Cross-linking of Selected Residues in the N- and C-terminal Domains of Escherichia coli Protein L7/L12 to Other Ribosomal Proteins and the Effect of Elongation Factor Tu*

Five different variants of protein L7/L12, each with a single cysteine substitution at a selected site, were produced, modified with125I-N-[4-(p-azidosalicylamido)-butyl]-3-(2′-pyridyldithio)propionamide, a radiolabeled, sulfhydryl-specific, heterobifunctional, cleavable photocross-linking reagent that transfers radiolabel to the target molecule upon reduction of the disulfide bond. The proteins were reconstituted with core particles depleted of wild type L7/L12 to yield 70 S ribosomes. Cross-linked molecules were identified and quantified by the radiolabel. No cross-linking of RNA was detected. Two sites in the dimeric N-terminal domain, Cys-12 and Cys-33, cross-linked strongly to L10 and in lower yield to L11 but to no other proteins. The three sites in the globular C-terminal domain all cross-linked strongly to L11 and, in lower yield, to L10. Weaker cross-linking to 50 S proteins L2 and L5 occurred from all three C-terminal domain locations. The 30 S ribosomal proteins S2, S3, S7, S14, S18 were also cross-linked from all three of these sites. Binding of the ternary complex [14C]Phe-tRNA·elongation factor Tu·guanyl-5′-yl imidodiphosphate) but not [14C]Phe-tRNA·elongation factor Tu·GDP·kirromycin increased labeling of L2, L5, and all of the 30 S proteins. These results imply the flexibility of L7/L12 and the transient proximity of three surfaces of the C-terminal domain with the base of the stalk, the peptidyl transferase domain, and the head of the 30 S subunit.

The Escherichia coli ribosomal protein, L7/L12, is the most extensively investigated representative of the small, four-copy, dimeric acidic proteins that are found in large ribosomal subunits of all organisms and exist as a conserved quaternary structural element in which two dimers are integrated into the ribosome through binding to a common anchoring protein (1)(2)(3). One or both of the L7/L12 dimers form a conspicuous morphological feature on the ribosome known in E. coli as the L7/L12 stalk (4). The proteins can be simply and selectively removed from and reconstituted into the ribosome (5), a property that greatly facilitates experiments of the type reported here in which genetically and biochemically modified proteins replace the wild type.
The L7/L12 polypeptide is composed of two distinct organized structural domains linked by a flexible hinge (6,7), as summarized in Fig. 1. The elongated, helical N-terminal do-main, residues 1-33, is responsible for dimer interaction (8), and the globular C-terminal domain, residues 53-120, is necessary for factor binding (9 -11). The C-terminal domains can be cross-linked to each other in different orientations yet retain full functional activity in supporting polypeptide synthesis (12). Flexibility of L7/L12 has been demonstrated in solution by fluorescence techniques (13) and in the ribosome by proton NMR (14,15), by electron microscopy (16), and with fluorescence probes attached to the C-terminal domains (13,17,18).
Immunoelectron microscopy with monoclonal antibodies (19) directly showed the presence of the C-terminal domain at the tip of the stalk and the N-terminal domain at the base of the stalk. This was consistent with earlier demonstrations that the N-terminal domain was responsible for binding of the fulllength L7/L12 to L10 and to the ribosome (9,10). It was shown that one dimer per particle was sufficient to form a visible stalk (20), despite earlier studies with polyclonal antibodies that had suggested that both dimers were present in the stalk (21). Different studies indicated that the presence of one L7/L12 dimer on the body of the 50 S particle in an extended conformation directed toward the central protuberance (22,23). Additional evidence that a C-terminal domain can occupy a location not only extended across the body of the ribosome but also near the base of the stalk came from cross-linking between a predetermined location in the C-terminal domain, Cys-89, and Cys-70 of L10 (24,25) and also from hinge deletion studies (26,27). A different heterobifunctional reagent, APDP 1 showed a cross-link between Cys-89 and L11 and also with L10 in a lesser extent near the EF-G binding site near the base of the stalk (28). The site-specific cross-linking experiments led to the proposal of a possible "bent" conformation for one of the dimers in which the C-terminal domain could lie on the body of the subunit near the N-terminal domain at the base of the stalk. We asked the question whether there was a preferred surface of the C-terminal domain that made these contacts and designed cysteine probe sites on other faces of the C-terminal domain. In addition we investigated whether any protein other than those at the base of the stalk made contact with the C-terminal domain. Previous experiments with 2-iminothiolane showed contact between an undefined lysine of L7/L12 with proteins L2 and L5 (22,29). Two probe sites in the N-terminal domain were also tested. The present approach permits us to define which domain makes these distant contacts. The effect of elongation factors on the pattern of cross-linking was also examined.
