Conservation of two distinct types of 100S ribosome in bacteria

Transcription

1 Conservation of two distinct types of ribosome in bacteria Masami Ueta1, Chieko Wada1*, Takashi Daifuku2, Yoshihiko Sako2, Yoshitaka Bessho3, Aya Kitamura3, Ryosuke L. Ohniwa4, Kazuya Morikawa4, Hideji Yoshida5, Takayuki Kato6, Tomoko Miyata6, Keiichi Namba6 and Akira Wada1* 1 Yoshida Biological Laboratory, Yamashina, Kyoto , Japan Laboratory of Marine Microbiology, Graduate School of Agriculture, Kyoto University, Kyoto , Japan 3 RIKEN SPring-8 Center, Harima Institute, Kouto, Sayo, Hyogo , Japan 4 Institute of Basic Medical Sciences, Graduate School of Comprehensive Human Sciences, University of Tsukuba, Tennodai, Tsukuba, Ibaraki, Japan 5 Department of Physics, Osaka Medical College, Takatsuki, Osaka, Japan 6 Graduate School of Frontier Bioscience, University of Osaka, Suita, Osaka, Japan 2 In bacteria, ribosomes (consisting of and subunits) dimerize to form ribosomes, which were first discovered in Escherichia coli. Ribosome modulation factor (RMF) and hibernation promoting factor (HPF) mediate this dimerization in stationary phase. The ribosome is translationally inactive, but it dissociates into two translationally active ribosomes after transfer from starvation to fresh medium. Therefore, the ribosome is called the hibernating ribosome. The gene encoding RMF is found widely throughout the Gammaproteobacteria class, but is not present in any other bacteria. In this study, ribosome formation in six species of Gammaproteobacteria and eight species belonging to other bacterial classes was compared. There were several marked differences between the two groups: (i) Formation of ribosomes was mediated by RMF and short HPF in Gammaproteobacteria species, similar to E. coli, whereas it was mediated only by long HPF in the other bacterial species; (ii) RMF/short HPF-mediated ribosome formation occurred specifically in stationary phase, whereas long HPF-mediated ribosome formation occurred in all growth phases; and (iii) ribosomes formed by long HPF were much more stable than those formed by RMF and short HPF. Introduction During the stationary phase of Escherichia coli cell growth, ribosome modulation factor (RMF; 55 amino acids) is specifically expressed, binds to ribosomes and induces their dimerization to form the ribosome (Wada et al. 1990). The ribosome lacks translational activity, because RMF binds to the ribosome, binds close to the peptidyl transferase center and peptide exit tunnel, and ribosome do not contain trna and mrna (Wada et al. 1995; Yoshida et al. 2002, 2004; Kato et al. 2010). RMF is essential for ribosome formation Communicated by: Hiroji Aiba *Correspondence: awada@yoshidabio.co.jp or cwada@yoshidabio.co.jp 554 in vivo (Yamagishi et al. 1993) and in vitro (Wada et al. 1995). The expression of RMF is positively regulated by ppgpp (Izutsu et al. 2001). In an rmf E. coli deletion mutant, the ribosome fails to form and the mutant survives for a shorter length of time than the wild-type strain during stationary phase (Yamagishi et al. 1993). Hibernation promoting factor (HPF; 95 amino acids) promotes formation of ribosomes. In vivo studies using hpf (yhbh) deletion mutants indicate that RMF binds to the ribosome, which dimerizes to form unfolded particles of approximately 90S. Thereafter, the particles are converted from 90S to by the binding of HPF (Ueta et al. 2005). These findings were confirmed by in vitro studies using purified HPF and RMF proteins (Ueta et al. 2008). DOI: /gtc.12057

2 Two types of ribosome in bacteria YfiA (113 amino acids), a paralog of HPF, is expressed specifically in stationary phase (Maki et al. 2000). Although YfiA shares 40% sequence homology with HPF, it interferes with ribosome formation (Ueta et al. 2005). The structure of HPF has been determined by multi-dimensional NMR (Sato et al. 2009). HPF uses a babbba-fold structure (Fig. S2B in Supporting Information) and contains a ribosome-binding region similar to YfiA (Fig. S2B in Supporting Information; Rak et al. 2002; Ye et al. 2002), as expected from their sequence homology. In contrast to YfiA, HPF contains structural modifications that may be involved in coordinating its activity with RMF. These include the absence of a C-terminal extension, the stabilization of the a2 helix and the conservation of acidic residues that are exposed at the rim of the common basic patch. RMF inhibits the binding of aminoacyl-trna to ribosomes and translational activity in vitro (Wada et al. 1995). Although HPF inhibits in vitro translation activity in a poly (U)-dependent phenylalanine incorporation assay, it does not inhibit normal in vitro translation activity in an MS2 mrna-dependent leucine incorporation assay (Ueta et al. 2008), as ribosome-bound HPF is removed by IF3 (Ueta et al. 2008; Yoshida et al. 2009). When E. coli cells are transferred from starvation to fresh medium, RMF and HPF are immediately released from the ribosome, and the dissociated ribosomes participate in translation (Wada 1998; Maki et al. 2000). This rapid process is completed within 1 min of the transfer (Aiso et al. 2005), and cells start to proliferate within 6 min. In E. coli, the ribosome represents a resting form within the ribosome cycle, and the process of ribosome formation has been termed ribosomal hibernation (Yoshida et al. 2002). Interconversion between the inactive ribosome and the active ribosome is important for regulating translational responses to environmental changes. Cryo-electron microscopy has showed that the ribosome comprises two trna-free ribosomes in E. coli, which have twofold symmetry (Kato et al. 2010). Formation of the ribosome occurs through interaction between the small subunit proteins S2, S3 and S5, which appear to be critical for the dimerization process. It is suggested that conformational changes around the S2 S5 region may inhibit binding of mrna after the formation of the ribosome. Ortiz et al. (2010) showed by cryoelectron tomography that native E. coli cells contain ribosomes. Krokowski et al. (2011) observed 110S ribosomes composed of ribosome dimers in nutrient-starved rat cells. HPF and RMF are involved in the formation of the ribosome. Although RMF homologues are only present in the Gammaproteobacteria class, HPF homologues are found in almost all bacteria (Ueta et al. 2008). Phylogenetic analysis has classified HPF homologues into four types: long HPF, short HPF, YfiA and plant plastid HPF (Ueta et al. 2008). All these homologues possess a common conserved region in the N-terminal half (amino acids 1 95). The HPF homologues found in the Gammaproteobacteria and Betaproteobacteria classes are short, whereas bacteria in other classes have long HPF homologues that are approximately double the molecular weight of short HPF. Long HPF homologues include YvyD of Bacillus subtilis (Drzewiecki et al. 1998; Nanamiya et al. 2004; Tam et al. 2006) and plant plastid PSRP-1 (Johnson et al. 1990; Yamaguchi & Subramanian 2003; Sharma et al. 2007), which both bind ribosomes. Staphylococcus aureus has no RMF homologue and forms ribosomes using only its long HPF homologue (SaHPF) (Ueta et al. 2010). These results suggest that long HPF might mediate ribosome formation in many bacteria that lack RMF homologues. In this study, we compared ribosome formation in six species belonging to the Gammaproteobacteria class and eight species belonging to other bacterial classes. This study reports in detail how ribosome formation mediated by long HPF homologues occurs in Lactobacillus paracasei and Thermus thermophilus. ribosome formation was also observed in five other bacterial strains possessing a long HPF homologue. The findings of this study suggest that ribosomes form in the majority of bacteria with a long HPF homologue. We propose that bacteria have evolved two distinct mechanisms by which ribosomes form, one involving RMF and short HPF, and the other involving only long HPF. Results ribosome formation in Gammaproteobacteria and Betaproteobacteria species Gammaproteobacteria have rmf and hpf genes. The formation of ribosomes mediated by RMF and HPF has been reported in detail in E. coli (Wada et al. 1990, 2000; Yamagishi et al. 1993; Wada 1998; Ueta et al. 2005). Here, ribosome formation 555

