|Home | About | Journals | Submit | Contact Us | Français|
The signal recognition particle (SRP)-dependent cotranslational targeting of proteins to the cytoplasmic membrane in bacteria or the endoplasmic reticulum membrane in eukaryotes is an essential process in most living organisms. Eukaryotic cells have been shown to respond to an impairment of the SRP pathway by (i) repressing ribosome biogenesis, resulting in decreased protein synthesis, and (ii) by increasing the expression of protein quality control mechanisms, such as chaperones and proteases. In the current study, we have analyzed how bacteria like Escherichia coli respond to a gradual depletion of FtsY, the bacterial SRP receptor. Our analyses using cell-free transcription/translation systems showed that FtsY depletion inhibits the translation of both SRP-dependent and SRP-independent proteins. This synthesis defect is the result of a multifaceted response that includes the upregulation of the ribosome-inactivating protein ribosome modulation factor (RMF). Although the consequences of these responses in E. coli are very similar to some of the effects also observed in eukaryotic cells, one striking difference is that E. coli obviously does not reduce the rate of protein synthesis by downregulating ribosome biogenesis. Instead, the upregulation of RMF leads to a direct and reversible inhibition of translation.
The delivery of cytosolically synthesized proteins to their correct cellular compartment is a crucial issue for every living cell, and accordingly, both eukaryotic and prokaryotic cells have developed sophisticated protein targeting and transport machineries. Although both eukaryotic and prokaryotic cells employ protein transport pathways which are not found outside their respective phylogenetic domains, some protein transport pathways are universally conserved (15, 32, 38, 41). One of these is the signal recognition particle (SRP)-dependent protein targeting pathway, which is involved in the cotranslational targeting of proteins to the Sec61 complex of the endoplasmic reticulum (ER) and the SecYEG complex of the bacterial cytoplasmic membrane (32, 49). The eukaryotic SRP is composed of six proteins, which are bound to the 7SL RNA and can be functionally divided into the S domain, which is responsible for substrate binding, and the Alu domain, which induces a transient elongation arrest upon binding to the ribosome (16). This elongation arrest is considered to provide a sufficiently long time window to allow targeting of the SRP-ribosome-nascent chain complex (SRP-RNC) to the membrane-bound SRP receptor (SR) (32). The elongation arrest is relieved upon binding of the SRP-RNC complex to the SR, and the RNC is subsequently transferred to the Sec61 complex.
In most prokaryotic cells, like Escherichia coli, the SRP consists of only the essential substrate binding subunit Ffh, which is bound to the 4.5S RNA (16). The Alu domain does not seem to be present in the prokaryotic SRP, and it is therefore currently unknown whether the prokaryotic SRP is able to induce an elongation arrest by an Alu domain-independent mechanism or whether elongation arrest is dispensable in prokaryotes (32). The bacterial SR consists also of only one subunit, which is called FtsY and which is homologous to the eukaryotic SR α-subunit (3, 4, 22, 34). In contrast to the eukaryotic SRP, which targets both secretory and membrane proteins to the endoplasmic reticulum, bacteria use the SRP pathway predominantly for the targeting of inner membrane proteins, while most periplasmic and outer membrane proteins are transported posttranslationally by the bacterium-specific SecA/SecB pathway (31).
Although the SRP-dependent protein targeting is considered to be essential for viability, some organisms, like Saccharomyces cerevisiae or the gram-positive bacterium Streptococcus mutans, are able to survive even in its absence. The inactivation of the SRP pathway in S. cerevisiae initiates a global response that leads to the repression of key ribosomal proteins and the induction of chaperones and proteases (39). A similar downregulation of protein synthesis genes and upregulation of chaperones is observed in S. mutans (23). This suggests that cells can adapt to impaired protein targeting by reducing protein synthesis and simultaneously increasing protein quality control mechanisms like protein folding and degradation. Unlike S. cerevisiae and S. mutans, E. coli is unable to survive in the absence of the SRP components (34). However, in E. coli strains which encode Ffh under the control of the arabinose promoter, a gradual depletion of Ffh also results in increased concentrations of the general chaperones DnaK and GroEL and the proteases Lon and ClpQ (9). Thus, enhanced protein folding and proteolysis of mislocalized proteins appears to be a general mechanism for coping with impaired protein targeting. However, whether E. coli also modulates protein synthesis in response to SRP or SR depletion is largely unknown. The results of initial in vivo studies using FtsY-depleted E. coli cells have indicated that the expression of SRP-dependent membrane proteins like SecY (42) or LacY (27) is specifically reduced, while the steady-state levels of the cytoplasmic protein β-galactosidase or the SRP-independent secretory protein β-lactamase (7) are unchanged under these conditions (42). This raises the possibility that E. coli is able to specifically reduce the synthesis of SRP substrates without impairment of SRP-independent substrates.