EXPERIMENTAL PROCEDURES APDP and Iodobeads TM were products of Pierce. 125 I was procured from Amersham Life Science, Inc. All other chemicals were of reagent grade of high purity. Kirromycin was a gift from G. Brocades, Delft, The Netherlands.
Preparations of Ribosomes and Ribosomal Core Particles-Ribosomes were prepared from midlog culture of E. coli MRE600 as described earlier (30). The ribosomes used in these experiments were more than 90% tight couple population as determined by sucrose gradient centrifugation. Core ribosomes completely lacking L7/L12 (P0 cores) protein were prepared from 70 S ribosomes following the method of Ref. 5 with modifications as described (21). They were stored in aliquots at Ϫ86°C.
Construction, Overexpression, and Purification of Cysteine-substituted Variants of L7/L12-Wild type L7/L12 lacks cysteine. The preparation of the Cys-89, Cys-63, Cys-33 variants was as described previously (26). The two additional amino acid substitutions, Ala-12 to Cys-12 and Ser-99 to Cys-99, were generated using the site-specific oligonucleotide-directed mutagenesis system from Amersham (12) Labeling of APDP with 125 I-APDP was labeled with 125 I using Iodobeads TM as described previously (28). The specific radioactivity of 125 I was adjusted to 10 -20 Ci/mmol by adding cold sodium iodide. Radioiodinated APDP was purified by TLC on silica plates (Silica Gel 60 F 254 ; Merck) in chloroform/benzene/ethyl acetate/acetic acid (10:10:10: 1). The plate was exposed for a few seconds with Kodak X-Omat film, and the position of 125 I-APDP was identified by comparing the UV quenching pattern with the radioautograph. Silica from the area containing 125 I-APDP was removed from the plate and extracted with acetonitrile. The solution of radioiodinated APDP was stored at Ϫ20°C in the dark for periods up to one month, during which time there was no detectable decomposition as judged by running analytical scale TLC followed by fluorography.
Modification of L7/L12 Variants with 125 I-APDP-The modification reactions were performed in fluorescent light as described previously (28). The protein concentrations varied from 0.4 to 2 mg/ml. A 5-40-fold molar excess of 125 I-APDP over protein was used in different preparations. Unreacted reagent was removed by passing the reaction mixture through NAP-5 columns (Pharmacia Biotech Inc.) equilibrated with 50 mM triethanolamine-HCl (pH 7.4). Fractions (100 l) were collected, and radioactivity and protein concentration were determined. The final specific radioactivity was very reproducible throughout all L7/L12 variants and routinely gave values above 90% of modification, assuming specific radioactivity of 125 I-APDP to be the same as initial Na[ 125 ]I. Modified proteins were stored at Ϫ20°C and used without additional treatment.
Reconstitution of Ribosomes with Radiolabeled L7/L12 Cysteine Variants-The 70 S P0 core particles (1 mg, 0.37 nmol) completely lacking wild type L7/L12, as judged by two-dimensional polyacrylamide gel electrophoresis, were first incubated at 37°C for 15 min in a buffer containing 10 mM Tris-HCl (pH 7.2), 100 mM NH 4 Cl,10 mM MgCl 2 , and 14 mM 2-mercaptoethanol, passed two times through Bio-Spin 6 column (Bio-Rad) for the complete removal of the reducing agent, and then reconstituted individually with each of the five different radiolabeled Cys variants as described previously (28). Excess, or unbound L7/L12 was removed by centrifugation of the particles through a 10% (w/v) sucrose cushion in the same buffer (without 2-mercaptoethanol) for 5 h at 58,000 rpm at 4°C in a Beckman Ti65 rotor. The reconstituted ribosomal pellet was resuspended in the same buffer (without the reducing agent) at a concentration of 5 mg/ml. Typically, reconstituted 70 S ribosomes in different experiments had between 2.9 and 3.7 copies of 125 I(azidophenyl)thio-L7/L12Cys. It had been shown previously that modification with APDP had no inhibitory effect on ribosomal activity in a poly(U)-dependent polyphenylalanine synthesis assay in which 15-20 Phe residues/ribosome were polymerized (28).
