Mutations at nucleotides G2251 and U2585 of 23 S rRNA perturb the peptidyl transferase center of the ribosome 1

JMB MS 2874 [24/1/97] J. Mol. Biol. (1997) 266, 40±50

Mutations at Nucleotides G2251 and U2585 of 23 S rRNA Perturb the Peptidyl Transferase Center of the Ribosome Rachel Green, Raymond R. Samaha and Harry F. Noller* Center for Molecular Biology of RNA, Sinsheimer Laboratories University of California, Santa Cruz, California, 95064, USA

Previous experiments have shown that the phylogenetically conserved G2252 of 23 S rRNA forms a Watson-Crick base-pair with C74 of peptidyl-tRNA. In the studies presented here, site-directed mutations were introduced at two other conserved positions in 23 S rRNA, G2251 and U2585, that were previously implicated in interaction of the CCA acceptor end of tRNA with the 50 S subunit P site. The mutant 23 S rRNAs were characterized by determining (1) the in vivo phenotypes, (2) the ability of mutant ribosomes to bind tRNA oligonucleotide fragments in vitro, using footprinting with allele-speci®c primer extension and (3) the ability of mutant ribosomes to catalyze peptide bond formation using a chimeric reconstitution approach. Mutations at either position confer a dominant lethal phenotype when the mutant 23 S rRNA is coexpressed with the endogenous wild-type 23 S rRNA. Mutations at 2585 disrupt binding of the wild-type (CCA) tRNA oligonucleotide fragment and cause a modest decrease in the peptidyl transferase activity of reconstituted ribosomes. By contrast, mutations at 2251 abolish both binding of the wild-type (CCA) tRNA fragment and peptidyl transferase activity using the wildtype tRNA fragment. In neither case was the loss of binding or peptidyl transferase activity suppressed by mutations in the tRNA oligonucleotide fragment. Chemical modi®cation analysis revealed that mutations at 2251 perturb the reactivity of bases 2584 to 2586, providing further evidence that the 2250 loop of 23 S rRNA interacts, either directly or indirectly, with the 2585 region in the central loop of domain V of 23 S rRNA. # 1997 Academic Press Limited

*Corresponding author

Keywords: ribosome; peptidyl transferase; tRNA binding; chimeric reconstitution; rRNA mutants

Introduction A central function of the large subunit of the ribosome is to catalyze the acyl transfer of activated amino acids to form peptide bonds (peptidyl transferase). In performing this function, the ribosome must bind and orient the appropriate templated aminoacylated tRNAs for catalysis. A variety of crosslinking and chemical footprinting studies have suggested that, in addition to its base-pairing with the mRNA codons, important contacts are made by tRNA with the 16 S and 23 S ribosomal RNAs (Prince et al., 1982; Barta et al., 1984; Moazed & Noller, 1986, 1989; Wower et al., 1989; Mitchell et al., 1993; von Ahsen & Noller, 1995). Chemical Present address: R. R. Samaha, Array Technologies, 460 Page Mill Road, Palo Alto, CA 94306, USA. 0022±2836/97/060040±11 $25.00/0/mb960780

footprinting data focused attention on several universally conserved nucleotides in domain V of 23 S rRNA as candidates for direct interactions with the similarly conserved CCA end of tRNA. Two invariant guanosine residues in 23 S RNA, G2252 and G2253, were identi®ed whose protection from kethoxal modi®cation by P-site (peptidyl or donor) bound tRNA was dependent on the presence of the CCA 30 terminus of tRNA (Moazed & Noller, 1989); a third invariant guanosine located in the 2250 loop, G2251, may be similarly protected by the CCA 30 terminus of P-site bound tRNA (R.G. and H.F.N., unpublished results). Three invariant uridine residues in 23 S rRNA, U2506, U2584 and U2585, were identi®ed whose protection from CMCT (carbodiimide) modi®cation by P-sitebound tRNA was dependent on the presence of the 30 terminal A of the invariant CCA sequence of # 1997 Academic Press Limited

JMB MS 2874 [24/1/97] Mutations at G2251 and U2585 of 23 S rRNA

41

Figure 1. A schematic of the secondary structure of domains V and VI of E. coli 23 S rRNA (Noller et al., 1981) showing the locations of nucleotides G2251 and U2585. The approximate cleavage positions at 2300 and 2530 used in the chimeric reconstitution experiments and the silent mutations introduced at allele-speci®c priming sites, I and II, that were used to monitor tRNA or oligonucleotide interactions at positions G2252-2253 or U2585, respectively, are indicated (Aagaard et al., 1991; Powers & Noller, 1993; Samaha et al., 1995).

tRNA (Moazed & Noller, 1989). These ®ndings suggested possible Watson-Crick pairing relationships between the protected bases in 23 S rRNA and the respective bases in tRNA. The physiological importance of G2252 and G2253 was examined by site-directed mutagenesis. Mutations at G2252 resulted in a dominant lethal phenotype in cells expressing the mutant 23 S rRNA. By contrast, two site-directed mutations at the 30 adjacent G2253 (A2253 and U2253) conferred only a recessive slow-growth phenotype, and ribosomes bearing

these mutations at position 2253 retain their ability to bind wild-type tRNA oligonucleotide fragments in vitro (Samaha et al., 1995). Parallel studies showed that the mutation C2253 has a signi®cantly stronger, dominant slow-growth phenotype (Gregory et al., 1994). Base-pairing between the CCA sequence of tRNA and these two conserved guanine residues was tested by in vitro genetic experiments that established that a Watson-Crick interaction between G2252 of 23 S RNA and C74 of tRNA is indeed critical for binding the tRNA sub-

