Cross-linked Amino Acids in the Protein Pairs L 3 - L l 9 and L23-L29 of Bacillus stearothermophilus Ribosomes after Treatment with Diepoxybutane”

Treatment of native 50 S ribosomal subunits of Ba- cillus stearothermophilus with the homobifunctional cross-linking reagent diepoxybutane generated two cross-linked protein pairs, L3-Ll9 and L23-L29, which were isolated and identified. The analysis of the cross-linking sites at the amino acid level in both protein pairs is presented. Using a combination of se- quence analysis and mass spectrometry it could be demonstrated that His-28 in protein L3 and the N- terminal amino acids Met-1, His-2, and His-3 in protein L19 are involved in forming the cross-link L3- L19. Within the pair L23-L29 Met-1 in protein L23 and Lys-4 in protein L29 were identified as cross- linking sites employing a similar approach. Comparison of our data with results derived from other cross- linking experiments showed that in general the structural organization of the ribosomes in eubacteria (the Gram-positive B. stearothermophilus and the Gram-negative Escherichia coli) has been conserved to quite an extent during evolution but that the fine structures differ slightly. By mass spectrometry the specificity of diepoxybutane and its cleaving mechanism using so- dium periodate could be examined. In addition the complete amino acid sequence of protein L19 of B. stearothermophilus has been

led to a substantial model of the 30 S subunit and for parts of the 50 S subunit which is still under investigation. In the last decade, sequence analysis of the ribosomal proteins derived from E. coli had been completed (Wittmann-Liebold, 1986) and for the Bacillus stearothermophilus ribosome 44 protein primary structures are available Arndt et al., 1991;Kruft et al., 1991). Therefore, it became possible to undertake topographical investigations by comparative cross-linking experiments in E. coli and in B. stearothermophilus ribosomes Kamp, 1986, 1988;Herwig, 1990). More recently these studies were extended to Haloarcula marismortui (Bergmann, 1992) since ribosomal particles from this organism are more suitable for x-ray crystallographic studies (Makowski et al., 1987;Von Bohlen et al., 1991;Yonath et al., 1980Yonath et al., , 1991. In addition for this organism, the analysis of the primary structures is not far from completion (for overview, see Arndt et al., 1991). So far only a few cross-linked protein pairs have been identified at the amino acid level in E. coli (Allen et al., 1979;Maassen et al., 1981;Pohl and Wittmann-Liebold, 1988) as well as in B. stearothermophilus (Brockmoller and Kamp, 1988) and argue for an overall conserved structure within Gram-positive and Gram-negative organisms, In this paper we describe the identification of the crosslinked amino acids in the protein pairs L3-Ll9 and L23-L29 of B. stearothermophilus ribosomes which have been formed after treatment of the native 50 S subunits with diepoxybutane. Using a combination of sequence analysis and mass spectrometry, we were able to exactly determine the crosslinking sites and to describe more precisely than known in the literature the specificity of diepoxybutane and the cleaving mechanism using sodium periodate. In order to assign the cross-linked amino acids in these protein pairs, the primary structures of proteins L3 and L19 from B. stearothermophilus were sequenced in the course of this work by protein-chemical sequence analysis (BstL19, this paper) and by sequencing of the corresponding genes, respectively (BstL3, Herwig et al., 1992). from Fluka (Buchs, Switzerland). Polyvinylpyrrolidone (PVP-40) was obtained from Sigma (Deisenhofen, F. R. G.). Vydac C4 reversedphase support was from The Separation Group (Hesperia, CA), and laboratory-filled columns were used. Lichrospher 100 CH-18/2 columns were from Merck (Darmstadt, F. R. G.), Immobilon PVDF membranes were from Waters, Millipore (Eschborn, F. R. G.) and PVDF-Sequelon AA membranes from MilliGen (Burlington, MA). All other chemicals were of pro analysis grade or higher quality obtained from Merck.
Preparation of Ribosomes-Growth of B. stearothermophilus (strain 799) and preparation of ribosomes and ribosomal subunits have already been described (Brockmoller and Kamp, 1986). Cross-linking of 50 S subunits with 1% (v/v) DEB was done as described (Brockmoller and Kamp, 1986). Total protein extracts were obtained by acetic acid extraction (Hardy et al., 1969).
