Forced extraction of targeted components from complex macromolecular assemblies

Abstract

When individual protein components of supramolecular complexes are required for assembly, determining whether they play additional structural or functional roles can be difficult. Removing a protein from the complex after assembly can circumvent this problem. Here, we show that an AAA+ unfoldase/protease can extract an essential assembly protein from the ribosome. Specifically, Mg²⁺ depletion allowed ClpXP to remove an ssrA-tagged variant of ribosomal protein L22 from the 50S subunit of E. coli ribosomes without disrupting either the structural integrity or hydrodynamic properties of the modified particle. Forced extraction using AAA+ enzymes and targeted component proteins should be broadly applicable to the study of macromolecular complexes.

It can be difficult to study the roles of individual components in macromolecular machines, especially when a component of interest is required for machine assembly. If some type of “molecular tweezers” could be used to pull a specific component out of the fully assembled complex, then the role of that molecule in the structural integrity and/or function of the machine might be possible. Here, we test this idea using the AAA+ unfoldase ClpX as tweezers and the Escherichia coli ribosome as the target machine.

ClpX is a hexameric AAA+ unfoldase, which uses the energy of ATP hydrolysis to unfold proteins by pulling on a terminal peptide sequence (1). ClpX has been shown to function as a disassembly chaperone by extracting a subunit of the MuA transposase from stable complexes of MuA tetramers with recombined DNA (2). ClpX also partners with the ClpP peptidase to form the energy-dependent ClpXP protease that degrades substrate proteins into small peptides. Any protein can be targeted to ClpX or ClpXP by the addition of the 11-residue C-terminal ssrA tag (3). We reasoned that appending an ssrA tag to a ribosomal protein might allow the selective removal of the fusion protein from purified ribosomes. In applying this strategy, a series of benchmark questions arise. Can the tagged component be recognized by ClpX and forcibly extracted? If the tagged molecule is extracted, does the remaining complex dissociate or remain intact? Finally, if the first two benchmarks are met, can the contribution of the extracted component to overall machine function be established?

E. coli ribosomes contain >50 proteins, the majority of which are required for cell viability (4). The assembly of each ribosomal subunit (i.e., 30S and 50S) entails the ordered addition of proteins to an RNA scaffold with the help of assembly chaperones (4, 5). For the large 50S subunit, a handful of component proteins play critical roles in assembly. Specifically, when combined with 23S rRNA, these primary assembly proteins form a precursor “core” particle, which serves as a scaffold for subsequent binding of other 50S proteins (6). L22 is a 12kDa, highly basic protein that is required for the assembly of the 50S subunit of bacterial ribosomes (7). L22 contacts all six domains of 23S rRNA and remains stably associated with inactive ribonucleoprotein particles after high-salt extraction (4, 8). To study the potential roles of L22 in ribosome integrity and activity, we purified ribosomes from an E. coli strain expressing an ssrA-tagged form of L22. We show that ClpXP can extract tagged L22 from Mg²⁺-depleted ribosomes, resulting in 50S subunits that remain intact but support little if any translational activity. However, the conditions required for forced extraction of tagged L22 also inactivated control ribosomes, precluding conclusions about the potential roles of L22 in translation.

Results

A Genetic System for L22 Modification.

To modify ribosomal protein L22, we developed a genetic system that allowed mutant variants to be expressed in cells as the only form of L22. E. coli L22 is encoded in the single-copy S10 operon with 10 other ribosomal proteins. We cloned this operon with its promoter and regulatory elements into a plasmid (pS10) and then replaced the chromosomal S10 locus with a kan^R-cassette (Fig. 1A). Cells supported only by pS10 doubled in 34 min, compared with 32 min for the parental strain. Hence, plasmid-encoded S10 genes support normal levels of ribosome function. Next, we modified the pS10 gene encoding L22 to encode H₆-L22-titin-ssrA, which contains an N-terminal His₆ tag and a C-terminal fusion to an ssrA-tagged variant of the I27 domain of human titin (9). In clpX- cells lacking the chromosomal S10 operon, pS10 encoding H₆-L22-titin-ssrA grew with a 35-min doubling time, suggesting that modification of L22 did not impair its function to any substantial degree. Southern blotting (Fig. 1B) and PCR (data not shown) verified that the gene encoding H₆-L22-titin-ssrA was the only version present in this strain. SDS/PAGE of purified ribosomes from clpX- cells containing only H₆-L22-titin-ssrA revealed a band of the size expected for H₆-L22-titin-ssrA (Fig. 1C). This protein cross-reacted with both anti-H₆ and anti-ssrA antibodies, and MALDI-TOF mass spectrometry revealed a single mass consistent with that calculated from the sequence of H₆-L22-titin-ssrA that lacks the N-terminal methionine (data not shown).

