DnaK-facilitated ribosome assembly in Escherichia coli revisited

INTRODUCTION

Extrinsic factors are likely to be involved in assembly of bacterial ribosome subunits (Alix 1993; for a recent extensive review of the 30S assembly, see Culver 2003). Such factors should be necessary to circumvent the heat-dependent conformational change of ribosomal precursor intermediates required during in vitro reconstitution of ribosomes (Nierhaus 1991; Williamson 2003). The chaperones DnaK/DnaJ/GrpE (Alix and Guerin 1993; Sbai and Alix 1998) and GroEL/GroES (El Hage et al. 2001) are prime candidates for such a function, but their role, direct or not, in ribosome assembly is still mysterious. A defective DnaK system does not affect ribosome assembly at 30°C but seriously disturbs assembly at temperatures above 35°C (El Hage et al. 2001). In vitro reconstitution of ribosomes at different temperatures in the presence or absence of chaperones offers a tool to explore the role of chaperones, and has been recently exploited (Maki et al. 2002). In these studies, the authors concluded that the DnaK chaperone system facilitates 30S ribosomal subunit assembly. In contrast to these results, we show here that addition of chaperones to in vitro reconstitution could not substitute the necessity of the heat-activating step, and that ribosomal subunits, rRNAs and ribosomal proteins purified from a mutant that contains a disrupted dnaK gene (ΔdnaK), are as thermostable and active as their dnaK⁺ counterparts in protein synthesis and in vitro reconstitution.

RESULTS AND DISCUSSION

In vitro reconstitution of 30S ribosomal subunits in the presence and absence of chaperones

We tested whether or not the presence of chaperones relieves the heat dependence of the 30S reconstitution process and/or accelerates the formation of active particles. To this end, the kinetics of total in vitro reconstitution of 30S ribosomal subunits at various temperatures (20°C, 30°C, and 40°C) were performed in the presence or absence of purified chaperones DnaK, DnaJ, GrpE, and ATP. The activity of the reconstituted 30S ribosomal subunits was assayed in a poly(Phe) synthesis system after complementing with native 50S subunits. The assay was performed at 20°C for 2 h, so that no further heat activation of the reconstituted particles would occur during poly(Phe) synthesis. Figure 1A demonstrates that reconstitution was efficient only at 40°C, in agreement with results reported by Traub and Nomura (1969), and that addition of chaperones and ATP to the reconstitution mixtures at all temperatures did not relieve the requirement for the heat activation of the reconstitution process or accelerate the formation of active particles. In fact, the rates in the presence and absence of chaperones were identical within the errors at each temperature and time point. rRNAs were not degraded during the reconstitution incubation, as indicated by the good activity of the particles formed and judged by SDS-polyacrylamide gel electrophoresis (data not shown), excluding the possibility of contaminating RNase(s) in the chaperone batches.

In an additional control experiment, we tested the activity of the chaperones used in the reconstitution assays concerning ATP hydrolysis. Figure 1B demonstrates that the DnaK system was highly active, with a turnover number of 1.8 min⁻¹, a value comparable to those reported by other groups (Ha et al. 1999).

To control for the possibility of endogenous DnaK present in the subunit preparation used for the preparation of TP30 and 16S rRNA, the subunits were analyzed by immunoblotting for the presence of DnaK. This was important because it had been reported that 50S subunits contain endogenous DnaK although 30S subunits did not (Vysokanov 1995). We nevertheless analyzed the DnaK content of the 30S subunits that were prepared from a dnaK⁺ Escherichia coli strain by immunoblotting. The first lane of Figure 2 shows the signal from 0.5 pmole of DnaK; in the second and third lanes, TP30 derived from 375 pmole of small subunits and TP50 from 160 pmole of large subunits, respectively, were applied. Scanning of the band intensities revealed that the 50S subunits contained only traces of DnaK (molar ratio DnaK:50S = 1:950), whereas DnaK was virtually absent in the 30S preparation (molar ratio 1:20,000; see legend to Fig. 2). The specific association of DnaK with the 50S subunits is reminiscent of the trigger-factor-binding specificity for the 50S subunit (Kramer et al. 2002a; Blaha et al. 2003). This subunit specificity of the binding is probably related to the cotranslational folding exerted by DnaK on nascent polypeptide chains emerging at a specific site (tunnel) on the cytoplasmic surface of the 50S ribosomal subunits, where the trigger factor was located recently (Kramer et al. 2002b; Blaha et al. 2003). The common or similar binding region is reflected by an overlapping function that is exemplified by the interesting interplay that exists between the trigger factor and DnaK. Deletion of either gene is tolerated by the cell, but deletion of both is lethal (Bukau et al. 2000).

