the proteins are EPSP synthase [52], HaeIIIDNA methyltransferase [53], initiation factor IF3 [54], andDNase I [55]. In contrast, no striking homologies werefound to the C-terminal domain of S8, although it doesshare some similarity with the b-roll structure in the CAPprotein [56].
DiscussionS8–RNA interactionsThere have been a number of biochemical and mutagene-sis experiments to investigate the interaction of S8 withRNA, and these results can now be assessed in light of thecrystal structure. Mougel et al. [57] chemically modifiedCys126 in the E. coli protein and demonstrated a loss ofRNA-binding activity. However, subsequent mutagenesisexperiments [58] suggested that this residue is not directly
Research Article Ribosomal protein S8 Davies et al. 1097
Figure 2
The overall structure of ribosomal protein S8.(a) A stereo Ca trace of the S8 backbonewith every tenth residue labelled and markedwith a closed circle. (b) A stereo ribbonrepresentation of S8 showing the elements ofsecondary structure. (Figure produced usingMOLSCRIPT [76].)
involved as it can be replaced by alanine with no loss ofactivity, although a serine does reduce binding. In the B.stearothermophilus protein, the corresponding residue is ahydrophobic-core alanine, and a modification or replace-ment by serine at this position presumably disrupts thelocal structure. In a more comprehensive study, Zimmer-mann and colleagues [59] carried out mutagenesis on E. coliS8 and then screened for mutants that were defective inRNA binding, based on the protein’s ability to associatewith the spc operon mRNA. A number of mutants wereidentified and characterized. When mapped onto the S8crystal structure, the majority of the mutants correspond to residues that are important for the structural integrity ofthe protein. However, several are within the putative RNA-binding sites. Using the B. stearothermophilus numbering,
residue 70 is within site 1, residues 86, 87 and 89 are in site2, and residues 79 and 80 are in site 3.
Recent S8–RNA crosslinking data provide additional cluesas to the RNA-binding surface of S8 (B Wittmann-Lieboldand H Urlaub, personal communication). A crosslink hasbeen identified in the E. coli system between Lys55 and16S rRNA. This residue is Gln56 in the B. stearother-mophilus protein (Fig. 3), but it is a conserved basic residuein most bacterial S8 sequences. It is located in loop 5 atthe ‘bottom’ of the N-terminal domain (Fig. 4), approxi-mately equidistant from sites 1 and 2 but at the oppositeend of the molecule from site 3. This result suggests thatan extended RNA molecule spans across S8 from thecrosslink site to one of the putative RNA-binding sites. A
1098 Structure 1996, Vol 4 No 9
Figure 3
1 10 20 30 40Bs V M T D P I A D M L T A I R N A N M V R H E K L E V P A S K I K R E I A E I L K R E G F I RTt M L T D P I A D M L T R I R N A T R V Y K E S T D V P A S R F K E E I L R I L A R E G F I KEc S M Q D P I A D M L T R I R N G Q A A N K A A V T M P S S K L K V A I A N V L K E E G F I EMl M T M T D P V A D M L T R L R N A N S A Y H D T V S M P S S K L K T R V A E I L K A E G Y I QNt M G R D T I A E I I T S I R N A D M D R K R V V R I A S T N I T E N I V Q I L L R E G F I EOs M G K D T I A D L L T S I R N A D M N K K G T V R V V S T N I T E N I V K I L L R E G F I EEv M G R D T I L E I I N S I R N A D R G R K R V V R I T S T N I T E N F V K I L F I E G F I EMp M G N D T I A N M I T S I R N A N L G K I K T V Q V P A T N I T R N I A K I L F Q E G F I DCp M V N D T I A D M L T G I R N A N L A K H K V A R V K A T K I T R C L A N V L K E E G L I Q loop1| α1 | loop2 | β1 |loop3| α2 | loop4 |
50 60 70 80Bs D Y E Y I E D N K Q G I L R I F L K Y G P N � � � � � � � � E R V I T G L K R I S K P G L R VTt G Y E R V D V D G K P Y L R V Y L K Y G P R R Q G P D P R P E Q V I H H I R R I S K P G R R VEc D F K � V E G D T K P E L E L T L K Y F Q G K A � � � � � � � � V V E S I Q R V S R P G L R IMl D W R E E E A E V G K K L T I D L K F G P Q R � � � � � � � E R A I A G L R R I S K P G L R VNt N V R K H R E K N K Y F L V L T L R H � � R R N � � R K R P Y R N I L N L K R I S R P G L R IOs S V R K H Q E S N R Y F L V S T L R H Q K R K T � � R K G I Y R T R T F L K R I S R P G L R IEv N A R K H R E K N K Y Y F T L T L R H � � R R N � � S K R P Y I N I L N L K R I S R P G L R IMp N F I D N K Q N T K D I L I L N L K Y Q G K K � � � � K K S Y I T T � � L R R I S K P G L R ICp N F E E I E N N L Q N E L L I S L K Y K G K K � � � � R Q P I I T A � � L K R I S K P G L R G β2 | loop5 | β3 | loop6 | β4 | loop7 |