3 M Ueta et al. was analyzed and compared in five Gammaproteobacteria species, namely Salmonella typhimurium, Proteus mirabilis, Serratia marcescens, Klebsiella pneumoniae and Pectobacterium carotovorum, and also in E. coli. E. coli, S. typhimurium and P. mirabilis were grown at 37 C, and Se. marcescens, K. pneumoniae and Pe. carotovorum were grown at 30 C in EP medium with shaking (Wada et al. 2000). Crude ribosome (CR) fractions were prepared from the various strains in exponential phase (turbidity of 50 Klett units) or stationary phase (after 2 or 3 days of culture). The ribosome profiles were analyzed by 5 20% sucrose density gradient (SDG) centrifugation (Fig. 1A). In E. coli, ribosomes are not detected during the majority of exponential phase, and low levels appear at the end of exponential phase. The level of ribosomes increases during stationary phase and is highest after 3 4 days of culture, after which they dissociate into ribosomes that are rapidly degraded (Wada 1998; Wada et al. 2000). Similarly, ribosomes were not detected in exponential phase but were detected in stationary phase in each of the five species tested. In S. typhimurium, P. mirabilis and Se. marcescens, ribosomes were most abundant after 2 4 days of culture, after which the level decreased (Fig. S1 in Supporting Information). RMF and HPF proteins were detected in each of the six species in stationary phase using radical-free and highly reducing two-dimensional polyacrylamide gel electrophoresis (RFHR 2-D PAGE) (Fig. 1B). The RMF and HPF protein sequences are highly conserved among the tested strains (Figs S2A,B in Supporting Information). ribosomes did not form in any growth phase in the Betaproteobacteria species Burkholderia multivorans (Fig. S3 in Supporting Information), which lacks RMF and has a short HPF homologue that differs from that found in Gammaproteobacteria species (Fig. S2B in Supporting Information). Time course of ribosome formation in Lactobacillus paracasei ribosome formation was examined in eight bacterial species that do not belong to the Gammaproteobacteria and Betaproteobacteria species, namely S. aureus, five Lactobacillaceae species, T. thermophilus and Synechocystis sp. PCC6803, which have neither genes for RMF nor short HPF but have a gene for long HPF. Among these species, S. aureus forms ribosomes. Unexpectedly, ribosomes are detected throughout exponential phase. The key protein that mediates ribosome formation is long HPF, the molecular weight of which is approximately twofold greater than that of short HPF (Ueta et al. 2010). We examined ribosome formation in L. paracasei. Bacteria were grown in modified MRS medium at 37 C (doubling time is approximately 2 h), and cell growth was measured by turbidity (Klett units) (Fig. 2A). Cells were harvested in exponential phase (after approximately 4 h of culture when turbidity reached 50 Klett units), the transition stage before entering stationary phase (hereafter referred to as transition phase ; after 8 h of culture) and stationary phase (after 16, 24 and 48 h of culture). CR fractions prepared from each sample were analyzed by 5 20% SDG centrifugation, and the ribosome profiles are shown in Fig. 2B. ribosomes were detected throughout both exponential phase and stationary phase (Fig. 2B). ribosomes accounted for 51% of the total ribosome fraction in exponential phase; this peaked during transition phase (68% after 8 h of culture) and then decreased during stationary phase (30% and 37% after 16 and 24 h of culture, respectively) (Fig. 2C). Similar results were obtained when cells were cultured in MRS medium (doubling time is approximately 90 min). These results show that ribosomes are present throughout the growth cycle in L. paracasei, similar to S. aureus (Fig. 2B). Analysis of ribosomal and ribosome-associated proteins in CR fractions from Lactobacillus paracasei A proteomics approach was used to search for key factors involved in ribosome formation in L. paracasei. Ribosomal proteins (r-proteins) and ribosome-associated proteins in the CR fractions of L. paracasei were analyzed using RFHR 2-D PAGE. CR, subunit and subunit fractions were prepared from bacteria in exponential phase. Proteins were extracted from each fraction using the acetic acid method and were analyzed by RFHR 2-D PAGE (Fig. 3A,B). After staining the gels with Coomassie Brilliant Blue (CBB), the spots were identified by matrix-assisted laser desorption/ionization time-offlight mass spectrometry (MALDI-TOF MS). The genome of L. paracasei contains 23 genes encoding small subunit r-proteins (S1 S21, two S14 proteins) and 33 genes encoding large-subunit r-proteins (L1 L36, including L33.2 and excluding L7, L8, L25 and L26) (Lactobacillus paracasei subsp. paracasei ATCC 25302, GOLD CARD: Gi03755). In RFHR 2-D PAGE analysis of the L. paracasei CR fraction,

