CURRENT MICROBIOLOGY Vol. 45 (2002), pp. 429 – 433 DOI: 10.1007/s00284-002-3684-y
Current Microbiology An International Journal © Springer-Verlag New York Inc. 2002
Isolation of Spectinomycin Resistance Mutations in the 16S rRNA of Salmonella enterica serovar Typhimurium and Expression in Escherichia coli and Salmonella Michael O’Connor, Albert E. Dahlberg J.W. Wilson Laboratory, Department of Molecular and Cellular Biology and Biochemistry, Brown University, 69 Brown Street, Providence RI 02912, USA Received: 28 September 2001 / Accepted: 26 March 2002
Abstract. Two single-base mutations in 16S rRNA conferring high-level resistance to spectinomycin were isolated on a plasmid-borne copy of the rrnD operon from Salmonella enterica serovar Typhimurium. Neither of the mutations (C1066U and C1192U) had appreciable effects on cell growth, but each had differential effects on resistance to spectinomycin and fusidic acid. Both mutations also conferred resistance to spectinomycin in Escherichia coli strains containing deletions of all seven chromosomal rrn operons and expressing plasmid-encoded Salmonella rRNA exclusively. In contrast, when expressed in E. coli strains containing intact chromosomal rrn operons, the strains were sensitive to spectinomycin. However, chromosomal mutations arose that allowed expression of the rRNA-dependent spectinomycin resistance phenotype. It is proposed that in heterogeneous rRNA populations, the native E. coli rRNA out-competes the heterologous Salmonella rRNA for binding to ribosomal proteins, translation factors, or ribosome assembly, thus limiting entry of the antibiotic-resistant 30S subunits into the functioning ribosome pool.
Analysis of complex macromolecules such as ribosomes has been facilitated greatly by the isolation of mutant forms of individual ribosomal components. While bacterial ribosomal proteins are typically encoded by single genes, many of the bacteria that are suitable for genetic analysis have multiple copies of the genes encoding 16S, 23S, and 5S rRNAs, thus hampering the genetic analysis of rRNA in bacteria. One approach that has been adopted to overcome this problem has been to construct and express the desired rRNA mutations on a plasmid copy of a selected rrn operon (reviewed in [23]). However, this approach results in a cell that contains a mixture of wild type, chromosomally encoded rRNAs and plasmidencoded mutant rRNAs. An important refinement of this methodology has been to introduce mutations conferring resistance to the antibiotics spectinomycin and erythromycin into the plasmid-associated cistrons encoding 16S and 23S rRNAs, respectively [28]. Addition of the relevant antibiotic to strains expressing such a mixture of Correspondence to: M. O’Connor; email: Michael_O’Connor@ Brown.edu
rRNAs ensures that only the plasmid-encoded rRNA is functional in protein synthesis. While the overwhelming majority of genetic analyses of translational components has been carried out in Escherichia coli, a considerable number of interesting mutant forms of elongation and release factors and ribosomal proteins also exist in Salmonella [11, 12, 24]. To facilitate the genetic analysis of rRNA in Salmonella, we have isolated and characterized mutations conferring resistance to spectinomycin on a plasmid copy of the Salmonella rrnD operon. In addition to providing a useful system for studying rRNA mutations in Salmonella, we have also used the spectinomycin-resistant rRNAs to analyze the antibiotic resistance and ribosome function of hybrid E. coli/Salmonella ribosomes that are produced when Salmonella rRNAs are expressed in E. coli. Materials and Methods Strains, plasmids, and growth media. LT2, the wild-type S. enterica serovar Typhimurium strain was used throughout. Escherichia coli strains DH1 (glnV44 gyrA96 recA1 endA1 gyrA96 thi-1 hsdR17 relA1 spoT1 rfbD), JM109 [⌬(lac-pro) glnV44 gyrA96 recA1 relA1 endA1 thi