90 100 110 120 130Bs Y V K A H E V P R V L N G L G I A I L S T S Q G V L T D K E A � R Q K G T G G E I I A Y V ITt Y V G V K E I P R V R R G L G I A I L S T S K G V L T D R E A � R K L G V G G E L I C E V WEc Y K R K D E L P K V M A G L G I A V V S T S K G V M T D R � A A R Q A G L G G E I I C Y V AMl Y A K S T N L P H V L G G L G I A I L S T S S G L L T N Q Q A A K K A G V G G E V L A Y V WNt Y S N Y Q R I P R I L G G M G I V I L S T S R G I M T D R E A � R L E G I G G E I L C Y I WOs Y A N Y Q G I P K V L G G M G I A I L S T S R G I M T D R E A � R L N R I G G E V L C Y I WEv Y S N S Q Q I P L I L G G I G I V I L Y T S R G I M T D R E A � R L K G I G G E L L C Y I WMp Y S N H K E I P K V L G G M G I V I L S T S R G I M T D R E A � R Q K K I G G E L L C Y V WCp Y A N H K E L P R V L G G L G I A I L S T S S G I M T D Q T A � R H K G C G G E V L C Y I W
An alignment of representative sequences of ribosomal protein S8from bacteria and chloroplasts. The abbreviations for each are asfollows: Bs, Bacillus stearothermophilus [43]; Tt, Thermusthermophilus [77]; Ec, E. coli [78]; Ml, Micrococcus luteus [79]; Nt,Nicotiana tabacum (tobacco) chloroplast [80]; Os, Oryza sativa (rice)chloroplast [81]; Ev, Epifagus virginiana plastid [82]; Mp, Marchantiapolymorpha chloroplast [83]; Cp, Cyanophora paradoxa [84]. The
numbering corresponds to the B. stearothermophilus protein and theregions of secondary structure of this protein are indicated. Blocks ofamino acids are colour coded as follows: orange, hydrophobic coreresidues; green, putative RNA-binding residues; yellow, exposedhydrophobic residues; purple, residues crosslinked to other ribosomalcomponents (see text for details). (The alignment was performed usingCLUSTAL V [85].)
model in which the RNA spans beneath the protein,between loop 5 and site 2, has several attractive features.First, site 2 is the most highly conserved of the three sitesand extends towards the crosslink site. Second, loop 5would be naturally close to an RNA molecule in this posi-tion and most easily crosslinked. Finally, the model isindirectly supported by the structural homology of S8 toHaeIII methyltransferase and DNase I [53,55]. In theseproteins, the regions that are homologous to S8 both inter-act with the bound DNA, and the DNA is positionedbelow the motif when viewed in the orientation shown in
Figure 6. Also, the protein–DNA interactions include theloops that correspond to loop 5 in S8.
S8–protein interactionsThe best candidate for a protein–protein interaction partnerwith S8, in the ribosome, is S5. The proteins are adjacentwithin the particle, as demonstrated by neutron diffraction[21] and protein–protein crosslinking [22], and a complexbetween the isolated proteins has been observed in solu-tion by sedimentation equilibrium studies [60]. As thestructure of S5 is known, we are in a position to search for
Research Article Ribosomal protein S8 Davies et al. 1099
Figure 4
A stereoview of ribosomal protein S8 showingthe overall distribution of residues believed tobe involved in mediating interactions withrRNA and other ribosomal proteins. Theresidues (in standard atom coloring) can beloosely grouped into three sites and these areindicated. The locations of S8–RNA andS8–S5 crosslinks are also shown. Theorientation of the molecule is identical to thatshown in Figure 2. (Figure produced usingMOLSCRIPT [76].)
Figure 5
Expanded views of the putative functionalsites of ribosomal protein S8 and the overallsurface electrostatic potential. (a) Site 1 inthe same orientation shown in Figure 4.(b) Site 2 viewed from below the moleculerelative to Figure 4. Gly123 and Gly122 areshown as red spheres. (c) Site 3 inapproximately the same orientation shown inFigure 4. (d) The surface electrostaticpotential calculated using GRASP [86]. Theextreme ranges of red (negative) and blue(positive) represent electrostatic potentials of<–9 to >+9 kbT, where kb is the Boltzmannconstant and T is the temperature. Whiterepresents non-polar residues. The moleculeis in the same orientation as shown in Figures2 and 4. Note that site 3 has a largehydrophobic patch flanked by positivepotential.
possible interacting surfaces on the two molecules. A guideto these surfaces has been provided by the work of Gualerziand colleagues [61] who obtained a crosslink betweenLys93 of S8 and Lys166 of S5 using E. coli proteins. The S8crosslink site is immediately adjacent to the conserved,concave hydrophobic patch within site 3 (Figs. 4,5c). TheS5 crosslink site is at the extreme C terminus, which is dis-ordered in the crystal structure [8]. However, there is a con-served convex hydrophobic patch in the vicinity of the S5 Cterminus that covers the lower surface of the C-terminaldomain b sheet. The S5 and S8 patches are approximatelyequal in size and generally complementary in shape.