JMB MS 2874 [24/1/97] 42 strate in the P site of the 50 S subunit and for peptidyl transferase activity (Samaha et al., 1995). No such relationship was found for G2253, however. In this study, we have investigated the effects of base mutations at G2251 and U2585, two of the other bases in 23 S rRNA whose protections depend on the CCA sequence of tRNA (Figure 1). Each set of mutant 23 S rRNAs was assessed for its level of function in several different assays: (1) the in vivo phenotype of cells expressing mutant 23 S rRNA; (2) the ability of mutant ribosomes to bind tRNA oligonucleotide fragments in vitro; and (3) the ability of reconstituted mutant ribosomes to catalyze peptide bond formation. Potential compensatory mutations in the CCA-containing oligonucleotide substrate (at positions C74, C75 and A76) were tested for their ability to suppress any observed binding or catalytic de®ciencies of the mutant ribosomes. Analysis of both sets of mutations relied on our ability to speci®cally generate and analyze mutant ribosome populations. For the analysis of tRNA binding to the mutant ribosomes (necessarily found in a mixed population with wild-type ribosomes), the allele-speci®c primer extension strategy was utilized. Here, silent mutations were incorporated into the cloned, mutant copy of 23 S rRNA allowing different DNA oligonucleotide primers to distinguish between the two populations (Aagaard et al., 1991; Powers & Noller, 1993; Samaha et al., 1995). For the peptidyl transferase assay, pure populations of mutant ribosomes were prepared using a recently developed chimeric in vitro reconstitution approach that allows in vitro transcribed partial 23 S rRNA transcripts to replace corresponding regions of Escherichia coli 23 S rRNA (Samaha et al., 1995; Green & Noller, 1996).

Results Site-directed mutations at G2251 and U2585 Mutations were introduced at positions G2251 and U2585 (Figure 1) of E. coli 23 S rRNA genes as part of an rrnB operon-containing plasmid construct under transcriptional control of the inducible lambda PL promoter. Expression of the mutant 23 S rRNA is induced by growth at 42 C. Cells expressing the plasmid-borne copy of 23 S rRNA contain a mixture of mutant and chromosomally encoded wild-type ribosomes. Dominant phenotypes are expressed on ampicillin-containing media. Recessive effects are observed in the presence of the antibiotic erythromycin, which speci®cally inhibits the activity of ribosomes bearing the chromosomal copies of 23 S rRNA; the mutant 23 S rRNA molecules carry the G2058 mutation conferring erythromycin resistance. A comparison of 23 S rRNA expression at 42 C, with or without the engineered priming sites, indicates that the priming site mutations are indeed silent (Figure 2(a) and (b), WT ‡/ÿ PS). By contrast, expression of 23 S rRNA containing mutations at G2251 (A2251 and

Mutations at G2251 and U2585 of 23 S rRNA

U2251) at 42 C caused a dramatic dominant inhibition of growth (Figure 2(b)) , whereas uninduced cells (30 C) grew normally (Figure 2(a)). A similar dominant growth phenotype was observed for all three mutations at U2585 (Figure 2(c) and (d)). Binding of tRNA oligonucleotide fragments Binding of full-length tRNA to 70 S ribosomes is determined primarily by its interactions with the 30 S subunit. To study interaction of the CCA acceptor end of the tRNA with the 50 S subunit P site, minimal tRNA substrates were substituted for intact tRNAs in the chemical footprinting analysis of the wild-type and mutant ribosomes. Early studies by Monro and colleagues demonstrated that Nblocked aminoacylated tRNA oligonucleotide substrates such as CACCA-F-Met could substitute in the P site for intact peptidyl tRNAs (Monro et al., 1968). These tRNA fragments are bound stably to the 50 S ribosomal P site in the presence of methanol or ethanol and the antibiotic sparsomycin, and are reactive with minimal A-site substrates such as puromycin in what is termed the fragment reaction (Monro et al., 1969). Furthermore, the set of 23 S rRNA nucleotides protected from chemical modi®cation by these minimal P-site substrates is virtually identical with those protected by intact tRNAs (Moazed & Noller, 1991). The footprinting patterns of the different tRNA oligonucleotide fragments on wild-type and mutant ribosomes are distinguished by taking advantage of silent mutations that have been introduced into the cloned copy of 23 S rRNA (the mutant version) to allow for allelespeci®c primer extension (Aagaard et al., 1991; Powers & Noller, 1993; Samaha et al., 1995: priming sites I and II are shown in Figure 1). As in our previous analysis of G2252 and G2253 (Samaha et al., 1995), carbodiimide (CMCT) modi®cation was used to monitor the reactivity of U2585, a nucleotide whose protection by the CCA end of tRNA is known to be dependent on the presence of the 30 -terminal A76 (present in the wild-type and mutant tRNAs tested). By monitoring tRNA binding at a site distinct from the site of mutation in 23 S rRNA, two potential ambiguities are avoided: ®rst, as the identity of the base of interest is changed, different chemical probes would be needed to study the protection pattern (e.g. when G2251 is changed to U, it would be necessary to switch from kethoxal to CMCT modi®cation), and second, any localized structural changes resulting from the mutation could perturb the primer extension patterns, making direct comparisons dif®cult. Either mutation at G2251 (A2251 or U2251) of 23 S rRNA abolishes binding of wild-type tRNA fragment CACCA-N-Ac-Phe. Mutation of either C74 or C75 caused loss of binding of tRNA fragments to wild-type ribosomes. Potential Watson-Crick compensatory mutations C74 and C75 (CAUCA-N-AcPhe and CACUA-N-Ac-Phe for A2251 and CAACA-N-Ac-Phe and CACAA-N-Ac-Phe for U2251) fail to suppress loss of binding (Figure 3(a)).

JMB MS 2874 [24/1/97] Mutations at G2251 and U2585 of 23 S rRNA

43

Figure 2. Expression of rRNA genes carrying mutations at positions G2251 and U2585 in 23 S rRNA. (a) and (b) DH1 cells transformed with wild-type plasmid pLK45 (WT(ÿPS)), wild-type plasmid containing allele-speci®c priming sites I and II (WT(‡PS)), and plasmid containing the priming sites in addition to mutations at G2251 (2251A and 2251U) were plated on solid medium at 30 C or at 42 C, respectively. (c) and (d) DH1 cells transformed with wildtype plasmid pLK45 containing allele-speci®c priming sites (WT(‡PS)) or plasmid containing allele-speci®c priming sites in addition to mutations at U2585 (2585A, 2585G and 2585C) were plated on solid medium at 30 C or 42 C, respectively.

