Reverse phase high performance liquid chromatography of Escherichia coli ribosomal proteins: standardization of 70 S, 50 S, and 30 S protein chromatograms.

We recently described the use of reverse phase high performance liquid chromatography for the separation of the proteins of the 30 S subunit of Escherichia coli ribosomes (Kerlavage, A. R., Kahan, L., and Cooperman, B. S. (1982) Anal. Biochem. 123, 342-348). In the present studies we report improvements in the technique and its extension to the separation of the proteins of the 50 S subunit and of 70 S ribosomes. Using an octadecasilyl silica column and a trifluoroacetic acid/acetonitrile solvent system, the 21 proteins of the 30 S subunit have been resolved into 17 peaks, the 33 proteins of the 50 S subunit into 22 peaks, and the 53 proteins of the 70 S ribosome into 31 peaks. The proteins present in each peak have been identified by polyacrylamide gel electrophoresis, by comparison with previously standardized chromatograms, and by calibration with authentic samples of purified proteins. All of the known ribosomal proteins have been identified on the chromatograms with the exception of L31 and its variant, L31'. Three protein peaks, not corresponding to known ribosomal proteins, have been observed in preparations from the total protein from 50 S subunits and 70 S ribosomes, but the significance of these peaks is unclear. The reverse phase high performance liquid chromatography technique has the potential for purifying all ribosomal proteins, as demonstrated by the increase in resolution we obtain when a peak isolated under standard gradient conditions and containing several proteins is reapplied to the column and eluted with a shallower gradient. Its utility in preparing proteins for functional studies is demonstrated by a reconstitution of active 30 S particles using 30 S proteins prepared by reverse phase high performance liquid chromatography.

The proteins derived from the small (30 S) subunit are designated S1-S21 and those derived from the large (50 S) subunit are designated Ll-L34. There are a total of 53 unique proteins; L8 is a stable complex of L7/L12 and L10 (1) and S20 and L26 are identical (a), one copy of this protein being partitioned between the 30 S and 50 S subunits (3).
The preparative and analytical separation of ribosomal proteins is of crucial importance to the study of the structure and function of the ribosome. Various approaches requiring purification or analysis of ribosomal proteins have been employed in these studies. These include reconstitution of ribosomal subunits from constituent components with the omission of a single protein (4,5), immunoelectron microscopy using antibodies to individual proteins (6,71, affinity labeling of the ribosome with derivatives of its various ligands (8-lo), neutron diffraction using selectively deuterated proteins (1 l ) , assembly mapping (reviewed in Ref. 12), protein cross-linking (13), and analysis of mutants with altered or missing proteins (reviewed in Ref. 14).
The major problem in separating the ribosomal proteins is that most of them are of similar molecular mass (6,000 to 30,000 Da) and isoelectric point (8 to 12) (15). The classical purification methods for ribosomal proteins include ion exchange and size exclusion chromatography steps. These procedures are quite laborious and yields range from 5% for the smaller, difficult to isolate proteins to 50% for the larger proteins (16)(17)(18)(19). The classical analytical methods for the separation of ribosomal proteins have been one-and twodimensional polyacrylamide gel electrophoresis. One-dimensional electrohoresis performed according to Leboy et al. (20) resolves the 21 proteins of the 30 S subunit into 15 bands and the 33 proteins of the 50 S subunit into 16 bands. For the case of analysis of radioactively labeled proteins, the time required for pouring, running, staining, destaining, and drying the gel and then oxidizing gel slices prior to scintillation counting is quite substantial. In addition, the recovery of covalently incorporated radioactivity is usually below 50% (21). Two-dimensional polyacrylamide gel electrophoresis, performed as introduced by Kaltschmidt and Wittmann (22) or using more recent modifications (23, 24), resolves almost all of the 30 S and 50 S proteins but with an even greater time expenditure and lower recovery of protein.
Reverse phase high performance liquid chromatography is well suited to separating the ribosomal proteins since the basis for retention is hydrophobicity rather than size or charge. RP-HPLC' also has the advantages of speed, high recovery yields, and high sensitivity. We previously reported on the separation of TP30 by RP-HPLC into a total of 15 peaks, and on the identification of the 30 S proteins within each of these peaks (25). In the present work we report several technical improvements in the RP-HPLC technique, use our improved methodology to separate TP50, TP30, and TP70 into large numbers of peaks, and identify the proteins eluting in each of the separated peaks. Our results clearly demonstrate the superiority of this technique to classical methods for both preparative and analytical separation of ribosomal proteins.

