Peptide Environment of the Peptidyl Transferase Center from Escherichia coli 70 S Ribosomes as Determined by Thermoaffinity Labeling with Dihydrospiramycin*

In an attempt to gain information about the peptidyl transferase center at the peptide level we cross-linked the spiramycin derivative dihydrospiramycin to its functional binding site in the 70 S ribosome of Escherichia coli . In this manner ribosomal proteins S12, S14, L17, L18, L27 and L35 were found specifically affinity-labeled. Proteolytic fragmentation of these proteins, separation by C 18 reversed-phase high performance liquid chromatography of the peptide mixtures, and sub- sequent sequence analysis of labeled peptides revealed peptide regions at positions Ala 1 -Lys 9 and Tyr 116 -Lys 119 of S12, Leu 47 -Asp 53 of protein S14, Ser 6 -Lys 35 of protein L17, Ala 57 -Lys 63 of protein L18, Ala 5 -Lys 18 and Val 66 Lys 71 of protein L27, and Thr 5 -Lys 11 of protein L35. This approach is a valuable tool to characterize the binding site of spiramycin as well as the peptidyl transferase center at the molecular level. Studies on ribosomal structure and function have been greatly aided of Such studies on functional of the ribosome characterize antibiotic

In an attempt to gain information about the peptidyl transferase center at the peptide level we cross-linked the spiramycin derivative dihydrospiramycin to its functional binding site in the 70 S ribosome of Escherichia coli. In this manner ribosomal proteins S12, S14, L17, L18, L27 and L35 were found specifically affinitylabeled. Proteolytic fragmentation of these proteins, separation by C 18 reversed-phase high performance liquid chromatography of the peptide mixtures, and subsequent sequence analysis of labeled peptides revealed peptide regions at positions Ala 1 -Lys 9 and Tyr 116 -Lys 119 of S12, Leu 47 -Asp 53 of protein S14, Ser 6 -Lys 35 of protein L17, Ala 57 -Lys 63 of protein L18, Ala 5 -Lys 18 and Val 66 -Lys 71 of protein L27, and Thr 5 -Lys 11 of protein L35. This approach is a valuable tool to characterize the binding site of spiramycin as well as the peptidyl transferase center at the molecular level.
Studies on ribosomal structure and function have been greatly aided by the use of antibiotics acting as inhibitors of protein biosynthesis. Such studies have shed light on functional domains of the ribosome and also have helped to characterize the antibiotic binding sites (Cundliffe, 1990;Cooperman et al., 1990).
The location of greatest functional and structural interest in the ribosome is the peptidyl transferase center, since it is the site where peptide bond formation takes place (Monro et al., 1969). To date, only limited data exist on the fine structure of the peptide moiety of this important enzymatic center, whereas detailed structural information is available on the rRNA moiety (Brimacombe et al., 1990).
A more minute insight into the topography of the peptidyl transferase center can be obtained by cross-linking ribosomal antibiotics to their functional sites (Cooperman et al., 1990). Prominent among these antibiotics are macrolides. They can be roughly classified structurally and functionally into two main groups represented by spiramycin and erythromycin, respectively (Vazquez, 1979). The erythromycin group has a 13-carbon lactone ring that is glycosylated in two positions by different monosaccharides. The second group consists of a 15-carbon lactone ring system, usually derivatized at only one position by a disaccharide. Members of the erythromycin family appar-ently block protein synthesis only after the formation of several peptide bonds due to steric hindrance by the drug molecule (Moazed and Noller, 1987;Contreras and Vazquez, 1977). The spiramycin group of macrolides exerts its effects in the very early stage of peptide bond formation by competing for the ribosomal A-site with the incoming charged tRNA as determined by kinetic analysis with puromycin (Dinos et al., 1993). In this respect the effect is comparable with that of typical peptidyl transferase inhibitors such as chloramphenicol. The dissimilar modes of action of macrolides have mainly been attributed to the different sugar moieties attached to the lactone ring. So far, the question remains to be answered whether the two groups of macrolides have overlapping or in fact different binding-sites (Vazquez, 1979).
