Ribosomal protein L15 as a probe of 50 S ribosomal subunit structure1

Article No. mb982236

J. Mol. Biol. (1998) 284, 1367±1378

Ribosomal Protein L15 as a Probe of 50 S Ribosomal Subunit Structure Kate R. Lieberman and Harry F. Noller* Center for the Molecular Biology of RNA, University of California, Santa Cruz CA 95064, USA

L15, a 15 kDa protein of the large ribosomal subunit, interacts with over ten other proteins during 50 S assembly in vitro. We have probed the interaction L15 with 23 S rRNA in 50 S ribosomal subunits by chemical footprinting, and have used localized hydroxyl radical probing, generated from Fe(II) tethered to unique sites of L15, to characterize the threedimensional 23 S rRNA environment of L15. Footprinting of L15 was done by reconstituting puri®ed, recombinant L15 with core particles derived from Escherichia coli 50 S subunits by treatment with 2 M LiCl. The cores migrate as compact 50 S-like particles in sucrose gradients, contain 23 S and 5 S rRNA, and lack a subset of the 50 S proteins, including L15. Using both Fe(II)
EDTA and dimethyl sulfate, we have identi®ed a strong footprint for L15 in the region spanning nucleotides 572-654 in domain II of 23 S rRNA. This footprint cannot be detected when L15 is incubated with ``naked'' 23 S rRNA, indicating that formation of the L15 binding site requires a partially assembled particle. Protein-tethered hydroxyl radical probing was done using mutants of L15 containing single cysteine residues at amino acid positions 68, 71 and 115. The mutant proteins were derivatized with 1-[p-(bromo-acetamido)benzyl]-EDTA
Fe(II), bound to core particles, and hydroxyl radical cleavage was initiated. Distinct but overlapping sets of cleavages were obtained in the footprinted region of domain II, and in speci®c regions of domains I, IV and V of 23 S rRNA. These data locate L15 in proximity to several 23 S rRNA elements that are dispersed in the secondary structure, consistent with its central role in the latter stages of 50 S subunit assembly. Furthermore, these results indicate the proximity of these rRNA regions to one another, providing constraints on the tertiary folding of 23 S rRNA. # 1998 Academic Press

*Corresponding author

Keywords: ribosome; 50 S subunit; protein-RNA interactions; chemical probing; hydroxyl radical probing

Introduction The 50 S ribosomal subunit is a complex ribonucleoprotein particle composed of two highly conserved RNA molecules (5 S and 23 S rRNA) and over 30 proteins. The 50 S subunit has a correspondingly complex functional repertoire, requiring intricate and highly coordinated interactions with the 30 S ribosomal subunit, elongation factors, and tRNA substrates as they move through the ribosome during protein synthesis. Catalysis of peptide bond formation is an integral activity of Abbreviations used: DMS, dimethyl sulfate; BABE, 1-[p-(bromo-acetamido)benzyl]-EDTA. E-mail address of the corresponding author: [email protected] 0022±2836/98/501367±12 $30.00/0

the 50 S subunit, and considerable evidence indicates that 23 S rRNA is a direct participant in peptidyl transferase function (Samaha et al., 1995; Nitta et al., 1998a,b; reviewed by Green & Noller, 1997). Precise three-dimensional folding of rRNA is undoubtedly critical for peptidyl transferase activity, and for integration of the catalytic step in the complex and dynamic process of protein synthesis. Thus a detailed understanding of the tertiary structure of 23 S rRNA and of rRNA-protein architecture in the 50 S subunit is essential for a mechanistic description of protein synthesis. However, in comparison with the less complex 30 S subunit, our knowledge of the higher-order structure of the 50 S subunit is considerably less advanced. # 1998 Academic Press

1368 A major contribution to our understanding of higher order interactions in the 30 S subunit came from a series of chemical probing experiments that determined the base-speci®c (Stern et al., 1988b) and RNA-backbone speci®c (Powers & Noller, 1995a) protein footprints on 16 S rRNA generated during de®ned stages of in vitro assembly of the Escherichia coli 30 S subunit. However, due to the signi®cantly greater complexity of in vitro reconstitution of 50 S subunits, (Dohme & Nierhaus, 1976) this approach has not yet been applied to structural studies of the 50 S subunit, and footprinting experiments have for the most part been restricted to those proteins that interact independently with 23 S or 5 S rRNA (Garrett & Noller, 1979; Egebjerg et al., 1987, 1990a, 1991; Leffers et al., 1988). Recently, site-directed hydroxyl radical probing experiments using Fe(II) tethered to speci®c positions of 30 S subunit proteins have produced strong constraints for the three-dimensional folding of 16 S rRNA in the 30 S subunit (Heilek et al., 1995; Heilek & Noller, 1996). This tethered probing approach permits the simultaneous detection of multiple rRNA elements and thus the assignment of rRNA tertiary proximities, as well as proteinrRNA proximities. Application of this methodology to the study of 50 S subunit structure may prove particularly valuable, as data from RNARNA crosslinking experiments (Mitchell et al., 1990) and phylogenetic tertiary covariations (Gutell & Woese, 1990) suggest an RNA architecture in the 50 S subunit that is rich in long-range, interdomain interactions. As an alternative approach to using intermediates of 50 S subunit in vitro assembly for protein footprinting and protein-tethered hydroxyl radical probing experiments, we chose to use 50 S subunits from which a subset of proteins had been selectively removed. In this study, we have probed the interaction of ribosomal protein L15 with 23 S rRNA by binding L15 to compact, 50 S-like core particles derived from 50 S subunits by treatment with LiCl (Moore et al., 1975). The core particles contain 23 S rRNA and 5 S rRNA, and lack a subset of the 50 S proteins, including L15. L15, a 15 kDa basic protein, is implicated in a central role in the latter stages of 50 S subunit assembly during in vitro reconstitution of Escherichia coli subunits, where it promotes the assembly of more than ten other 50 S subunit proteins (RoÈhl & Nierhaus, 1982; Herold & Nierhaus, 1987). Recently, the L15 homologue from Thermus aquaticus has been identi®ed as one of only eight proteins that remain associated with 23 S rRNA in catalytically active particles derived from 50 S subunits by treatment with SDS, proteolysis and extensive extraction with phenol, leading to the suggestion that these proteins are ``caged'' by the rRNA when it is folded in an active conformation (Noller et al., 1992; Khaitovich et al., 1998). L15 therefore seemed an excellent candidate for probing 50 S subunit structure. To our knowledge, no