With regard to the 30S assembly, the practically complete absence of contaminating endogenous DnaK on this subunit means that no traces of DnaK (if not added) were present during the reconstitution of this subunit. It follows that the DnaK chaperone family is not sufficient to facilitate reconstitution of 30S subunits at temperatures at or below 40°C, in striking contrast to the results of Culver and colleagues (Maki et al. 2002).

Thermostability of ribosomes and reconstitution competence of ribosomal components from a ΔdnaK mutant grown at 30°C

In the next experiments, we made use of an E. coli dnaK-null mutant (BB1553) that contains a disrupted gene (ΔdnaK52::cm^R), resulting in a lack of DnaK (Fig. 2) and also drastically reduced levels of DnaJ, probably because the transcription of the distal dnaJ gene is interrupted (Sell et al. 1990).

Wild-type E. coli cells containing normal amounts of RNase I are an unsuitable source for ribosomes that should be used for a reconstitution analysis, because this RNase binds tightly to the ribosome, copurifies with the protein fraction, and destroys the rRNA during the reconstitution incubation (Dohme and Nierhaus 1976). Therefore, we set out to transduce the rna-19 allele encoding an inactive RNase I into a dnaK-null mutant (and its wild-type isogene; Materials and Methods) to study the in vitro characteristics of their 30S and 50S ribosomal subunits at normal and high temperatures. The resulting strains, MC41.14 (rna-19, dnaK⁺) and BB11 (rna-19, ΔdnaK52::cm^R), were grown at 30°C, because BB11 is temperature sensitive at 37°C (Table 1A), and their 30S and 50S ribosomal subunits were isolated. Their activities in poly(U)-dependent [¹⁴C]-poly(Phe) synthesis were the same, irrespective of the dnaK status of the strain (Table 1B). Similarly, their ribosomal protein composition (TP50 and TP30 patterns) was identical (Fig. 3). It could be argued that the 30S and 50S subunits from the ΔdnaK strain appear perfectly normal, just because they were purified as 70S ribosomes and thus selected as normal ones. However, this is unlikely, because we have previously shown that the corresponding crude extract (S30) shows the same radioactive RNA profile after sucrose-gradient centrifugation as that seen with a wild-type strain, as long as the ΔdnaK strain was grown at temperatures not higher than 30°C (Fig. 1F of El Hage et al. 2001). In other words, no accumulation of assembly intermediates was observed, because of the absence of DnaK. Our findings further agree with the previous observation that no noticeable difference in protein synthesis levels exists between dnaK⁺ and ΔdnaK52 cells grown at 30°C, except a very low fraction (0.5%–1%) of total proteins aggregated in ΔdnaK52 cells (Tomoyasu et al. 2001).

To check whether the ribosomes formed in the absence of DnaK/DnaJ were as thermostable as their counterparts from a dnaK⁺ strain, 70S ribosomes from both strains ΔdnaK and dnaK⁺ were subjected to a severe thermal injury (up to 70°C) for 10 min in high salt conditions (1 mM MgCl₂, 200 mM NH₄Cl), before subjecting them to sucrose-gradient centrifugation under the same ionic conditions. Figure 4 shows that ribosomal subunits from both strains are equally stable until 50°C (Fig. 4A,B), but that thermodestruction occurs at 60°C for 30S subunits (but not for 50S subunits, Fig. 4C,D) and at 70°C for both subunits (Fig. 4E,F), in agreement with studies on wild-type ribosomes by Tal et al. (1977). Ribosomes of the ΔdnaK strain are as thermostable as those from the wild-type strain. Once assembled at 30°C, ribosomes do not need DnaK for thermoprotection, neither in vivo (Alix and Guerin 1993) nor in vitro (Fig. 4).