In the present study, we used in vitro transcription/translation systems and found no indication that FtsY depletion selectively inhibits the in vitro synthesis of SRP substrates. Instead, our data show a general protein synthesis defect affecting cytosolic, secretory, and inner membrane proteins. This synthesis defect is a result of a multifaceted response of E. coli upon FtsY depletion that leads to the inactivation of ribosomes by the ribosome modulation factor (RMF).
E. coli MC4100 and DH5α were used as wild-type E. coli strains and routinely grown on LB medium. The conditional FtsY depletion strains N4156(pAra14-FtsY′) (N4156) (34) and IY28 (kindly provided by Eitan Bibi) were routinely grown on LB medium supplemented with 0.4% arabinose and, in the case of N4156(pAra14-FtsY′), with 50 μg/ml ampicillin. For FtsY depletion, E. coli N4156(pAra14-FtsY′) or IY28 was grown overnight on medium supplemented with 0.2% arabinose. After harvesting by centrifugation, cells were washed twice with medium lacking arabinose and used to inoculate cultures containing either 0.2% arabinose to induce FtsY expression or 0.2% fructose for FtsY depletion. Growth was monitored by measuring the optical density at 600 nm. When indicated, cells from the fructose-grown culture were used to inoculate a second passage on fructose-containing medium.
An S135 cell extract (supernatant after centrifugation at 135,000 × g) was prepared from E. coli cells as described previously (36). In brief, after cell breakage using a French pressure cell at 8,000 lb/in2 (Thermo-Scientific Corp., Bad Homburg, Germany), unbroken cells and large debris were removed by centrifugation at 15,500 rpm for 30 min at 4°C using a Sorvall SS34 rotor. The resulting S30 supernatant was then subjected to a readout of endogenous mRNA and subsequently centrifuged at 90,000 rpm for 13 min at 4°C using a Beckmann TLA 100.2 rotor. The S135 supernatant was then used for coupled in vitro transcription/translation assays as described previously (36), using plasmid pDMB (8) or pKSM717-MtlA (31). The reconstituted in vitro system and its components were isolated as described previously (31). After in vitro synthesis, samples were separated on 13% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) gels. Radiolabeled proteins were visualized and quantified using a Storm720 phosphorimager (GE Healthcare, Munich, Germany) and Imagequant software. The S135 cell extract of heat-shocked cells was prepared from cells which were shifted for 3 h from 37°C to 45°C.
E. coli cells for ribosome isolation were grown on S150 medium (37) supplemented with glucose (0.1%; wild type), arabinose (0.2%; FtsY-containing N4156 cells), or fructose (0.2%; FtsY-depleted N4156 cells). After preparation of an S30 cell extract (see above), the S30 cell extract was centrifuged for 2.5 h at 150,000 × g in a Beckmann 50Ti rotor to yield a ribosome- and membrane-containing pellet. The pellet was resuspended in high-salt buffer (50 mM triethanolamine acetate, pH 7.5, 1 M potassium acetate, pH 7.5, 5 mM magnesium acetate, 1 mM dithiothreitol), and about 0.5 ml of the resuspended pellet was loaded onto a 10-to-40% sucrose gradient in high-salt buffer (gradient volume, 12.5 ml). The gradients were prepared by using an automatic Teledyne Isco gradient former. After centrifugation for 17 h at 29,000 rpm in a Beckmann SW40 rotor, the gradient was fractionated by using a density gradient fractionation system driven by Peak-track software (Teledyne Isco Corp., Lincoln, NE). Fractions with high levels of absorption at 256 nm were further analyzed on a 15% SDS-PAGE gel. The fractions containing ribosomal proteins were pooled and centrifuged for 60 min at 90,000 rpm and 4°C. The ribosomal pellet was resuspended in CTF buffer (50 mM triethanolamine acetate, pH 7.5, 50 mM potassium acetate, pH 7.5, 5 mM magnesium acetate, 1 mM dithiothreitol), and the concentration was measured spectrophotometrically. Absorption of 1.0 at 256 nm corresponds to a concentration of 23 nM. Inner membrane vesicles were isolated by density gradient centrifugation as described previously (31, 35). Inclusion bodies were isolated according to published protocols (33). For immunoblot analyses, proteins were electroblotted onto nitrocellulose membranes or Immobilon-PSQ membranes (for detection of RMF; Millipore Corp., Schwalbach, Germany). Mono- and polyclonal antibodies were used with horseradish peroxidase-conjugated goat anti-rabbit, goat anti-mouse, or sheep anti-goat secondary antibodies (Caltag Laboratories, Burlingame, CA) and ECL detection reagents as the detection substrate (GE Healthcare, Munich, Germany). Monoclonal antibodies directed against E. coli RpoS (σS) were purchased from NeoClone Biotechnology (Madison, WI).