L15 as a Probe of 50 S Ribosomal Subunit Structure

previous footprinting or RNA-protein crosslinking data have been reported for this protein. When L15 was assembled with protein-depleted, LiCl core particles, we identi®ed a strong footprint for L15 in the region spanning nucleotides 572-654 in domain II of 23 S rRNA. This footprint cannot be detected when L15 is incubated with ``naked'' 23 S rRNA, indicating that formation of the L15binding site requires a partially assembled particle. We have also used localized hydroxyl radical probing, generated from Fe(II) tethered to three different positions of L15, to characterize its threedimensional 23 S rRNA environment. While protections by L15 from chemical probes delivered from solution are restricted to domain II of 23 S rRNA, the results of tethered probing experiments indicate that speci®c regions of domains I, II, IV and V, distal to one another in the secondary structure of 23 S rRNA, are proximal to L15 and to one another in the three-dimensional structure of this 50 S particle.

Results Chemical footprinting of L15 Methods for removing subsets of proteins from 50 S subunits using CsCl or LiCl have previously been employed to study the protein requirements for reconstitution of 50 S subunit function (Staehelin et al., 1969; Nierhaus & Nierhaus, 1973; Kazemie, 1975; Moore et al., 1975; Tate et al., 1983), and as an initial step in 50 S protein puri®cation (Wystup et al., 1979). Treatment of E. coli 50 S subunits with 2 M LiCl, followed by high-speed centrifugation, yields subunit cores that migrate as compact 50 S-like particles in sucrose gradients (Figure 1(a)). The cores contain 23 S and 5 S rRNA, and a subset of 50 S proteins, including stoichiometric amounts of L2, L3, L4, L13, L14, L17, L19, L21, L22, L23, L24, L29 and L30. The supernatant contains the remaining proteins, or ``split proteins'', in which L6, L10, L11, L7/L12, L15, L16, L26, L27, L28, L32 and L33 are represented stoichiometrically. The proteins L1, L5, L9, L18 and L25 are partially extracted from the cores and are thus substoichiometrically present in both the cores and in the split protein fraction (Moore et al., 1975; Wystup et al., 1979). Core particles lack peptidyl transferase activity, which can be partially reconstituted by incubation with the split proteins (Figure 1(b)). As demonstrated (Moore et al., 1975), catalytic activity can be partially restored by the addition of puri®ed L16 (data not shown). L15 is among those proteins removed from the core particles. We assembled core particles with puri®ed recombinant L15, and probed the accessibility of 23 S and 5 S rRNA in the reconstituted particles to sugar-phosphate backbone cleavage by hydroxyl radicals generated from Fe(II)
EDTA (Figure 2(a)), or to chemical modi®cation (at N-1 of adenine and N-3 of cytosine residues) by dimethyl

L15 as a Probe of 50 S Ribosomal Subunit Structure

Figure 1. (a) Sedimentation pro®les (A254) of 50 S ribosomal subunits, 2 M LiCl cores, and cores reconstituted with the 2 M LiCl split protein (SP) fraction. (b) Peptidyl transferase activity of (~) 50 S ribosomal subunits, (^) 2 M LiCl cores, and (*) cores reconstituted with the 2 M LiCl split protein fraction. Peptidyl transferase activity was measured using the fragment reaction (see Materials and Methods). Values plotted are the average from four experiments.

sulfate (DMS; Figure 2(b) and (c)). We compared the resulting modi®cation patterns to those obtained with cores alone, with cores reconstituted with split proteins, or with intact 50 S subunits by primer extension analysis of all of 23 S and 5 S rRNA.