In the next experiment, the reconstitution competence of the ribosomal proteins and rRNAs components from the ΔdnaK strain BB11 was tested. 16SrRNAs, (23S + 5S) rRNAs, TP30, and TP50 were extracted from the ribosomal subunits of both dnaK⁺ and ΔdnaK strains. Homologous reconstitutions (ribosomal proteins and rRNAs from the same strain) and heterologous reconstitutions (ribosomal proteins from one strain and rRNAs from the other, and vice versa) were performed using both types of subunits (50S and 30S). The results shown in Table 1, C and D, reveal that 16S rRNA, (23S + 5S) rRNA, TP50, and TP30 from a ΔdnaK52::cm^R strain were as active as their wild-type counterparts, in all different possible combinations. Therefore, we conclude that DnaK is not required for biosynthesis of ribosomal subunits, at least at 30°C, nor for the capabilities of ribosomal components to reconstitute active ribosomal subunits.

Wireman and Sypherd (1975) detected disparity between the physical formation of 30S particles and the acquisition of activity when the reconstitution is performed with mature 16S rRNA: formation of 30S particles occurs at temperatures 15°C lower than the appearance of active subunits. Both features were reported to be improved by the DnaK system (Maki et al. 2002), that is, reconstituted particles became more homogeneous as shown with sucrose-gradient profiles, and tRNA binding occurred in the absence of a heat-activation step. However, tRNA binding was used as the sole functional assay of the reconstituted particles, and the conditions of the applied test include an incubation at 37°C for 15 min (Culver and Noller 1999). Because this incubation could easily provide the heat-activation energy of 38 kcal/mole required for in vitro reconstitution (Traub and Nomura 1969), it raises doubt as to whether the DnaK system does actively stimulate the formation of active particles.

Overexpression of ribosomal protein S4 partially rescued the thermosensitive phenotype of a dnaK756-ts mutant and showed more wild-type profiles of ribosome patterns in sucrose-gradient analyses, and these observations were thought to “suggest that DnaK can functionally interact with ribosomal components in vivo and that this interaction has an effect on ribosome assembly” (Maki et al. 2002). We see an alternative explanation based on the following findings: (1) Binding of the primary (such as S4) and secondary ribosomal proteins to the nascent 16S rRNA is DnaK-independent over a wide range of temperatures tested (El Hage et al. 2001). (2) S4 is a translational autorepressor such that overexpression will decrease expression of other ribosomal proteins in the same operon (Zengel and Lindahl 1994) and also will stimulate rRNA synthesis (Takebe et al. 1985; Torres et al. 2001). It follows that overexpression of S4 does not foster the assembly step from precursors to active particles, but rather has the opposite effect, that is, to increase the amount of precursor particles. The observed rescue of the dnaK756-ts phenotype by overexpression thus asks for another explanation. One possibility is a titration of the misfolded or “sticky” DnaK756 chaperone by S4, which otherwise might sequester one or several cellular components at high temperatures. In line with this interpretation is the observation that the ribosome assembly defect of a dnaK756-ts mutant could not be cured by a roughly equimolar expression of wild-type DnaK (Sbai and Alix 1998).

In any case, the results presented here do not support a role for DnaK during the in vitro formation of functional 30S subunits. Our observations are in agreement with in vivo data demonstrating that biogenesis of the 30S and 50S ribosomal subunits is unaffected in a strain lacking DnaK and grown at 30°C, but is partially and totally defective at 37°C and 42°C, respectively (El Hage et al. 2001). The DnaK chaperone system is therefore necessary for in vivo ribosome assembly at high temperatures, but not at 30°C. Just the opposite would be expected, if DnaK would help to bypass the heat-activation step, namely, a requirement at low temperatures such as 30°C or less. The use of a more sophisticated in vitro reconstitution system, starting from nascent rather than mature ribosomal components, that is, in vitro rRNA transcripts (Semrad and Green 2002), will perhaps be necessary to assign a defined role to extrinsic factors in ribosome assembly.

MATERIALS AND METHODS

Strain BB11 (ΔdnaK52) is chloramphenicol-resistant (cm^R, growth in LB medium in the presence of 25 μg/mL chloramphenicol) and temperature-sensitive (ts, does not grow at 37°C), whereas strain MC41.14 (dnaK⁺) is sensitive to chloramphenicol and thermoresistant (at 30°C–44°C).