Fig. 1.

A strain using H₆-L22-titin-ssrA (A) Strains expressing wild-type L22 or H₆-L22-titin-ssrA from chromosomal or plasmid-borne S10 operons. (B) Southern blot probed with an oligonucleotide complementary to the antisense strand of the L22 gene (rplV) reveals the wild-type 508 bp EcoRI fragment in SM1090 and SM1110, and the mutant 874 bp EcoRI fragment in SM1145. (C) Coomassie-stained SDS gel shows that ribosomes purified from SM1145 contain H₆-L22-titin-ssrA. Wild-type L22 (SM1090) is not resolved from other ribosomal proteins.

Ribosomal H₆-L22-titin-ssrA Is Refractory to Complete ClpXP Degradation.

In an effort to extract and degrade the tagged L22 protein, we incubated H₆-L22-titin-ssrA ribosomes with ClpXP and ATP. At selected times, samples were withdrawn, diluted into 6 M GuHCl containing an unrelated H₆-protein as a “recovery control,” and H₆-tagged proteins were purified using Ni⁺⁺-affinity chromatography and analyzed by SDS/PAGE. ClpXP degradation generated a truncated form of H₆-L22-titin-ssrA, which retained the H₆ tag and, thus, was an N-terminal fragment (Fig. 2A). Moreover, the total amount of full-length H₆-L22-titin-ssrA plus this fragment was essentially constant over the time course of degradation (Fig. 2B), indicating that little, if any, H₆-L22-titin-ssrA was degraded to species other than the major truncation product. ClpXP completely degraded ribosome-free L22 protein with an ssrA tag (data not shown), indicating that ribosome association impedes complete degradation of H₆-L22-titin-ssrA.

Fig. 2.

ClpXP truncation of H₆-L22-titin-ssrA in ribosomes (A) Coomassie-stained SDS gel of H₆-tagged proteins after incubation of H₆-L22-titin-ssrA ribosomes (3 μM) with 0.5 μM ClpX₆, 1 μM ClpP₁₄, and 2 mM ATP. Positions are marked for full-length H₆-L22-titin-ssrA, a truncated product, and a His₆-tagged protein added as a “recovery control” before Ni⁺⁺-NTA purification. Left lane, protein standards and molecular weights (kDa). (B) Kinetics of H₆-L22-titin-ssrA loss, truncated-product appearance, and the sum of both species after quantification of bands from the panel-A gel and normalization for recovery and length. (C) Model for truncation product formation. ClpXP stalls when it encounters the ribosome surface during degradation of H₆-L22-titin-ssrA liberating a truncation product. Because each ClpXP enzyme truncates approximately five H6-L22-titin-ssrA molecules, the protease must be released from the stalled complex.

The truncated protein showed MALDI-TOF masses consistent with molecules containing all of the H₆-L22 portion of the fusion protein plus 38 or 40 residues of the titin domain. These titin tails are long enough to span the distance from the entry of the ClpX pore to the active sites in the proteolytic chamber of ClpP and are similar in length to tails found in other multidomain proteins that are only partially digested by ClpXP (10–12). Hence, ClpXP degrades the C-terminal portion of H₆-L22-titin-ssrA but appears to stall when it encounters L22 embedded in the ribosome (Fig. 2C). The same partial degradation of tagged L22 was also observed using isolated 50S subunits as a substrate, more ClpXP, longer incubations, higher temperature, and with a variety of destabilizing agents (e.g., 1 M NH₄Cl, 2 M urea, 0.5 M LiCl, 10% ethanol, 25% dimethyl sulfoxide, and 25% formamide) (data not shown). These results suggest that ClpXP, under many conditions, is unable to exert enough force to break the extensive contacts between L22 and 23S ribosomal RNA.