The S8 environment within the ribosomeIt is now clear that the principal role of ribosomal proteinsis to promote the folding of rRNA and to stabilize its finaltertiary structure. Consistent with this idea, the structuresof S5 [8], S17 [10], L9 [5] and L14 [6] display two apparentRNA-binding regions that can potentially knit togethertwo RNA loci. This idea has now been confirmed by theidentification of the two rRNA sites for proteins S5 [11],S17 (HF Noller, personal communication) and L9 [46]. S8is also known to interact with several 16S rRNA sites thatinclude its principal site in the 596–597/641–643 stem-loopregion and two others close by in the 585/755 and 825/875regions [28]. These multiple RNA-binding sites are con-sistent with the role of S8 in organizing the centraldomain. They also agree with the latest 3D models of the
16S rRNA which show that the S8 location coincides witha complicated region of the nucleic-acid structure [12].
Site 2 is the best candidate for the primary RNA-bindingsite of S8. It is the most conserved, and the extendedstem-loop structure of the rRNA site would easily span tothe crosslink location within loop 5. Site 3 is an ideal sitefor interacting with ribosomal protein S5. Not only doesS5 have a complementary hydrophobic patch on its C-ter-minal domain (see above), but this domain of S5 is alsoknown to interact with RNA [11]. The curious hydropho-bic/basic nature of site 3 in S8 (Fig. 5c,d) may reflect adocking site for the S5–rRNA complex. This leaves site 1to bind to a second region of rRNA, and the most likelyregion is the 825/875 helix where a strong interaction hasbeen observed [28].
S8 is located between ribosomal proteins S5 and S17 whichsit above and below the protein, respectively, in current30S models [12]. The putative interactions of S8 within theribosome described above would dictate that the C-termi-nal domain points upwards towards S5 and the head, andthat the N-terminal domain points down towards S17. Theback surface of S8, which is extremely polar and poorlyconserved, would presumably point towards the outside ofthe ribosome. The detailed crystal structure of S8 will nowpermit specific mutations to be made to further investigatehow the protein interacts with other ribosomal components.
1100 Structure 1996, Vol 4 No 9
Figure 6
The structural homology of the N-terminaldomain of ribosomal protein S8 to regions ofother proteins. (a) Residues 1–65 of S8 areshown together with the homologous regionsof (b) DNase I (residues 13–88, Ca rms3.44 Å); (c) HaeIII DNA methyltransferase(residues 84–164, Ca rms 3.08 Å); (d) EPSPsynthase (residues 21–77, Ca rms 2.68 Å);and (e) initiation factor IF3 (residues 93–170,Ca rms 3.2 Å). (f) A topological diagram ofthe common motif. In DNase I and HaeIII DNAmethyltransferase, the DNA binds below themotif as shown in the figure. The segmentswere placed in the same orientation using theO program [75], and displayed usingMOLSCRIPT [76].
Such studies will be crucial for mapping the environmentof S8 which is pivotal to the structure of the whole 30Ssubunit. Also, as the structures of S5 [8] and S17 [10] areboth known and their RNA contacts have been well deter-mined, the opportunity now exists to model the wholeS5–S8–S17–rRNA region, which represents a considerableand crucial segment of the small subunit.
Homology and evolution of S8Although none of the prokaryotic ribosomal proteins can bedescribed as being large by normal protein standards, thelarger ones whose structures have been determined, namelyL1, L6, L9 and S5, all have two independently foldeddomains. The L6 molecule has obviously evolved by geneduplication [4], and L1 [7], L9 [5] and S5 [8] appear to bethe result of a fusion between two different proteins. Thisidea is supported by the fact that the individual domains arein most cases structurally homologous to other proteins.Domain I of L1, each domain of L6 and the N-terminaldomain of L9 all conform to the split b–a–b motif that isalso found in L7/L12 [2], L30 [3], S6 [9] and the RNA-recognition motif [47,62]. Domain II of L1 has a Rossmann-fold topology, and the fold of the N-terminal domain of S5has been found to occur in the double-stranded RNA-binding proteins staufen [49] and RNase III [50]. It shouldalso be noted that S17 is a member of the OB-fold family ofproteins [48]. In an evolutionary sense, ribosomal proteinsmust represent some of the earliest protein molecules, andthe observation that their folds are present in a wide varietyof proteins, many associated with RNA and DNA, givescredence to this general idea.
S8 also clearly follows this pattern, and in general terms itmost closely resembles S5. Like its neighbour in the smallsubunit, S8 has two structurally distinct domains that arefused end-to-end, but they interact so closely through ahydrophobic interface that they form a single globularmolecule. Also, as described above, the N-terminal half ofS8 is present in a number of other proteins where it repre-sents a central folding substructure. In EPSP synthase, theentire molecule is essentially six of these units arranged asa pair of trimers [52]. In IF3 it is a distinct domain [54],and in HaeIII methyltransferase [53] and DNase I [55],the unit is embedded within the larger structures and con-tributes to the DNA-binding site.
Biological implicationsThe ribosome is the center of protein synthesis in allcells, and for over 30 years efforts to understand its struc-ture and mechanism have been ongoing. Such an aimrepresents a considerable challenge because the typicalbacterial ribosome contains three large RNA moleculesand over 50 proteins. It is generally accepted that the par-ticle is a huge ribozyme in which the ribosomal RNA ismaintained in a particular folded conformation by itsassociation with the ribosomal proteins.