430 hsdR17 F⬘ traD36, lacIq ⌬(lacZ)M15], and the ⌬7 prrn strain AVS69009 (⌬rrnE ⌬rrnB ⌬rrnA ⌬rrnH ⌬rrnG::lacZ ⌬rrnC::cat ⌬rrnD::cat ⌬recA/pTRNA66, pHKrrnC; [29]) were used to test pST1derived plasmids for spectinomycin resistance. The tetracycline-resistant, pBR322-derived plasmid pST1 [5] carrying the Salmonella rrnD operon was introduced into Salmonella by CaCl2-mediated transformation [17]. The E. coli rrnB plasmid pKK3535 and its derivatives containing the C1066U and C1192U mutations have been described previously [16, 22]. Luria Bertaini (LB) medium was used for routine cultivation and was supplemented where necessary with tetracycline (25 mg/L), ampicillin (200 mg/L), and spectinomycin (60 mg/L). Minimal inhibitory concentrations (MIC) of antibiotics were determined as described previously [19]. Growth rates of Salmonella strains containing pST1 plasmids were measured by diluting overnight cultures into fresh LB medium containing tetracycline and following the increases in turbidity thereafter with a Klett-Summerson colorimeter. Mutagenesis and characterization of rRNA mutations. Mutagenesis with N-methyl-N⬘-nitro-N-nitrosoguanidine was carried out as described by Miller [17]. Initially, spectinomycin resistance mutations were identified by sequencing the regions corresponding to the 1190 and 1060 regions of 16S rRNA on the plasmid, by using end-labeled primers [7]. Subsequently, restriction fragments encoding 16S rRNA from each mutant plasmid were subcloned into M13 mp18 and sequenced with Sequenase (USB Corporation, Cleveland, OH) according to the manufacturer’s instructions. RNA was extracted from cells by detergent lysis [9]. The percentage of plasmid-derived rRNA in the cell was assessed by a modified primer extension analysis [26] using a primer complementary to bases 1215 to 1193 of 16S rRNA. In this method, the extension mix contains a mixture of deoxy and dideoxy nucleotides; extension by reverse transcriptase on wild-type and mutant rRNA templates is halted at different positions depending on the nature of the rRNA mutations. These primer extension products were separated on a 15% acrylamide/urea gel and quantitated with a Fuji BAS2500 PhosphorImager. The relative proportions of mutant and wild type rRNA were expressed as the percentage of radioactivity incorporated by each primer extension product divided by the sum of the radioactivities incorporated by both products.
Results and Discussion Isolation of spectinomycin-resistant rRNA mutations. The Salmonella enterica serovar Typhimurium strain LT2 transformed with the plasmid pST1, carrying a copy of the Salmonella rrnD operon [5], was mutagenized with N-methyl-N⬘-nitro-N-nitrosoguanidine [17]. Mutagenized cultures were allowed to grow overnight and were then plated on medium containing tetracycline and spectinomycin. Several hundred spectinomycin-resistant colonies appeared after 48 h of growth. In order to separate plasmid-associated, spectinomycin-resistant mutants from chromosomally encoded ribosomal protein mutants, LT2 was transformed with plasmid DNA prepared from cells scraped from the spectinomycin-containing plates, and transformants were selected on plates containing both tetracycline and spectinomycin. Four spectinomycin-resistant transformants were obtained. Plasmid DNA was prepared from cultures of each individual spectinomycin-resistant transformant; LT2 was
CURRENT MICROBIOLOGY Vol. 45 (2002)
Fig. 1. Secondary structure of 16S rRNA with the helix 34 region boxed (left-hand panel) and the sequence and secondary structure of helix 34 of Salmonella serovar Typhimurium 16S rRNA with the base changes conferring spectinomycin resistance indicated (right-hand panel).