Isolation of Proteins-Isolation of the protein pairs L3-Ll9 and L23-L29 has been described in detail (Herwig, 1990). Different to the described purification procedure for L23-L29 the protein pair was alternatively separated on preparative one-dimensional high Tris-SDS-gel electrophoresis (Fling and Gregerson, 1986) following the chromatography on laboratory packed Vydac C4 reversed-phase columns (250 X 4.6 mm, 5-pm particle size, 300-A pore size). Subsequent blotting onto Immobilon PVDF membranes (0.45-pm pore size) was carried out at 150 mA for 1 h and 650 mA for 6.5 h (Choli and Wittmann-Liebold, 1990). The membranes were stained with Ponceau S (0.5% (w/v)) in 1% (v/v) acetic acid (Salinovich and Montelaro, 1986) and destained with water.
Protein L19 was isolated to homogeneity from B. stearothermophilus 50 S total protein mixture by one-step reversed-phase chromatography on an analytical Vydac C4 column using a gradient of 0.1% trifluoroacetic acid with acetonitrile as eluent.
Periodate Cleauage-Purified protein and peptide samples resulting from RP-HPLC separations were cleaved for 15 min in 100 ml 0.1% (v/v) trifluoroacetic acid at pH 2 containing 10 mM freshly dissolved sodium(meta)periodate. Oxidation was stopped by direct injection onto RP-HPLC columns.
For digestion of blotted proteins, the corresponding bands were excised out of the PVDF membranes, cut into small pieces, and incubated for 30 min in 1 ml of PVP-40 (0.2% (w/v) in MeOH Uvasol) at room temperature. Excess PVP-40 was removed by washing the membranes three times with water. A final wash was done with the digestion buffer. Enzymatic cleavages were then carried out as described above. After cleavage the digestion supernatant was transferred into a second Eppendorf tube and the membranes washed twice with 80% formic acid and twice with water. All washing solutions were added to the digestion mixture dried by vacuum centrifugation and stored at -20 "C for further HPLC analysis.
Separation of Peptides-Peptides were separated by RP-HPLC on Lichrospher 100 CH-18/2-columns (particle size 5 pm, pore size 100 A, 250 X 4.6 mm) with a linear gradient of acetonitrile in 0.1% trifluoroacetic acid.
N-terminal Sequencing-N-terminal sequence determination was performed on an Applied Biosystems pulsed liquid-phase sequencer, model 477A, equipped with a model 120 PTH amino acid analyzer. Samples were dissolved in 100% trifluoroacetic acid and applied to a trifluoroacetic acid-treatedpolybrene-coated glass filter that had been precycled, as described previously (Hewick et al., 1981).
Sequence analysis was also done in the Knauer Modular Protein Sequencer model 810 (Knauer, Berlin) according to Wittmann-Liebold (1988) and Herfurth et al. (1991a). Glu-C-digested peptides were attached via their C-terminal glutamic acid onto arylamine derivatized PVDF membranes (Sequelon AA, Herfurth et al., 1991b). For attachment the peptides (50-500 pmol) were dissolved in 50% (v/v) acetonitrile in water, applied onto the membrane, and dried for 5 min at 55 "C in the reaction chamber. Coupling occurred after the addition of 10 pl of carbodiimide (10 mg/ml) dissolved in 0.1 M MES, pH 5, in 15% acetonitrile for 20 min at ambient temperature.
Mass Spectrometry-The mass spectra were recorded on a Bio-IonTM 20 plasma desorption mass spectrometer (Applied Biosystems, Foster City, CA). The spectra were acquired for 1 X lo6 fission events corresponding to 10 min. For sample application a nitrocellulose matrix was employed prepared by electrospraying 50 pl of a nitrocel-lulose solution in acetone (2 mg/ml) onto a Mylar foil. Samples were dissolved in 20 pl of 0.1% trifluoroacetic acid in 20% acetonitrile (v/ v) and applied onto the target by the spin drying technique (Nielsen et al., 1988).

RESULTS
Treatment of 50 S Ribosomal Subunits with Diepoxybutane-After treatment of native 50 S ribosomal subunits of B. stearothermophilus with DEB, two new protein spots have been detected in the higher molecular mass area of the twodimensional gel pattern of total protein extract corresponding to the protein pairs L3-Ll9 and L23-L29. L3-Ll9 is always detectable as a double spot, which both contain the crosslinked protein pair as deduced from sequence analysis. The isolation and identification of both protein pairs has been described in detail (Herwig, 1990;Brockmoller and Kamp, 1986). As given under "Experimental Procedures," we had to slightly modify the purification procedure for L23-L29 which necessitated in situ enzymatic and chemical cleavages of the protein pair directly on blots.