Our contribution to this effort has been the determinationof the structures of important ribosomal proteins. Eachstructure reveals potential sites of interaction with otherribosomal components, and these, when combined withbiochemical data, enable models to be constructed of thelocal ribosome environment. These models can then beincorporated into the increasingly higher resolutionimages of the ribosome provided by electron microscopy.The protein structures also permit further experiments tobe designed to refine the ribosome models.
Here we describe the crystal structure of protein S8which is a central organizing component of the smallribosomal subunit. It contains two domains, and potentialsites of interaction with two known regions of RNA andprotein S5 have been identified. S8 is also an importantRNA-binding protein outside the ribosome. Its principalRNA-binding site mediates an interaction with poly-cistronic mRNA in controlling the synthesis of ribosomalcomponents. A considerable amount of biochemical datahave been accumulated on the S8–RNA interaction, andthe crystal structure will be invaluable for interpretingthese results. Finally, similar to other ribosomal proteins,one domain of S8 is homologous to substructures within anumber of other proteins, including two that bind DNA.
Materials and methodsGene cloning and sequencingThe procedure for cloning, sequencing and expressing the S8 gene fromB. stearothermophilus using the T7 system [63] was the same as thatdescribed previously [6,15].
Protein purificationE. coli cells incorporating the T7 expression system (BL21(DE3)) andcontaining the S8 expression vector (pET13) were grown in a mediumthat contained 25g l-1 of Luria Broth (Gibco) and 25mg l-1 of kanamycin.Two 1l batches were inoculated with 3ml of an overnight culture, andwhen the OD550 had reached 0.6, the cells were induced by adding isopropyl-b-D-thiogalactopyranoside (IPTG) to a final concentration of0.4mM. After 3h, the cells were harvested by centrifugation at 2000rpm,and resuspended in 40ml of 50mM Tris, pH8.4, 0.5mM EDTA, 50mMNaCl, 1mM sodium azide and protease inhibitors. The sample wasfrozen at –20°C after adding 2mg each of lysozyme and DNase to facili-tate cell lysis and DNA cleavage. To purify S8, the cell sample wasdefrosted and spun at 18000 rpm for 30min, and the resulting super-natant was applied to an S-sepharose column (Pharmacia) equilibratedin the same resuspension buffer. S8 bound to the column under theseconditions, and was eluted with a salt gradient of 0–1M NaCl. S8 elutedat 0.2M NaCl, and the appropriate fractions were determined by theirabsorbance at 280 nm and then confirmed by SDS-PAGE gels. Theconditions required to incorporate selenomethionine into S8 were thesame as those described previously [4,5].
CrystallizationFollowing the elution from the S-sepharose column, the protein wasconcentrated to 30mg ml–1 using Centricon-10 microconcentrators(Amicon). The extinction coefficient was estimated at 0.45 based on theamino-acid sequence. The crystallization trials used the hanging dropmethod [64], initially with the crystal screen protocols I and II developedby Jankarik and Kim [65] but later with additional conditions. In theseexperiments, 3ml of the protein solution was mixed with 3ml of wellsolution, and the dishes were stored at 22°C.
Research Article Ribosomal protein S8 Davies et al. 1101
Data collectionDiffraction data were collected on an Rigaku RAXIS-II image platesystem with Yale mirror optics (Molecular Structures Corporation)mounted on a Rigaku RU-300 X-ray generator operating at 40kV and80 mA. Data were collected at room temperature by the standard oscil-lation method [66] using a crystal-to-plate distance of 130mm. Typi-cally, the crystal was rotated a total of 160° with the spindle axisapproximately parallel to the b* axis using increments of 2° and an expo-sure time of 6 min per degree. These parameters were adjusted witheach crystal according to its size and orientation to avoid saturation andminimize reflection overlap. The cell dimensions and space group weredetermined using the RAXIS-II autoindexing software. The data wereintegrated, merged and scaled using the HKL processing software [67].The Friedel pairs from each derivative data set were not merged to allowsubsequent evaluation of the anomalous signal. For model refinement, anative dataset was collected at 1.9 Å resolution using a crystal-to-platedistance of 90mm, 8 min per degree exposure times, and 2° incrementsin less crowded regions of reciprocal space and 1.4° increments inmore crowded regions.
PhasingA search for derivatives was made by soaking crystals in solutions ofheavy-atom compounds in the stabilizing buffer 4.3 M NaCl, 100mMHEPES, pH 7.5. This search was hindered by the propensity of mostheavy-atom compounds to precipitate in the high-salt solution. After datacollection and processing, potential derivatives were identified by Patter-son methods. Three derivatives were found in this way: K2Pt(CN)4,GdCl2 and Pb(CH3COOH)2. A fourth derivative was obtained by themetabolic incorporation of selenomethionine. The platinum differencePatterson showed two clear peaks and was solved by manual inspec-tion, and an initial phase set was calculated using both the isomorphousand anomalous information. These phases were used in the calculationof cross difference Fourier maps which revealed the precise heavy-atompositions in the selenium, gadolinium and lead derivatives. These posi-tions correlated well with relatively weak but clear peaks in the individualdifference Patterson maps. Following heavy-atom refinement, all fourderivatives were included in a final phase calculation which included theanomalous data from the platinum and gadolinium derivatives. Theresulting phases were improved by solvent flattening assuming a solventcontent of 51%. All calculations were carried out using the PHASESpackage [68].