4 Two types of ribosome in bacteria (A) E. coli S. typhimurium P. mirabilis P. carotovorum K. pneumoniae S. marcescens Exp. Stat. Absorbance 260 nm (B) E. coli S. typhimurium P. mirabilis P. carotovorum K. pneumoniae S. marcescens S10 S10 S10 S10 HPF S10 S10 HPF L30 RMF S21 HPF L30 RMF S21 HPF L30 RMF S21 HPF L30 RMF S21 L30 RMF S21 HPF? L30 RMF S21 Figure 1 Sucrose density gradient (SDG) centrifugation of ribosomes and the location of ribosome modulation factor (RMF) and hibernation promoting factor (HPF) on RFHR 2-D PAGE gels. Escherichia coli, Salmonella typhimurium, Proteus mirabilis, Serratia marcescens, Klebsiella pneumoniae and Pectobacterium carotovorum were cultured in EP medium at 30 C or 37 C, and cells were harvested at log phase (50 Klett units) or stationary phase (after 2 or 3 days of culture). Crude ribosome (CR) fractions from each sample were prepared and analyzed (150 pmol per sample) by 5 20% SDG centrifugation. The profiles of ribosomes are shown (A). CR fractions from bacteria in stationary phase were analyzed by RFHR 2-D PAGE and the location of RMF and HPF in each species is shown (B). Turbidity (Klett units) (A) Absorbance 260nm (B) 4 h 8 h 16 h 24 h 48 h (C) 100 in ribosome particles (%) Time (h) Time (h) Figure 2 ribosome formation in Lactobacillus paracasei. (A) Growth curve of L. paracasei cultured in modified MRS medium at 37 C. Cell growth was monitored in Klett units. Numbered arrows indicate exponential phase (1, 4 h), transition phase (2, 8 h) and stationary phase (3, 16 h; 4, 24 h; 5, 48 h). Turbidity at stages 1, 2, 3, 4 and 5 was 47, 184, 374, 448 and 467 Klett units, respectively. (B) Time course of ribosome formation in L. paracasei. Samples were taken at 4, 8, 16, 24 and 48 h (i.e. the time points indicated by arrows in A), and crude ribosome fractions were prepared and analyzed (150 pmol per sample) by 5 20% sucrose density gradient (SDG) centrifugation. Ribosome profiles are shown. Optical density at 260 nm is plotted on the y-axis, and the x-axis shows the various SDG fractions. (C) Percentage of ribosomes in total ribosomal particles. Mean standard deviation is plotted (n = 5 or more). The percentage of ribosomes was calculated from the SDG centrifugation patterns (Fig. 1B), and standard deviation was calculated using Excel. 557

5 M Ueta et al. (A) CR S1 S2 L7/L12 L5 L1 L10 L4 S4 L3 L2 S3 S5 L6 S7 1(LpHPF) L11 L13 S8 L9 L14 S6 L22 L16 S9 S12 L17L15 L18 S11 S13 L23 S10 2 L19 L24 L20 S19 S15+L31B S17 S16 S18 L27 S20 L27 L30 3 L32 L33.1 S14 L28 S21 L35 CR Acidic region S S2 6 L34 L7/L12 L10 1(LpHPF) S5 L5 (B) L7/L12 L10 L5 L1 L4 L3 L2 L13 L6 L11 L14 L15 L16 L22 L18 L17 L19 L23 L31B L24 L20 L27 L30 L27 L33.1 L32 L28 L35 S6 S2 S5 S4 S3 LpHPF S8 S7 S11 S9 S12 S10 S19 S13 S15 S17 S16 S20 S18 S14 S21 L34 Acidic region L7/L S1 10 S2 9 L10 1(LpHPF) S5 L5 Figure 3 Proteome analysis of Lactobacillus paracasei ribosomal fractions using RFHR 2-D PAGE. (A) RFHR 2-D PAGE analysis of crude ribosome (CR) fractions (left) from L. paracasei cells in exponential phase (4 h). CR fraction prepared from cells cultured for 4 h was analyzed by RFHR 2-D PAGE. The gel on the left shows all proteins in the CR. The gel on the right shows an enlargement of the region analyzed by electrophoresis of acidic proteins (denoted by the region inside the dashed line in the gel on the left). Each spot was identified by MALDI-TOF MS. Spots labeled S2 S21 and L1 L35 (except L8, L21, L25, L29 and L33.2) correspond to r-proteins of the and subunits, respectively. Spots labeled 1 10 correspond to ribosome-associated proteins. The results are summarized in Table 1 and Table S2 in Supporting Information. (B) RFHR 2-D PAGE analysis of the and subunits prepared from L. paracasei at transition phase. Samples of cells were taken after 8 h of culture, and a crude ribosome (CR) fraction was prepared. The and subunits were prepared from high salt-washed ribosomes of the CR fraction. Proteins were analyzed by RFHR 2-D PAGE, and each spot was identified by MALDI-TOF MS. The results are summarized in Table 1 and Table S2 in Supporting Information. of the small subunit r-proteins (all except for the two S14 proteins) and 29 of the large-subunit r-proteins (all except for L21, L29, L33.2 and L36) were identified (Fig. 3A,B, and Table S2 in Supporting Information). L9 was found in the CR fraction (Fig. 3A), but not in the ribosome fraction washed with high salt buffer (Table S2 in Supporting Information), which suggests that this protein binds to ribosome particles more weakly than the other r-proteins. Proteins were extracted from the CR fractions of cells at growth stages 1, 2, 3 and 4, corresponding to culture for 4, 8, 16 and 24 h, respectively (Fig. 2A), and analyzed by RFHR 2-D PAGE. In addition to the r-proteins, ten CR-associated proteins were 558