again transformed with each of these DNAs and plated on tetracycline-containing medium. All tetracycline-resistant transformants were also spectinomycin resistant. Identification of mutations in 16S rRNA. Previous analyses of rRNA-associated spectinomycin resistance in E. coli and various organelles [4, 8] had shown that the great majority of such mutations were located in helix 34 of the small ribosomal subunit rRNA (Fig. 1). In addition, chemical protection experiments and recent highresolution X-ray structures of ribosomes complexed with the antibiotic indicates that spectinomycin interacts with the G1064 and C1192 regions of helix 34 [6, 18]. Sequence analysis of the four spectinomycin-resistant, pST1-derived plasmids yielded three C1192U mutations and a single C1066U mutation in helix 34 (Fig. 1). The mutant plasmids were designated pST1192U and pST1066U, respectively. Both of these mutations have been reported previously to confer resistance to spectinomycin in E. coli [13, 16, 25]. Both the C1066U and C1192U spectinomycin resistance mutations, as well as other resistance mutations in adjacent nucleotides [4], disrupt conserved base pairs in helix 34, and this has led to the proposal that disruption of base pairing in the spectinomycin binding region of helix 34 is alone sufficient to confer resistance. However, extensive mutagenesis of the 1064 –1192 base pair showed that a G at position 1064 and a C at 1192 (irrespective of the presence or absence of a base pair at these positions) were required for spectinomycin binding [4]. Expression of plasmid-encoded mutant rRNA. Analysis of rRNA-mediated antibiotic resistance mutations has shown that expression of the resistance phenotype requires that a majority of the cellular rRNA be of the mutant form [23, 28]. The percentage of the total cellular
M. O’Connor and A.E. Dahlberg: Spectinomycin Resistance in Salmonella rRNA Table 1. Doubling times and MICs of wild-type and spectinomycinresistant strains
Strain/Plasmid
Doubling time (min)
MIC (g/ml) Spectinomycin
MIC (g/ml) Fusidic acid
LT2 pST1 (wt) LT2 pST1192U LT2 pST1066U
33 ⫾ 1 38 ⫾ 3 39 ⫾ 3
40 ⬎20,000 2,500
1,250 1,250 625
Growth rates were determined in LB medium supplemented with tetracycline (12.5 mg/L) at 37°C. MICs were determined as described in the text [19]. Each growth rate and MIC represents the mean ⫾ SE of at least three independent determinations.
rRNA containing the C1192U mutation was assayed by primer extension analysis [26]. In Salmonella strains containing the pST1192U plasmid, 82% of the rRNA was plasmid encoded. Similar levels of plasmid-based rRNA expression have been determined previously for cloned E. coli rrn operons [21]. Resistance to spectinomycin and fusidic acid in Salmonella. The C1192U mutation conferred resistance to high levels of spectinomycin (MIC ⬎ 20,000 mg/L), while strains expressing the C1066U mutation were eightfold less resistant to the antibiotic (MIC ⫽ 2500 mg/L; Table 1). Spectinomycin is believed to inhibit the EF-G-mediated translocation step of protein synthesis [3], and EF-G- and EF-2-dependent crosslinks and protections of the helix 34 region of the small subunit rRNAs have been described [10, 30]. Fusidic acid is also believed to inhibit the translocation reaction by a mechanism distinct from that of spectinomycin [27], and mutations in EF-G conferring resistance to this antibiotic have been isolated [12]. Salmonella strains expressing wild-type rRNAs or the C1192U mutation in 16S rRNA was equally resistant to fusidic acid (Table 1), while expression of C1066U mutation caused a twofold decrease in the level of resistance, as had previously been observed with this same mutation in E. coli [13]. These data indicated that, while both the C1192U and C1066U mutations conferred resistance to spectinomycin, they exerted a differential interaction with fusidic acid. Minimal effects of the 16S rRNA mutations on growth rate. Despite their effects on antibiotic resistance, the C1066U and C1192U mutations had little effect on cell growth rates (Table 1). This is consistent with the observed lack of effect of the C1192U mutation on the individual steps of translation when mutant E. coli ribosomes were analyzed in vitro [3]. However, we have previously noted that E. coli strains expressing the C1066U mutation exhibit a decreased requirement for 4.5S RNA [22], while expression of the C1192U muta-
431