For N-terminal sequencing, samples were blotted onto polyvinylidene difluoride membranes and the bands were analyzed by Edman degradation in a gas phase sequencer (model 477A; Applied Biosystems, Foster City, CA) with online identification of the amino acid phenylthiohydantoins.
Amounts of 100 μl of trichloromethane and 200 μl of methanol were mixed with 80 μl of crude membranes, and after homogenization, another 100 μl of trichloromethane was added and mixed for 30 s. After the addition of 100 μl of distilled water and mixing again for 30 s, samples were centrifuged for 10 min at 4°C and 3,000 rpm. The lipid phase was transferred into a fresh tube and dried under a continuous stream of nitrogen. The dry lipid pellet was resolved in trichloromethane to yield a final lipid concentration of 50 μg/μl. Samples were stored under a nitrogen atmosphere at −20°C in sealed tubes for further analysis. For thin-layer chromatography (TLC), 2-μl (100 μg lipid) amounts of the samples were applied on 20- by 20-cm TLC glass plates (TLC silica gel 60 F254; Merck, Darmstadt, Germany) and separated with a mixture of chloroform/methanol/acetic acid (80:10:10) for 45 min. After the TLC plate was dried, lipid spots were visualized by using molybdatophosphoric acid with subsequent incubation for 5 to 15 min at 135°C.
Microscopy was performed on an Olympus BX51 microscope, and images were acquired with a charge-coupled device camera (F-View; Olympus, Hamburg, Germany) driven by cell*F software (Olympus Soft Imaging Solutions, Münster, Germany). Cells were grown in LB medium supplemented with 0.2% arabinose (FtsY-containing cells) or 0.2% fructose (FtsY-depleted cells) and immobilized on a microscope slide with low-melting-point agarose or poly-l-lysine. For electron microscopy samples, bacterial cell pellets were fixed in a solution consisting of 4% paraformaldehyde and 1% glutaraldehyde in 0.1 M phosphate buffer (PB) for 20 min and rinsed in PB. Fixed tissue was osmicated (1% OsO4, 6.8% saccharose in PB for 40 min), dehydrated in graded ethanol using 1% uranyl acetate and 1% wolfram phosphoric acid in 70% ethanol overnight at 4°C, and embedded in epoxy resin (Durcupan; Fluka, Sigma, Taufkirchen, Germany). Ultrathin sections (85 nm) were picked up on Formvar-coated (0.35% Formvar diluted in dichlorethane) nickel slot grids. Sections were then incubated in 50 mM Tris-buffered saline (TBS, pH 7.6) containing 0.1% Triton X-100 (TBST), 50 mM glycine, and 0.1% sodium borohydrate in 0.05 M TBST for 10 min. This was followed by incubation with primary antibody (in TBST containing 2% human serum albumin) overnight at room temperature. As primary antibodies, rabbit anti-YidC (1:200 dilution) and goat anti-S2 and goat anti-L2 (1:200) were used on a given section. Sections were kept in a humidified chamber overnight at room temperature. After several rinses in TBST for 30 min and blocking in 2% HSA in TBST for 10 min, sections were incubated with goat anti-rabbit immunoglobulin G and sheep anti-goat immunoglobulin G, respectively, coupled with 10-nm gold particles (diluted 1:20; British BioCell, Cardiff, United Kingdom) in TBST containing 2% human serum albumin for 2 h at room temperature. After several washes in distilled water, sections were counterstained with 1% aqueous uranyl acetate for 30 min, followed by rinses in distilled water and staining with lead citrate. Sections were digitally photographed with a Zeiss Leo 906 E electron microscope.