6 Two types of ribosome in bacteria Table 1 Summary of ribosome-binding proteins and/or proteins that co-sedimented with the crude ribosome (CR) fraction from Lactobacillus paracasei Spot density on RFHR 2-D gel Spot No. Accession No. Protein name 4 h CR 8 h CR 16 h CR 24 h CR 4 h HSR 8 h 8 h 1 gi Ribosome-associated inhibitor protein Y (LpHPF) +++ (++) +++ (+++) gi Bacterial nucleoid protein Hbs gi Conserved hypothetical protein gi Pyruvate dehydrogenase complex, E2 component, dihydrolipoamide acetyltransferase gi Myosin cross-reactive antigen gi Pyruvate dehydrogenase complex E3 component, dihydrolipoamide dehydrogenase 7 gi Trigger factor gi Enolase (2-phosphoglycerate dehydratase) gi Pyruvate dehydrogenase complex E component, alpha subunit 10 gi Pyruvate dehydrogenase complex E component, beta subunit The ribosome-associated proteins in CR fractions from cells in exponential phase (4 h), transition phase (8 h) and stationary phase (16 or 24 h) were identified on RFHR 2-D PAGE gels (see Fig. 2A,B). In addition to r-proteins, 10 other proteins were found in the CR fractions. The genome sequences of Lactobacillus paracasei subsp. paracasei ATCC 25302, L. paracasei subsp. paracasei 8700:2 and L. paracasei were used to identify these proteins. The NCBI accession number of each gene is shown. Data from high salt-washed ribosomes (HSR) from cells in exponential phase (4 h) and dissociated subunits ( and ) from cells in transition phase (8 h) are shown in the right column [+++, copy number >0.3; ++, copy number ; +, copy number <0.1;, spots not detected on the RFHR 2-D PAGE gel; and (++), (+++), copy number in ribosomes]. 559

7 M Ueta et al. detected and identified by MALDI-TOF MS (Fig. 3A,B, Table 1). Stained spot densities were measured using a densitometer, and the copy number of each protein was calculated by comparison with ribosomal proteins with a known copy number. Spots 2 10 were released from the ribosomal fraction by high salt washing, whereas spot 1 was not (Table 1). This persistence of spot 1 is consistent with the fact that ribosomes of L. paracasei do not dissociate into ribosome even with high salt washing (Fig. 8B). In addition, spot 1 was located in the subunit but not in the subunit of the and ribosome fractions (Fig. 3B, Table 1). Spot 1 was identified as the long HPF homologue (accession number; gi , ribosome-associated inhibitor protein Y) of L. paracasei and named LpHPF (accession number, AB744219). The molecular weight of LpHPF (185 amino acid, MW = Da, pi = 6.19) is almost double that of its E. coli homologue (HPF; 95 amino acid, MW = Da, pi = 6.50). As expected from the pi values, LpHPF was located to the lower right of L10 (168 amino acid, MW = Da, pi = 5.17) on the RFHR 2-D PAGE gels (Fig. 3A). These results strongly suggest that LpHPF is a key factor for ribosome formation in L. paracasei. Correlation between ribosome content and the copy number of LpHPF Lactobacillus paracasei cells were harvested after culture for 4, 8, 16, 24 and 48 h, and the ribosome content and LpHPF copy number were measured. The profiles of ribosome particles in CR fractions were analyzed by SDG centrifugation (Fig. 2B). The percentage of ribosomes in the total ribosome particles at different time points is shown in Fig. 2C. Proteins were extracted from each CR fraction and separated by RFHR 2-D PAGE, and the density of the stained spots was measured with a densitometer to calculate the protein copy number. LpHPF was detected during exponential phase (4 h), similar to SaHPF of S. aureus. The level of LpHPF increased in transition phase (8 h) and peaked during early stationary phase (16 h) (Fig. 4A). The increase in the level of LpHPF up to 8 h correlated with an increase in the ribosome content (Fig. 4A). The abundance of ribosomes decreased during stationary phase (16, 24 and 48 h), although the LpHPF copy number remained high (Fig. 4A). The profiles of ribosome particles in cell extract fractions were analyzed by SDG centrifugation, and the levels of LpHPF in each fraction (soluble, subunit, subunit, ribosome and ribosome) were monitored by Western blotting (Fig. S4 in Supporting Information). The ribosome profiles of cell extracts and CR fractions were similar in various growth phases (compare Fig. 2B with Fig. S4 in Supporting Information), and LpHPF was present in and ribosomes but not in the subunit, subunit or soluble fractions (Fig. S4 in Supporting Information). This finding shows that LpHPF binds ribosomes as well as ribosomes. LpHPF is present in ribosomes during exponential phase To investigate the localization of LpHPF in detail, CR fractions prepared from cultures grown for 4 or 8 h were separated into ribosome, ribosome and + subunit fractions using preparative SDG centrifugation. Proteins extracted from each fraction were analyzed by RFHR 2-D PAGE. In exponential phase (4 h), LpHPF was present in the ribosome fraction (0.25 copies) and the ribosome fraction (0.83 copies), indicating that LpHPF was mainly present in the ribosome fraction (Fig. 4C and Fig. S4 in Supporting Information). In transition phase (8 h), LpHPF was more evenly distributed between the and ribosome fractions (0.80 and 0.97 copies, respectively; Fig. 4B,C and Fig. S4 in Supporting Information) and was not detected in the + subunit fraction (Fig. 4B and Fig. S4 in Supporting Information). These results indicate that LpHPF preferentially is present in ribosomes during the exponential phase, whereas it binds both and ribosomes during transition phase (8 h). The data indicate that the molar ratio of LpHPF to ribosome in the ribosome is 1 : 1 (Fig. 4C) and that two LpHPF molecules are present in each ribosome. The position of LpHPF on RFHR 2-D PAGE gels was the same at each time point examined. This indicates that the net charge and molecular weight of LpHPF did not change, which suggests that this protein is neither modified nor cleaved during any growth phase. ribosome formation in vitro using purified LpHPF To examine the role of LpHPF in ribosome formation further, in vitro experiments were carried out in which purified LpHPF was mixed with ribosomes prepared from cells at exponential phase 560