Thus, while G2251 is universally conserved and its identity is clearly essential to a functionally intact P site, the function of ribosomes containing mutations at this position is not rescued by nucleotide substitutions at positions C74 or C75 in the tRNA oligonucleotide substrate. Next, we examined the effect of mutations at U2585 (A2585, C2585 and G2585) on tRNA fragment binding. As above, to assay binding we monitored the reactivity of a tRNA-protected site distinct from the site of mutation; in this case, kethoxal protection of G2252-G2253 was used as a diagnostic probe. In this experiment, tRNA oligonucleotide fragments bearing mutations at position A76 were tested for their ability to suppress observed losses in binding ef®ciency. Wild-type (U2585) ribosomes were footprinted by the wildtype tRNA fragment CCACCA-N-Ac-Met but not

by the A76 mutant fragments CCACCC-N-Ac-Met or CCACCU-N-Ac-Met; the mutant (A2585) ribosomes were not footprinted by either wild-type or mutant fragments (Figure 3(b)). The dominant negative in vivo phenotype conferred by the A2585 mutation may result from a failure to bind the tRNA fragment appropriately. Elongator methionine tRNA was used for the construction of A76 mutant tRNAs because of the dif®culties encountered in aminoacylating A76 mutant phenylalanine tRNAs; in the case of the CCG and CCU tRNA mutants, essentially all of the input tRNA was converted to a product that comigrated with CACCAN-Ac-Phe by paper electrophoresis at pH 3.5. This conversion is likely to be the result of CCA repair enzymes present in the S100 extract (Deutscher, 1983) and the phenylalanyl tRNA synthetase preparation used.

JMB MS 2874 [24/1/97] 44

Mutations at G2251 and U2585 of 23 S rRNA

Figure 4. (a) and (b) Chimeric reconstitution approach for analysis of mutations at G2251 and U2585, respectively. Activity refers to the peptidyl transferase activity of the chimeric reconstitution reaction using wild-type 23 S rRNAs; ‡ ‡ , high activity; ‡, 10 to 30-fold lower activity; ÿ, no detectable activity.

Figure 3. Binding of wild-type and mutant aminoacyloligonucleotides to the P site of 50 S ribosomal subunits containing wild-type or mutant 23 S rRNA, assayed by protection of U2585 from attack by CMCT or by protection of G2252/G2253 from attack by kethoxal, respectively. WT indicates wild-type 23 S rRNA, and A2251, U2251 and A2585 are the corresponding mutant 23 S rRNAs. A and G indicate sequencing lanes, K is unmodi®ed 23 S rRNA, sp indicates the presence (‡) or absence (ÿ) of sparsomycin. CCA indicates the wildtype tRNA oligonucleotide fragment CACCA-(N-AcPhe), and ACA, CAA, UCA, CUA the corresponding mutant versions of the oligonucleotide in (a), and CCACCA-(N-Ac-Met), and CCC, CCU the corresponding mutant versions of the oligonucleotide in (b). Chemical reactivity of U2585 was monitored by primer extension from priming site II and reactivity of G2252/ G2253 was monitored from priming site I, respectively, as described for Figure 1.

Peptidyl transferase activity The peptidyl transferase activity of the mutant ribosomes was assayed using a recently developed

chimeric reconstitution approach (Samaha et al., 1995; Green & Noller, 1996). This is necessary because pure mutant rRNA cannot be expressed in vivo due to the existence of multiple chromosomal copies of 23 S rRNA, and in vitro transcribed 23 S rRNA is not functional in in vitro reconstitution reactions. However, because only a small segment of natural 23 S rRNA (extending from positions ca 2445 to 2523) is required for in vitro reconstitution, fragments of natural 23 S rRNA (containing the critical segment) can be complemented by in vitro transcripts of the remaining regions of 23 S rRNA to form functional particles (Green & Noller, 1996). Mutations are thus readily incorporated into the partial in vitro 23 S rRNA fragment. Another important feature of our approach is the use of a highly sensitive peptidyl transferase assay that is compatible with the low ef®ciency of the chimeric reconstitution procedure. This assay uses the standard fragment reaction, in which puromycin, an aminoacyl-tRNA analog, acts as a nucleophile to attack the N-blocked aminoacylated fragment of tRNA in the P site, in this case the 30 terminus of elongator tRNA methionine, CCACCA-N-Ac-[35S]Met, to form as the product N-Ac-

JMB MS 2874 [24/1/97] Mutations at G2251 and U2585 of 23 S rRNA

Figure 5. Peptidyl transferase assay. Phosphorimager exposure of paper electrophoresis analysis of the peptidyl transferase reaction. (a) Catalyzed by the wild-type (G2251-natural (nat) or in vitro transcribed (T7)) and mutant (A2251-T7 and U2251-T7) 23 S rRNA chimeric reconstitution products using the wild-type (CCACCAN-Ac-[35S]Met) and mutant (CCACAA-N-Ac-[35S]Met and CCACUA-N-Ac-[35S]Met) tRNA oligonucleotide substrates. (b) Catalyzed by the wild-type (U2585) and mutant (A2585, C2585 and G2585) 23 S rRNA chimeric reconstitution products using the wild-type (CCACCAN-Ac-[35S]Met) and mutant (CCACCC-N-Ac-[35S]Met, CCACCG-N-Ac-[35S]Met and CCACCU-N-Ac-[35S]Met) tRNA oligonucleotide substrates. Spots represent the product N-Ac-[35S]Met-puromycin.