A direct way to test both alternatives is to cross-link these molecules to their functional sites. Previous affinity labeling has revealed protein L22 as the major cross-linking site for erythromycin (Arévalo et al., 1988) and protein L27 for spiramycin (Tejedor and Ballesta, 1986), thus favoring the notion of different binding sites for the two groups. The amino acid residues involved in the respective cross-links have not been characterized so far but are of great interest in terms of elucidating the topography of the peptidyl transferase center.
In this communication we report peptide regions of the small and large ribosomal subunit proteins cross-linked to spiramycin in the Escherichia coli ribosome that were identified by direct sequence analysis. The results present details of the peptide environment close to the peptidyl transferase center. The cross-linking data are integrated with the available data concerning the peptidyl transferase center topography.
All chemicals were proanalysis grade or of higher quality, obtained from Merck (Darmstadt, Germany). Endoproteinase Lys-C was purchased from Boehringer Mannheim (Germany).
[ 3 H]Dihydrospiramycin was prepared from spiramycin (Sigma Deisenhofen, Germany) by reduction of the aldehyde group in position C 18 with [ 3 H]borohydride (DuPont NEN). Completion of the reaction was checked by TLC. [ 3 H]Dihydrospiramycin was repurified by RP-HPLC 1 on a laboratory-packed Vydac C 18 column (5 m, 300 Å) to remove decomposition products less than 30 days prior to use. The final product was analyzed by infrared and mass spectroscopy as well as by 1 H and 13 C NMR (data not shown).
Preparation of Ribosomes-Cells of E. coli strain MRE600 were grown and 70 S ribosomes were isolated as detailed by Brockmöller and Kamp (1986). 70 S ribosomes were extensively dialyzed against BKM10 at 4°C prior to affinity labeling. * This work was supported by Grant SFB 344 TP YE 6 from the Deutsche Forschungsgemeinschaft (to B. W.-L.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
‡  (Bischof et al., 1994). Thermal incorporation of 1 M [ 3 H]dihydrospiramycin (0.14 Ci/mmol) into activated 70 S ribosomes (50 A 260 units/ml) was performed in BKM10 buffer at 37°C for 30 min under constant mild agitation in a waterbath. Immediately after radiolabeling, the ribosomes were precipitated with 2 volumes of ethanol, 2-mercaptoethanol (9:1) for 2 h at Ϫ20°C. Reprecipitation of the ribosomes four times removed more than 90% of nonincorporated spiramycin (data not shown). The final pellet was redissolved in BKM10 buffer (340 A 260 units/ml), and proteins were extracted for 5 min with 50% acetic acid (Bischof et al., 1994). The precipitated rRNA was pelleted for 20 min at 13,000 rpm in an HB6-Sorvall rotor at 4°C. Supernatants were portioned into 30 A 260 units and stored at Ϫ80°C. Radioactivity of ribosomal proteins and rRNA was monitored by liquid scintillation counting.
Preparation and Purification of Labeled Ribosomal Proteins-Total proteins of 70 S ribosomes were separated by RP-HPLC on a laboratorypacked Vydac C 4 column (5 m, 300 Å) employing a gradient of acetonitrile in water both containing 0.1% trifluoroacetic acid. An aliquot of each protein fraction was counted in a liquid scintillation counter, the remainder lyophilized and stored at Ϫ80°C.
Peptide Mapping-Protein fractions containing radioactivity were digested with endoproteinase Lys-C at an enzyme:substrate (total protein) ratio of 1:50 in 400 l of 50 mM NH 4 HCO 3 , pH 7.8, for 12 h. The digests were stopped by injection of the reaction mixture into the HPLC system. All peptide separations were performed on a Vydac C 18 column as described by  employing buffer A, 0.1% trifluoroacetic acid in water, and buffer B, 0.1% trifluoroacetic acid in acetonitrile. Peak fractions were monitored for radioactivity. Fractions containing labeled peptides were dried under vacuum and sequenced in a model 477A pulsed liquid phase sequencer equipped with a model 120A phenylthiohydantoin derivative analyzer (Applied Biosystems, Foster City, CA). Sequences were identified by comparison with sequences in the NBRF, Swiss Prot and RIBO data banks. 2

RESULTS
Binding Specificity of Dihydrospiramycin-Specificity of dihydrospiramycin binding to 70 S ribosomes was tested in a saturation experiment. As is demonstrated in Fig. 1, increasing amounts of spiramycin competed with [ 3 H]dihydrospiramycin binding, indicating a specific binding of dihydrospiramycin to the ribosome.