tion suppresses certain temperature-sensitive EF-G mutants [14]. Thus, while both spectinomycin resistance mutations have little detectable effect on cell growth and ribosome function, interaction of the mutant ribosomes with antibiotics and other ligands is affected in subtle ways. The availability of a cloned Salmonella rrn operon carrying an antibiotic resistance marker that permits the selective usage of plasmid-encoded 16S rRNA should now facilitate genetic analysis of rRNA function in Salmonella. Expression of mutant Salmonella rRNAs in E. coli. Sequence analysis has revealed that many of the genomes of E. coli and Salmonella (including their rRNAs) share extensive sequence identity, and complementation analyses have indicated that many of the genes from one organism are functional in the other [15]. The 16S rRNA of Salmonella differs from that of E. coli at 50 out of a total of 1542 positions (3% difference). The C1192U and C1066U mutations have previously been constructed in an E. coli rrn operon in this laboratory [16, 22]. Introduction of rrn plasmids containing these mutations into the wild-type Salmonella strain LT2 yielded transformants that grew on 60 mg/L of spectinomycin, while the wild-type control plasmid, pKK3535, showed no growth on this level of antibiotic. Surprisingly, however, the converse experiment of transforming a wild-type E. coli strain (DH1) with pST1 plasmids carrying the Salmonella C1066U and C1192U rRNA mutations did not yield any spectinomycin-resistant transformants. Similar results were obtained when the E. coli strain JM109 was used as a host for transformation. Primer extension analyses of RNA samples extracted from duplicate DH1 pST1192U transformants indicated that 78% of the total rRNA was plasmid derived. A similar analysis of RNA from Salmonella strains carrying the same pST1192U plasmid indicated that 82% of the rRNA was plasmid encoded (Table 2). Finally, analysis of RNA from DH1 expressing the C1192U mutation in the E. coli rrnB operon on plasmid pKK1192U indicated that 83% of the rRNA was plasmid encoded. Together, these results show that the failure of the Salmonella C1192U mutation to confer spectinomycin resistance in E. coli does not derive from lack of expression of the heterologous rRNA. The construction of a strain of E. coli containing deletions in all seven rrn operons (⌬7 prrn) and in which all rRNA is plasmid-encoded has facilitated the genetic analysis of rRNA [1]. Recently, it was shown that the wild-type Salmonella rrn plasmid, pST1, used here could replace the resident E. coli rrn plasmid in the ⌬7 prrn strain, and consequently, the hybrid ribosomes from such a strain contained only E. coli ribosomal proteins and
432
CURRENT MICROBIOLOGY Vol. 45 (2002)
Table 2. Expression of 16S rRNA, spectinomycin resistance mutations in Salmonella and E. coli Spectinomycin resistance/sensitivity conferred by rRNA plasmid (% plasmid-encoded rRNA) Strain Salmonella LT2 E. coli AVS69009 E. coli DH1 E. coli DH1 psr-5
pST1066U resistant (82) resistant (100) sensitive (78) resistant (81)
pST1192U
pKK1192U
resistant resistant sensitive resistant
resistant resistant resistant (83) ND
Antibiotic resistance was assessed by streaking on LB plates containing 60 mg/L of spectinomycin. ND, not determined. The E. coli strain DH1 contains seven intact chromosomal rrn operons, whereas AVS69009 contains deletions in all seven rrn operons, and rRNA is expressed exclusively from a plasmid rrn operon.
only Salmonella rRNAs [1]. When the E. coli ⌬7 prrn strain AVS69009 was transformed with pST1192U or pST1066U, spectinomycin-resistant transformants were obtained readily and at the same frequency as with the E. coli rrnB plasmid pKK3535 carrying the C1192U or C1066U mutations. Moreover, all of these transformants had displaced the resident kanamycin-resistant rrn plasmid, pHKrrnC, indicating that in the pST1192U and pST1066U transformants, only Salmonella rRNA was being expressed. These results indicate that, while in heterogeneous populations of ribosomes the Salmonella 16S rRNA C1192U and C1066U mutations did not confer resistance to spectinomycin, resistance was expressed when the same mutant hybrid ribosomes constituted 100% of the population. There are some 50 nucleotide differences between the Salmonella and E. coli 16S rRNAs, and one possible explanation for the anomalous behavior of the hybrid ribosomes is that in heterogeneous populations, as a consequence of some of these nucleotide