1369 The patterns of rRNA reactivity in core particles with Fe(II)
EDTA and DMS were identical with rRNA probed in 50 S subunits in most regions of the 23 S and 5 S rRNA chains. Notably, however, discrete regions of 23 S rRNA, some of which may correspond to binding sites for proteins removed from the cores, were more exposed in the cores. L15 binding to core particles protected nucleotides 625-652 in domain II of 23 S rRNA from cleavage by Fe(II)
EDTA (Figure 2(a)), and strongly protected nucleotides A572, A626, A643, A644 and A645 from modi®cation by DMS (Figure 2(b)). Residue A631 was moderately protected from DMS by L15 binding, and position A654 was very weakly protected. While residues A586 and A587 are very weakly modi®ed by DMS in core particles, they are completely protected from modi®cation in 50 S subunits and in core particles that have been reconstituted with either split proteins or puri®ed L15. In addition, binding of L15 to cores conferred strong protection of position A990 in domain II from modi®cation by DMS (Figure 2(c)). No L15-induced chemical protection was observed elsewhere in 23 S rRNA, or in 5 S rRNA. The relative extent of the protections was estimated by visual inspection of band intensities. The L15 footprinting sites are summarized on a map of the secondary structure of domain II of 23 S rRNA (Figure 3). In these regions of domain II of 23 S rRNA, binding of L15 alone to the core particles yields a pattern of protection from Fe(II)
EDTA cleavage that is indistinguishable from that obtained in intact 50 S subunits, or when core particles are reconstituted with the complete split protein fraction. Similarly, L15 binding alone reproduced most of the pattern of reactivity toward DMS obtained in complete 50 S particles in this region, with some exceptions. Nucleotide A574 is unreactive toward DMS in 50 S subunits, but moderately reactive in core particles. Curiously, the reactivity of this residue was enhanced in core particles reconstituted with split proteins relative to cores alone, but unchanged in particles reconstituted with puri®ed L15. A574 thus represents one of the few sites in 23 S rRNA where reconstitution of cores with split proteins does not restore the pattern of nucleotide accessibility obtained in 50 S subunits. Protection of residues A631 and A632 from DMS modi®cation by L15 alone is not as strong as that obtained in cores reconstituted with split proteins, or in 50 S subunits. Finally, the DMS reactivity of A633 in 50 S subunits is enhanced relative to core particles, and this enhancement is nearly, but not completely restored by reconstitution with either split proteins or L15 alone. Since it has been reported that L15 can bind directly to 23 S rRNA in the absence of other ribosomal proteins (Littlechild et al., 1977; Marquardt et al., 1979), we tested whether L15 would confer protection of naked 23 S rRNA from chemical probes in the 570-655 region. We prepared rRNA from 50 S subunits according to a method that

1370

L15 as a Probe of 50 S Ribosomal Subunit Structure

Figure 2. Primer extension analysis of chemical modi®cation of 23 S rRNA with (a) Fe(II)
EDTA (primer 806) or DMS ((b) primer 806; (c), primer 1001) in (1) 50 S subunits, (2) 2 M LiCl cores, (3) cores reconstituted with total split proteins, or with (4) puri®ed protein L15. and A and G, Dideoxy sequencing lanes; (K1), 50 S subunits, unmodi®ed; K2, cores, unmodi®ed.

yields rRNA competent for total 50 S reconstitution (Nierhaus, 1990). We incubated L15 with 23 S rRNA under conditions of either the ®rst step, the second step, or the ®rst followed by the second step of total 50 S reconstitution (Nierhaus, 1990), and then probed with Fe(II)
EDTA or DMS. We were unable to detect a footprint for L15 in this region of domain II using any of these conditions (data not shown), suggesting that formation of the L15 binding site requires a partially assembled 50 S particle. Site-directed hydroxyl radical probing with L15 The reagent 1-[p-(bromo-acetamido)benzyl]EDTA (BABE) has been employed to tether Fe(II) to speci®c sites on small subunit ribosomal pro-

teins via alkylation of cysteine thiol groups (Heilek et al., 1995; Heilek & Noller, 1996). To generate sites for attachment of BABE
Fe(II) in L15, we introduced mutations encoding single cysteine residues at positions 9, 35, 68, 71, 74, 85, 115, 117, 129 and 142. Since the three-dimensional structure of L15 has not been determined, we used an amino acid sequence alignment of L15 homologues from 14 organisms (S. Mian, B. Weiser & H.F.N., unpublished results) to chose amino acid residues with high phylogenetic variability as sites for cysteine mutations. Mutant L15 proteins were overexpressed, puri®ed, and modi®ed with Fe(II)
BABE (see Materials and Methods). The derivatized proteins were incubated with core particles under reconstitution conditions, and binding of the modi-

L15 as a Probe of 50 S Ribosomal Subunit Structure Table 1. Preincubation of 2 M LiCl Cores with L15 or Fe(II)
BABE-modi®ed L15 mutants stimulates activity of reconstituted particles Preincubation (50 C, 60 minutes) None None None None Cores Cores Cores ‡ L15 (wt, unmodified) Cores ‡ L15-68C-Fe(II)
BABE Cores ‡ L15-71C-Fe(II)
BABE Cores ‡ L15-115C-Fe(II)
BABE

Reconstitution Peptidyl incubation transferase (50 C, 90 minutes) activity (%) 50 S Cores Cores‡L15 Cores‡split proteins No further addition Add split proteins Add split proteins Add split proteins Add split proteins Add split proteins

100 1  0.2 2  0.2 28  3 1  0.1 29  3 44  3 46  5 41  4 45  4

Peptidyl transferase activity was determined for cores reconstituted with split proteins, with or without preincubation with wild-type (wt), unmodi®ed L15 or Fe(II)
BABE modi®ed L15 mutants. Core particles were incubated under the indicated conditions and the reconstituted products were assayed for peptidyl transferase activity as described in Materials and Methods. All rates were determined from time-points in the linear range of the reaction, and are expressed relative to the activity of 50 S subunits.