Cross-linking and Sample Processing-Cross-linking was initiated by irradiation of the reconstituted ribosome samples kept in standard 1.5-ml Eppendorf tubes at 4°C for 5 min with a UV light source (Black Ray model B-100A, Fisher; UV light Ͼ300 nm). The source was mounted to provide a vertical beam of light centered over an opened sample tube at a distance of 10 cm. After the cross-linking reaction, samples were made 1% with 2-mercaptoethanol and incubated at 37°C for 1 h to cleave the disulfide bond in the cross-link bridge. For final identification, 50 pmol of radioactive ribosomes were mixed with 125 pmol of cold ribosomes in a buffer containing 1% of 2-mercaptoethanol. The total protein was extracted with 66% acetic acid (31), and the extracted proteins were extensively dialyzed against 6% acetic acid at 4°C and lyophilized. Samples of approximately 140 g were dissolved in 25 l of sample buffer containing 20 mM Bis-Tris acetate (pH 3.7), 8 M urea, and 1% 2-mercaptoethanol and analyzed by two-dimensional acid/urea gel electrophoresis (32). The gels were stained with Coomassie R-250 and dried on Whatmann 3MM paper.
Analysis of Cross-linked Ribosomal Subunits by Sucrose Gradient Centrifugation-Ribosomes containing L7/L12Cys63 modified with 125 I-APDP were analyzed by sucrose gradient centrifugation before photocross-linking, after cross-linking and reduction, in the absence of elongation factors, and in the presence of EF-Tu ternary complexes containing GMP-P(NH)P or GDP⅐kirromycin. In these experiments, excess L7/L12 used for reconstitution was not removed. Each aliquot contained 4 A 260 units of 70 S ribosomes and 644,000 cpm of the 125 I-APDP radiolabel. Centrifugation was carried out in an SW-40 rotor for 12 h at 25,000 rpm. The gradient buffer from 5-20% sucrose contained 1 mM MgCl 2 , 100 mM KCl, 10 mM Tris-HCl (pH 7.2), and 0.2% 2-mercaptoethanol. Fractions (0.45 ml) were collected, and A 260 and radioactivity were determined for each. The relative amount of radiolabel attached to ribosomal particles is difficult to quantify accurately due to the presence of the preponderant fraction at the top of the gradient. Approximately 2% of the label was found in the ribosome fractions, a yield consistent with nitrene-generating cross-linking reagents (33). For this reason it was impossible to compare quantitatively the different variants.
Identification of Cross-linked Proteins and Quantification of Crosslinking Patterns-Stained and dried gels were exposed with high sensitivity screens of a Molecular Imager TM System (GS-250, Bio-Rad) followed by quantitative analysis of the screen according to the procedures recommended by the manufacturer. The images were processed and quantified using Molecular Analyst TM software (Bio-Rad, Version 1.1). The exposure times were from 3 to 14 h depending on the amount of radioactivity in the gels. The instrument has a broad range of linearity, and all images were within this response range. The image size on the computer monitor was set to represent the actual size of the stained gel. Qualitative identification of cross-linked proteins was performed by superimposition of the patterns of radioactivity with a transparency of  (36). The extent of ternary complex formation was 40%, assayed by adsorption filtration through nitrocellulose filters (35). After incubation, the samples were irradiated, and the distribution of cross-linked proteins was analyzed as described above. To measure the subunit distribution and cross-linking yield of ribosomes with bound EF-Tu, reduced samples were analyzed by sucrose gradient centrifugation as described above.

Selection and Location of Sites for Cysteine Substitution and
Photocross-linking-There are no cysteine residues in wild type L7/L12 from E. coli (37). Fig. 1 shows the locations of the five single cysteine substitutions used in the study described here in the non-staggered, parallel arrangement of the dimer (3). Three are in the C-terminal domain, the domain required in elongation factor binding, and two are in the N-terminal domain, the domain required for dimerization and binding to the ribosome through interaction with L10. The N-terminal domain in Fig. 1A is a simplified representation that indicates the positions of residues 12 and 33. Recent, NMR structure revealed that each monomer is a hairpin of two parallel ␣-helices, the dimer having a four-stranded ␣-helical structure (38).