Strategy for the Identification of Cross-linked Amino Acids-Preliminary identification of cross-linked peptides was done by HPLC analysis. Peaks exhibiting an altered elution behavior after periodate cleavage were candidates for cross-linked peptides and further investigated by sequence and MS analysis. The localization of the cross-linking site was based on the known sequence of the involved proteins. Due to the modification the cross-linked amino acids are not identifiable as PTH amino acids in the HPLC trace of the automatic Edman degradation since they are still linked to the other peptide chain via the cross-link. Monovalently modified amino acid residues also yielded gaps in the sequence because the modification has altered the elution position of the PTH derivative. Therefore, cross-linked peptides had to be confirmed by mass spectrometry.
The Cross-link L3-L19-5 nmol of L3-Ll9 were digested with S. aureus protease (Glu-C I digest) and the resulting peptide mixture separated by RP-HPLC (Fig. lA). Periodate cleavage and rechromatography of each single peak revealed that only one peak eluting at 34.5% acetonitrile ( Fig. 1B) was cleavable and resulted in five new peaks of shifted retention times (Fig. IC). N-terminal sequencing of the peptides yielded fragments from both proteins, L3 and L19, resulting from this cleavage (Table I). These fragments were considered as candidates for cross-linked peptides.
Another Glu-C digest of L3-Ll9 (Glu-C I1 digest), chromatographically analyzed prior to (Fig. 2a) and after periodate cleavage (Fig. 2b) of the whole mixture, showed altered elution behavior of two original peaks after periodate cleavage. The assignment of the resulting fragments (A = A1 + A2; B = B1 + B2) has been confirmed by periodate cleavage and rechromatography of the single peaks A and B. Analysis of the original peak A by sequencing, MS analysis and HPLC separation of chymotryptic peptides prior t o and after periodate treatment (data not shown) revealed that both peptides merely coelute and that no cross-linking site exists between them. Sequence analysis of the original peak B gave two peptides already shown to elute together in the Glu-C I digest:

NGDLIPVTVIxATPNVVLQKKTIE.
While histidine at position 11 of the L3 peptide could not be detected at all, both histidines of the L19 peptide were found in a very reduced rate. Additionally, both peptides were not detectable in equimolar amounts as would be expected for cross-linked peptides due to partial blockage of the L19 peptide. Periodate cleavage of peak B resulted in two new peaks (Fig. 2b). While the L3 peptide has scarcely changed its  (peak B l ) . Due to the minute amount of cleavage of the diepoxybutane bridge (Fig. 3). Further masses cross-linked peptides we were not able to confirm the mass of (931.6, 967.3, and 1010.6 daltons) result from uptake and the cross-linked peptide in the original peak B. Nevertheless, release of HzO. Additionally, partial fragmentation may have we analyzed the periodate-cleaved products B1 and B2 by occurred (903.8, 860.3, and 799.5 daltons). In peak B2 the mass spectrometry. In peak B1 we found the mass of the L19 mass of the L3 peptide as well as the mass of the overlapping peptide modified with one (949.5 Da) and two formylmethyl peptide, both modified by a formylmethyl group (2642.2 and  . 1C) resulting fromperiodate cleaoage of the peak eluting at 34.5% acetonitrile in the Glu-CZ digest (Fig. I, A and B 3220.2 daltons) could be confirmed (Fig. 4). These results strongly support that His-28 in protein L3 is cross-linked to one of the N-terminal amino acids Met-1, His-2, and His-3 of protein L19 (Fig. 5) and that an additional monovalent modification also exists in the N-terminal area of the L19 peptide. The results argue for a changing contribution of all three Nterminal amino acids to the cross-link formation.