®ed proteins was assayed by chemical probing of 23 S rRNA with DMS. L15 derivatized at positions 68, 71 and 115 protected the same nucleotides as did wild-type, unmodi®ed L15, although the footprints were not as strong as those obtained with the unmodi®ed protein (data not shown). L15 proteins modi®ed at positions 9, 35, 74, 85, 117, 129 and 142, however, all failed to yield a footprint in domain II of 23 S rRNA, suggesting that these positions of L15 may be important for assembly into 50 S subunits. Although binding of L15 alone to core particles does not restore peptidyl transferase activity, preincubation of the cores with L15 prior to reconstitution with the complete split protein fraction leads to a small but reproducible stimulation of the activity of reconstituted particles, and L15 mutant proteins derivatized with BABE at positions 68, 71 or 115 stimulated activity of the reconstituted particles to the same extent as wild-type, unmodi®ed L15 (Table 1). Since there is some degree of heterogeneity in the protein composition of the core particles, this stimulation may re¯ect an L15-induced change in a subpopulation of particles that renders them more competent for subsequent assembly steps. To probe the rRNA environment of L15, the Fe(II)
BABE-modi®ed proteins were bound to core particles, and unbound protein was removed with Microcon 100 ultra®ltration units. Wild-type L15, which contains no cysteine residue, was subjected to mock modi®cation with Fe(II)
BABE and reconstituted with core particles, as a negative control in the probing experiments. Puri®ed reconstituted particles were recovered and hydroxyl radicals were generated by addition of hydrogen peroxide and ascorbic acid. Sites of cleavage of the rRNA

1371 backbone were identi®ed by primer extension analysis of 23 S and 5 S rRNA. Strong cleavages have been shown to occur at backbone positions within Ê of the tethered probe, while medium and 0-22 A weak cleavages are obtained within ranges of Ê and 20-44 A Ê of the probe, respectively 12-36 A (Joseph et al., 1997), in good agreement with previously observed ranges (Han & Dervan, 1994). Primer extension gels showing regions of 23 S rRNA in which Fe(II) tethered to L15 produced rRNA strand scission are shown in Figure 4. In the region of domain II footprinted by L15 in the experiments described above, sets of overlapping cleavages were obtained from the three Fe(II)-derivatized L15 positions. Nucleotides 631-638, 640-655 and 665-667 were weakly attacked when Fe(II) was tethered to position 68 of L15, and weak cleavages were obtained spanning nucleotides 591-594, 622638, 640-656 and 665-667 from position 71. Nucleotides 652-655 were strongly targeted, and nucleotides 601-604, 635-638, 640-646, 650-651 and 656661 were more weakly cleaved, when Fe(II) was tethered to position 115 (Figure 4(b)). Hydroxyl radicals generated from position 115 also attacked nucleotides 842-855, in a separate region of domain II (Figure 4(c)). In domain I of 23 S rRNA, L15 derivatized at positions 68 and 71, but not 115, yielded moderately strong cleavages in the region spanning nucleotides 242-258 (Figure 4(a)). In domain IV, nucleotides 1870-1873 were targeted from all three derivatized positions of L15 (Figure 4(d)), although the cleavages from positions 71 and 115 were weaker than those generated from position 68. In domain V, Fe(II) attached to position 71 weakly attacked nucleotides 2360-2363, while nucleotides 2401-2410 and 2413-2418 were weakly cleaved from both positions 68 and 71. The cleavages from position 115 in domain V were distinct from, and interdigitated with, the cleavages from the other two L15 positions. L15 derivatized at position 115 weakly targeted nucleotides 2349-2352, 2365-2371 and 2401-2402 (Figure 4(e)). No cleavage was observed in 5 S rRNA. The results of the tethered probing experiments are summarized on the secondary structure of 23 S rRNA (Figure 5). All of the observed rRNA strand scissions resulted from speci®cally bound L15, as all were abolished by inclusion of a fourfold molar excess of unmodi®ed, wild-type L15 (over the concentration of modi®ed protein) in the reconstitution reactions (data not shown). Identical cleavage patterns were obtained when cores were reconstituted with a combination of BABE-modi®ed L15 and unmodi®ed L16, which yields a partially active particle (data not shown). When the cores were reconstituted with BABE-modi®ed L15 and split proteins to form complete particles, all of the tethered cleavages were signi®cantly diminished, such that only the very strongest (for example, those in the 650-655 region from position 115) were still readily detectable (data not shown). We cannot distinguish whether the diminished cleavage inten-

1372

L15 as a Probe of 50 S Ribosomal Subunit Structure

Figure 3. Diagrams of the secondary structure of domain II of E. coli 23 S rRNA showing the nucleotides protected from (a) cleavage by Fe(II)
EDTA or (b) modi®cation by DMS by binding L15 to 2 M LiCl core particles. The relative extent of the protections was estimated as strong, medium or weak by visual comparison of band intensities and is indicated by large, medium or small dots, respectively.

sities result from competition for binding of derivatized L15 by the wild-type, unmodi®ed L15 present in the split proteins, or whether the assembly of the split proteins shields rRNA regions, previously exposed in the cores, from hydroxyl radical attack.

Discussion We have probed the interaction of ribosomal protein L15 with 23 S rRNA using both chemical footprinting and directed hydroxyl radical probing from L15-tethered Fe(II), by binding L15 to core particles derived from 50 S subunits by treatment with LiCl. In domain II of 23 S rRNA, L15 chemical footprints and L15-tethered cleavages encompass nearly the entire cloverleaf-like secondary stuctural element spanned by nucleotides 600-655, implicating this element as a major determinant of L15-rRNA interaction. Interestingly, no footprint or

directed cleavage was observed in most of the right-hand stem-loop of this structure (nucleotides 607-622). The loop end of this element (nucleotides 613-617) is the site of one of two crosslinks obtained between ribosomal protein L4 and 23 S rRNA in 50 S subunits using 2-iminothiolane (Gulle et al., 1988). In domain I of 23 S rRNA, we observed cleavages of nucleotides 242-258 from positions 68 and 71 of L15. Nearby, L4 has been crosslinked to nucleotides 320-325 of domain I (Gulle et al., 1988). L4 is present in the core particles and, indeed, in vitro assembly of L15 into 50 S subunits is dependent upon the presence of L4 (RoÈhl & Nierhaus, 1982). In a model for the arrangement of the 50 S subunit proteins based on immunoelectron microscopy and protein-protein crosslinks (Walleczek et al., 1988), L4 has been predicted to be proximal to L15. The results of our probing experiments thus indicate that L15, L4, and speci®c