Residues Cys-12 and -33 are distal and proximal to the hinge, respectively. The location of residues 63, 89, and 99 are represented in the model based on the high resolution crystallographic structure of the C-terminal domain (39) as shown in Fig. 1B. Residue 89 is located in the turn between the ␣B helix and the ␤B sheet, residue 63 is in the turn between the ␤A sheet and the ␣A helix, and residue 99 is in the turn between the ␤B sheet and the ␣C helix. The three sites are on exposed surfaces facing in different directions. Residues 63, 89, and 99 (␣Cs) are 30.3, 23.9, and 5.2 Å, respectively, from residue 53 (␣C), which is located at the junction of the globular head and the flexible hinge. Cys-99 is near the junction of the C-terminal domain with the hinge. Residue 63 is 13.9 Å from residue 89.
Analysis of APDP Cross-linking-UV-induced reaction between 125 I-APDP attached to L7/L12 and target molecules leads to disulfide-bridged cross-links. Subsequent reduction breaks the bridge, with the concomitant transfer of label to the target. The subunit specificity and UV dependence of radiolabeling was analyzed by sucrose gradient centrifugation. The samples were reduced, and the gradient contained reducing agent. Any labeling of subunits indicates transfer of radiolabel from L7/L12 to a different neighboring location. Fig. 2 shows these results for L7/L12 labeled at Cys-63, the location that consistently gave the highest overall level of labeling. Panel A shows that there is radioactivity coincident with the 50 S peak and possibly with the 30 S subunit. The A 260 -absorbing material at the top of the gradient is due to free (azidophenyl)thio resulting from reaction of the photoactive moiety with H 2 O. Both (azidophenyl)thio and protein-bound APDP contribute to the radioactivity at the top of the gradient, the presence of which makes it difficult to estimate any label in the 30 S subunits. Panel B indicates that the labeling of the 50 S subunit was largely dependent upon irradiation. It was difficult to exclude completely background light. That the major target of labeling was the 50 S subunit is to be expected from the location of L7/L12 and, as seen later, from the fact that a major labeled species is L7/L12, itself due to transfer from one L7/L12 subunit to another ("homolabeling"). When SDS was included in the samples and gradient buffer, all radiolabel was found at the top of the gradient. There was no label coincident with the 23 and 16 S ribosomal RNA peaks. This was the case for all five variants (results not shown). Accordingly, characterization of the radiolabeled ribosomal proteins was pursued. Fig. 3 shows analysis of ribosomal proteins by two-dimensional gel electrophoresis. Panel 3A shows the stained gel of 70 S proteins. The gel system separates most 70 S ribosomal proteins with good resolution, and this is the case for the labeled proteins. Panel 3B shows the distribution and identities of radiolabeled proteins with the 125 I-APDP probe attached at Cys-63, the probe location that gave maximal labeling of proteins other than L7/L12 itself. The ellipsoids indicate how the Molecular Imager analysis was conducted as detailed under "Experimental Procedures." Proteins L7/L12, L10, L11 are the major labeled proteins. Proteins L2 and L5 are labeled to a less, but clearly significant extent. In addition to 50 S proteins, 30 S ribosomal proteins S2, S3, S7, S14, and S18 are clearly labeled. Protein S4 is visible in this gel but is generally found only at a very low level and only with Cys63. Only a low level of labeling occurred in the absence of irradiation, due to difficulty in completely excluding stray light, as indicated by the sucrose gradient analysis. Only L7/L12 itself was labeled when irradiation was performed before reconstitution. The result rules out labeling by disulfide interchange. When modified, L7/L12 was irradiated after mixing with 30 S subunits, and a low level of S2 labeling occurred in addition to the homolabeling.
Panels C, D, E, and F show the patterns of labeling for variants Cys89, Cys99, Cys12 and Cys33, respectively. The patterns for the C-terminal variants Cys89 and Cys99 are qualitatively similar to Cys63. The patterns for the N-terminal variants Cys12 and Cys33 are notably different. Labeling is confined to the L7/L12, L10, L11 regions, and there is no labeling of 30 S subunit proteins.
Equal amounts of total ribosomal protein from ribosomes reconstituted with the different C-terminal domain variants radiolabeled with the same specific activity were analyzed. When homolabeling was excluded there were differences in the overall yield of labeling of target proteins for the three variants. Cys63 gave the highest yield, and Cys99 gave the lowest.
The Molecular Imager was programmed to quantify the radiolabel present in each protein spot relative to the total detected on the gel. The results for all five variants are shown in Fig. 4. The height of the bars provides a linear depiction of the relative distribution of the extent of labeling among the total proteins. The 10 proteins listed qualitatively from the gels shown in Fig. 3 appear as bars of different heights. For all of the variants except Cys12, L7/L12 itself most heavily labeled species; therefore, the software was programmed to normalize each protein to L7/L12, taken as 1 in the bar graph. The extent of homolabeling (transfer of label to L7/L12 itself) was compared for each variant in the free and ribosome-bound state. Homolabeling in solution for the two N-terminal domain variants was much greater than for the C-terminal domain variants, with Cys63 distinctly lower than Cys89 and Cys99. Ribosome binding had no effect on the C-terminal domain sites but greatly lowered the Cys12 homolabeling and reduced Cys33 homolabeling (data not shown).