differences, the native E. coli rRNA may out-compete the Salmonella rRNA for binding to ribosomal proteins or factors. Alternatively, the assembled, hybrid 30S subunits may associate less well with 50S subunits than with native E. coli 30S particles. In each case, the hybrid ribosomes containing the resistance mutations are predicted to be under-represented in the functioning pool of 70S ribosomes. In homogeneous populations of ribosomes, no such competition between Salmonella and E. coli rRNAs exists, and the resistance phenotype is observed. Upon prolonged incubation (1–2 weeks) of DH1 pST1192U transformants on plates containing 60 mg/L of spectinomycin, resistant colonies eventually arose. To explore the basis of this resistance, four of the spectinomycin-resistant isolates were cured of the pST1192U
plasmid by growth in the absence of selection and screening for loss of plasmid-borne tetracycline resistance. Plasmid-free derivatives were then tested on spectinomycin-containing medium (60 mg/L), and all four were sensitive to the antibiotic. When plasmid pST1192U or pST1066U was then re-introduced back into one of the cured strains (DH1 psr-5), this strain again became resistant to spectinomycin. Finally, when DH1 was transformed with plasmid DNA isolated from the spectinomycin-resistant DH1 isolates carrying pST1192U, no spectinomycin-resistant transformants were obtained. Together, these results indicate that resistance was achieved through secondary mutations on the chromosome and that these chromosomal mutations did not themselves confer resistance to spectinomycin, but, rather, they allowed expression of the Salmonella 16S rRNA-associated spectinomycin resistance phenotype. These mutations were designated psr (for potentiation of spectinomycin resistance). In the context of the model presented above, the psr mutations could be in ribosomal protein or translation factor genes such that the mutant ribosomal protein or factor no longer interacts differentially with Salmonella and E. coli 16S rRNAs. Primer extension analysis of the 16S rRNA isolated from the spectinomycin-resistant DH1 psr pST1192U strains showed that the secondary mutations did not affect the level of plasmid-encoded rRNA (Table 2). The identity of these psr mutations is currently being investigated. While the sequences of the rRNAs of Salmonella and E. coli are almost identical, some interesting differences do exist: the 23S rRNA of Salmonella is interrupted by an intron-like element that is removed during endonucleolytic processing [5]. Moreover, while searches for mutant forms of tRNAs and translation factors affecting frameshifting have yielded some identical mutations in both organisms [2], several of the 23S rRNA mutations that affect reading frame maintenance in E. coli [20] failed to exert this effect in Salmonella (M. O’ Connor and A.E. Dahlberg, unpublished observations). The reduced ability of the Salmonella rRNAs to interact with the rest of the E. coli translational apparatus (at least in the context of a heterogeneous population of ribosomes) derives, presumably, from the co-evolution of the rRNAs, ribosomal proteins, and factors in each organism. An important consequence of such differences between related organisms may be to limit expression of components of ribosomes (or other macromolecular complexes) inherited by lateral gene transfer. ACKNOWLEDGMENTS We are indebted to Dr. Norman Pace for supplying plasmid pST1 and to Dr. George Q Pennabble for his saturnine scuttlebutt. This work was
M. O’Connor and A.E. Dahlberg: Spectinomycin Resistance in Salmonella rRNA supported by grant GM19756 to A. E. Dahlberg from the U.S. National Institutes of Health.
Literature Cited 1. Asai T, Zaporojets D, Squires C, Squires CL (1999) An Escherichia coli strain with all chromosomal rRNA operons inactivated: complete exchange of rRNA genes between bacteria. Proc Natl Acad Sci USA 96:1971–1976 2. Atkins JF, Weiss RB, Thompson S, Gesteland RF (1991) Towards a genetic dissection of the basis of triplet decoding and its natural subversion: programmed reading frame shifts and hops. Annu Rev Genet 25:201–228 3. Bilgin N, Richter AA, Ehrenberg M, Dahlberg AE, Kurland CG (1990) Ribosomal RNA and protein mutants resistant to spectinomycin. EMBO J 9:735–739 4. Brink MF, Brink G, Verbeet MP, de Boer HA (1994) Spectinomycin interacts specifically with the residues G1064 and C1192 in 16S rRNA thereby potentially freezing