Identification of Dihydrospiramycin-labeled Ribosomal Proteins and Peptides-Incorporation of dihydrospiramycin into 70 S ribosomes is a rather fast reaction. Approximately 50% of the total incorporation takes place within the first 2 min. The reaction reaches saturation after about 60 min. 2.6% of total radioactivity was specifically incorporated into ribosomes. 72% of this radioactivity was found in ribosomal proteins, and 17% was found in rRNA.
[ 3 H]Dihydrospiramycin-labeled total protein extracted from 70 S ribosomes was fractionated by C 4 RP-HPLC. Five major radiolabeled peaks were detected in the chromatogram (Fig. 2). Most of the radioactivity was found in peaks 2, 3, and 4. Pooling and N-terminal sequence analysis of radioactive peaks revealed the protein composition of these pools. Table I summarizes these results. No obvious difference in the elution pattern of unlabeled and labeled total protein of 70 S ribosomes was observed (data not shown). Subsequent proteolytic fragmentation of labeled protein pools with endoproteinase Lys-C and separation of the resulting peptide mixtures by C 18 RP-HPLC revealed only one peptide being labeled in the case of peaks 3, 4, and 13, while two and three peptides were found radiolabeled in peaks 20 and 2, respectively. Two examples of RP-HPLC separations of Lys-C-digested protein pools are presented in Fig. 3.

Sequence Analysis of [ 3 H]Dihydrospiramycin-labeled
Peptides-During sequence analysis of the purified affinity-labeled peptides a gap at the position of the modified amino acid is expected. However, for all residues determined the predicted phenylthiohydantoin-derivatives were found. Therefore, the af-  finity cross-link product is obviously not stable under the harsh conditions of Edman chemistry, and indeed acid lability of the dehydrospiramycin cross-link was suggested by Tejedor and Ballesta (1986). We proved this assumption correct by extended exposure of affinity-labeled ribosomes in 20% trifluoroacetic acid, 10% trichloroacetic acid, and 2% acetic acid at sequencerlike temperatures (i.e. 48°C). Here, we found that most of the originally incorporated radioactivity was subsequently lost during the heat procedure (data not shown). Table II lists the sequences of all [ 3 H]dihydrospiramycin cross-linked peptides together with the parental ribosomal proteins. It should be noted at this point that the sequence of the labeled S14 peptide (peak 13 in Table II) abruptly terminated at position Asp-53. Eventually, this might be an indication for a modification. The peptides obtained in this study can be roughly divided into three classes according to the degree of their incorporated radioactivity: (i) high label, peptide region Ala 1 -Lys 9 of protein S12; (ii) medium label, peptide region Thr 5 -Lys 11 of protein L35, Ser 6 -Lys 35 of protein L17, Leu 47 -Asp 53 of protein S14, Tyr 116 -Lys 119 of S12, and Ala 5 -Lys 18 as well as Val 66 -Lys 71 of protein L27; and (iii) low label, peptide region Ala 57 -Lys 63 of protein L18.

DISCUSSION
Affinity labeling with (dihydro)spiramycin is ideally suited to investigate structural components of the peptidyl transferase center in the ribosome. Similar to spiramycin, dihydrospiramycin inhibits peptide bond formation and is competing with puromycin for an overlapping binding site (Dinos et al., 1993). Our experimental approach was designed to identify peptide-(dihydro)spiramycin cross-links by direct peptide sequencing, thereby establishing data concerning the molecular environment of (dihydro)spiramycin in its functional binding site.
Approximately 72% of dihydrospiramycin incorporated into ribosomal proteins. Extraction of labeled 70 S ribosomes and separation of ribosomal proteins by RP-HPLC resulted in 5 (dihydro)spiramycin-labeled protein fractions. Proteolytic digestion of these fractions and subsequent peptide sequence analysis of labeled peptides yielded peptide sequences from six different ribosomal proteins i.e. proteins S12, S14, L17, L18, L27, and L35 (Table II).