Mg²⁺ Depletion Allows H₆-L22-titin-ssrA Extraction and Degradation.

Mg²⁺ is important for ribosome stability (13–15). In an effort to destabilize ribosomes and allow L22 extraction, we incubated H₆-L22-titin-ssrA ribosomes with ClpXP in buffer containing 5.4 mM Mg²⁺ and increasing amounts of EDTA, a strong Mg²⁺ chelator (16). At EDTA concentrations up to 3 mM, the full-length H₆-L22-titin-ssrA protein was still converted to the truncated form by ClpXP (Fig. 3), indicating that ClpXP could not dislodge embedded L22. Strikingly, however, ClpXP degraded ≈60% of H₆-L22-titin-ssrA to completion and 40% to the truncated fragment when 3.5 mM EDTA was present and ≈95% to completion and 5% to the fragment when 4 mM EDTA was present. Thus, reducing the free Mg²⁺ levels during the degradation reaction weakened the ribosome sufficiently to allow ClpXP to extract and degrade L22 from the large subunit.

Fig. 3.

Mg²⁺ depletion allows complete H₆-L22-titin-ssrA degradation by ClpXP Coomassie-stained SDS gel of H₆-tagged proteins recovered after ClpXP degradation of H₆-L22-titin-ssrA ribosomes in the presence of EDTA. Reaction conditions were the same as in Fig. 2A, except 0.3 μM ribosome were used and degradation proceeded for 10 min. The H₆-L22-titin-ssrA and truncated-product bands were quantified as in Fig. 2B and shown as the percent L22 remaining relative to the “No ClpX” lane.

L22 Is Not Required for the Integrity of the 50S Subunit.

To determine whether L22 extraction resulted in a major disruption of the large subunit, we used sucrose-gradient centrifugation and SDS/PAGE to analyze 50S subunits after ClpXP extraction of ≈95% of H₆-L22-titin-ssrA in 4 mM EDTA. The sedimentation profile of 50S subunits treated with ClpXP in the absence of EDTA was indistinguishable from untreated subunits (Fig. 4A, top traces). Thus, ClpXP truncation of H₆-L22-titin-ssrA does not alter the sedimentation properties of the resulting particle. Subunits from the EDTA-treated sample lacking ClpXP migrated primarily at 50S with a shoulder of slower-sedimenting material (Fig. 4A, trace 3). Importantly, the sedimentation profile of the sample containing both ClpXP and EDTA was indistinguishable from the control sample lacking protease (Fig. 4A, bottom traces). Thus, forced extraction of L22 from the large ribosomal subunit causes little, if any, change in the hydrodynamic properties of the resulting particle.

Fig. 4.

Sedimentation of 50S subunits after H₆-L22-titin-ssrA extraction (A) Purified 50S subunits were treated ± EDTA and ± ClpX and sedimented through sucrose gradients. All samples contained ClpP. (B) Peak fractions of each sample were pooled and analyzed by SDS/PAGE and Coomassie staining. In lane 2, ClpX and ClpP, which comigrates with the L22 truncation product, appear to copurify with 50S subunits.

Proteins from the peak sedimentation fractions were analyzed by SDS/PAGE. Subunits treated with ClpXP but no EDTA had a protein content similar to mock-treated particles, except for conversion of full-length H₆-L22-titin-ssrA to the truncated form (Fig. 4B, lanes 1 and 2). Some ClpX sedimented with the 50S subunit, suggesting that it remained bound to the truncated protein. The protein content of subunits treated only with EDTA was similar to untreated particles (Fig. 4B, lanes 1 and 3), consistent with previous reports (14, 17). Significantly, except for the absence of H₆-L22-titin-ssrA, 50S subunits treated with EDTA and ClpXP had a protein content very similar to subunits treated only with EDTA (Fig. 4B, lanes 3 and 4). Hence, after ClpXP extraction of H₆-L22-titin-ssrA, the 50S subunit of the ribosome maintains an essentially unaltered composition and shape.