Noncrystallographic symmetry averagingAlthough the initial solvent flattened MIR map was of sufficiently highquality to build a model, it was decided to see whether the phasescould be improved further using non crystallographic symmetry (NCS)averaging. A region of the electron-density map was extracted that con-tained the bulk of the asymmetric unit, and the NCS axis which isalmost colinear with the b axis was refined. Despite a high correlationcoefficient of 0.68 at 3.0 Å, there was no noticeable improvement inthe electron-density map after averaging. These calculations were alsoperformed using PHASES [68].
Model building and refinementMIR electron-density maps were calculated using CCP4 [69] and dis-played using the FRODO program [70]. A backbone model was initiallybuilt into the electron density and this was followed by the fitting of theamino-acid sequence. The other molecule in the asymmetric unit wasgenerated by applying the NCS transformation matrix. The model wasrefined by alternating rounds of simulated annealing using XPLOR [71]and manual rebuilding. The correctness of the final model was verifiedby examining its stereochemistry in PROCHECK [72], its 3D–1D profile[73] and the Ramachandran plot. The secondary structure was analyzedusing the program PROMOTIF [74].
Accession numbersThe coordinates have been deposited with the Brookhaven Protein DataBank, accession code 1sei.
AcknowledgementsWe would like to thank Sue Ellen Gerchman and Stephanie Porter for techni-cal assistance, Richard Brimacombe for helpful discussions and Dr Hee-WonPark for invaluable help in the crystallographic calculations and data collec-tion. This work was supported by grant GM44973 from the National Institutesof Health (to SWW and VR).
References1. Ramakrishnan, V., et al., & White, S.W. (1995). Structures of
prokaryotic ribosomal proteins: Implications for RNA-binding andevolution. Biochem. and Cell Biol. 73, 979–986.
2. Leijonmarck, M. & Liljas, A. (1987). Structure of the C-terminal domainof the ribosomal protein L7/L12 from Escherichia coli at 1.7 Å. J. Mol.Biol. 195, 555–580.
3. Wilson, K.S., Appelt, K., Badger, J., Tanaka, I. & White, S.W. (1986).Crystal structure of a prokaryotic ribosomal protein. Proc. Natl. Acad.Sci. USA 83, 7251–7255.
4. Golden, B.L., Ramakrishnan, V. & White, S.W. (1993). Ribosomalprotein L6: Structural evidence of gene duplication from a primitiveRNA-binding protein. EMBO J. 12, 4901–4908.
5. Hoffman, D.W., et al., & Ramakrishnan, V. (1994). Crystal structure ofprokaryotic ribosomal protein L9: A bi-lobed RNA-binding protein.EMBO J. 13, 205–212.
6. Davies, C., White, S.W. & Ramakrishnan, V. (1996). The crystalstructure of ribosomal protein L14 reveals an important organizationalcomponent of the translational apparatus. Structure 4, 55–66.
7. Nikonov, S., et al., & Liljas, A. (1996). Crystal structure of the RNA-binding ribosomal protein L1 from Thermus thermophilus. EMBO J.15, 1350–1359.
8. Ramakrishnan, V. & White, S.W. (1992). The structure of ribosomalprotein S5 reveals sites of interaction with 16S rRNA. Nature 358,768–771.
9. Lindahl, M., et al., & Amons, R. (1994). Crystal structure of theribosomal protein S6 from Thermus thermophilus. EMBO J. 13,1249–1254.
10. Jaishree, T.N., Ramakrishnan, V. & White, S.W. (1996). Solutionstructure of prokaryotic ribosomal protein S17 by high resolution NMRspectroscopy. Biochemistry 35, 2845–2853.
11. Heilek, G.M. & Noller, H.F. (1996). Site-directed hydroxyl radicalprobing of the rRNA neighbourhood of ribosomal protein S5. Science272, 1659–1662.
12. Stark, H., et al., & van Heel, M. (1995). The 70S Escherichia coliribosome at 23 Å resolution: fitting the ribosomal RNA. Structure 3,815–821.
13. Frank, J., et al., & Agrawal, R.K. (1995). A model of protein synthesisbased on cryo-electron microscopy of the E. coli ribosome. Nature376, 441–444.
14. Yonath, A. (1992). Approaching atomic resolution in crystallography ofribosomes. Annu. Rev. Biophys. Biomol. Struct. 21, 77–93.
15. Ramakrishnan, V. & Gerchman, S.E. (1991). Cloning, sequencing andoverexpression of genes for ribosomal proteins from Bacillusstearothermophilus. J. Biol. Chem. 266, 880–885.
16. Held, W.A., Ballou, B., Mizushima, S. & Nomura, M. (1974). Assemblymapping of 30S ribosomal proteins from Escherichia coli. Furtherstudies. J. Biol. Chem. 249, 3103–3111.
17. Zimmermann, R.A. & Singh-Bergmann, K. (1979). Binding sites forribosomal proteins S8 and S15 in the 16S rRNA of Escherichia coli.Biochim. Biophys. Acta 563, 422–431.
18. Gregory, R.J., et al., & Zimmermann, R.A. (1984). Interaction ofribosomal proteins S6, S8, S15 and S18 with the central domain of16S ribosomal RNA from Escherichia coli. J. Mol. Biol. 178, 287–302.