The RMF gene was deleted in IY28 via the λ-red recombination method (11). In brief, IY28 was first transformed with plasmid pKD46, encoding the Red recombinase of the λ phage. A kanamycin cassette containing rmf-specific sequences at both ends was generated via PCR using the plasmid pKD4 as template and the following primer pair: RMFfw (5′-ATCACTGTTTTCTTTTCCACCAGAAACCAGTATGAGGGAAACGAGGCATGGTGTAGGCTGGAGCTGCTTCG-3′) and RMFrev (5′-TGCGGAGGTTTCTTTTTAAAGAGACAGAATCAGGCCATTACCCTGTCCATATGAATATCCTCCTTAGT-3′). The PCR product was electroporated into IY28 pKD46, and screening for kanamycin resistance was performed on LB plates at 37°C. Chromosomal DNA of kanamycin-resistant colonies was isolated, and the insertion of the kanamycin cassette into the RMF coding sequence was verified by two separate PCRs. The first PCR used primers that were specific for the flanking regions of rmf (RMF-fw-1, 5′-GCTTAACTGTGATTGCACAT-3′, and RMF-rev-1, 5′-AAGGCGAAACCTCCGCAATG-3′). The second PCR was performed with a primer pair specific for the kanamycin cassette (k2, 5′-CGGTGCCCTGAATGAACTGC-3′, and kt, 5′-CGGCCACAGTCGATGAATCC-3′).
For analyzing the effect of FtsY depletion on protein synthesis, we performed in vitro transcription/translation assays using a soluble cellular extract (S135) derived from the conditional E. coli FtsY mutant N4156(pAra14-FtsY′) (N4156). In this strain, the expression of ftsY is under the control of the arabinose promoter (34) and, thus, only in the presence of arabinose does the strain display wild type-like growth (Fig. (Fig.1A,1A, Ara). If arabinose was replaced by fructose, growth was significantly reduced (Fig. (Fig.1A,1A, Fru-1). This growth defect was even more pronounced if the fructose-grown culture was back-diluted into fresh fructose-containing medium (Fig. (Fig.1A,1A, Fruc-2).
The FtsY content in arabinose- or fructose-grown N4156 cells was monitored by Western blotting of the S135 extracts using polyclonal antibodies against FtsY. In arabinose-grown cells, FtsY antibodies detected two bands (Fig. (Fig.1B).1B). The upper band corresponds to full-length FtsY, which has a predicted molecular mass of 56 kDa but runs on SDS-PAGE gels at about 100 kDa (22, 34), while the lower band is an N-terminally truncated isoform (FtsY-14) which lacks the first 14 amino acids and runs on SDS-PAGE at about 75 kDa (34, 48). In fructose-grown cells, the full-length FtsY was significantly reduced in Fru-1 cells and almost completely absent in Fru-2 cells. Simultaneously with the disappearance of full-length FtsY, the FtsY-14 isoform started to accumulate in fructose-grown cells and disappeared only after further incubation in the absence of arabinose (data not shown), results which are in agreement with previously published data (34, 48). The exact function of the FtsY-14 isoform is currently unknown, but even its accumulation does not suppress the diminished growth of FtsY-depleted cells, suggesting that it cannot act as a full substitute for full-length FtsY (17, 43, 48).
The S135 extracts were then analyzed for their ability to in vitro synthesize either the SRP-dependent membrane protein mannitol permease (MtlA) or the SRP-independent secretory protein OmpA. Phosphorimaging of the radioactively labeled substrates revealed that both substrates were efficiently in vitro synthesized if the S135 extract was prepared from FtsY-containing cells (Fig. (Fig.1C).1C). In contrast, when the S135 extract was prepared from cells depleted of FtsY, we observed reduced protein synthesis for the partially depleted extract (Fru-1) and no protein synthesis with the more-extensively depleted extract (Fig. (Fig.1C,1C, Fru-2). These data indicate that FtsY-depleted cell extracts loose their ability for in vitro protein synthesis of both SRP-dependent (MtlA) and SRP-independent (OmpA) substrates. The same synthesis defect was observed when the cytosolic protein Ffh, the secretory protein β-lactamase, or the inner membrane protein YidC were in vitro synthesized in these S135 extracts (data not shown). Thus, although our data support the previous observation (42) that the depletion of the bacterial SR blocks protein synthesis, this synthesis defect does not exclusively affect SRP substrates, at least in vitro. Importantly, the protein synthesis defect could not be rescued by the addition of purified FtsY (Fig. (Fig.1C),1C), suggesting that this defect is not simply the result of lower FtsY concentrations in the in vitro reaction mixtures.