Cross-linking from the N-terminal domain sites to proteins other than L7/L12 was limited to proteins L10 and L11, with L10 the strongest. For Cys12, L10 labeling exceeded that of L7/L12. The labeling of L10 from both Cys12 and Cys33 greatly exceeded L10 labeling from any of the C-terminal domain sites. Protein L11 was labeled to a lesser extent than L10, and the probe from Cys33 gave more L11 labeling than Cys12. The different labeling patterns for the different Cys residues again reinforces the specificity of the photochemical reaction and the absence of labeling by chemical disulfide interchange. Crosslinking from the three C-terminal domain sites was strongest FIG. 5. Sucrose gradient analysis of ribosomal complexes containing EF-Tu. A, samples represent aliquots of the same ribosomes that contain 125 I-APDP-Cys63 L7/L12 shown in Fig. 2. A, ternary complex with GMP-P(NH)P. B, ternary complex with GDP⅐kirromycin.
FIG. 6. Quantitative analysis of the distribution of label among target proteins as described for Fig. 4. All proteins were normalized to protein L11 taken as 1. Open bars, no factor present; gray bars, EF-Tu ternary complex with GMP-P(NH)P; black bars, EF-Tu ternary complex with kirromycin⅐GDP.
to L11, a result reported previously for Cys89 (28). Crosslinking to L10 was lower than to L11. L11 was labeled more from the C-terminal domain sites than from the N-terminal domain sites; the converse was the case for L10. Homolabeling was of approximately of the same magnitude as L10 and L11 (results not shown because of the normalization). Weaker, but significant labeling of L2 and L5, cross-links found previously with 2-iminothiolane (22), was detected from all three C-terminal domain sites. The C-terminal domain sites cross-linked to certain 30 S proteins, more strongly to S3, but significantly to S2, S7, S14, and S18, the extent of labeling being comparable to that of L2 and L5.
Effect of Elongation Factors on the Cross-linking Pattern-The distribution of cross-links for each of the five L7/L12 variants was determined in the presence of EF-G⅐GDP⅐fusidic acid, EF-G⅐GMP-P(NH)P and the EF-Tu⅐GMP-P(NH)P⅐Phe-tRNA, and the EF-Tu⅐GDP⅐kirromycin ternary complexes. Neither of the EF-G complexes gave any consistent, major alteration in the distribution of cross-links (results not shown). The EF-Tu complexes made with Cys63 subunits were first analyzed by sucrose gradient centrifugation and are shown in Fig. 5. The kirromycin complex shows a lower extent of 50 S labeling than the GMP-P(NH)P complex and resembled that with no factor present. The results were the same for the other variants (results not shown). The pattern of protein labeling was analyzed and quantified as before. These results with the EF-Tu ternary complexes are shown in Fig. 6, which compares the extent of labeling of the 10 major proteins in the absence and presence of the two EF-Tu complexes. Because there were significant differences in homolabeling, results were normalized to L11, taken as 1. Little difference can be discerned between the kirromycin complex and the pattern with no factor present. For the GMP-P(NH)P complex, there is a consistent increase in the labeling of L2, L5, and small subunit proteins S2, S3, S7, and S14. In addition, the homolabeling is increased. Fig. 7 shows the location on the ribosome of the proteins cross-linked from the various sites in L7/L12. The evidence for the location of 50 S cross-linked proteins L10, L11, L2, and L5 has been presented previously (40). The locations of the 30 S proteins is taken from the immunoelectron microscopy studies of Lake and co-workers (41) and Stöffler-Meilicke et al. (42,43). The 30 S subunit is shown with the platform on the left, opposite the L7/L12 stalk. All of the 50 S proteins labeled as well as S7 and S18 are on the subunit interface surface of the subunit; S2, S3, and S14 are on the exposed head and neck of the 30 S subunit. The roman numbers indicate alternative locations in the 50 S subunit for the C-terminal domain of L7/L12 as determined by immunoelectron microscopy and protein cross-linking (40). The letter A indicates a fourth site, implied by the results presented here. The curved arrows suggest movements or conformations of L7/L12 consistent with cross-linking to the target proteins.