EXPERIMENTAL PROCEDURES
Materials-Trifluoroacetic acid and HPLC grade acetonitrile were purchased from Fisher. Lyophilization jars were siliconized with Prosil-28 (PCR Chemicals). Poly(U) was purchased from Miles. ["C] Phe.tRNAPhe was prepared from [14C]Phe (400-450 Ci/mol, New England Nuclear) and bulk-stripped tRNA (Grand Island Biological) as described by Ravel and Shorey (26). Coomassie brilliant blue G-250 and bovine serum albumin were purchased from Sigma. All other chemicals were reagent grade.
Purified E. coli 50 S ribosomal proteins were a generous gift of Dr. M. Nomura, University of Wisconsin. Ribosomal proteins L7/L12 were a generous gift of Dr. A. Dahlberg, Brown University.
Isolation of Ribosomal Proteins-70 S ribosomes were prepared from E. coli Q13 bacteria harvested in mid or late log phase using the modification of the Traub et al. (27) procedure previously described (21). Ribosomal subunits were prepared by sucrose gradient centrifugation as described previously (21) using buffer A (50 mM Tris-HC1 (pH 7.6 at 4 "C), 50 mM KCl, 1 mM MgClZ, 6 mM 2-mercaptoethanol). Protein was extracted from 70 S, 50 S, and 30 S particles using the Me-acetic acid procedure of Hardy et al. (16) and was precipitated with acetone (28). Precipitates were dissolved in buffer B (6 M urea, 150 mM LiCl, 10 mM (adjusted to pH 8.0 with methylamine), 3 mM 2-mercaptoethanol) prior to HPLC injection.
High Performance Liquid Chromatography-The HPLC system consisted of one 6000 A pump, one "45 pump, a 660 programmer, and a U6K universal injector, all from Waters Associates. Column eluates were monitored for UV absorbance using a Waters extended wavelength module (214 nm) and a model 440 absorbance detector (280 nm) connected in series. Each was equipped with a 15.5-pl cell with a 1-cm path length. All separations were performed on a Synchropak RP-P Cls-silica column (6.5 pm silica, 300-A pore, 4.1 X 250 mm; SynChrom, Inc.).
Proteins were eluted at room temperature using a gradient (as described in Figs. 1, 4, and 6) from 0.1% (w/v) F3CCOOH in H20, pH 2.14 (solvent A) to 0.1% (w/v) F3CCOOH in CHSCN (solvent B). The "45 was used to pump solvent A and the 6000 A was used to pump solvent B at a combined constant flow rate of 0.7 ml/min with column pressure between 500 and 1500 p.s.i. Solvent A was prepared from deionized, reversed osmosis-purified HzO which was filtered twice under aspirator vacuum through 0.45pm Metricel filters (Gelman Sciences) placed in a 0.5-pm sintered glass Millipore vacuum filter. FaCCOOH was added after filtration. CH&N was filtered once through 0.2-pm nylon 66 filters (Rainin Instruments) prior to addition of F3CCOOH. Solvents were stirred constantly during runs and degassed approximately every 24 h by filtering as above. Columns were returned to initial conditions by a 10-min linear gradient at a flow rate of 2 ml/min and equilibrated for an additional 10 min at 2 ml/min.
Little change in the chromatographic pattern obtained was observed on variation of the total protein applied in a single run between 10 pg and 2 mg. Minor losses in resolution were observed on application of up to 5 mg of protein. The overall yield of protein recovered from a typical HPLC run was 85%, in agreement with our previous result (25). Protein determinations were made according to Bradford (29) using bovine serum albumin as a standard.
Polyacrylamide Gel Electrophoresis-One-dimensional urea-polyacrylamide gels were run according to Leboy et al. (20). Two-dimensional urea-polyacrylamide gels were run according to Kenny et al. (24). Assignment of 50 S and 30 S protein migrations within the gels followed those of Mora et al. (30) and Rummel and Noller (31), respectively. Fractions from HPLC runs were lyophilized and redissolved in running buffer (8 M urea, 10 mM 2-mercaptoethanol) before application to the gels. 30 S Reconstitution-Non-HPLCor HPLC-treated TP30 (2.8 nmol) was reconstituted with 16 S RNA (2.3 nmol) as described by Traub et al. (27). TP30 was prepared either by MP-acetic acid extraction, with the acetone precipitate redissolved in 8 M urea, or by LiC1-urea extraction (17). In either case, the urea suspension of protein was exhaustively dialyzed against buffer C (5 mM H3PO4 (adjusted to pH 7.5 with KOH at 4 "C), 20 mM MgC12, 1 M KCl, 6 mM 2-mercaptoethanol) prior to reconstitution with 16 S RNA. HPLC-treated protein was prepared by applying an aliquot of TP30 in buffer C to the Cls-silica column and eluting the protein as described above. The total eluate was collected in a siliconized jar and lyophilized. The lyophilized protein was taken up in 4 M urea and 2 M LiCl and dialyzed against buffer C prior to reconstitution with 16 S RNA.
Sucrose Density Gradient Centrifugation-Reconstituted and native 30 S subunits (80-200 pmol) were centrifuged through linear 15-30% sucrose gradients made up in buffer A in a Beckman VTi50 rotor at 4 "C as described elsewhere (33).