Translation Activity.

Ribosomes from the ClpXP extraction reactions shown in Fig. 3 or from mock reactions were assayed for their ability to translate mRNA. Ribosomes treated with >2.5 mM EDTA lost translation activity, but ClpXP treatment caused little additional loss of activity (Fig. 5A), even under conditions where H₆-L22-titin-ssrA was almost entirely extracted/degraded. When intact ribosomes were inactivated by EDTA, translation activity could be restored by addition of untreated 50S subunits but not by 30S subunits (data not shown). Similarly, when isolated 50S subunits were inactivated with EDTA, addition of untreated 30S subunits did not restore activity. Thus, Mg²⁺ depletion allows ClpXP to extract H₆-L22-titin-ssrA from the 50S subunit of ribosomes but also damages this subunit, causing loss of translation activity.

Fig. 5.

Translation activity of ribosomes after H₆-L22-titin-ssrA extraction (A) Ribosomes from ClpXP (+) or mock (−) reactions in Fig. 3 were supplemented with Mg²⁺ to compensate for EDTA, heated, and added to complete a coupled system in which an mRNA encoding SspB and β-lactamase was transcribed by T7 RNA polymerase. After 1 h of transcription/translation in the presence of ¹⁴C-labeled amino acids, samples were electrophoresed on an SDS gel, which was fixed, dried, and exposed to film. Intensities are shown under each band as a percentage of the untreated control activity in the left lane. (B) Ribosomes containing His₆-L22-titin-ssrA were subjected to degradation by ClpXP ± 3.5 mM EDTA and anti-H₆ antibodies were then used to immunodeplete ribosomes containing remaining full-length or truncated His₆-L22-titin-ssrA. H₆-tagged proteins remaining in the supernatant after immunoprecipitation or mock treatment without antibody were purified and analyzed by SDS/PAGE to determine the percent of ribosomes containing L22 (Upper). The supernatants were also tested for mRNA translation activity (Lower).

Because EDTA treatment reduced translation activity roughly in concert with successful L22 extraction, we considered two models: (i) L22 is only extracted/degraded from a subpopulation of ribosomes inactivated by EDTA treatment; or (ii) EDTA both inactivates ribosomes and allows L22 extraction, but these events occur independently, resulting in L22-depleted ribosomes that retain translation activity. To test these possibilities, we used anti-H₆ antibodies to deplete ribosomes still containing either full-length or truncated His₆-L22-titin-ssrA after ClpXP extraction or a mock reaction. The immunodepletion procedure neither removed nor altered the activity of ribosomes lacking H₆-tagged L22 (data not shown). Without EDTA, ClpXP converted all H₆-L22-titin-ssrA to the truncated form, and immunodepletion removed ≈85% of the resulting ribosomes containing the truncated substrate (Fig. 5B, lanes 1 and 2). In the presence of 3.5 mM EDTA, ClpXP extracted and completely degraded ≈80% of the H₆-L22-titin-ssrA, and immunodepletion removed virtually all ribosomes still containing H₆-L22 (Fig. 5B, lanes 3 and 4). Assays of mRNA and polyU translation revealed that activity correlated with ribosomes that still contained L22 (Fig. 5B; data not shown). Importantly, immunodepletion removed almost all activity from the sample treated with EDTA and ClpXP, even though ≈80% of ribosomes in this sample had been stripped of H₆-L22 and were insensitive to immunodepletion. We conclude that the translation activity remaining after EDTA and ClpXP treatment arises from ribosomes that had not been weakened sufficiently to allow complete extraction/degradation of tagged L22 by ClpXP.

Structural Footprinting.