At least analysis of all peptides resulting from Glu-C digest of L3-Ll9 by mass spectrometry revealed that no €-amino groups of lysines have reacted with DEB since they are still protonated at pH 7.9. Only two sites, both located in the Nterminal region of the two proteins, displayed monovalent modification by DEB (Table 11). One site could be identified as the N-terminal threonine of protein L3 due to blockage of the peptide. But parts of the corresponding peptide could be sequenced without exhibiting any modification at Thr-1 (see Table 11). MS analysis revealed the occurrence of unmodified as well as monovalently modified peptide eluting in one single peak. Hence, only the unblocked portion was accessible to sequence analysis. The other site of DEB modification was located in the N-terminal part of L19 whereas at least two of the three N-terminal amino acids were shown to be monovalently modified. In both proteins no other site of monovalent modification, even of histidines, has been detected. The Cross-link L23-L29-3-5 nmol of L23-L29 were digested directly from blots with chymotrypsin and Glu-C enzyme, respectively. Sequence analysis of peptides resulting from Glu-C digest yielded all L23 peptides except the Nterminal fragment whereas L29 eluted as a bulk of undigested protein material at higher retention times. After periodate treatment of the whole mixture, the missing N-terminal fragment of L23 appeared in the HPLC chromatogram. Redigestion of the bulk material and sequencing of the resulting peptides revealed most peptides of protein L29, except the Nterminal and C-terminal fragments. As a result most peptides of L23-L29 could be detected as single peptides either by sequencing or by MS analysis (Fig. 6). Furthermore, both proteins were sequenced up to positions 33 (L29) and 34 (L23) in the cross-link. While lysine in position 4 of protein L29 could not be detected as PTH amino acid lysine of position 2 was found in protein L23 after the first coupling reaction instead of the N-terminal methionine. From these results we conclude that the N-terminal methionine of protein L23 is cross-linked to methionine in position 4 of L29. Based on the putative cross-linking site, we reanalyzed the masses obtained from chymotryptic peptides of L23-L29. In fact we were now able to confirm the cross-linking site by mass analysis of the cross-linked peptide which is calculated to be 1221.36 Da and is in accordance with chymotryptic cleavage sites (data not shown).
Sequence Determination of Protein L19 from Bacillus stearothermophilus-Protein L19 was isolated from acetic acid protein extract of 50 S ribosomal subunits by one-step RP-HPLC on a Vydac C4 column and identified by two-dimensional gel electrophoresis (Geyl et al., 1981) and N-terminal sequencing. The whole sequence has been achieved by sequence analysis of overlapping peptides derived from enzymatic digests with Glu-C and Lys-C proteases (Fig. 7). The protein contains 116 amino acids and is typically basic with a net charge of +18. The positive charged domain is mainly located in the C-terminal half. The calculated molecular mass is 13446 Da. Comparison of the sequence with the corresponding protein from E. coli (Brosius and Arfsten, 1978) revealed 56% identical residues without introducing any gaps and gave an alignment score of 37.2 standard deviation units. Conserved regions are symmetrically distributed over the whole sequence (Fig. 8).

DISCUSSION
In order to obtain more precise data about the structural organization of the ribosome, we have made cross-linking experiments within the 50 S subunit of B. stearothermophilus ribosomes and identified the cross-linked amino acids in the two protein pairs L3-Ll9 and L23-L29. We chose diepoxy-   (Kimura et al., 1985). Single peptides, resulting from chymotryptic digest of the intact protein pair, could be confirmed by plasma desorption mass spectrometry (dotted lines). Continuous lines indicate peptides derived from S. aureus protease digest which could be confirmed by sequence analysis. butane as the cross-linking reagent which is already well-tried in ribosomal research (Baumert et al., 1978;Brockmoller and Kamp, 1986;Pohl and Wittmann-Liebold, 1988) and, due to its short span (4 A), generates only few cross-linked protein pairs.
L3-Ll9 which has been formed in highest yield in B. stearothermophilus was also obtained in E. coli after treatment with different cr$ss-linking reagents spanning distances between 4 and 12 A, respectively (Traut et al., 1980;Walleczek et al., 1989). The close location of both proteins was confirmed earlier by immunoelectron microscopy in combination with cross-linking data (Walleczek et al., 1988) and positions both proteins in the seat region of the "crown projection" slightly displaced from the central point of the 50 S particle toward the L7/L12 stalk. These results are in agreement with RNAprotein binding- (Leffers et al., 1988) and cross-linking studies (Brimacombe et al., 1990) in which the L3-binding site lies close to the location of the 3'-end of the 23 S RNA. Recently, 50 protein distances have been determined by neutron scattering applying the strategy of the "glassy ribosome" (May et al., 1992). The protein distances between the corresponding mass centers was determined to be 74 A for L3 and L19, which seems to be fairly high compared to our cross-linking data.