8 Two types of ribosome in bacteria (A) Copy number of LpHPF in crude ribosome ( ) Time (h) in crude ribosome (%)( ) (B) Absorbance 260nm + S2 S2 L10 L10 L7/12 LpHPF L7/12 LpHPF S6 S2 L10 L7/12 LpHPF S6 (C) Copy number of LpHPF h 8 h LpHPF to ( fraction) LpHPF to ( fraction) Figure 4 Relationship between ribosome formation and LpHPF copy number at various phases of Lactobacillus paracasei growth. (A) Correlation between LpHPF copy number and ribosome formation during culture of Lactobacillus paracasei. Samples of cells were taken at various time points, and the proportion of ribosomes and the LpHPF copy number in the crude ribosome (CR) were measured. The left axis indicates the LpHPF copy number (opened circles). The right axis indicates the percentage of ribosomes (opened squares) (this data were taken from Fig. 2C). Mean standard deviation is plotted (from at least three samples and three gels). Standard deviation was calculated using Excel. (B) LpHPF is present in both the and ribosomal fractions, but is not present in the + subunit fraction. Lactobacillus paracasei cells were harvested at transition phase (after 8 h of culture). The CR fraction (150 pmol) was separated into + subunit, ribosome and ribosome fractions by 5 20% sucrose density gradient centrifugation (upper panel). The proteins in each fraction were extracted using the acetic acid method, and acidic proteins were analyzed by RFHR 2-D PAGE (lower panels show an enlargement of this region). Arrows indicate LpHPF spots, and the S2, S6, L7/12 and L10 spots are labeled. (C) One copy of LpHPF is present in each of the ribosomes in the ribosome. The copy numbers of LpHPF in the (white) and (black) ribosomal fractions from cells in exponential phase (4 h) and transition phase (8 h) are shown. (4 h), which contain a low copy number of LpHPF (0.25 copies) (Fig. 4C and Fig. S4 in Supporting Information). The mixture was incubated for 30 min at 37 C and was then centrifuged and analyzed by SDG centrifugation. The percentage of ribosomes increased from 19% to 36% after the addition of LpHPF (Fig. 5A). Thus, purified LpHPF converted free ribosomes into ribosomes in vitro. This suggests that LpHPF is a key factor for ribosome formation in L. paracasei. SaHPF is essential for the formation of ribosomes in Staphylococcus aureus Recently, long HPF (SaHPF) was shown to be a key protein for ribosome formation in S. aureus (Ueta et al. 2010). However, the Sahpf gene product has not been shown to be essential for ribosome formation. We studied the function of Sahpf using a deletion mutant (DSahpf), which could not generate ribosomes (Fig. 5B). Ribosomes isolated from DSahpf cells were mixed with purified SaHPF protein in vitro and incubated for 30 min at 37 C. The incubation mixture was centrifuged and analyzed by SDG centrifugation. ribosomes formed after the addition of SaHPF, whereas no ribosomes were detected in the absence of SaHPF (Fig. 5C). Furthermore, immunoblotting showed that SaHPF bound both and ribosomes (Fig. 5C). These results show that SaHPF is essential for the formation of ribosomes in S. aureus. TtHPF is essential for the formation of ribosomes in Thermus thermophilus Thermus thermophilus is remote from Lactobacillus and Staphylococcus in the phylogenetic tree (Battistuzzi & Hedges 2009), but it does contain a gene encoding a long HPF homologue. Thermus thermophilus HB8 was originally isolated from a thermal vent in a hot spring in Izu (Oshima & Imahori 1974) and is an extremely thermophilic, gram-negative and obligate aerobic bacterium that can grow at temperatures of up to 85 C. We examined ribosome formation in T. thermophilus HB8; the strain was cultured at 75 C, and the cells were harvested at exponential phase, transition phase and stationary phase (Fig. 6A). The profiles of ribosome particles in CR fractions 561

9 M Ueta et al. (A) Absorbance 260 nm (B) Absorbance 260 nm (C) Absorbance 260 nm 4h fraction Wild type + LpHPF ΔSahpf + Buf.I + SaHPF SaHPF SaHPF Figure 5 SaHPF and LpHPF mediate the formation of ribosomes. (A) Formation of ribosomes mediated by LpHPF in vitro. Purified LpHPF (molar radio of LpHPF to ribosome was 1.5 : 1 or buffer I was mixed with ribosomes obtained from cells in exponential phase (4 h of growth) and incubated for 30 min at 37 C. The mixture was precipitated and analyzed by 5 20% sucrose density gradient (SDG) centrifugation. The panels show the patterns when ribosomes were mixed with buffer I (left) and LpHPF (right). Each sample contained 150 pmol ribosomes. (B) In Staphylococcus aureus, the Sahpf gene product is essential for the formation of ribosomes in vivo. Cell extracts prepared from wild-type N315 and DSahpf cells in transition phase (OD 260 nm = 5 per sample) were analyzed by 5 20% SDG centrifugation. Ribosome profiles are shown. (C) SaHPF mediates the dimerization of ribosomes from DSahpf cells in vitro. High salt-washed dissociated ribosomes isolated from DSahpf cells were incubated with buffer I or SaHPF (molar radio of LpHPF to ribosome was 1.5 : 1) for 30 min at 37 C. The incubation mixtures were precipitated, and 150 pmol per sample analyzed by 5 20% SDG centrifugation. Protein samples were separated into 18 SDG fractions, precipitated with final 10% TCA, separated on 12% SDS-PAGE gels and analyzed by immunoblotting with an anti-sahpf antibody. The Western blots show that level of SaHPF protein in each of the corresponding fractions in the graphs. Purified SaHPF (10 ng) was loaded as a control on the right. OD, optical density. prepared from each growth phase were analyzed by SDG centrifugation. ribosomes were observed in all growth phases, similar to Lactobacillus and Staphylococcus (Fig. 6B,C, upper columns). The long HPF homologue was identified as accession number: gi , ribosomal subunit interface protein by proteome analysis of the CR fractions using RFHR 2-D PAGE and MALDI-TOF MS and named TtHPF (accession number; BR001021) (Fig. 6B lower column). The copy number of TtHPF was 0.97 (from two experiments) as determined by measuring the density of the stained TtHPF and L9 spots on RFHR 2-D PAGE gels (Fig. 6B (2) lower column). The copy number of TtHPF was reduced to 0.57, and the level of ribosomes decreased after high salt washing (Fig. 6B (3)). The function of TtHPF was studied using a deletion mutant (DTthpf: SP2011A), which could not generate ribosomes (Fig. 6C upper right column). Ribosomes prepared from DTthpf cells were mixed with purified TtHPF protein and incubated for 30 min at 60 C. The incubation mixture was centrifuged and analyzed by SDG centrifugation. ribosomes were detected after the addition of TtHPF, whereas they were not detected in the absence of TtHPF (Fig. 6C lower column). In vitro ribosome formation occurred most efficiently at 60 C (Fig. 6D) and when the molar ratio of TtHPF to ribosomes was 1.5 : 1 (Fig. 6E). The level of ribosomes formed decreased when more TtHPF protein was added (Fig. 6E). Formation of ribosomes in four Lactobacillus species and Synechocystis sp. PCC6803 In addition to L. paracasei and T. thermophilus, ribosomes were detected in CR fractions from all growth phases (exponential, transition and stationary phase) in Lactobacillus casei, Lactobacillus acidophilus, Lactobacillus delbruekii and Lactobacillus plantarum (Fig. S5 in Supporting Information). According to phylogenic analysis, L. paracasei and L. casei (L. casei group), L. acidophilus and L. delbruekii (L. delbruekii group), and L. plantarum (L. plantarum group) belong to different branches (Makarova et al. 2006). In addition, ribosomes were detected in Synechocystis sp. PCC6803, which has a long HPF that is similar to SaHPF, LpHPF and TtHPF (Fig. S5 in Supporting Information). The HPF proteins (short and long) in the examined bacterial species showed high sequence homology, particularly in the N-terminal and C-terminal domains (Fig. S6 in Supporting Information). Long HPF is also widely conserved among the various bacteria that 562