19. Svensson, P., Changchien, L.-M., Craven, G.R. & Noller, H.F. (1988).Interaction of ribosomal proteins, S6, S8, S15 and S18 with thecentral domain of 16S ribosomal RNA. J. Mol. Biol. 200, 301–308.
20. Geyl, D., Böck, A. & Wittmann, H.G. (1977). Cold-sensitive growth ofa mutant of Escherichia coli with an altered ribosomal protein S8:analysis of revertants. Mol. Gen. Genet. 152, 331–336.
21. Capel, M.S., et al., & Moore, P.B. (1987). A complete mapping of theproteins in the small ribosomal subunit of Escherichia coli. Science238, 1403–1406.
22. Lambert, J.M., Boileau, G., Cover, J.A. & Traut, R.R. (1983). Cross-links between ribosomal proteins of 30S subunits in 70S tight couplesand in 30S subunits. Biochemistry 22, 3913–3920.
23. Wiener, L., Schuler, D. & Brimacombe, R. (1988). Protein bindingsites on Escherichia coli 16S ribosomal RNA; RNA regions that areprotected by proteins S7, S9 and S19, and by proteins S8, S15 and
1102 Structure 1996, Vol 4 No 9
S17. Nucleic Acids Res. 16, 1233–1250.24. Thurlow, D.L., Ehresmann, C. & Ehresmann, B. (1983). Nucleotides in
16S rRNA that are required in unmodified form for featuresrecognized by ribosomal protein S8. Nucleic Acids Res. 11,6787–6802.
25. Wower, I. & Brimacombe, R. (1983). The localization of multiple siteson 16S rRNA which are cross-linked to proteins S7 and S8 inEscherichia coli 30S ribosomal subunits by treatment with 2-iminothiolane. Nucleic Acids Res. 11, 1419–1437.
26. Osswald, M., Greuer, B., Brimacombe, R., Stöffler, G., Baumert, H. &Fasold, H. (1987). RNA–protein cross-linking in Escherichia coli 30Sribosomal subunits; determination of sites on 16S RNA that are cross-linked to proteins S3, S4, S5, S7, S8, S9, S11, S13, S19 and S21 bytreatment with methyl p-azidophenyl acetimidate. Nucleic Acids Res.15, 3221–3240.
27. Chiaruttini, C., Milet, M., Hayes, D.H. & Expert-Bezançon, A. (1989).Crosslinking of ribosomal proteins S4, S5, S7, S8, S11, S12 and S18to domains 1 and 2 of 16S rRNA in the Escherichia coli 30S particle.Biochimie 71, 839–852.
28. Powers, T. & Noller, H.F. (1995). Hydroxyl radical footprinting ofribosomal proteins on 16S rRNA. RNA 1, 194–209.
30. Gregory, R.J. & Zimmermann, R.A. (1986). Site-directedmutagenesis of the binding site for ribosomal protein S8 within 16Sribosomal RNA from Escherichia coli. Nucleic Acids Res. 14,5761–5776.
31. Mougel, M., et al., & Ehresmann, C. (1987). Binding of Escherichiacoli ribosomal protein S8 to 16S rRNA. A model for the interactionand the tertiary structure of the RNA binding site. J. Mol. Biol. 198,91–107.
32. Gregory, R.J., Cahill, P.B.F., Thurlow, D.L. & Zimmermann, R.A.(1988). Interaction of Escherichia coli ribosomal protein S8 with itsbinding sites in ribosomal RNA and messenger RNA. J. Mol. Biol.204, 295–307.
33. Ree, H.K., Cao, K., Thurlow, D.L. & Zimmermann, R.A. (1989). Thestructure and organization of the 16S ribosomal RNA gene from thearchaebacterium Thermoplasma acidophilum. Can. J. Microbiol. 35,124–133.
34. Mougel, M., Allmang, C., Eyermann, F., Cachia, C., Ehresmann, B. &Ehresmann, C. (1993). Minimal 16S rRNA binding site and role ofconserved nucleotides in Escherichia coli ribosomal protein S8recognition. Eur. J. Biochem. 215, 787–792.
35. Wu, H., Jiang, L. & Zimmermann, R.A. (1994). The binding site forribosomal protein S8 in 16S rRNA and spc mRNA from Escherichiacoli: minimum structural requirements and the effects of single bulgedbases on S8–RNA interaction. Nucleic Acids Res. 22, 1687–1695.
36. Allmang, C., Mougel, M., Westhof, E., Ehresmann, B. & Ehresmann, C.(1994). Role of conserved nucleotides in building the 16S rRNAbinding site of E. coli ribosomal protein S8. Nucleic Acids Res. 22,3708–3714.
37. Yates, J.L., Arfsten, A.E. & Nomura, M. (1980). In vitro expression ofEscherichia coli ribosomal protein genes: autogenous inhibition oftranslation. Proc. Natl. Acad. Sci. USA 77, 1837–1841.
38. Dean, D., Yates, J.L. & Nomura, M. (1981). Escherichia coli ribosomalprotein S8 feedback regulates part of the spc operon. Nature 289,89–91.
39. Mattheakis, L., Vu, L., Sor, F. & Nomura, M. (1989). Retroregulation ofthe synthesis of ribosomal proteins L14 and L24 by feedbackrepressor S8 in Escherichia coli. Proc. Natl. Acad. Sci. USA 86,448–452.