L15 as a Probe of 50 S Ribosomal Subunit Structure

1373

Figure 4. Primer extension gels showing hydroxyl radical cleavage in 23 S rRNA from Fe(II) tethered to positions 68, 71 or 115 of L15 reconstituted into 2 M LiCl core particles. (a) Primer 454; (b) primer 806; (c) primer 1001; (d) primer 1945; (e) primer 2493. Directed probing was done using cores reconstituted with (1 and 5) wild-type (cysteinefree) L15 subjected to the Fe(II)
BABE derivatization reaction; (2) Fe(II)-L15-68C; (3) Fe(II)-L15-71C; (4) Fe(II)-L15115C. A and G, Dideoxy sequencing lanes; K1 and K2, cores alone. Complexes in lanes K2-5 were treated with H2O2 and ascorbic acid.

elements of both domains I and II of 23 S rRNA are intimately associated in the 50 S subunit. Probing from all three L15 positions targeted the region spanning nucleotides 2349-2418 of domain V. Cleavages from both positions 68 and 71 covered nearly the entire stem-loop structure containing nucleotides 2402-2419, while, as discussed above, Fe(II) tethered to these two L15 positions yielded cleavages in domain I. Phylogenetic sequence covariation between nucleotides 413 and 416 in domain I and between nucleotides 2407 and

2410 in domain V provides evidence for interaction between these two regions of 23 S rRNA (Egebjerg et al., 1990b; Gutell & Woese, 1990). Directed cleavage from all three L15 positions targeted nucleotides 1870-1873 in domain IV. Mounting evidence supports the intimate association of this part of 23 S rRNA with domain V, where we observed directed cleavages in the 2400 region (described above). This evidence includes several intra-subunit RNA-RNA crosslinks (Mitchell et al., 1990), phylogenetic sequence covar-

1374

L15 as a Probe of 50 S Ribosomal Subunit Structure

Fig. 5(a) and (b) (legend opposite)

L15 as a Probe of 50 S Ribosomal Subunit Structure

1375

Figure 5. Diagrams of the secondary structure of E. coli 23 S rRNA showing sites of directed hydroxyl radical cleavage from Fe(II) tethered to positions (a) 68, (b) 71 and (c) 115 of L15. Cleavages were rated as strong, medium or weak (indicated by large, medium or small dots, respectively) by visual comparison of the intensity of gel bands corresponding to the tethered cleavages with the intensity of bands in adjacent sequencing lanes (Joseph et al., 1997).

iation (Gutell & Woese, 1990), protection from chemical modi®cation by tRNA of nucleotides in both domains (Moazed & Noller, 1989), and speci®c crosslinking of aminoacyl-tRNA derivatives to both domains (Barta et al., 1984; Wower et al., 1989; Mitchell et al., 1993). Strikingly, nucleotides in both domains IV and V targeted by directed hydroxyl radical cleavage from L15 overlap 23 S rRNA positions cleaved by directed probing using tRNA derivatized with Fe(II) at its 50 end bound to the E (exit) site of 70 S ribosomes (Joseph & Noller, 1996); furthermore, the directed cleavages from L15-tethered Fe(II) observed in the 2400 region of domain V ¯ank position 2394, which is strongly protected from DMS modi®cation by E-site tRNA (Moazed & Noller, 1989). This suggests that L15, along with the regions of domains I and II to which it is proximal, are near the path of tRNA as it exits the 50 S subunit during protein synthesis. In Thermus aquaticus 50 S subunits, L15 is one of eight proteins that display strong resistance to both extraction with phenol and proteolysis when associated with folded, intact 23 S rRNA in catalytically active particles; when these particles are treated with ribonuclease, these proteins become protease-sensitive (Noller et al., 1992; Khaitovich, et al., 1998). Thus these authors have suggested

that the protease-resistant proteins may be ``caged'' in a highly folded rRNA structure. The results of the footprinting and directed hydroxyl radical probing experiments in this study indicating that L15, a ribosomal protein of only 15 kDa, is proximal to speci®c elements of no less than four of the six 23 S rRNA secondary structural domains, is consistent with this proposal. There is also a suggestive parallel with the behavior of L15 in the in vitro assembly of E. coli 50 S subunits, where it promotes the assembly of more than ten other proteins (RoÈhl & Nierhaus, 1982; Herold & Nierhaus, 1987), although there are 50 S subunit assembly pathways that can bypass the requirement for L15 (Lotti et al., 1983; Franceschi & Nierhaus, 1990). Our studies suggest that L15 is anchored to an rRNA structure in the 600-655 region of domain II following some earlier assembly steps, and becomes surrounded in active, mature 50 S subunits by an interdomain 23 S rRNA network held together by higher-order interactions that remain to be identi®ed. Our ®nding that elements of domains I, II, IV and V are proximal to L15 when it is bound to compact, 50 S-like core particles demonstrates the proximity of these distal rRNA secondary structural elements to one another, providing strong constraints for the three-dimensional folding of 23 S rRNA in the 50 S subunit.