10 Two types of ribosome in bacteria belong to other groups aside from the Gammaproteobacteria and Betaproteobacteria classes. These results suggest that ribosomes are present in almost all bacteria that have a wild-type long HPF gene. Long HPF (SaHPF or LpHPF) can mediate dimerization of E. coli ribosomes In stationary phase of E. coli, formation of ribosomes is mediated by RMF and short HPF. We generated the quadruple deletion mutant YB1008 (W3110 DyfiA, Dhpf, Drmf, DompT::Km). This deletion mutant does not express YfiA, HPF or RMF, which are involved in ribosome formation in E. coli, and it cannot form ribosomes in stationary phase (Fig. 7). When the CR fraction prepared from bacteria in transition phase is washed with high salt buffer to remove ribosome-associated proteins and then treated with dissociation buffer, the ribosomes dissociate into the and subunits. Purified SaHPF or LpHPF protein was mixed with the treated E. coli ribosomes and incubated for 30 min at 37 C. The incubation mixture was centrifuged and analyzed by SDG centrifugation. Interestingly, ribosomes formed after the addition of SaHPF or LpHPF, and both proteins were detected in the and ribosome fractions (Fig. 7). These results show that both SaHPF and LpHPF can mediate the dimerization of E. coli ribosomes. ribosomes formed by long HPF are more stable than those formed by RMF and short HPF The stabilities of ribosomes formed by RMF and short HPF or by long HPF were compared using SDG centrifugation. When 150 pmol [equivalent to 50 ll of a sample with an optical density (OD) at 260 nm of 100] of CR fraction prepared from E. coli in stationary phase was analyzed by SDG centrifugation, the concentration of ribosomes in the ribosome peak was approximately M (OD at 260 nm of 1). The ribosome peak was stable; however, when the amount of CR fraction was reduced to 30 and 15 pmol, the level of ribosomes was reduced and the sedimentation coefficient was altered indicating a change from tight ribosomes to unfolded 90S ribosomes. When 6 pmol of the CR fraction was applied, the dimer dissociated into monomers with shoulder [Fig. 8A (3); Kato et al. 2010]. The ribosomes in S. typhimurium, Pe. carotovorum and particularly in K. pneumoniae were more unstable than those in E. coli (Fig. S7 in Supporting Information). In K. pneumoniae, ribosomes were not detected when 30 pmol of the CR fraction was applied, whereas they were detected as a shoulder of the ribosome peak when 150, 300 or 600 pmol was applied (Fig. 1A and Fig. S7 in Supporting Information). In contrast, ribosome generated by long HPF (LpHPF, SaHPF and TtHPF) were completely stable even when 6 or 15 pmol of the CR fraction was applied [Fig. 8A (1), 8A (2) and Fig. S7 (1) in Supporting Information]. This suggests that ribosomes formed by long HPF are much more stable than those generated by RMF and short HPF, with the dissociation constants (Kd) differing by at least one order of magnitude. Interestingly, E. coli ribosomes that formed after the addition of SaHPF or LpHPF (Fig. 7) did not dissociate when low amounts of the CR fraction were applied, similar to those formed by long HPF in vivo [Fig. 8A (4), 8A (5)]. These results suggest that RMF/short HPF and long HPF determine the stability of ribosomes. It appears that long HPF mediates strong interactions between ribosomes, whereas the interaction mediated by RMF/short HPF is weaker. LpHPF binds tightly to subunits By high salt washing, CRs release ribosome-associated proteins, translational factors, mrna and trna to generate high salt-washed ribosomes. When the CR fraction prepared from E. coli in stationary phase is washed with a high salt buffer, RMF is released and the ribosome dissociates into two ribosomes (Wada et al. 1990). In S. aureus, treatment of the CR fraction with a high salt buffer releases SaHPF and causes the ribosome to dissociate (Ueta et al. 2010). However, LpHPF was not released from the ribosome in L. paracasei after high salt washing, and the ribosome did not dissociate (Fig. 8B). Treatment of high salt-washed ribosomes with dissociation buffer led to the dissociation of and subunits; however, LpHPF remain associated with the subunit (Fig. 8C). ribosomes efficiently reformed when the and fractions were mixed together in association buffer, which suggests that LpHPF is highly efficient at mediating the formation of ribosomes (Fig. 8D). Effects of long HPF on in vitro translation An in vitro translation experiment was carried out to examine the physiological role of long HPF in trans- 563

11 M Ueta et al. (A) 10 OD 600nm Time (min) (B) (1) Exponential (2) Stationary (3) Stationary HSR Absorbance 260 nm S5 L9 TtHPF S5 L9 TtHPF L9 S5 TtHPF (D) 50 (C) Absorbance 260 nm Absorbance 260 nm Transition + Buf. I ΔTthpf + TtHPF formation (%) (E) formation (%) Temperature ( C) TtHPF/ribosome (fold) 564