The proteins thus identified to be in close contact with the spiramycin molecule have been implicated to be close to or at the peptidyl transferase region by various other experimental approaches (Walleczek et al., 1988;Traut et al., 1986). Our results are also in good agreement with previous results of Tejedor and Ballesta (1986), who could unambiguously identify L27 as being the primary cross-linked product in the 50 S subunit and tentatively identify S9/S12 as those that are labeled to the highest degree in the small subunit. In addition, with L35 we could identify a ribosomal protein that had not yet been placed in the current ribosomal models.
In our affinity-labeling approach with (dihydro)spiramycin the N-terminal region of S12 (positions Ala 1 -Lys 9 ) was found to be labeled with the highest intensity. Additionally, we obtained a second cross-link with medium intensity in the C terminus of the protein (positions Tyr 116 -Lys 119 ). These findings confirm speculations of Tejedor and Ballesta (1986) who proposed that S12 is involved in spiramycin binding, but these authors were FIG. 3. Two representative C 18 RP-HPLC chromatograms of Lys-C digests from dihydrospiramycin-labeled protein fractions and their corresponding radioactivity profile. A, pool 3 containing labeled protein S12. B, pool 2 containing labeled proteins S12 and L27 (see Fig. 2 and Table I) of a preparative dihydrospiramycin cross-link experiment were digested with endoproteinase Lys-C, and the resulting peptide mixtures were separated by C 18 RP-HPLC employing the following gradient: isocratic elution for 5 min at 6% solvent B (acetonitrile in 0.1% trifluoroacetic acid) followed by a linear gradient from 6 to 12% solvent B in 25 min, 12-35% solvent B in 270 min, 35-55% solvent B in 30 min, and 75-99% solvent B in 10 min. The absorption range was 0.04 absorbance units at full scale, and the flow rate 0.03 ml/min. The eluate was monitored at 220 nm. Peptides were monitored for radioactivity by scintillation counting. The radioactivity contained in single fractions is depicted as bars. The dashed line indicates the background radioactivity. Resulting peptides are indicated by numbers.  (1975) not able to provide unambiguous experimental proof for their assumption.
The results presented here for S12 are a bit surprising since macrolides were thought to bind only to the 50 S subunit (Vazquez, 1979). However, previous experiments with a macrolide-resistant L4 mutant in E. coli have demonstrated that additional mutations in proteins S12 and S5 of the 30 S subunit can mask this effect, thereby leading to a macrolide-sensitive phenotype again (Saltzman and Apirion, 1976). These findings shed light on the intricate influence both subunits have on the binding site as well as on the mode of action of macrolides. In the case of S12 this influence can be easily explained since the N-terminal region that we found labeled in our approach was also observed to be in close contact with the 16 S rRNA (Urlaub et al., 1995), most likely with the 530 loop or 900 stem loop (Stern et al., 1988). In addition, Alexander et al. (1994) identified S12 as a major neighbor to 16 S rRNA nucleotides 518 -526. Accordingly, conformational changes in the N-terminal region of the protein can be transferred through the 16 S rRNA molecule, and by this means the overall topography of the macrolide binding site can be modulated.
Immunoelectron microscopy (Stöffler-Meilicke and Stöffler, 1990) as well as neutron-scattering data (Capel et al., 1987) have placed S12 on the cytosolic side, close to the neck of the 30 S subunit. However, since the functional binding site of spiramycin is located directly in the peptidyl transferase center, which is believed to be located at the interface side of the 30 and 50 S subunit close to the base of the central protuberance (Oakes et al., 1990), at least the N-terminal as well as the C-terminal region of the protein have to be exposed at the peptidyl transferase center. This notion is supported by intraprotein cross-linking results of S12 and L16 (Traut et al., 1986). The latter is a well known constituent of the peptidyl transferase center and forms an intraprotein cross-link to L27, a protein that was also labeled in our approach. Another consequence that arises from the above results is that the 530 loop and/or the stem loop 900 of the 16 S rRNA must be exposed at the luminal side of the ribosome.
The second constituent of the spiramycin binding site on the 30 S subunit is S14. This finding further demonstrates the importance the small subunit has on the macrolide binding site. The peptide stretch that was labeled (positions Leu 47 -Asp 53 ) is adjacent to a peptide region that we found crosslinked to puromycin. This confirms that the midsection of S14 is a major component of the peptidyl transferase center.