1376

Materials and Methods Preparation of 50 S subunits, 2M LiCl cores, and split proteins The 50 S ribosomal subunits were prepared from E. coli MRE600 cells by zonal sucrose-gradient centrifugation (Nierhaus, 1990) using the buffer conditions described (Moazed et al., 1986). The 2 M LiCl cores and split proteins were prepared from 50 S subunits as described (Moore et al., 1975). The 50 S reconstitution reactions and peptidyl transferase assays Reconstitution of 50 S subunits from 2 M LiCl cores and the 2 M LiCl split protein fraction was done essentially as described (Moore et al., 1975) with minor modi®cations. Cores (40 pmol) were incubated at 50 C for 90 minutes in 50 ml of 20 mM Tris HCl (pH 7.5), 400 mM NH4Cl, 20 mM MgCl2, 4 mM b-mercaptoethanol (reconstitution buffer), with protein additions as indicated. The concentration of the split protein fraction required for reconstitution of maximum peptidyl transferase activity was determined by titration. Aliquots (8 pmol) of the reconstitution reactions were assayed for peptidyl transferase activity under fragment reaction conditions (Noller et al., 1992) using CACCA-N-Ac-[35S]Met as P-site substrate and 1 mM puromycin as A-site substrate. Aliquots were withdrawn during the linear time range of the reaction, unreacted fragment substrate and the methyl ester side product hydrolysed by incubation in 1 M NaOH for 20 minutes at 37 C, after which the solution was brought to pH 5.5 by the addition of 0.3 M sodium acetate saturated with MgSO4. The product N-Ac-[35S]Met-puromycin was selectively extracted with ethyl acetate followed by scintillation counting of the ethyl acetate phase. Sucrose-gradient analysis The 2 M LiCl cores (40 pmol), cores reconstituted with split proteins, or 50 S subunits were analyzed by centrifugation through 5.5 ml 10%-40% (w/v) sucrose gradients formed in reconstitution buffer. Gradients were spun at 28,000 rpm in an SW41 rotor for 18.5 hours and eluted with an ISCO gradient fractionator equipped with a UV detector set at 254 nm. Cloning, mutagenesis, and purification of L15 The coding sequence of the gene for E. coli ribosomal protein L15 was ampli®ed by PCR using primers determined from the published nucleotide sequence (Cerretti et al., 1983). The L15 ampli®cation product was cloned into the NdeI-BamHI sites of pET-24b (Novogen), under transcriptional control of the phage T7 RNA polymerase promoter, and sequenced in its entirety. Mutations encoding cysteine residues were introduced into the L15 gene at positions 68, 71 or 115 by site-directed mutagenesis (Kunkel, 1985), and con®rmed by sequencing. The mutant and wild-type versions of L15 were overexpressed in E. coli strain BL21(DE3) (F-ompT hsdSB(rB-mB-) gal dcm) following induction with isopropylthio-b-Dgalactoside. Cells were lysed by sonication, inclusion bodies containing L15 were recovered by centrifugation and solublilized in 20 mM Tris-HCl (pH 7.5), 20 mM KCl, 4 mM b-mercaptoethanol containing 6 M urea. L15

L15 as a Probe of 50 S Ribosomal Subunit Structure was puri®ed by cation-exchange chromatography on a Resource S FPLC column (16 cm  30 cm Pharmacia) using a linear gradient of 20 mM-350 mM KCl in the above buffer. Urea was removed by dialysis against core reconstitution buffer and proteins were stored in small aliquots at ÿ70 C. Chemical probing and primer extension analysis The 2 M LiCl cores (40 pmol) were incubated in 50 ml with split proteins or with 80 pmol of L15 under reconstitution conditions. The 50 S subunits were subjected to a mock reconstitution incubation. Hydroxyl radical probing with Fe(II)
EDTA, DMS probing, isolation of modi®ed rRNA, and primer extension of 23 S and 5 S rRNA were done as described (Stern et al., 1988a; Powers & Noller, 1995b). Numbers assigned to primers represent the rRNA template position complementary to the ®rst incorporated nucleotide of the synthesized cDNA chain. BABE modification of L15 mutants and L15 tethered probing The Fe(II)
BABE chelation complex was formed as described (Heilek et al., 1995). Two nanomoles of each of the mutant L15 proteins, as well as wild-type L15 (containing no cysteine) was reacted with Fe(II)
BABE (®nal concentration 140 mM) in 100 ml of reconstitution buffer at 37 C for 30 minutes, after which unreacted Fe(II)
BABE was removed using Microcon 3 ultra®tration units (Amicon). Modi®ed proteins (160 pmol) were reconstituted in 50 ml with 40 pmol of core particles, after which unbound protein was removed using Microcon 100 ultra®ltration units (Amicon). Hydroxyl radical cleavage was initiated by addition of 1ml each of 250 mM ascorbic acid and 2.5% (v/v) H2O2. After ten minutes on ice, the cleavage reaction was quenched by addition of 50 ml of 100 mM thiourea. RNA was isolated and subjected to primer extension as described (Stern et al., 1988a).

Acknowledgements We thank Gloria Culver and Rachel Green for critical reading of the manuscript, Chuck Merryman and Simpson Joseph for helpful suggestions for the probing experiments, Vernita Ares for oligonucleotide synthesis, and members of the Noller laboratory for discussions. This work was supported by NIH grant GM17129 to H.F.N., an NIH postdoctoral fellowship to K.R.L., and a grant from the Lucille P. Markey Charitable Trust to the Center for Molecular Biology of RNA.