[35S]Met-puromycin (NMP). By resolving the ribosome-catalyzed product (NMP) from an excess of non-enzymatic hydrolysis and methanolysis products, the signal-to-noise ratio has been optimized to allow detection of peptidyl transferase activities ®ve to six orders of magnitude below that of native 50 S subunits (Green & Noller, 1996). This sensitivity is suf®cient for analysis of the chimeric reconstitutions presented below. Mutations at G2251 were characterized using the chimeric reconstitution system described previously for the G2252 mutations (Samaha et al., 1995: and see Figure 4(a)). Brie¯y, 50 and 30 RNase H fragments of natural 23 S rRNA (extending from nucleotides 1 to 2294 and 2315 to 2904) were generated by targeting a complementary DNA oligonucleotide to positions 2295 to 2314 of 23 S rRNA. When the two natural, puri®ed fragments of 23 S rRNA are combined in a standard in vitro reconstitution reaction with 5 S rRNA and proteins from the large subunit (TP50), catalytically active 50 S

45 subunits are obtained. The peptidyl transferase activity of the crude reconstitution mixture is measured to assess the ef®ciency of the reconstitution reaction. Overall, the ef®ciency of the fragmented 23 S rRNA reconstitution reaction (both pieces natural) is low (ca 1%) when compared with that of native 50 S subunits. This loss in activity is only partially the result of fragmentation; much of the loss results from the denaturing treatments that the natural RNAs must undergo in order to be separated from one another on sucrose gradients (Green & Noller, 1996). Importantly, the background activity observed for the 30 natural fragment reconstituted alone is below the limits of detection (data not shown). For the analysis of mutant 23 S rRNAs, the 50 fragment of 23 S rRNA extending from nucleotides 1 to 2314 was substituted with wild-type or mutant in vitro partial transcripts. When the wild-type 50 in vitro transcript is combined with the wild-type 30 natural fragment of 23 S rRNA in a reconstitution reaction, the peptidyl transferase activity of the resulting mixture is 30% of that of the analogous reconstitution with two natural fragments (Figure 5(a)). Chimeric reconstitution reactions containing the mutant 50 in vitro transcripts (A2251 and U2251) were severely de®cient in peptidyl transferase activity; wild-type tRNA oligonucleotide fragments were utilized at least 200-fold less ef®ciently by both mutant ribosomes (A2251 and U2251) than by the corresponding wild-type (G2251) ribosomes (Fig. 5a). No suppression of this decrease in peptidyl transferase activity was observed using the C75 mutant tRNA fragments. The observed loss in peptidyl transferase activity is consistent with the severe dominant lethal phenotype conferred by these mutations in vivo, the decreased representation of the mutant ribosomes in polysomes (data not shown), and with the loss of a tRNA fragment footprint observed for A2251 and U2251 mutant 50 S subunits. Because the mutant ribosomes are well represented in the population of tight-couple 70 S ribosomes in the cell (data not shown), it is unlikely that the observed de®ciency in peptidyl transferase activity is the result of poor in vitro reconstitution by these mutant rRNAs. To test the peptidyl transferase activity of 50 S subunits containing mutations at U2585, a similar strategy was employed (Figure 4(b)). Use of a different RNase H target was necessary because in the previously described chimeric reconstitution U2585 was contained within the 30 natural fragment. Using a complementary DNA oligonucleotide targeted to positions 2524 to 2538, 50 and 30 natural fragments of 23 S rRNA were generated and puri®ed on sucrose gradients as above. Because of a tendency for these two natural fragments to remain tightly associated, two successive sucrose-gradient puri®cations were required to obtain 50 natural fragment that was suf®ciently free of 30 natural fragment to give low background levels of peptidyl transferase activity when reconstituted alone (<1% relative to the two-piece natu-

JMB MS 2874 [24/1/97] 46 ral reconstitution). As above, the two natural 23 S rRNA fragments are combined in a reconstitution reaction with 5 S rRNA and TP50, and the peptidyl transferase activity of the resulting reconstitution determined. The ef®ciency of this two-piece natural reconstitution containing a break at position 2530 is relatively high (5% of the activity of native 50 S subunits) when compared with a number of different fragmentation sites that have been tested (data not shown). Next, wild-type and mutant 30 in vitro transcripts were prepared. In this two-piece system, when the 30 natural fragment is substituted with the corresponding 30 in vitro transcript, the peptidyl transferase activity is 66% of that of the corresponding reaction with both natural fragments (Green & Noller, 1996). To test the activities of the U2585 mutant ribosomes, the different mutant 30 in vitro fragments were combined in a reconstitution reaction with the natural 50 fragment and the peptidyl transferase activity was measured using wild-type and A76 mutant tRNA oligonucleotide fragments. Each of the U2585 mutations has some effect on peptidyl transferase activity with wild-type fragment; A2585 is 10(‡/ÿ1)% as active as U2585 (wt), C2585 is 30(‡/ÿ15)% as active and G2585 is 30(‡/ÿ5)% as active (Figure 5(b)). While these are signi®cant decreases in activity, the severity does not compare with that of the peptidyl transferase de®ciency observed with mutations at 2251 (or 2252). No compensatory activity is observed with the A76 mutant tRNA fragments and the observed decreased levels of peptidyl transferase activity are consistent with the inability of A2585 mutant ribosomes to bind wild-type tRNA fragment (Figure 3(b)). Chemical probing analysis of 2251 mutant ribosomes Because of the severe effect of the 2251 mutations on both fragment binding and on peptidyl transferase activity, we tested for possible structural perturbations by examining the reactivities of uridine residues in the peptidyl transferase center using the carbodiimide reagent, CMCT. Primer extension from the two allele-speci®c priming sites (I and II) located at positions ca 2310 and 2700, respectively, was used to monitor the modi®cation pattern of two different segments of 23 S rRNA, from positions 2100 to 2250 and 2500 to 2650, which comprise much of the peptidyl transferase-associated region of 23 S rRNA. This analysis showed that the RNA in mutant (A2251 and U2251) ribosomes has a pattern of reactivity that is generally indistinguishable from that of wild-type ribosomes. However, there are signi®cant differences speci®cally in the reactivities of uridine residues 2584 to 2586 between mutant and wild-type ribosomes (Figure 6). In both the A2251 and U2251 mutant ribosomes, the reactivity of U2586 increases substantially, and the normally unreactive U2584 becomes reactive toward CMCT. Thus, mutations at G2251 affect the

Mutations at G2251 and U2585 of 23 S rRNA

Figure 6. Modi®cation of 2584 to 2586 in 23 S rRNA by CMCT in wild-type (WT) and mutant (A2251 and U2251) ribosomes. A and G indicate sequencing lanes, K is unmodi®ed 23 S rRNA and the time of modi®cation (in minutes) is indicated.

chemical accessibility of nucleotides 2584 to 2586, located more than 300 nucleotides away in the primary sequence. These results provide evidence for interaction between the 2250 loop and the 2585 region of the central loop of domain V. We cannot determine whether this interaction is direct or indirect.