To characterize the effects on the 50S subunit of EDTA treatment and/or L22 extraction in greater detail, we performed structural footprinting using dimethyl sulfate (DMS). In one set of reactions, we evaluated the reactivity of bases which form part of the peptidyltransferase center (PTC) (18). The DMS reactivity of residues A2439 and A2248–A2453 increased substantially and equivalently in samples treated with 4 mM EDTA with or without ClpX extraction (Fig. 6A, right two lanes). This region includes residue A2451, an essential component of the PTC, which becomes more DMS reactive in inactive conformations of the large subunit (19, 20). Therefore, EDTA treatment irreversibly perturbs the structure of the PTC, explaining the loss of translation activity, but removal of L22 caused no additional alterations of base reactivity.

Fig. 6.

Structural footprinting After treatment with EDTA and/or ClpXP, ribosomes were treated with DMS, and RNA was purified and used as template in primer-extension reactions. (A) An olionucleotide complementary to bases 2458–2485 of 23S rRNA was used to prime extension reactions that extend through part of the peptidyltransferase center. A control reaction using unmodified RNA is shown in the left lane. Base positions corresponding to prominent extension products are labeled. (B) Extension products from reactions using a primer complementary to 23S rRNA residues 1620–1650. A sequencing reaction containing dideoxy thymidine was used to assign base positions far from the primer (lane ddT). Bases corresponding to selected extension products are labeled.

We also evaluated the DMS reactivity of bases in a section of domain III that makes some contacts with L22 (21). EDTA treatment, with or without ClpXP extraction of L22, caused increased reactivity of bases A1614–C1617, C1607, C1606, C1585, and A1570–A1572 (Fig. 6B). The remaining DMS modifications detectable in this experiment were similar in control samples without EDTA or ClpXP treatment, indicating, for the most part, that these regions of the 50S subunit remained properly folded. The DMS reactivity of residue A1614 appeared to increase somewhat more after EDTA and ClpX treatment, but otherwise the major changes seemed to be caused by EDTA alone.

Gesteland (14) originally showed that Mg²⁺ depletion using EDTA alters ribosomal subunits. We found that EDTA-dependent ClpXP extraction of tagged L22 and loss of translation activity were correlated, but differed in their reversibility. Although we were unable to identify conditions that restored translation activity after EDTA treatment, addition of compensating Mg²⁺ to EDTA-treated ribosomes completely prevented ClpXP extraction of tagged L22 (data not shown). Thus, different conformational changes in the ribosome must be responsible for translational inactivation and for L22 resistance to forced extraction.

Discussion

Ribosomal protein L22 is required for assembly of the 50S subunit of bacterial ribosomes and is a genetically essential protein (6, 22, 23). We generated an E. coli strain that exclusively expresses an ssrA-tagged form of L22 and used ribosomes purified from this strain as substrates for ClpXP degradation in vitro. ClpXP largely extracted and degraded the tagged L22 variant when Mg²⁺ was depleted but only generated a truncated form of L22 that remained embedded in the ribosome under other conditions tested. Importantly, removal of L22 from the large subunit did not cause dissociation of other ribosomal proteins. Moreover, although Mg²⁺ depletion perturbed the DMS reactivity of bases in some regions of 23S RNA domains III and V, additional extraction of L22 did not cause significant additional changes in reactivity in these regions. Additional RNA residues outside of the examined regions that contact L22 directly would likely exhibit differential DMS reactivity upon L22 extraction as was seen for residue 1614.

Mg²⁺ depletion damages ribosomes as well as allowing ClpXP extraction of ssrA-tagged L22 from ribosomes. Indeed, studies have shown that EDTA treatment can change the conformation of both ribosomal subunits, albeit without causing protein dissociation (14, 17). In our studies, ribosome populations from which ≈95% of the L22 had been removed/degraded by ClpXP were only slightly less active in translation than mock-treated ribosomes with a full complement of L22. However, when we depleted ribosomes still containing L22, the translation activity of the remaining L22-free ribosomes was negligible. Moreover, in polyU and mRNA translation assays, EDTA treatment decreased translation activity at concentrations lower than those in which L22 extraction by ClpXP could be detected. These observations and EDTA-mediated increases in DMS reactivity of bases in domains III and V suggest that Mg²⁺ depletion caused the large subunit to assume an inactive conformation before L22 was extracted by ClpXP. Hence, we are currently unable to determine if L22 plays a role in translation. It remains to be determined whether L22 could be forcibly extracted from ribosomes under conditions that do not lead to irreversible inactivation of translation. In one attempt, we deleted a buried L22 loop that forms part of the exit channel but is not required for assembly or translation in vivo (24, 25). However, ribosomes containing H₆-L22-titin-ssrA with a complete deletion of this loop also resisted ClpXP extraction until EDTA inactivation occurred (data not shown).