DISCUSSION
The labeling of L11 and L10 from the two sites selected in the N-terminal domain is consistent with the role of the N-terminal domain in binding directly to L10 and the established proximity of L10 and L11 (44). In the C-terminal domain, Cys63 is maximally exposed and more distant from the N terminus than Cys99, and its formation of cross-links in highest yield is consistent with this. Nevertheless, all three probe locations in the C-terminal domain label L11, and to a lesser extent L10, defining site II. This reinforces the conclusion of a bent conformation for one of the dimers, facilitated by the flexible hinge (28). The labeling from all three residues, 63, 89, and 99 implies that there is no preferred tight interaction of a single C-terminal domain with L10 and L11 in a defined orientation, since it is unlikely that all three regions could simultaneously make the necessary contacts. The labeling from all three sites is more likely facilitated by the presence of two C-terminal domains at site II and their capacity to retain functional activity even when locked in disparate orientations by zero-length disulfide cross-linking (12). The reagent used here maximally extends 21 Å, which also plays a role in facilitating proximity from all three residues. The labeling of proteins L2 and L5 in peptidyl transferase site (site III) confirms earlier results with 2-iminothiolane (22) and defines the C-terminal domain as the relevant domain. In that study, L7/L12-L5 cross-link was identified in 50 S but not 70 S ribosomes, and it was suggested that competition for L5 lysines by 30 S interface proteins could account for this finding. The present result with 70 S ribosomes implies that 30 S binding does not block access across the interface cavity.
The cross-linking of 30 S proteins was an unexpected result, since none had been detected with 2-iminothiolane (22). The lysine specificity of that approach versus the nonspecificity of the photocross-linking used here may explain the different results. Proteins S2, S3, and S14 are clustered on the exterior of the head of the 30 S subunit on the same side of the 70 S particle as the L7/L12 stalk at site A. The location of this group of proteins coincides nearly perfectly with that determined by immunoelectron microscopy for EF-Tu (45,46). Previous studies indicated that the conformation of L7/L12 is modulated by interaction of elongation factors with the ribosomes and depends on hydrolysis of GTP (47). Movement of the flexible stalk dimer could bring the C-terminal domain near these proteins. The results suggest the participation of the C-terminal domain of L7/L12 in EF-Tu binding at the R (recognition) site and thus imply a role for L7/L12 in factor binding at a site other than the site at the base of the stalk where EF-G binds. The stimulation of labeling of 30 S proteins by EF-Tu is consistent with a conformational change that brings L7/L12 in closer proximity to sites on the small subunit. The proteins S7 and S18 are far from each other and far from the S2-S3-S14 cluster. Protein S7 is located near L5 and L2, and the extended orientation (site III) can also account for S7 labeling. The labeling of S18 is not readily explained in the static model depicted. The increase in S18 labeling caused by EF-Tu suggests a conformational change that brings the S18 site nearer the interface cavity, where contact with the inner dimer could take place.
This study confirms three possible locations for the C-terminal globular domains (40). Cross-linking of L11 and L10 confirms the bent conformation with the C-terminal domain at the base of the stalk (site II). Cross-linking of L2 and L5 confirms an extended orientation across the body of the 50 S subunit (site III). In cryoelectron microscopic studies, L7/L12 is also visualized as an extended structure within the 50 S body. 2 Cross-linking of S2, S3, and S14 shows a location of the stalk (site I), stretching across the subunit interface to reach the EF-Tu binding site A on the small subunit. These three locations appear to represent specific, stable conformations. The probe does not sweep out a swath of labeled proteins around a radius from its anchoring site or from the junction of the hinge and anchoring N-terminal domain. Site II is clearly of functional significance, since it coincides with the well established EF-G binding location. The results suggest a functional role for site I, the extended stalk. It may make an initial contact with EF-Tu in the 30 S site and be involved in its movement to site II at which EF-Tu and EF-G have overlapping sites and compete.
Determining more precisely the ribosomal location of the functionally important C-terminal domains and distinguishing the locations and dynamics of the two dimers is a goal of future work. Previous work showed the presence of L7/L12 in a crosslinked complex containing in addition L10 and EF-Tu (48). Work is in progress to estimate direct cross-links between factors and the L7/L12 variants described here. The most recent work on cryoelectron microscopy of EF-Tu-bound ribosomes indicated a direct contact of the factor from the L7/L12 stalk (49).