Discussion In this study we examined the effects of singlebase changes at two universally conserved bases, G2251 and U2585, which have been localized to the peptidyl transferase center of 23 S rRNA in the 50 S subunit of the ribosome. G2251 is of interest for a number of reasons: (1) it is universally conserved; (2) it is the site of a conserved post-transcriptional 20 -O-methyl modi®cation implicated in 50 S subunit assembly in yeast mitochondria (Sirum-Connolly & Mason, 1993); (3) it is protected from modi®cation by P-site bound tRNA (Moazed & Noller, 1989); and (4) the adjacent base, G2252, has recently been shown to form a Watson-Crick base-pair with C74 in the CCA acceptor end of Psite bound tRNA (Samaha et al., 1995). Because of its proximity to G2252, we tested the possibility that base 2251 might interact with position C75 of tRNA in a Watson-Crick manner. Two mutations (A2251 and U2251) were introduced at position 2251 and their effects were examined in vivo and in vitro. In an in vivo expression system where mutant and wild-type ribosomes are coexpressed, both mutations at 2251 exhibit a dominant lethal phenotype. In this system, such phenotypes are most easily explained by the interference of impaired mutant ribosomes with functional wild-type ribosomes on shared polysomes; indeed, A2251

JMB MS 2874 [24/1/97] Mutations at G2251 and U2585 of 23 S rRNA

and U2251 mutant ribosomes were underrepresented on polysomes relative to the wild-type plasmid-encoded version (data not shown). Consistent with this explanation, two different in vitro assays showed that the A2251 and U2251 mutant ribosomes were de®cient in peptidyl transferase-related functions. Although the dominant lethal in vivo phenotype of the 2251 mutations could formally be explained as a temperature-sensitive conditional lethality at 42 C, the observed biochemical de®ciencies in two different in vitro assays performed at 4 C suggests that this is not the case. First, chemical protection analysis using allele-speci®c primer extension showed that the A2251 and U2251 mutant ribosomes are unable to bind the wild-type tRNA oligonucleotide fragment in the P site. Potential compensatory mutations at positions C74 and C75 in the CCA acceptor end of tRNA failed to suppress the observed binding de®ciency. Second, a peptidyl transferase assay using wildtype and mutant (C75) tRNA oligonucleotide fragments indicated that the mutant ribosomes (A2251 and U2251) are de®cient in catalyzing peptide bond formation. The peptidyl transferase assay, performed under subsaturating conditions for the P-site substrate, is unable to distinguish the relative contributions of kcat and KM to this de®ciency; however, it is likely that loss of activity is at least partly the result of impaired binding of the wild-type and mutant fragments to the mutant ribosomes (i.e. a KM effect). Whether it is also the result of a disturbed catalytic center (kcat) is unknown. Chemical probing experiments using the carbodiimide reagent CMCT indicate that the conformation of at least one region of 23 S rRNA in the peptidyl transferase center is perturbed in the 2251 mutant ribosomes. Elsewhere, uridine reactivities of mutant ribosomes were indistinguishable from those of wild-type ribosomes, indicating that the overall structure of the 50 S subunits was probably unaffected by the mutations at G2251. Consistent with this observation is the fact that the mutant ribosomes used for the footprinting experiments were isolated as tight-couple 70 S ribosomes, indicating that the mutant 50 S subunits are structurally intact and able to associate with 30 S subunits. Interestingly, however, the CMCT reactivities of uridine residues 2584 to 2586 are signi®cantly enhanced in the A2251 and U2251 mutant ribosomes (Figure 6). In a previous study, it was shown that mutations at G2252 also affected CMCT modi®cation at U2585; in that study, the rate and extent of modi®cation of U2585, as well as its protection by tRNA, were diminished in A2252 mutant ribosomes (Samaha et al., 1995). The observation of such localized and speci®c changes in the chemical accessibility of nucleotides located more than 300 nucleotides from the sites of these mutations provides strong evidence for interaction, either direct or indirect, between these two regions. The affected nucleotides around position 2585 have themselves been placed in proximity to the 30 end of peptidyl-tRNA by two separate lines of evi-

47 dence. U2585 is protected from CMCT modi®cation by tRNA bound in the 50 S P site. Its protection is speci®cally dependent on the presence of the 30 -terminal A76 of tRNA (Moazed & Noller, 1989, 1991). Crosslinking studies have indicated that U2585 is in close proximity to the 30 -linked acyl moiety of peptidyl-tRNA (Barta et al., 1984). The experiments presented here provide evidence for a structural link between this region, likely proximal to the 30 end of the tRNA, and the 2250 loop, known to interact directly with C74 of the CCA end of tRNA (Samaha et al., 1995). Unlike our ®ndings for C74 and G2252 (Samaha et al., 1995), base mutations at C74 and C75 fail to suppress the de®ciencies of G2251 mutant ribosomes. Thus, of three universally conserved guanosine residues in the 2250 loop, whose P-site tRNA protections depend on the presence of the terminal CA sequence of tRNA, only one, G2252, has been found to interact directly with tRNA. Clearly, however, the identity of G2251 is critical to the function of the ribosome. The absence of Watson-Crick suppression does not, of course, exclude the possibility that other moieties such as the phosphate or ribose of the RNA backbone are involved in direct interactions between 23 S rRNA and tRNA. Indeed, binding the acceptor end of P-site tRNA is likely to be much more complex than simple Watson-Crick base-pairing. Not only are the bases of C74 and C75 protected from dimethyl sulfate modi®cation when bound to the P site of the 70 S ribosome (Peattie & Herr, 1981; Douthwaite et al., 1983), but in addition, the CCA backbone is protected from hydroxyl radicals when bound in the same position (Huttenhofer & Noller, 1992). The data have been interpreted to mean that the CCA end of the tRNA is buried at the subunit interface, inaccessible to aqueous contact, a feature consistent with the role of the ribosome in excluding water from the activated tRNA substrates as they undergo peptide bond formation. Rather than providing a contact for tRNA binding to the P site of the ribosome, it is possible that G2251 is involved in a long-range tertiary interaction, drawing together two important regions of 23 S rRNA; disruption of this crucial contact (i.e. in the 2251 mutants) would thus result in loss of substrate binding and catalysis. The altered chemical modi®cation pattern observed at positions 2584 to 2586 could be a manifestation of such a disruption. Alternatively, the identity of G2251 could be essential for maintaining the local conformation of the 2250 loop, and in particular for correct orientation of G2252, to promote its interaction with C74 of tRNA (Samaha et al., 1995). Mutations were introduced at nucleotide U2585 to test whether the A76-dependent P-site tRNA protection of U2585 from CMCT modi®cation was the result of A76-U2585 base-pairing. In the in vivo expression system, all three mutations exhibited an unambiguous dominant lethal phenotype. In an in vitro assay for tRNA oligonucleotide fragment binding, A2585 mutant ribosomes did not bind