Preparations of crude 70S ribosomes and 50S and 30S ribosomal subunits followed Bommer et al. (1996). The ribosomes were derived from the strain CAN20-12E (dnaK⁺; lacking RNases BN, I, II, and D; Deutscher et al. 1984; Zaniewski et al. 1984) unless otherwise specified. rRNA polyacrylamide-SDS gel electrophoresis, two-dimensional polyacrylamide gel electrophoresis of ribosomal proteins, and poly(U)-directed [¹⁴C]-poly(Phe) synthesis were carried out as described (Nierhaus 1990). The presence or absence of RNase I encoded by the rna gene was checked by the methyl-green procedure (Wright 1971). E. coli chaperones DnaK, DnaJ, and GrpE were purchased from StressGen Biotechnologies Corp. The ATPase activity of these chaperones was monitored by the hydrolysis of [γ-³²P]ATP according to Seals et al. (1978) with the modifications described by Benaroudj et al. (1994).

The rna-19 allele encoding an inactive form of RNase I and present in strain D10 (Pfennig and Flower 2001) was introduced into strains MC4100 (dnaK⁺) and BB1553 (MC4100 ΔdnaK52::cm^R, sidB1 = mutant rpoH; Bukau and Walker 1990) in two steps. First, the nearby tet^R genetic marker zba-601::Tn10 was transduced from the wild-type strain CAG12149 (Singer et al. 1989) into strain D10 with phage P1, selecting for tetracycline resistance, and a tet^R, rna-19 P1 transductant was chosen and named D10TET. Second, the rna-19 allele was readily transduced from strain D10TET into MC4100 and BB1553, taking advantage of the tet^R marker adjacent to it, yielding strains MC41.14 (dnaK⁺, tet^R, rna-19) and BB11 (ΔdnaK52::cm^R, ts, tet^R, rna-19), respectively.

Total reconstitution experiments for both 30S and 50S subunits were performed as described (Nierhaus 1990). For reconstitution kinetics (Fig. 1A), 30S reconstitution mixtures were prepared in an ice bath, each containing 4 A₂₆₀ units of 16S rRNA and 5.6 equivalent units (=0.7 A₂₃₀ units) of TP30, both extracted from 30S subunits of the strain CAN20-12E (dnaK⁺) in a final volume of 300 μL, in the presence or absence of 1 mM ATP, 2 μg of DnaK, 1.4 μg of DnaJ, and 1.6 μg of GrpE. Negative controls without 16S rRNA, or TP30, or both, and a positive control with native 30S ribosomal subunits were included. The samples were then incubated at 20°C, 30°C, or 40°C; at the indicated times; aliquots of 30 μL were withdrawn and checked for activity in a poly(U)-directed [¹⁴C]-poly(Phe) synthesis assay containing an excess of native 50S ribosomal subunits at 20°C for 2 h.

Quantitative immunodetection of DnaK in native 50S and 30S ribosomal subunits was performed as follows. Protein samples were prepared by boiling aliquots of DnaK, 30S, or 50S ribosomal subunits with SDS gel-loading buffer, and subjected to an electrophoresis on SDS-polyacrylamide (8%) gels. Serial dilutions of the samples allowed quantification of the signals in the linear range as demonstrated before (Lopes-Ferreira and Alix 2002). Immunoblots were then prepared as described in Sambrook et al. (1989). The blots were probed with polyclonal rabbit antibody raised against E. coli HSP70 (DnaK; from Dako, S.A.), and reactive bands were visualized with [¹²⁵I]-labeled Protein A (Amersham). Protein concentrations in the bands were quantified using the PhosphorImager.

Sedimentation profiles of heat-treated 70S ribosomes

For sedimentation profiles, 5 A₂₆₀ units of crude 70S ribosomes prepared from strains MC41.14 (dnaK⁺) and BB11 (ΔdnaK52::cm^R) were diluted in the buffer H₂₀M₁N₂₀₀SH₄ (20 mM HEPES-KOH at pH 7.6 and 0°C, 1 mM MgCl₂, 200 mM NH₄Cl, 4 mM β-mercaptoethanol), which promotes dissociation of ribosomal subunits, and incubated at 0°C, 30°C, 40°C, 50°C, 60°C, or 70°C for 10 min. They were then loaded on 12-mL 10% → 30% linear sucrose gradients prepared in the same buffer. After centrifugation at 4°C in a Beckmann SW40 rotor for 24 h at 18,000 rpm, the optical density (A₂₆₀/mL) profile of each gradient was recorded.