References Barta, A., Steiner, G., Brosius, J., Noller, H. F. & Kuechler, E. (1984). Identi®cation of a site on 23 S ribosomal RNA located at the peptidyl transferase center. Proc. Natl Acad. Sci. USA, 81, 3607± 3611. Cerretti, D. P., Dean, D., Davis, G. R., Bedwell, D. M. & Nomura, M. (1983). The spc ribosomal protein operon of Escherichia coli: sequence and cotranscription of the ribosomal protein genes and a protein export gene. Nucl. Acids Res. 11, 2599± 2616.

L15 as a Probe of 50 S Ribosomal Subunit Structure Dohme, F. & Nierhaus, K. H. (1976). Total reconstitution and assembly of 50 S subunits from Escherichia coli ribosomes in vitro. J. Mol. Biol. 107, 585± 599. Egebjerg, J., Leffers, H., Christensen, A., Andersen, H. & Garrett, R. A. (1987). Structure and accessibility of domain I of Escherichia coli 23 S RNA in free RNA, in the L24-RNA complex and in 50 S subunits. J. Mol. Biol. 196, 125±136. Egebjerg, J., Douthwaite, S., Liljas, A. & Garrett, R. A. (1990a). Characterization of the binding sites of protein L11 and the L10(L12)4 pentameric complex in the GTPase domain of 23 S ribosomal RNA from Escherichia coli. J. Mol. Biol. 213, 275± 288. Egebjerg, J., Larsen, N. & Garrett, R. A. (1990b). Structural map of 23 S rRNA. In The Ribosome: Structure, Function and Evolution (Hill, W. E., Moore, P. B., Dahlberg, A., Schlessinger, D., Garrett, R. A. & Warner, J. R., eds), pp. 168± 179, ASM Press, Washington, DC. Egebjerg, J., Christiansen, J. & Garrett, R. A. (1991). Attachment of primary binding proteins L1, L2 & L23 on 23 S ribosomal RNA of Escherichia coli. J. Mol. Biol. 222, 251±264. Franceschi, F. J. & Nierhaus, K. H. (1990). Ribosomal proteins L15 and L16 are mere late assembly proteins of the large ribosomal subunit. J. Biol. Chem. 265, 16676± 16682. Garrett, R. A. & Noller, H. F. (1979). Structures of complexes of 5 S RNA with ribosomal proteins L5, L18, and L25 from Escherichia coli: identi®cation of kethoxal reactive sites on the 5 S RNA. J. Mol. Biol, 132, 637± 648. Green, R. & Noller, H. F. (1997). Ribosomes and translation. Annu. Rev. Biochem. 66, 679± 716. Gulle, H., Hoppe, E., Oûwald, M., Greuer, B., Brimacombe, R. & StoÈf¯er, G. (1988). RNA-protein cross-linking in Escherichia coli 50 S subunits; determination of sites on 23 S RNA that are cross-linked to proteins L2, L4, L24, and L27 by treatment with 2-iminothiolane. Nucl. Acids Res. 16, 815± 832. Gutell, R. R. & Woese, C. R. (1990). Higher order structural elements in ribosomal RNAs: pseudo-knots and the use of noncanonical pairs. Proc. Natl Acad. Sci. USA, 87, 663± 667. Han, H. & Dervan, P. B. (1994). Visualization of RNA tertiary structure by RNA-EDTA
Fe(II) autocleavage: analysis of tRNAPhe with uridine-EDTA
Fe(II) at position 47. Proc. Natl Acad. Sci. USA, 91, 4955± 4959. Heilek, G. M. & Noller, H. F. (1996). Site-directed hydroxyl radical probing of the rRNA neighborhood of ribosomal protein S5. Science, 272, 1659± 1662. Heilek, G. M., Marusak, R., Meares, C. F. & Noller, H. F. (1995). Directed hydroxyl radical probing of 16 S rRNA using Fe(II) tethered to ribosomal protein S4. Proc. Natl Acad. Sci. USA, 92, 1113± 1116. Herold, M. & Nierhaus, K. H. (1987). Incorporation of six additional proteins to complete the assembly map of the 50 S subunit from Escherichia coli ribosomes. J. Biol. Chem. 262, 8826± 8833. Joseph, S. & Noller, H. F. (1996). Mapping the rRNA neighborhood of the acceptor end of tRNA in the ribosome. EMBO J. 15, 910± 16. Joseph, S., Weiser, B. & Noller, H. F. (1997). Mapping the inside of the ribosome with an RNA helical ruler. Science, 278, 1093± 1098. Kazemie, M. (1975). The importance of Escherichia coli ribosomal proteins L1, L11 and L16 for the associ-