40. Olins, P. O. & Nomura, M. (1981). Translational regulation byribosomal protein S8 in Escherichia coli: structural homology betweenrRNA binding site and feedback target on mRNA. Nucleic Acids Res.9, 1757–1764.
41. Cerretti, D.P., Mattheakis, L.C., Kearney, K.R., Vu, L. & Nomura, M.(1988). Translational regulation of the spc operon in Escherichia coli.Identification and structural analysis of the target site for S8 repressorprotein. J. Mol. Biol. 204, 309–329.
42. Matthews, B.W. (1968). Solvent content of protein crystals. J. Mol.Biol. 33, 491–497.
43. Arndt, E., Scholzen, T., Kromer, W., Hatakeyama, T. & Kimura, M.(1991). Primary structures of ribosomal proteins from thearchaebacterium Halobacterium marismortui and the eubacteriumBacillus stearothermophilus. Biochimie 73, 657–668.
44. Brünger, A.T. (1992). Free R value: a novel statistical quantity for
assessing the accuracy of crystal structures. Nature 355, 472–475.45. Urlaub, H., Kruft, V., Bischof, O., Müller, E.-C. & Wittmann-Liebold, B.
(1995). Protein-rRNA binding features and their structural andfunctional implications in ribosomes as determined by cross-linkingstudies. EMBO J. 14, 4578–4588.
46. Adamski, F.M., Atkins, J.F. & Gesteland, R.F. (1996). Ribosomalprotein L9 interactions with 23S rRNA: The use of translationalbypass assay to study the effect of amino acid substitutions. J. Mol.Biol. 261, in press.
47. Hoffman, D.W., Query, C.C., Golden, B.L., White, S.W. & Keene, J.D.(1991). RNA-binding domain of the A protein component of the U1small nuclear ribonucleoprotein analyzed by NMR spectroscopy isstructurally similar to ribosomal proteins. Proc. Natl. Acad. Sci. USA88, 2495–2499.
48. Murzin, A.G. (1993). OB (oligonucleotide/oligosaccharide binding)-fold: common structural and functional solution for non-homologoussequences. EMBO J. 12, 861–867.
49. Bycroft, M., Grunert, S., Murzin, A.G., Proctor, M. & St Johnson, D.(1995). NMR solution structure of a dsRNA binding domain fromDrosophila staufen protein reveals homology to the N-terminal domainof ribosomal protein S5. EMBO J. 14, 3563–3571.
50. Kharrat, A., Macias, M.J., Gibson, T.J., Nilges, M. & Pastore, A. (1995).Structure of the dsRNA binding domain of E. coli RNase III. EMBO J.14, 3572–3584.
51. Holm, L. & Sander, C. (1995). Dali: a network tool for protein structurecomparison. TIBS 20, 478–480.
52. Stallings, W.C., et al., & Kishore, G.M. (1991). Structure andtopological symmetry of the glyphosate target 5-enol-pyruvylshikimate-3-phosphate synthase: A distinctive protein fold. Proc. Natl. Acad. Sci.USA 88, 5046–5050.
53. Reinisch, K.M., Chen, L., Verdine, G.L. & Lipscomb, W.N. (1995). Thecrystal structure of HaeIII methyltransferase covalently complexed toDNA: An extrahelical cytosine and rearranged base pairing. Cell 82,143–153.
54. Biou, V., Shu, F. & Ramakrishnan, V. (1995). X-ray crystallographyshows that translational initiation factor IF3 consists of two compacta/b domains linked by an a-helix. EMBO J. 14, 4056–4064.
55. Weston, S.A., Lahm, A. & Suck, D. (1992). X-ray structure of theDNase I-d(GGTATACC)2 complex at 2.3 Å resolution. J. Mol. Biol.226, 1237–1256.
56. Weber, I. & Steitz, T.A. (1987). Structure of a complex of catabolitegene activator protein and cyclic AMP refined at 2.5 Å resolution. J.Mol. Biol. 198, 311–326.
57. Mougel, M., Ehresmann, B. & Ehresmann, C. (1986). Binding ofEscherichia coli ribosomal protein S8 to 16S rRNA: kinetic andthermodynamic characterization. Biochemistry 25, 2756–2765.
58. Wu, H., Wower, I. & Zimmermann, R.A. (1993). Mutagenesis ofribosomal protein S8 from Escherichia coli: Expression, stability, andRNA-binding properties of S8 mutants. Biochemistry 32, 4761–4768.
59. Wower, I., Kowaleski, M.P., Sears, L.E. & Zimmermann, R.A. (1992).Mutagenesis of ribosomal protein S8 from Escherichia coli: Defects inregulation of the spc operon. J. Bacteriol. 174, 1213–1221.
60. Tindall, S.H. & Aune, K.C. (1981). Assessment by sedimentationequilibrium analysis of a heterologous macromolecular interaction inthe presence of self-association: interaction of S5 with S8.Biochemistry 20, 4861–4866.
61. Allen, G., Capasso, R. & Gualerzi, C. (1979). Identification of the aminoacid residues of proteins S5 and S8 adjacent to each other in the 30Sribosomal subunit of Escherichia coli. J. Biol. Chem. 254, 9800–9806.