RESULTS AND DISCUSSION
Since our previous work on the separation of TP30 by RP-HPLC (25), four important technical improvements have been made: (a) utilization of a 300-A pore Cle-silica column (Synchropak RP-P, SynChrom, Inc.) replacing the 125-A pore column (p-Bondpak, Waters Associates) used earlier; (b) changing the gradient shape from linear to convex; (c) addition of 0.1% F,CCOOH to the acetonitrile solvent; ( d ) dual wavelength monitoring at 214 and 280 nm. The results of applying our improved procedure to separations of TP50, TP30, and TP70 are discussed in turn below. 50 S Proteins-Using our improved procedure, the thirtythree 50 S proteins were resolved into at least 22 peaks as shown in Fig. 1. The relative retentions ( a ) of the peaks are listed in Table I Table I.   Tables I and 11, tR, was chosen as the retention time of L26 (= S20), the only protein common to both subunits. The retention times used to calculate n values in both tables were measured from a TP70 elution profile. The n values reported were reproducible to k0.01.
Peak F contains oxidized L27. When peak F was incubated at 42 "C in buffer B containing 100 mM 2-mercaptoethanol for 2 h, a new peak eluted in the position of L27. The peak also showed a spot corresponding to L27 in two-dimensional PAGE.

~ ~~
temperature and the composition of the column and the mobile phase, and is independent of column dimensions, instrument dead volume, sample size, and flow rate. The identities of the proteins in the peaks, summarized in Table  I, were determined by one or more of the following methods: ( a ) one-dimensional PAGE; (b) two-dimensional PAGE; (c) calibration with authentic samples of purified 50 S proteins. In the analyses using purified proteins, relatively high concentrations of protein standards were applied along with a low concentration of TP50, which served to calibrate the chromatogram. In the two-dimensional PAGE analyses, relatively high concentrations of proteins from HPLC peaks were run against a low concentration of background TP30, which served to calibrate the gel. A typical two-dimensional PAGE analysis is presented in Fig. 2, in which peak Q from Fig. 1 is shown to contain proteins L22 and L23. A trace of L18, from the neighboringpeak R, is also seen.
A steep gradient step at the end of the HPLC run was required to elute proteins L4 and L10 (peak X), L12 (peak Y ) , and L7 (peak 2) (Fig. 1). Resolution of the latter two proteins demonstrates that the L7/L12 dimer is disrupted under the elution conditions. Proteins L7/L12 are observed in reduced quantities in TP50 compared to TP70 (see Fig. 5 below). Recovery of 50 S subunits from sucrose gradients by high speed pelleting rather than ethanol precipitation afforded higher yields of L7/L12 in accord with previous results (3,34).
All of the previously identified TP50 proteins have been located in the HPLC chromatogram shown in Fig. 1 with the exceptions of proteins L31 and L31', the latter described by Fanning and Traut (35). These proteins are stained very weakly by Coomassie G-250, and would have been difficult to identify by the PAGE procedures discussed above. Furthermore, they have previously been isolated in only small amounts (3). Conversely all of the peaks in Fig. 1 have been identified as containing known ribosomal proteins except for peaks L), G, and H. In contrast to the other peaks in Fig. 1, the relative amounts of peaks D, C, and H varied with conditions of TP50 preparation and storage, an indication of their greater susceptibility to degradation. Peak G was the most unstable. When it was lyophilized, redissolved in buffer B, and rechromatographed, no peak eluted in the original position of peak G. Instead, ten earlier eluting peaks were observed. Peaks D and H were unstable to long term storage at -20 "C, but could be stored at -78 "C. On rechromatography of both peaks D and H , the major peaks eluted in the original peak positions, although several minor, earlier eluting peaks were seen with peak D. A composite schematic summary of two-dimensional PAGE runs showing positions due to peaks D and H is shown in Fig. 3. Peak D gave rise to three spots, two of which migrated with the dye front in the area of proteins L32 and L33, and one which was slightly slower in the second dimension. Peak H gave rise to two spots, one migrating near L33 and the other in an area not corresponding to a known ribosomal protein. It remains to be established whether peaks D, G, and H correspond to new 50 S proteins discovered as a consequence of our new HPLC method of analysis, or to degradation products of previously identified 50 S proteins, or to some other procedural artifacts.