JMB MS 2874 [24/1/97] 48 wild-type CCA tRNA fragment and A76 mutations (CCC and CCU) failed to compensate for this binding de®ciency. In peptidyl transferase assays using chimerically reconstituted ribosomes, activity was diminished, but not abolished, in each mutant (A2585 10%, C2585 - 30% and G2585 30%). No suppression of the observed decreases in activity was observed when mutant tRNA fragments were supplied as P-site substrates, suggesting that protection of U2585 is not the result of simple basepairing with A76 of tRNA. Again, the data do not exclude the possibility of interactions between the sugar-phosphate backbones of the respective RNAs. Alternatively, and keeping in mind that the assays used in this study focus on very limited aspects of ribosomal function, the strong in vivo phenotype of the U2585 mutants may be the result of a failure at some other step in the complex process of translation. A number of potential pairing candidates for A76 of tRNA remain and are currently being tested using approaches similar to those used in this study. As explained above, the possibility that the dominant lethal in vivo phenotypes of U2585 mutations are the result of ts conditional lethality at 42 C seems unlikely in view of the observed de®ciencies of mutant ribosomes in in vitro assays performed at 4 C. It is clear that the regions of 23 S rRNA targeted here are of fundamental importance to catalysis of peptide bond formation by the ribosome. Indeed, a recent study focusing on the lower half (positions 2493 to 2606) of the central loop of domain V reported that 13 of 21 randomly chosen mutations in this region exhibited compromised peptidyl transferase activity (Porse & Garrett, 1995). We are hopeful that a systematic structural and functional analysis of site-directed mutants in 23 S rRNA will provide further insight into the nature of this catalytic function of the ribosome.

Materials and Methods Construction of mutants Oligonucleotide-directed mutations were constructed in pBS23S as described (Samaha et al., 1995), using the following mutagenic primers: 2251(A/T), 50 -GGG-TAGTTT-GAC-TG(T/A)-GGC-GGT-CTC-C-30 ; 2585(A/C/G), 50 -CAC-GAC-GTT-CTA-(C/G/T)AC-CCA-GCT-CGC-G30 . Expression plasmids were constructed by digestion of mutant derivatives of pBS23S with Asp718 and ligation to BamHI linkers. The resulting linear plasmid was digested with BamHI and SphI and introduced into plasmid pLK45 containing the rrnB operon under control of phage lambda PL promoter (Samaha et al., 1995). Growth phenotypes Mutant plasmids were transformed into E. coli strain DH1 containing plasmid pcI857, which encodes a thermolabile allele of the lambda repressor. Resulting pLK plasmids containing the respective mutations and engineered priming sites, ‡PS, or lacking engineered priming sites, ÿPS, were selected on plates containing ampicillin (40 mg/l) and kanamycin (50 mg/l). Overnight cultures

Mutations at G2251 and U2585 of 23 S rRNA were grown at 30 C, diluted to 10ÿ4, 10ÿ5 and 10ÿ6 in LB medium, and 12 ml of each dilution spotted on Amp40Kan50 plates and grown overnight at either 30 C or 42 C. tRNA oligonucleotide fragment footprinting Templates for transcription of mutant and wild-type tRNAs were prepared by polymerase chain reaction (PCR) ampli®cation either of plasmid CF23 containing the gene for tRNAPhe from E. coli (a gift from O. Uhlenbeck) or of E. coli genomic DNA for elongator methionine tRNA (Samaha et al., 1995). Ampli®ed DNA was transcribed in vitro with T7 RNA polymerase (Milligan et al., 1987). tRNA transcripts were puri®ed on a denaturing 10%(w/v) polyacrylamide gel and eluted overnight in 0.3 M sodium acetate at 4 C; eluted tRNAs were recovered by precipitation in ethanol and resuspended in water. Before use, tRNAs were renatured by heating at 95 C for two minutes followed by slow cooling to room temperature; MgCl2 was then added to a ®nal concentration of 10 mM. Charging and N-acetylation were performed as described (Breitmeyer & Noller, 1976; Moazed & Noller, 1989). tRNA oligonucleotide fragments containing 20 30 -linked N-acetyl-Phe were prepared by RNase T1 digestion as described (Moazed & Noller, 1991). CMCT and kethoxal probing, isolation of the modi®ed RNA, and allele-speci®c primer extension were performed essentially as described (Samaha et al., 1995); in Figure 3(a), tight-couple 70 S were at a concentration of 0.2 mM and tRNA oligonucleotide fragments at 2 mM, whereas in Figure 3(b), 70 S were at a concentration of 0.2 mM and tRNA fragments were at 0.4 mM. Peptidyl transferase assay Aminoacylated wild-type and mutant rRNA fragments were prepared essentially as described (Moazed & Noller, 1991) except that elongator tRNAs were transcribed in vitro by phage T7 RNA polymerase from PCR DNA ampli®ed from E. coli genomic DNA (Milligan et al., 1987). Peptidyl transferase activity was measured essentially as described (Samaha et al., 1995). The 50 S subunits (10 pmol), or an equivalent amount of an in vitro reconstitution reaction, were incubated with CCACCA(N-Ac-[35S]Met) or a mutant version (2.5 pmol) in 100 ml of 0.4 M potassium acetate, 50 mM Tris (pH 8.3) and 60 mM MgCl2 on ice for ®ve minutes. The reaction was initiated on ice by the addition of puromycin (®nal concentration 1 mM) and 50 ml of methanol. Aliquots of 35 ml were removed at various times and placed in 10 ml of 2 M KOH and quick-frozen. At completion of the time-course, aliquots were incubated at 37 C for 20 minutes, then 150 ml of 0.3 M sodium acetate (pH 5.5) saturated with MgSO4 (and containing 0.02% (w/v) xylene cyanol) were added. The product, N-acetyl-methionine[35S]puromycin (NMP), was selectively removed by extraction with 1 ml ethyl acetate. The dried organic sample was spotted on Whatman 3 MM paper, and subjected to high-voltage paper electrophoresis in 0.5 M formic acid (pH 2.0) at 3000 V for 40 minutes. Radioactivity was visualized using a Molecular Dynamics Phosphorimager. Preparation of natural and in vitro 23 S rRNA partial transcripts RNase H cleavage was performed by incubating equimolar amounts of 23 S rRNA and cDNA oligonucleotide