The main constituents of the spiramycin binding site on the 50 S subunit are proteins L17, L27, and L35. The location of L17 in the ribosome has been a subject of controversy. Walleczek et al. (1988) place it distant from the peptidyl transferase center, while Traut et al. (1986) position it closer to the peptidyl transferase center since it can be cross-linked to protein L16. Our results favor the Traut location of L17, at least of the N-terminal region (positions Ser 6 -Lys 35 ). The discrepancies of the L17 location in both models could be brought into accordance if L17 possesses a more elongated structure. The finding that protein L17 is a major neighbor of nucleotides 2497-2505 of domain V supports this assumption. 3 In the case of L27 there is no doubt about the localization of this protein in the ribosomal models. Protein-protein crosslinking studies (Traut et al., 1986) as well as immunoelectron microscopy (Stöffler-Meilicke and Stöffler, 1990) all place it slightly beyond and left of the base of the central protuberance toward the L1 arm. Since L27 is affinity-labeled by several photoreactive tRNA derivatives (Wower et al., 1993) as well as by chloramphenicol (Cooperman et al., 1990) and by tiamulin (Högenauer et al., 1981) it must be considered a central element of the tRNA binding site. In addition, it can be crosslinked to numerous nucleotides in domain VII of the 23 S rRNA, which is in close neighborhood to domain V, the latter being an essential constituent of the peptidyl transferase center (Brimacombe et al., 1990).
A fair amount of labeling was found in L35. So far, no structural or functional information is available for this protein. However, it is obvious from our results that the N-terminal part of the protein must be part of the spiramycin binding pocket and therefore be exposed at the peptidyl transferase center. Additional data are necessary to clarify the significance of our results.
The protein that was labeled to the weakest extent in our approach was L18 (positions Ala 57 -Lys 63 ). It binds to the 5 S rRNA-complex and is implicated to be part of the peptidyl transferase center as proved by affinity-labeling (Cooperman et al., 1990) as well as by reconstitution studies (Hampl et al., 1981). Furthermore, L18 was one of the proteins we recently identified as being cross-linked to puromycin (Bischof et al., 1994). Both cross-link positions are adjacent to each other (puromycin, Tyr 64 -Lys 68 ). Taking the S14 data into account this result indicates that at least part of the spiramycin and puromycin binding sites are close to each other.
Another discussion point is noteworthy. As briefly mentioned in the discussion of the S12 results, a correlation of antibiotic and rRNA binding sites in ribosomal proteins was found (Urlaub et al., 1995). This general idea is supported by the existence of various other peptide-antibiotic cross-links that 3 R. Alexander, personal communication.

FIG. 4. Localization of dihydrospiramycin-labeled proteins (black circles) in the 70 S ribosome of Stö ffler-Meilicke and
Stö ffler (1990). The view presents the 30 S subunit from the cytosolic site and the 50 S subunit from the luminal site. The circle indicates the most probable position of L35 (the most likely location of L35 in the circle is depicted by arrows). Dihydrospiramycin-labeled peptide regions of the respective ribosomal proteins have to be placed toward the interface gap. The two-way arrow depicts the discrepancy in the positioning of L17 in the Traut (Traut et al., 1986) and Walleczek model (Walleczek et al., 1988). are found to be affinity-labeled in the same peptide region that is also in close neighborhood to the rRNA (Table III). Most of the peptide regions listed in Table III are highly conserved and predominantly contain basic residues and/or aromatic residues that can contribute to a protein-rRNA interaction that is prone to disruption by the respective antibiotic. These findings shed light on the importance ribosomal proteins have on the correct folding of functional ribosomes. Our observations indicate that ribosomal proteins in many cases determine the formation of well defined structural and functional domains to which equally well defined ribosomal substrates such as tRNAs or antibiotics can bind. The results discussed above prove that the labeled peptide stretches of proteins S12, S14, L17, L18, L27, and L35 are part of the spiramycin binding site in the ribosome (Fig. 4).
Cross-linking of antibiotics to their functional site in combination with isolation and sequencing of the peptide-antibiotic cross-links is a direct approach to establish data concerning the peptide environment of these antibiotics in their functional binding site. Similar studies on the binding site of other antibiotics and ribosomal substrates at the peptide level, as recently described for puromycin (Bischof et al., 1994), lead to a more detailed insight into ribosomal structure-function correlation.