this molecule into an inactive conformation. Nucleic Acids Res 22:325–331 5. Burgin AB, Parodos K, Lane DJ, Pace NR (1990) The excision of intervening sequences from Salmonella 23S ribosomal RNA. Cell 60:405– 414 6. Carter AP, Clemons WM, Brodersen DE, Morgan-Warren RJ, Wimberly BT, Ramakrishnan V (2000) Functional insights from the structure of the 30S ribosomal subunit and its interactions with antibiotics. Nature 407:340 –348 7. Chen EY, Seehurg PH (1985) Supercoil sequencing: a fast and simple method for sequencing plasmid DNA. DNA 4:165–170 8. Fromm H, Edelman M, Aviv D, Galun E (1987) The molecular basis for rRNA-dependent spectinomycin resistance in Nicotiana chloroplasts. EMBO J 6:3233–3237 9. Godson GN, Sinsheimer RL (1967) Lysis of Escherichia coli with a neutral detergent. Biochim Biophys Acta 149:476 – 488 10. Holmberg L, Nygard O (1994) Interaction sites of ribosome-bound eukaryotic elongation factor 2 in 18S and 28S rRNA. Biochemistry 33:15159 –15167 11. Hughes D, Atkins JF, Thompson S (1987) Mutants of elongation factor Tu promote ribosomal frameshifting and nonsense readthrough. EMBO J 6:4235– 4239 12. Johanson U, Hughes D (1994) Fusidic acid-resistant mutants define three regions in elongation factor G of Salmonella typhimurium. Gene 143:55–59 13. Johanson U, Hughes D (1995) A new mutation in 16S rRNA of Escherichia coli conferring spectinomycin resistance. Nucleic Acids Res 23:464 – 466 14. Koosha H, Cameron D, Andrews K, Dahlberg AE, March PE (2000) Alterations in the peptidylytransferase and decoding domains of ribosomal RNA suppress mutations in the elongation factor G gene. RNA 6:1166 –1173 15. Krawiec S, Riley M (1990) Organization of the bacterial chromosome. Microbiol Rev 54:502–539
433
16. Makosky PC, Dahlberg AE (1987) Spectinomycin resistance at site 1192 in 16S ribosomal RNA of E. coli: an analysis of three mutants. Biochimie 69:885– 889 17. Miller JH (1991) A short course in bacterial genetics. Cold Spring Harbor, New York: Cold Spring Harbor Laboratory Press 18. Moazed D, Noller HF (1987) Interaction of antibiotics with functional sites in 16S ribosomal RNA. Nature 327:389 –394 19. National Committee for Clinical Laboratory Standards (2000) Methods for dilution antimicrobial susceptibility disk tests for bacteria that grow aerobically, 5th edn., Vol. 20, No. 2. Approved standard M7-A5. National Committee for Clinical Laboratory Standards, Wayne, PA 20. O’Connor M, Dahlberg AE (1995) The involvement of two distinct regions of 23S ribosomal RNA in tRNA selection. J Mol Biol 254:838 – 847 21. O’Connor M, Goringer HU, Dahlberg AE (1992) A ribosomal ambiguity mutation in the 530 loop of E. coli 16S rRNA. Nucleic Acids Res 20:4221– 4227 22. O’Connor M, Brunelli CA, Firpo MA, Gregory ST, Lieberman KR, Lodmell JS, Moine H, Van Ryk DI, Dahlberg AE (1995) Genetic probes of ribosomal RNA function. Biochem Cell Biol 73:859 – 868 23. O’Connor M, Bayfield MA, Gregory ST, Lee W-CM, Lodmell JS, Mankad A, Thompson JR, Vila-Sanjurgo A, Squires CL, Dahlberg AE (1999) In: The ribosome: structure, function, antibiotics and cellular interactions. Garrett RA, Douthwaite SR, Liljas A, Matheson AT, Moore PB, Noller HF (eds). Washington, DC: American Society for Microbiology, pp 217–228 24. Olafsson O, Ericson JU, VanBogelen R, Bjork GR (1996) Mutation in the structural gene for release factor 1 (RF-1) of Salmonella typhimurium inhibits cell division. J Bacteriol 178:3829 –3839 25. Sigmund CD, Ettayebi M, Morgan EA (1984) Antibiotic resistance mutations in 16S and 23S ribosomal RNA genes of Escherichia coli. Nucleic Acids Res 12:4653– 4663 26. Sigmund CD, Ettayebi M, Borden A, Morgan EA (1988) Antibiotic resistance mutations in ribosomal RNA genes of Escherichia coli. Methods Enzymol 164:673– 690 27. Tanaka N, Kinoshita T, Masukawa H (1968) Mechanism of protein synthesis inhibition by fusidic acid and related antibiotics. Biochem Biophys Res Commun 30:278 –283 28. Triman K, Becker E, Dammel C, Katz J, Mori H, Douthwaite S, Yapijakis C, Yoast S, Noller HF (1989) Isolation of temperaturesensitive mutants of 16S rRNA in Escherichia coli. J Mol Biol 209:645– 653 29. Vila-Sanjurjo A, Squires CL, Dahlberg AE (1999) Isolation of kasugamycin resistant mutants in the 16S ribosomal RNA of Escherichia coli. J Mol Biol 293:1– 8 30. Wilson KS, Noller HF (1998) Mapping the position of translational elongation factor EF-G in the ribosome by directed hydroxyl radical probing. Cell 92:131–139