12 Two types of ribosome in bacteria Figure 6 TtHPF is essential for the formation of ribosomes in Thermus thermophilus. (A) Growth curve of T. thermophilus cultured at 75 C. Cell growth was monitored by optical density at 600 nm. Numbered arrows indicate exponential phase (1, 200 min), transition phase (2, 500 min) and stationary phase (3, 700 min). (B) ribosome formation in T. thermophilus. Cell cultures were sampled at 200 and 700 min (arrows of Fig. 7A), and crude ribosome fractions were prepared and analyzed (150 pmol per sample) by 5 20% sucrose density gradient (SDG) centrifugation. The ribosome profiles are shown. (C) ribosome formation in T. thermophilus and DTthpf cells at 500 min (upper). Purified TtHPF mediates the dimerization of ribosomes from DTthpf cells in vitro (lower). High salt-washed dissociated ribosomes isolated from DTthpf cells were incubated with buffer I or TtHPF (molar radio of TtHPF to ribosome was 1.5 : 1) for 30 min at 60 C. The incubation mixtures were precipitated and analyzed by 5 20% SDG centrifugation. The ribosome profiles are shown (lower). (D) Effect of temperature on ribosome formation in vitro. The experiment was carried out as described in (C), except the incubation was carried out at 37, 50, 60 or 75 C. (E) Effect of the amount of TtHPF on ribosome formation in vitro. The experiment was carried out as described in (C), except the molar ratio of TtHPF to ribosomes was 0.5 : 1, 1 : 1, 1.5 : 1, 3.0 : 1 or 5 : 1. All experiments were carried out three times. lation. GST protein made by translation of gst mrna was quantified by immunoblotting. When a 10 : 1 molar ratio of LpHPF to ribosomes was added to the reaction mixture, translation was inhibited by approximately 65% (Fig. 9), indicating LpHPF inhibits translation in vitro. Discussion The ribosome was first identified in E. coli as a fourth type of ribosomal particle in addition to the and subunits and ribosomes (Wada et al. 1990; Wada 1998). RMF mediates the dimerization of ribosomes in E. coli (Wada et al. 1990). Short HPF supports the function of RMF and stabilizes the dimer, thereby allowing conversion from an unfolded 90S ribosome to a folded ribosome (Ueta et al. 2005). Expression of short HPF and RMF is induced during stationary phase. ribosomes are not detected in E. coli during exponential phase; they begin to form from early stationary phase and are most abundant during stationary phase (after 3 4 days of culture). The ribosome is the most abundant ribosomal particle in E. coli in stationary phase. The ribosome dissociates into two ribosomes as cell viability falls (after 5 6 days of culture), and these ribosomes are finally degraded (Wada et al. 2000). ribosomes were specifically generated in stationary phase in each of the five Gammaproteobacteria species tested (Fig. 1A and Fig. S1 in Supporting Information). The genes encoding RMF and short HPF are highly conserved in these species (Fig. S2A, S2B in Supporting Information), and the RMF and short HPF proteins were found in ribosomal particles in each of these strains (Fig. 1B). Bacteria belonging to other groups aside from the Gammaproteobacteria and Betaproteobacteria classes also form ribosomes. Long HPF mediates the formation of ribosomes in eight strains tested. Unexpectedly, ribosomes formed by long HPF were observed throughout all growth phases including exponential phase (compare Fig. 1A with Figs 2B and 6B,C, and Fig. S5 in Supporting Information). The formation of ribosomes specifically during stationary phase in Gammaproteobacteria may reflect changes in the expression levels of RMF and short + Buf.I + SaHPF + LpHPF Absorbance 260 nm SaHPF SaHPF LpHPF Figure 7 ribosomes are formed when long HPF (SaHPF or LpHPF) is mixed with Escherichia coli ribosomes. High saltwashed dissociated ribosomes isolated from YB1008 cells at transition phase were mixed with buffer I, SaHPF or LpHPF and incubated for 30 min at 37 C. The mixtures were precipitated and analyzed (150 pmol per sample) by 5 20% sucrose density gradient centrifugation. The ribosome profiles are shown. The Western blots show the level of SaHPF or LpHPF protein in each of the corresponding fractions in the graph. Purified SaHPF (10 ng) or LpHPF (7 ng) protein was loaded as a control on the right. 565

13 M Ueta et al. (A) (1) Staphylococcus aureus Absorbance 260nm Absorbance 260nm Absorbance 260nm 150 pmol 30 pmol 15 pmol 6 pmol (2) Lactobacillus paracasei 150 pmol 30 pmol 15 pmol 6 pmol (3) Escherichia coli 150 pmol 30 pmol 15 pmol 6 pmol (4) E. coli HSR + SaHPF 150 pmol 30 pmol 15 pmol 6 pmol Absorbance 260nm Absorbance 260nm (5) E. coli HSR + LpHPF 150 pmol 30 pmol 15 pmol 6 pmol (B) Absorbance 260 nm High-salt washing S S2 L10 LpHPF (C) LpHPF S2 S5 L10 S2 S5 L5 (D) Association 566

14 Two types of ribosome in bacteria Figure 8 ribosomes formed by long HPF are stable. (A) ribosomes formed by LpHPF or SaHPF do not dissociate when the ribosome concentration is low. Ribosomes prepared from Lactobacillus paracasei or Staphylococcus aureus cells at transition phase or stationary phase, and Escherichia coli high salt-washed dissociated ribosomes mixed with SaHPF or LpHPF (Fig. 7) were analyzed by 5 20% sucrose density gradient (SDG) centrifugation at various different concentrations (150, 30, 15 and 6 pmol). The ribosome profiles are shown. The y-axes of the 150, 30, 15 and 6 pmol ribosome profiles indicate absorbance at 260 nm, with axes going to a maximum of 2, 0.5, 0.25 and 0.15 units, respectively. (B) Effects of high salt washing on ribosomes in L. paracasei. Following high salt washing of crude ribosome fractions from L. paracasei, LpHPF was not released from the ribosome (right, RFHR 2-D PAGE pattern) and ribosomes did not dissociate. (C) LpHPF is not released after dissociation of ribosomes into the and subunits. High salt-washed ribosomes were dissociated into and subunits, and the proteins in each fraction were analyzed by RFHR 2-D PAGE. LpHPF remained associated with the subunit (left, RFHR 2-D PAGE pattern). (D) ribosomes reform from and subunit fractions. The separated and subunit fractions from (C) were combined, incubated for 30 min at 37 C and analyzed by 5 20% SDG centrifugation. The ribosome profile shows that ribosomes form. (A) LpHPF/ribosome (fold) (B) Relative activity Control - mrna 1 LpHPF LpHPF/ribosome (fold) Figure 9 Effect of LpHPF on in vitro translation activity. In vitro translation was measured in the presence of various amounts of LpHPF (the amount of LpHPF was 1-, 2.5-, 5- or 10-fold higher than that of ribosomes). (A) GST protein synthesized in vitro was detected by immunoblotting with an anti- GST antibody. The experiment was repeated at least six times. (B) The graph shows the mean in vitro translation activity standard deviation. Standard deviation was calculated using Excel. The y-axis shows translation activity relative to that without LpHPF. The x-axis shows the amount of LpHPF relative to that of ribosomes (fold). GST HPF. Expression of RMF is induced during stationary phase due to an increase in the level of ppgpp, whereas ppgpp is not expressed during exponential phase meaning RMF expression is not induced (Izutsu et al. 2001). The mechanism controlling expression of long HPF is unknown; however, long HPF is continually expressed throughout all growth phases. The DSahpf and DTthpf mutants constructed in this study could not generate ribosomes (Figs 5B and 6C). Moreover, ribosomes isolated from DSahpf or DTthpf cells dimerized to form ribosomes in the presence of purified SaHPF or TtHPF in vitro, respectively (Figs 5C and 6C). These results show that the long HPF homologues SaHPF and TtHPF are essential for ribosome formation in S. aureus and T. thermophilus, respectively. ribosomes may form in all bacteria with a long HPF homologue; genes encoding long HPF are widely conserved in bacteria, except for Gammaproteobacteria and Betaproteobacteria, and the N-terminal and C-terminal domains of long HPF are highly homologous among species (Fig. S6 in Supporting Information). However, only a low level of ribosomes were formed when ribosomes isolated from DSahpf cells were mixed with purified SaHPF protein in vitro (Fig. 5). There may be unknown factors that stimulate formation of ribosomes; however, we were unable to identify any such factors in CR fractions from S. aureus by RFHR 2-D PAGE analysis. There may other unidentified reasons why the level of S. aureus ribosome formation was low in vitro. The stability of ribosomes formed by RMF and short HPF differed from those formed by long HPF (Fig. 8 and Fig. S7 in Supporting Information). The behavior of ribosomes in SDG centrifugation faithfully reflects the Kd of the association between two ribosomes (i.e. the stability of the ribosome). When SDG centrifugation analysis were carried out using pmol of the CR fraction, changes in the relative abundances of ribosomal particles were observed reflecting the dissociation and unfolding of dimers, and these changes differed between the various Gammaproteobacteria species. A ribo- 567