S Proteins-The improved
RP-HPLC procedure described above has resulted in an increased resolution of the twenty-one 30 S proteins into 17 peaks (Fig. 4), as compared with the 15 peaks reported earlier (25). The identities of the proteins present in each peak, as summarized in Table 11, were determined by one or more of the following methods: ( a ) comparison with the previously standardized TP30 chromatogram (25); ( b ) calibration with authentic samples of purified 30 S proteins; ( c ) one-dimensional PAGE analysis.
In contrast to the TP50 results, all of the known TP30 proteins have been identified and all of the major peaks correspond to known 30 S proteins. The smaller unlabeled peaks seen in Fig. 4 are 50 S proteins, present due to minor contamination of our 30 S subunits by 50 S subunits. The intensity of peak e, corresponding to S20, when compared to peak K (L26) (Fig. 1) confirms the previously reported result that this single protein partitions mostly to the 30 S subunit (3). Two further points should be noted. Proteins S1 and S2 were observed in only small amounts reflecting losses during subunit preparation (3), and the shoulder on peak p (arrow, Fig. 4) contains the protein A described by Subramanian et al. (36).
70 S Proteins-The fifty-three 70 S proteins were resolved into at least 31 peaks by our improved RP-HPLC procedure (Fig. 5). The highly reproducible nature of the chromatograms enabled peak assignment to be made largely on the basis of profiles. In certain cases, confirmations were obtained using one-dimensional PAGE analysis.
A noteworthy point is that the peak eluting at 135 min during the final steep gradient portion of the chromatogram was found by subsequent two-dimensional PAGE analysis to contain several 50 S proteins, in the relative amounts L1, L9 > L4 > L3, L5, L10 > L15. When this peak was lyophilized, redissolved in buffer B, and rechromatographed, greater than 80% of the absorbance at 214 nm eluted in its original position. Thus, it is likely that this peak contains proteins which are denatured and/or aggregated. It should be emphasized that the amount of any of the proteins found in this peak is small (<lo% even for proteins L1 and L9) compared to the amount eluted in the major peak of that protein as defined in Table I. Further studies designed to understand the basis for this apparent partitioning of some ribosomal proteins into two regions of the chromatogram are underway and will be reported elsewhere.
Resolution    Table 11. The arrow corresponds  Table I. umn, and eluted with a gradient identical with that in Fig. 1 (dashed line). The upper and lower truces are absorbances at 214 nm and 280 nm, respectively. The upper case letters refer to the TP50 proteins in Table I   an unresolved peak to the column and re-elution with a shallower gradient. For example, peak M from Fig. 1, containing proteins L14, L19, and L30, was resolved into three peaks ( Fig. 6) when reapplied to the column at 35% solvent B and eluted with a 60-min linear gradient to 40% solvent B. The proteins were identified by one-dimensional PAGE. In cases where elution with a shallower gradient does not give full resolution, a column having a functional group with a polarity different than C18-silica (e.g. NH2-, CN-, or phenyl-silica), might be employed in the second step.

See Footnote a in
Functionality of HPLC-purified Proteins-The results already presented make clear the utility of the RP-HPLC technique for analysis of ribosomal proteins. A separate question concerns the utility of this method for preparation of proteins which could be used in functional studies. The sucrose density gradient results presented in Fig. 7 provide a clear demonstration that HPLC-purified proteins are capable of being reconstituted into 30 S subunits on recombination with 16 S RNA. The functionality of these subunits was tested  by their ability to bind Phe. tRNAPh" in a poly(U)-dependent manner. The activity of reconstituted 30 S subunits prepared from proteins extracted using either the acetic acid (16)or LiC1-urea (17) procedures, and then subjected to HPLC treat-