Using degradation tags to mediate the extraction of specific proteins from assembled macromolecular complexes provides a new biochemical tool. Although the experiments presented here were designed to study a ribosomal protein, the technology described offers a conceptual framework that should be broadly applicable for studying the structural and functional roles of proteins that are components of many macromolecular machines.

Materials and Methods

Plasmid and Strain Construction.

Plasmid pS10 was constructed using PCR products containing the S10 operon amplified from E. coli strain MG1655, a ColA replication origin from pCOLADuet-2 (Novagen), and a tetracycline-resistance gene from pACYC184. DNA sequencing established that the cloned S10 operon had no mutations. Strain DY378 (D. Court, NIH) transformed with pS10 was temperature shifted to induce recombination enzymes (26), and a kanamycin-resistance cassette (27), containing 50-bp tails that were complementary to regions flanking the chromosomal S10 operon, was introduced by electroporation. Kan^R colonies (SM1041) were screened by PCR to verify the chromosomal S10 deletion.

The H₆-L22-titin-ssrA protein had MSDH₆LQ at the N terminus, TS and the V13P variant of the I27 domain of human titin (9) after the normal L22 C terminus, and a C-terminal QLHHRPAANDENYALAA (ssrA tag in italics). Strain SM1090 was generated from E. coli strain X90 (28) by serially deleting the clpX, clpA, and rna genes by targeted recombination followed by removal of kanamycin cassettes using FLP recombinase (27). Strain SM1145 (ΔS10::kan, clpX–, clpA–, rna–, pS10-H₆-L22-titin-ssrA) was generated by transforming SM1090 with pS10-H₆-L22-titin-ssrA and then P1 transducing the S10::kan marker from SM1041 to delete the chromosomal S10 operon. pS10-H₆-L22-titin-ssrA was stably maintained in SM1145 without antibiotic selection. The absence of the rplV gene encoding wild-type L22 in SM1145 was verified by PCR using primers complementary to upstream and downstream flanking regions, and by digestion of total DNA with EcoRI and Southern blotting with an oligonucleotide complementary to the antisense strand of rplV.

Ribosome Preparation.

Ribosomes were purified from cell lysates using a combination of differential centrifugation and hydrophobic interaction chromatography using methods adapted from a previous publication (29) as described in supporting information (SI) Text. Ribosomes were stored in degradation buffer (20 mM Hepes-Tris (pH 7.5), 100 mM K-glutamate, 6.1 mM Mg-OAc, 0.1 mM EDTA, 0.05% Tween-20, 14 mM 2-mercaptoethanol, and 10% glycerol). Details are in SI.

ClpXP Degradation.

We found that treatment of wild-type ribosomes with wild-type E. coli ClpXP reduced translation activity ≈2-fold (data not shown), suggesting damage to a ribosomal protein or factor necessary for full activity. This effect was largely eliminated when ClpX variants lacking the N-terminal domain were used. All experiments shown in this article were performed using purified Bacillus subtilis ClpX-ΔN and ClpP (gifts from Jim Butler, MIT), which were slightly more active than ΔN variants of E. coli ClpXP in extracting H₆-L22-titin-ssrA from ribosomes.