15 M Ueta et al. some peak was not detected when large amounts of the CR fractions (600 pmol) were applied for K. pneumoniae but detected when 150 pmol was applied for the other species, whereas ribosomes dissociated into monomers when 6 pmol of the CR fractions was applied. ribosomes were stable in E. coli, in which the ribosome peak at the end of the centrifugation step corresponded to a concentration of approximately M. The ribosomes formed by long HPF were also analyzed by SDG centrifugation; these ribosomes were most stable even when only 6 pmol of the CR fraction was applied. Therefore, the Kd of ribosomes formed by long HPF must be lower than that of ribosomes formed by RMF/ short HPF in E. coli by at least one order of magnitude. E. coli ribosomes were formed in vitro using the long HPF SaHPF or LpHPF. These ribosomes were as stable as those formed by long HPF in vivo. Therefore, the stability of ribosomes is determined by the factors that mediate their formation, and the binding between two ribosomes may occur differently in ribosomes generated by these two distinct mechanisms. Escherichia coli HPF contains 95 amino acids and has been classified as a short HPF, whereas LpHPF, SaHPF and TtHPF are classified as long HPFs as they contain 185 amino acids. The E. coli HPF sequence was aligned with long HPF sequences from bacteria in which ribosome formation has already been confirmed, that is, S. aureus, T. thermophilus, five Lactobacillaceae species and Synechocystis sp. PCC6803 (Fig. S6 in Supporting Information). The N-terminal sequences (amino acid 1 95) of these proteins are very well conserved, as are five of the six amino acids involved in binding between YfiA and the ribosome (Lys22, Lys25, Lys79, Arg82 and Lys86) (Vila-Sanjurjo et al. 2004). The C-terminus-specific regions of long HPF also show high homology. As E. coli short HPF cannot dimerize ribosomes in vitro (Ueta et al. 2008), the C-terminal sequence of long HPF might be involved in the dimerization of ribosomes, a function that is carried out by RMF in E. coli. Escherichia coli RMF inhibits in vitro translational activity in an MS2 mrna-dependent leucine incorporation assay, whereas E. coli short HPF inhibits poly (U)-phenylalanine translational activity (Wada et al. 1995; Ueta et al. 2008). We examined the effect of long HPF on the translational activity of E. coli ribosomes using an in vitro translation assay. LpHPF protein was added to the in vitro translation system in which GST protein was synthesized from purified gst mrna. When the molar ratio of LpHPF to ribosomes was 10 : 1, GST synthesis was reduced to 35% that of the control (Fig. 9A,B). The N-terminal domain of LpHPF is similar to that of YfiA (Y protein) of E. coli, which competes with conserved translation initiation factors to fill the trna- and mrna-binding channel of the small ribosomal subunit (Vila-Sanjurjo et al. 2004). It is possible that the N-terminal domain of LpHPF prevents binding of trna and mrna to ribosomes and thereby inhibits translation. Phylogenic analysis of HPF homologues shows that long HPF is an ancestor of the HPF homologue, which is widely conserved in bacteria and plant plastids (Ueta et al. 2008). A phylogenetic tree of prokaryotes (Battistuzzi & Hedges 2009) shows that the Alphaproteobacteria class and a common ancestor of the Gammaproteobacteria and Betaproteobacteria classes diverged first and then the Gammaproteobacteria and Betaproteobacteria classes diverged many years later. During this first divergence of the Alphaproteobacteria class, which has a long HPF, the common ancestor appears to have lost the C-terminal region of long HPF and with it ribosomes. The Gammaproteobacteria class diverged from the Betaproteobacteria class, acquired the rmf gene and acquired a different system to generate ribosomes. Here, we used B. multivorans to confirm that the Betaproteobacteria class lacks ribosomes (Fig. S3 in Supporting Information). Thus, it appears that Gammaproteobacteria have a distinct system in which formation of ribosomes is mediated by RMF, HPF and YfiA, not by long HPF. In summary, we propose there are two distinct mechanisms by which ribosomes can form in bacteria. There are three major conclusions of this study: (i) Formation of ribosomes is mediated by RMF and short HPF in the Gammaproteobacteria class, whereas it is mediated by long HPF in other types of bacteria. The Betaproteobacteria class has neither RMF nor long HPF and cannot form ribosomes; (ii) ribosomes formed by RMF and short HPF are generated specifically during stationary phase, whereas those formed by long HPF are generated in all growth phases; and (iii) The association between ribosomes is much weaker in ribosomes formed by RMF and short HPF than in those formed by long HPF. Experimental procedures Bacterial strains and plasmids All strains used in this study are shown in Table S1 in Supporting Information, except E. coli K12 W3110 and E. coli 568