JMB MS 2874 [24/1/97] 49

Mutations at G2251 and U2585 of 23 S rRNA (either 2295 (50 -TGA-TGT-CCG-ACC-AGG-ATT-AG-30 ) or 2524 (50 -GAC-CTA-CTT-CAG-CCC-30 )) in a volume of water such that ®nal concentrations are approximately 4 mM. The 23 S rRNA and DNA were incubated on ice for ®ve minutes and then at 42 C for ®ve minutes. Next, the volume was adjusted such that the ®nal RNA concentration was approximately 1 mM, and the salts adjusted to 40 mM Tris (pH 7.9), 10 mM MgCl2, 60 mM KCl, and 1 mM DTT (Tapprich & Hill, 1986). Cleavage was initiated by the addition of 0.02 units of RNase H/ml reaction (Wako Pharmaceuticals) and allowed to proceed for 20 minutes at 42 C. Reaction was stopped by extracting with neutralized phenol, and precipitating with two volumes of ethanol. The digested RNA was resuspended in 10 mM Tris (pH 7.6), 50 mM EDTA, heated at 80 C for four minutes, and loaded directly onto 5% to 25% (w/v) sucrose gradients in 10 mM Tris (pH 7.6), 50 mM KCl. SW41 gradients were spun at 30000 rpm for 16 hours, fractionated, and the appropriate peaks collected and ethanol-precipitated for use in reconstitution reactions. It is worth noting that the natural 23 S rRNA fragments obtained by RNase H digestion have heterogeneous edges resulting from the enzymatic treatment and that, moreover, the two natural pieces combined lack a total of ca 20 nucleotides covered by the cDNA oligonucleotide (data not shown). Notably, the in vitro transcribed partial rRNA 50 fragment includes the ca 20 nucleotides removed from natural 23 S rRNA during the RNase H cleavage reaction; the inclusion of these nucleotides increases the activity of the two-piece T7/ natural reconstitution ca twofold over the equivalent T7 fragment missing those 20 nucleotides (data not shown). In vitro transcripts were transcribed by T7 RNA polymerase using PCR DNA templates (Milligan et al., 1987; Saiki et al., 1988) obtained by amplifying wild-type and mutant rRNA sequences found on plasmids; pLK45(‡PS), pLK45(ÿPS), pLK45-A2251(ÿPS), pLK45U2251(ÿPS), pLK45-A2585(‡PS), pLK45-C2585(‡PS) and pLK45-G2585(‡PS). The primers used for ampli®cation of the 50 partial in vitro transcript at the 2300 fragmentation site were as follows: 39.3, 50 -TAA-TAC-GACTCA-CTA-TAG-GTT-AAG-CGA-CTA-AGC-GTA-CAC3 0 and 2395, 5 0 -TGA-TGT-CCG-ACC-AGG-ATT-AG-30 . By using 23 S plasmid constructs not containing engineered priming sites (in particular the engineered priming site located at 2300), the 2310 region was compatible with the 30 natural fragment of 23 S rRNA. The primers used for ampli®cation of the 30 partial in vitro transcript at the 2530 fragmentation site were as follows: 35.3, 50 TAA-TAC-GAC-TCA-CTA-TAG-GGC-TGA-AGT-AGGTCC-CA-3 0 and 21.1, 5 0 -AAG-GTT-AAG-CCT-CACGGT-TCA-30 . Transcripts were separated from unincorporated nucleotides on a G50-Sephadex (Pharmacia) gel ®ltration column (1 cm  20 cm) run in 10 mM Tris (pH 7.5), 4 mM magnesium acetate. Recovered RNA was precipitated with 0.5 M ammonium acetate (pH 6.0) and resuspended in 10 mM Tris (pH 7.5). In vitro reconstitution reactions In vitro reconstitution was performed essentially as described (Nierhaus, 1990). Brie¯y, in a total volume of 20 ml, 0.5 A260 unit of full-length 23 S rRNA (or 0.4 A260 unit of the 50 domain and 0.1 A260 unit of the 30 domain resulting from cleavage at 2310, or 0.43 A260 unit of the 50 domain and 0.07 A260 unit of the 30 domain resulting from cleavage at 2530) and 0.02 A260 unit of 5 S rRNA (Boehringer Mannheim) were incubated with 1.2 equivalents of TP50 (total protein from 50 S subunits) in 20 mM

Tris (pH 7.4), 4 mM magnesium acetate, 0.4 M NH4Cl, 0.2 mM EDTA and 5 mM 2-mercaptoethanol at 44 C for 30 minutes. Next, the concentration of magnesium acetate was raised to 20 mM and a second incubation step was carried out at 50 C for 90 minutes. The resulting mixture was added directly to a peptidyl transferase reaction and the products analyzed as described above. CMCT probing of 2251 mutant ribosomes Wild-type or mutant E. coli 70 S tight-couple ribosomes (100 pmol) were incubated in 500 ml 70 mM potassium borate, (pH 8.0), 100 mM NH4Cl, 20 mM MgCl2 and 6 mM DTT at 37 C for ten minutes. An aliquot (100 ml) was removed as an unmodi®ed control (K), mixed with 10 ml of 3 M sodium acetate (pH 5.5) and quick-frozen. Next, an equal volume (400 ml) of CMCT at a concentration of 42 mg/ml in the same buffer was added to the ribosome mixture at 37 C. Aliquots of 100 ml were removed after three minutes and nine minutes and the reaction stopped with the addition of 10 ml of 3 M sodium acetate (pH 5.5) and quick-freezing. Isolation of the modi®ed rRNA and allele-speci®c primer extension were performed as described (Samaha et al., 1995).