62. Oubridge, C., Ito, N., Evans, P.R., Teo, C.H. & Nagai, K. (1994).Crystal structure at 1.92 Å resolution of the RNA-binding domain ofthe U1A spliceosomal protein complexed with an RNA hairpin. Nature372, 432–438.
63. Studier, F.W., Rosenberg, A.H., Dunn, J.J. & Dubendorff, J.W. (1990).Use of T7 RNA polymerase to direct the expression of cloned genes.Methods Enzymol. 185, 60–89.
65. Jancarik, J. & Kim, S.H. (1991). Sparse matrix sampling: a screeningmethod for crystallization of proteins. J. Appl. Cryst. 24, 409–411.
66. Wonacott, A.J. (1977). Geometry of the rotation method. In TheRotation Method in Crystallography. (Arndt, U.W. & Wonacott, A.J.eds), pp. 75–105, North-Holland Publishing, Oxford, UK.
67. Otwinowski, Z. (1993). Data collection and processing. In Proceedingsof the CCP4 Study Weekend. (Sawyer, L., Isaacs, N. & Bailey, S. eds),pp. 56–62, SERC Daresbury Laboratory, Warrington, UK .
68. Furey, W. & Swaminathan, S. (1990). PHASES. Am. Cryst. Assoc.
Research Article Ribosomal protein S8 Davies et al. 1103
Annu. Mtg. Program. Abstr. 18, 73.69. Collaborative Computational Project, Number 4. (1994). The CCP4
suite: programs for protein crystallography. Acta Cryst. D 50,760–763.
71. Brünger, A.T., Kuriyan, J. & Karplus, M. (1987). Crystallographic R-factor refinement by molecular dynamics. J. Mol. Biol. 203, 803–816.
72. Laskowski, R.A., MacArthur, M.W., Moss, D.S. & Thornton, J.M.(1993). PROCHECK: a program to check the stereochemical qualityof protein structures. J. Appl. Cryst. 26, 283–291.
73. Luthy, R., Bowie, J.U. & Eisenberg, D. (1992). Assessment of proteinmodels with three-dimensional profiles. Nature 356, 83–85.
74. Hutchinson, E.G. & Thornton, J.M. (1996). PROMOTIF: A program toidentify and analyse structural motifs in proteins. Protein Sci. 5,212–220.
75. Jones, T.A., Zou, J.Y., Cowan, S.W. & Kjeldgaard, M. (1991). Improvedmethods for building protein models in electron density maps and thelocation of errors in these models. Acta Cryst. A 47, 110–119.
76. Kraulis, P.J. (1991). MOLSCRIPT: a program to produce both detailedand schematic plots of protein structures. J. Appl. Cryst. 24,946–950.
77. Reinbolt, J., Eliseikina, I., Sedelnilkova, S., Garber, M., Ehresmann, C.& Ehresmann, B. (1993). The complete amino acid sequence ofribosomal protein S8 from Thermus thermophilus. J. Protein Chem.12, 79–83.
78. Stadler, H. & Wittmann-Liebold, B. (1976). Determination of theamino-acid sequence of the ribosomal protein S8 of Escherichia coli.Eur. J. Biochem. 66, 49–56.
79. Ohama, T., Muto, A. & Osawa, S. (1989). Spectinomycin operon ofMicrococcus luteus: evolutionary implications of organization andnovel codon usage. J. Mol. Evol. 29, 381–395.
80. Tanaka, M., Wakasugi, T., Sugita, M., Shinozaki, K. & Sugiura, M.(1986). Genes for the eight ribosomal proteins are clustered on thechloroplast genome of tobacco (Nicotiana tabacum): similarity to theS10 and spc operons of Escherichia coli. Proc. Natl. Acad. Sci. USA83, 6030–6034.
81. Hiratsuka, J., et al., & Sugiura, M. (1989). The complete sequence ofthe rice (Oryza sativa) chloroplast genome: intermolecularrecombination between distinct tRNA genes accounts for a majorplastid DNA inversion during the evolution of the cereals. Mol. Gen.Genet. 217, 185–194.
82. Wolfe, K.H., Morden, C.W., Ems, S.C. & Palmer, J.D. (1992). Rapidevolution of the plastid translational apparatus in a nonphotosyntheticplant: loss or accelerated sequence evolution of tRNA and ribosomalprotein genes. J. Mol. Evol. 35, 304–317.
83. Fukuzawa, H., et al., & Ohyama, K. (1988). Structure and organizationof Marchantia polymorpha chloroplast genome. III. Gene organizationof the large single copy region from rbcL to trnI(CAU). J. Mol. Biol.203, 333–351.
84. Michalowski, C.B., Pfanzagl, B., Loeffelhardt, W. & Bohnert, H.J.(1990). The cyanelle S10 spc ribosomal protein gene operon fromCyanophora paradoxa. Mol. Gen. Genet. 224, 222–231.
85. Higgins, D.G. (1994). CLUSTAL V: multiple alignment of DNA andprotein sequences. Methods Mol. Biol. 25, 307–318.
86. Nicholls, A., Sharp, K.A. & Honig, B. (1991). Protein folding andassociation: insights from the interfacial and thermodynamicproperties of hydrocarbons. Proteins 11, 281–296.