Cross-linking of L23-L29 has also been demonstrated in E. coli after treatment with the almost eqyally short cross-linking reagent o-phenylenedimaleimide (5 A) while no cross-link formation has been observed with DEB (Walleczek et al., 1989). Interestingly, the same proteins could also be crosslinked in H . marismortui uspg DEB and dithiobis(succinimidy1propionate) (12 A) as cross-linking reagents (Bergmann, 1991). The same cross-linking site as for B. stearothermophilus could be identified in protein L23.
These results display the close vicinity of proteins L23 and L29 in the different organisms and support the placement of protein L23 near the base of the 50 S subunit on the side of the L1 protuberance (Hack1 and Stoffler, 1988). This result is in accordance with RNA-protein cross-linking data in which E. coli ribosomal proteins L23 and L29 are positioned to adjacent sites in the 23 S RNA (Wower et al., 1981;Brimacombe et al., 1990). Nevertheless, reconstitution experiments of E. coli ribosomal subunits containing puromycinmodified L23 also argue for the positioning of L23 close to the peptidyl-transferase center (Weitzmann and Cooperman, 1990). An elongated shape of L23 may therefore be assumed.
The combination of sequencing and MS analysis proved to be necessary because the digests of L3-Ll9 resulted in variable amounts of overlapping peptides. Glu-Arg and Glu-Lys bonds especially were relatively stable against Glu-C digest. In addition the high specificity of DEB was confirmed since the only monovalent modification occurred in the N-terminal part of both proteins. Taking into account that the C-terminal parts of proteins L3 and L19, which are predominantly basic, are rather involved in 23 S RNA binding the N-terminal part may be more accessible for the reagent.
As also deduced from MS analysis most of the methionines were present in their oxidized form after cross-linking. This effect is increased after treatment of the peptides with periodate (Yamasaki et al., 1982) and may explain slight displacements of unmodified peptides after periodate cleavage in the HPLC chromatogram (see Fig. IC). This could also be proved by MS analysis of periodate-treated peptides: a formylmethyl group remained bound to the amino acid residue.
Within the protein pair L23-L29, Met-1 in protein L23 and Lys-4 in protein L29 were identified as cross-linking sites using the same approach as for L3-Ll9 except that enzymatic digests were carried out in situ on PVDF membranes. When Glu-C was used for fragmentation only peptides of protein L23 appeared in the HPLC chromatogram while L29 eluted as intact protein. Digestion of protein L29 failed when attached to the membrane due to the protein's high hydrophobicity.
The putative cross-linking site in the N-terminal area of L23-L29 was confirmed by N-terminal sequencing of the intact cross-link. While Lys-4 in L29 has not been detected 10 20 30 40

FIG. 8. Amino acid sequence comparison of ribosomal proteins L19 from B. stearothermophilus (BstLZ9) and E. coli
as PTH amino acid the sequence of L23 started with Lys instead of the N-terminal Met. Such an abnormal degradation due to cyclization in alkaline conditions and partial degradation of the succeeding amino acid residue has already been described (Chang, 1978). For ribosomal proteins S11, L33, L16, and the initiation factor IF-3, the direct formation of the thiohydantoin ring at the coupling reaction (in alkaline solution) has been observed by reaction of DABITC (or phenylisothiocyanate) with the N-terminal N-monomethylated amino acid. Hence, the first amino acid was cleaved off at the coupling in the first degradation step and extracted into the n-heptanelethylacetate solution together with excess DA-BITC/phenylisothiocyanate and by-products. The completeness of the cleavage obviously depends on the following sequence. Most likely, the same mechanism may explain the loss of N-terminal methionine during coupling when alkylated by DEB reaction. This phenomena requires further investigation. Despite our findings that DEB does not tend to react with €-amino groups of lysines at pH 7.9 Lys-4 in protein L29 has been identified as a cross-linking site. In this case, however, a glutamic acid which is deprotonated at pH 7.9 (pK 4.25) is located at position 5 and may attract a proton of the lysine tamino group which thereupon becomes more nucleophilic.
The results demonstrate that the structural organization of the ribosomes in different organisms has been conserved to quite an extent during evolution although the fine structures may slightly differ. It has also been shown that cross-linking is a useful means for structural elucidation of the ribosome. As deduced from our data the cross-linked amino acid residues in the corresponding protein pairs are not more than 4 A apart.