Acknowledgements We thank B. Cormack and G. Culver for critically reading the manuscript; B. Weiser for Figure 1; O. Uhlenbeck for supplying phenylalanyl tRNA synthetase; and S. Joseph and K. Lieberman for helpful discussions. This work was supported by grants from the NIH, the NSF, the Lucille P. Markey Charitable Trust to the Center for Molecular Biology of RNA, and a postdoctoral fellowship from the Damon Runyon-Walter Winchell Foundation to R.G.

References Aagaard, C., Rosendahl, G., Dam, M., Powers, T. & Douthwaite, S. (1991). Speci®c structural probing of plasmid-coded ribosomal RNAs from Escherichia coli. Biochimie, 73, 1439± 1444. 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. Breitmeyer, J. B. & Noller, H. F. (1976). Af®nity labeling of speci®c regions of 23 S RNA by reaction of Nbromoacetyl-phenylalanyl-transfer RNA with Escherichia coli ribosomes. J. Mol. Biol. 101, 297± 306. Deutscher, M. P. (1983). Enzymes of Nucleic Acid Synthesis and Modi®cation, vol 2, pp. 159, CRC Press, Boca Raton, FL. Douthwaite, S., Garrett, R. A. & Wagner, R. (1983). Comparison of Escherichia coli tRNAPhe in the free state, in the ternary complex and in the ribosomal A and P sites by chemical probing. Eur. J. Biochem. 131, 261± 269. Green, R. & Noller, H. F. (1996). In vitro complementation analysis localizes 23 S rRNA posttranscriptional modi®cations that are required for Escherichia coli 50 S ribosomal subunit assembly and function. RNA, 2, 1011± 1021. Gregory, S. T., Lieberman, K. R. & Dahlberg, A. E. (1994). Mutations in the peptidyl transferase region

JMB MS 2874 [24/1/97] 50

Mutations at G2251 and U2585 of 23 S rRNA

of E. coli 23 S rRNA affecting translational accuracy. Nucl. Acids Res. 22(3), 279 ± 284. Huttenhofer, A. & Noller, H. F. (1992). Hydroxyl radical cleavage of tRNA in the ribosomal P site. Proc. Natl Acad. Sci. USA, 89, 7851± 7855. Milligan, J. F., Groebe, D. R., Witerell, G. W. & Uhlenbeck, O. C. (1987). Oligoribonucleotide synthesis using T7 RNA polymerase and synthetic DNA templates. Nucl. Acids. Res. 15(21), 8783± 8798. Mitchell, P., Stade, K., Osswald, 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(4), 887 ± 896. Moazed, D. & Noller, H. F. (1986). Transfer RNA shields speci®c nucleotides in 16 S ribosomal RNA from attack by chemical probes. Cell, 47, 985 ± 994. 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. & Noller, H. F. (1991). Sites of interaction of the CCA end of peptidyl-tRNA with 23 S rRNA. Proc. Natl Acad. Sci. USA, 888, 3725± 3728. Monro, R. E., Cerna, J. & Marcker, K. A. (1968). Ribosome-catalyzed peptidyl transfer: Substrate speci®city at the P-site. Biochemistry, 61, 1042± 49. Monro, R. E., Celma, M. L. & Vazquez, D. (1969). Action of sparsomycin on ribosome-catalyzed peptidyl transfer. Nature, 222, 356 ±358. Nierhaus, K. (1990). Reconstitution of ribosomes. In Ribosomes and Protein Synthesis: A Practical Approach, pp. 161 ± 189, IRL Press, Oxford. Noller, H. F., Kop, J. A., Wheaton, V., Brosius, J., Gutell, R. R., Kopylov, A. M., Dohme, F., Herr, W., Stahl, D. A., Gupta, R. & Woese, C. R. (1981). Secondary structure model for 23 S ribosomal RNA. Nucl. Acids Res. 9, 6167± 6189. Peattie, D. A. & Herr, W. (1981). Chemical probing of the tRNA-ribosome complex. Proc. Natl. Acad. Sci. USA, 78(4), 2273± 2277.

Porse, B. T. & Garrett, R. A. (1995). Mapping important nucleotides in the peptidyl transferase centre of 23 S rRNA using a random mutagenesis approach. J. Mol. Biol. 249, 1 ± 10. Powers, T. & Noller, H. F. (1993). Allele-speci®c structure probing of plasmid-derived 16 S ribosomal RNA from Escherichia coli. Gene, 123, 75± 80. Prince, J. B., Taylor, B. H., Thurlow, D. L., Ofengand, J. & Zimmerman, R. A. (1982). Covalent crosslinking of tRNAVal to 16 S RNA at the ribosomal P site: identi®cation of crosslinked residues. Proc. Natl Acad. Sci. USA, 79, 5450± 5454. Saiki, R. K., Gelfand, D. H., Stoffel, S., Scharf, S. J., Higuchi, R., Horn, G. T., Mullis, K. B. & Erlich, H. A. (1988). Primer-directed enzymatic ampli®cation of DNA with a thermostable DNA polymerase. Science, 239, 487± 491. Samaha, R. R., Green, R. & Noller, H. F. (1995). A base pair between tRNA and 23 S rRNA in the peptidyl transferase center of the ribosome. Nature, 377, 309± 314. Sirum-Connolly, K. & Mason, T. L. (1993). Functional requirement of a site-speci®c ribose methylation in ribosomal RNA. Science, 262, 1886± 1889. Tapprich, W. E. & Hill, W. E. (1986). Involvement of bases 787-795 of Escherichia coli 16 S ribosomal RNA in ribosomal subunit association. Proc. Natl Acad. Sci. USA, 83, 556± 560. von Ahsen, U. & Noller, H. F. (1995). Identi®cation of bases in 16 S rRNA essential for tRNA binding at the 30 S ribosomal P site. Science, 267, 234± 237. 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.

Edited by P. E. Wright (Received 30 May 1996; received in revised form 14 November 1996; accepted 14 November 1996)