1377 ation of ribosomal subunits and the formation of the 70 S initiation complex. Eur. J. Biochem. 58, 501± 510. Khaitovich, P., Green, R., Lancaster, L., Noller, H. F. & Mankin, A. (1998). Characterization of functionally active subribosomal particles from Thermus aquaticus. Proc. Natl Acad. Sci. USA, in the press. Kunkel, T. A. (1985). Rapid and ef®cient site-speci®c mutagenesis without phenotypic selection. Proc. Natl Acad. Sci. USA, 82, 488± 492. Leffers, H., Egebjerg, J., Andersen, A., Christiansen, T. & Garrett, R. A. (1988). Domain VI of Escherichia coli 23 S ribosomal RNA: structure, function and assembly. J. Mol. Biol. 204, 507± 522. Littlechild, J., Dijk, J. & Garrett, R. A. (1977). The identi®cation of new RNA-binding proteins in the Escherichia coli ribosome. FEBS Letters, 74, 292± 294. Lotti, M., Dabbs, E. R., Hasenbank, R., StoÈf¯er-Meilicke, M. & StoÈf¯er, G. (1983). Characterization of a mutant from E. coli lacking protein L15 and localization of protein L15 by immuno-electron microscopy. Mol. Gen. Genet. 192, 295±300. Marquardt, O., Roth, H. E., Wystup, G. & Nierhaus, K. H. (1979). Binding of Escherichia coli ribosomal proteins to 23 S RNA under reconstitution conditions for the 50 S subunit. Nucl. Acids Res. 6, 3641± 3650. Mitchell, P., Oûwald, M., Schueler, D. & Brimacombe, R. (1990). Selective isolation and detailed analysis of intra-RNA cross-links induced in the large ribosomal subunit of E. coli: a model for the tertiary structure of the tRNA binding domain in 23 S rRNA. Nucl. Acids Res. 18, 4325 ±4333. Mitchell, P., Stade, K., Oûwald, M. & Brimacombe, R. (1993). Site-directed cross-linking studies on the E. coli tRNA-ribosome complex: determination of sites labelled with an aromatic azide attached to the variable loop or aminoacyl group of tRNA. Nucl. Acids Res. 21, 887± 896. Moazed, D. & Noller, H. F. (1989). Interaction of tRNA with 23 S rRNA in the ribosomal A, P, and E sites. Cell, 57, 585± 597. Moazed, D., Stern, S. & Noller, H. F. (1986). Rapid chemical probing of conformation in 16 S ribosomal RNA and 30 S ribosomal subunits using primer extension. J. Mol. Biol. 187, 399± 416. Moore, V. G., Atchison, R. E., Thomas, G., Moran, M. & Noller, H. F. (1975). Identi®cation of a ribosomal protein essential for peptidyl transferase activity. Proc. Natl Acad. Sci. USA, 72, 844± 848. Nierhaus, D. & Nierhaus, K. H. (1973). Identi®cation of the chloramphenicol-binding protein in Escherichia coli ribosomes by partial reconstitution. Proc. Natl Acad. Sci. USA, 70, 2224± 2228. Nierhaus, K. H. (1990). Reconstitution of ribosomes. In Ribosomes and Protein Synthesis: A Practical Approach (Spedding, G., ed.), pp. 161± 189, IRL Press, Oxford. Nitta, I., Kamada, Y., Noda, H., Ueda, T. & Watanabe, K. (1998a). Reconstitution of peptide bond formation with Escherichia coli 23 S ribosomal RNA domains. Science, 281, 666± 669. Nitta, I., Ueda, T. & Watanabe, K. (1998b). Possible involvement of Escherichia coli 23 S ribosomal RNA in peptide bond formation. RNA, 4, 257±267. Noller, H. F., Hoffarth, V. & Zimniak, L. (1992). Unusual resistance of peptidyl transferase activity to protein extraction procedures. Science, 256, 1416± 1419.

1378

L15 as a Probe of 50 S Ribosomal Subunit Structure

Powers, T. & Noller, H. F. (1995a). Hydroxyl radical footprinting of ribosomal proteins on 16 S rRNA. RNA, 1, 194± 209. Powers, T. & Noller, H. F. (1995b). A temperaturedependent conformational rearrangement in the ribosomal protein S4-16 S rRNA complex. J. Biol. Chem. 270, 1238± 1242. RoÈhl, R. & Nierhaus, K. (1982). Assembly map of the large subunit (50 S) of Escherichia coli ribosomes. Proc. Natl Acad. Sci. USA, 79, 729± 733. Samaha, R. R., Green, R. & Noller, H. F. (1995). A base pair between tRNA and 23 S rRNA in the peptidyl transferase centre of the ribosome. Nature, 377, 309± 314. Staehelin, R., Maglott, D. & Monro, R. E. (1969). On the catalytic center of peptidyl transfer: a part of the 50 S ribosome structure. Cold Spring Harbor Symp. Quant. Biol. 34, 39 ±48. Stern, S., Moazed, D. & Noller, H. F. (1988a). Stuctural analysis of RNA using chemical and enzymatic probing monitored by primer extension. Methods Enzymol. 164, 481± 489. Stern, S., Weiser, B. & Noller, H. F. (1988b). Model for the three-dimensional folding of 16 S ribosomal RNA. J. Mol. Biol. 204, 447± 481.

Tate, W. P., Schulze, H. & Nierhaus, K. H. (1983). The importance of the Escherichia coli ribosomal protein L16 for the reconstitution of the peptidyl-tRNA hydrolysis activity of peptide chain termination. J. Biol. Chem. 258, 12810± 12815. Walleczek, J., Schuler, D., StoÈf¯er-Meilicke, M., Brimacombe, R. & StoÈf¯er, G. (1988). A model for the spatial arrangement of the proteins in the large subunit of the Escherichia coli ribosome. EMBO J. 7, 3571± 3576. Wower, J., Hixson, S. S. & Zimmerman, R. A. (1989). Labeling the peptidyltransferase center of the Escherichia coli ribosome with photoreactive tRNAPhe derivatives containing azidoadenosine at the 30 end of the acceptor arm: a model of the tRNA-ribosome complex. Proc. Natl Acad. Sci. USA, 86, 5232± 5236. Wystup, G., Teraoka, H., Schulze, H., Hampl, H. & Nierhaus, K. H. (1979). 50-S subunit from Escherichia coli ribosomes: isolation of active ribosomal proteins and protein complexes. Eur. J. Biochem. 100, 101± 113.

Edited by D. Draper (Received 14 July 1998; received in revised form 16 September 1998; accepted 17 September 1998)