Degradation reactions were performed by mixing components (all in degradation buffer) with 1/10 volume of a Mg²⁺-free ATP-regeneration system (5.4 mM Mg²⁺ 5.4 final). Each reaction contained 0.5 μM ClpX₆, 1 μM ClpP₁₄, 2 mM ATP, 16 mM creatine phosphate, 5 μg/ml creatine phosphokinase (Sigma), 0.4 units/ml Superase-In (Ambion), and either 0.3 or 3 μM ribosomes. EDTA, when present, was added to a premixture lacking ribosomes and ClpX; ribosomes were then added, incubated for 5 min, and ClpX was added last to initiate degradation. After degradation, a portion of the reaction was diluted 2-fold in buffer containing sufficient Mg²⁺ to balance any EDTA and subsequently frozen. Before translation and DMS/primer extension assays, the samples were heated in a thermal cycler at 42 °C for 20 min, cooled to room temperature, 1/9 volume of 3 M ammonium acetate plus 100 mM Mg-OAc was added, and a final 90-min incubation at 50 °C was performed. This treatment, based on a protocol that improved in vitro assembly (30), increased the polyU translation activity of ribosomes damaged by Mg²⁺ depletion 2- to 3-fold and had no effect on normal ribosomes. Another portion of the degradation reaction (15 μl for reactions with 3 μM ribosomes; 140 μl for 0.3 μM ribosomes) was diluted into a 10-fold excess of 6 M GuHCl, 5 mM Mg-OAc, 2 mM imidazole, 50 mM Na-PO₄, and 10 mM Tris-Cl (pH 8); 1 μl of His₆-tagged λ cI N-domain protein (30 μM) was added to control for recovery in subsequent steps (31). Ni⁺⁺-NTA resin was added, the mixture was agitated for 30 min, and resin was collected by centrifugation, washed 3 times with guanidine buffer containing 12 mM imidazole and 3 times with 10 mM Tris-Cl (pH 8). After removal of the supernatant, 2 μl of a solution containing 1% SDS, 144 mM 2-mercaptoethanol, and 25 mM EDTA was mixed with approximately 10–20 μl resin slurry. Total protein was extracted using 3 × 100 μl washes of 50% acetonitrile and 0.1% trifluoroacetic acid, dried under vacuum, and resuspended in SDS loading buffer supplemented with 25 mM NaOH. The total sample was used for SDS/PAGE.

Sedimentation.

Samples (100 μl) were layered on 10–30% sucrose gradients containing 10 mM Mg-OAc prepared for an SW-41 rotor. After centrifugation for 4 h (41,000 rpm, 4 °C), the gradients were fractionated using a piston fractionator (BioComp) and the absorbance spectrum of each fraction was measured. Peak fractions were pooled, 75 μg of linear polyacrylamide was added as a carrier, and the entire sample was precipitated with 1.2 volumes of isopropanol. The precipitate was washed with ethanol, dried, and resuspended in loading buffer for SDS/PAGE.

Transcription Reactions.

Coupled transcription/translation was accomplished by adding ribosomes to ribosome-free translation extracts prepared from strain SM1200 (SM1090, S10::kan supported by pS10 encoding L22 with a FLAG epitope) and was based on published protocols (32–34). A template plasmid from which mRNAs for SspB and β-lactamase are transcribed from a T7 promoter was used to direct synthesis of ¹⁴C-labeled proteins (35) as described in SI Text.

Immunodepletion.

Degradation and heat recovery were performed as described above except that ribosomes contained a longer N-terminal epitope (MSDH₆GPTSLQ) and 2-mercaptoethanol was omitted. Aliquots of 240 μl were mixed with 35 μl buffer with or without monoclonal anti-His₆ antibody (Roche) exchanged into HT-6. After 30 min on ice, washed Pansorbin cells (Calbiochem) at 2× concentration (70 μl) were added and incubated on ice for another 60 min. The cell/antibody/ribosome complexes were pelleted by centrifugation and the supernatant was recovered. Two hundred-eighty microliters of each supernatant were used for Ni-NTA recovery of His₆-tagged proteins (described above) and the remainder was frozen at −80°C before translation reactions.

DMS Modification and Primer Extension.

Degradation reactions were exchanged into 50 mM K-cacodylate, 100 mM KCl, and 10 mM Mg-OAc, pH 7.2, before modification with DMS following a published protocol (36). After quenching the reaction, RNA was purified and used as templates for reverse transcription at 53°C using SuperScript III (Invitrogen) and a primer with a fluorescent label on its 5′ end. See SI Text for detailed methods.