Streptococcus mutans has been shown to respond to an inhibition of the SRP cycle by downregulating ribosome biogenesis (23), which in principle could explain the protein synthesis defect observed in FtsY-depleted cell extracts. The S135 cell extract used for in vitro protein synthesis is routinely prepared by ultracentrifugation of a crude cell extract (S30 extract) which contains all soluble and membrane-bound components of the cell, including ribosomes. Western blotting using antibodies against either the large ribosomal subunit L2 or the small ribosomal subunit S2 did not reveal significant differences in the total amount of ribosomes in FtsY-containing or FtsY-depleted S30 cell extracts (Fig. (Fig.2A).2A). In contrast, when the S135 cell extracts were analyzed, the concentrations of both the large and the small ribosomal subunits were significantly reduced (Fig. (Fig.2B).2B). These data indicate that it is not the total ribosome concentration that is changed upon FtsY depletion but, rather, the distribution of ribosomes between the soluble S135 fraction and the insoluble fraction, which is removed during the preparation of the S135 cell extract.
The lack of ribosomes in FtsY-depleted S135 cell extracts implies that the addition of ribosomes should restore in vitro protein synthesis. Adding purified wild-type ribosomes to FtsY-containing S135 extracts did not significantly stimulate in vitro synthesis, indicating that under wild-type conditions, e.g., in the presence of FtsY, the S135 extract contains enough ribosomes for efficient in vitro protein synthesis (Fig. (Fig.2C).2C). In contrast, adding ribosomes to the otherwise-inactive FtsY-depleted S135 cell extract completely restored protein synthesis; in the case of MtlA, protein synthesis was even more efficient than in the control S135 extract (Fig. (Fig.2C).2C). In summary, these data indicate that FtsY-depleted S135 cell extracts are inactive because they lack sufficient amounts of ribosomes.
To analyze the altered ribosome distribution in more detail, we analyzed the pellet fraction of the S135 cell extract ultracentrifugation (containing membranes and pelleted cytosolic components) on a 20-to-70% sucrose gradient. The distribution of the large ribosomal subunit within the gradient was then determined by Western blotting. In the pellet fraction of FtsY-containing extracts, we were able to detect the L2 subunit in fractions 1 to 8 of the gradient (corresponding to 20 to 40% sucrose) (Fig. (Fig.3A).3A). These fractions comprise the cytoplasmic membrane of E. coli (30) plus the membrane-bound ribosomes, which constitute about 10% of the total ribosome content of E. coli (26). However, when the pellet fractions of FtsY-depleted extracts were separated by sucrose gradient centrifugation, the L2 subunit was detected not only in the first fractions but also in the high-density fractions of the gradient (fractions 10 to 20) (Fig. (Fig.3A),3A), indicating either the formation of large ribosomal aggregates or binding of ribosomes to larger cellular structures. This was further analyzed by subjecting these high-density fractions (e.g., fractions 10 to 20 of the gradients) to Coomassie blue staining. In the high-density fractions of FtsY-depleted cell extracts, we observed a significant increase in protein bands compared to the numbers of bands in FtsY-containing extracts (Fig. (Fig.3B).3B). In particular, two distinct protein bands appeared to accumulate upon FtsY depletion, one running at about 40 kDa and a second running at about 15 kDa. By N-terminal sequencing, the 40-kDa protein was identified as YjhC, a protein of unknown function which exhibits homology to oxidoreductases, while the 15-kDa band corresponds to IbpA/IbpB, a small heat shock protein that is associated with the formation of inclusion bodies in E. coli (2). The three bands which are present in both FtsY-containing and FtsY-depleted extracts correspond to the outer membrane proteins OmpF, OmpT, and OmpX.
The presence of IbpA/IbpB in the high-sucrose density fractions is indicative of the formation of inclusion bodies, and we therefore analyzed FtsY-depleted cells microscopically. FtsY-depleted cells displayed a filamentous and somewhat bulged appearance (Fig. (Fig.3C),3C), which has been observed before (21, 34). In addition, we noticed in these cells nontransparent structures that were located close to the cytoplasmic membrane (Fig. (Fig.3C).3C). These structures were further analyzed by electron microscopy. In FtsY-depleted cells, we observed electrodense, amorphous structures without an obvious surrounding barrier (Fig. (Fig.3D).3D). These structures were distributed over the whole cell and are most likely identical to the nontransparent structures detected by light microscopy. The morphological features of these internal structures suggest the presence of protein inclusion bodies (39), which is in line with the upregulation of the inclusion body-associated protein IbpA/IbpB upon FtsY depletion (Fig. (Fig.3B).3B). Inclusion bodies are normally removed by the first centrifugation that results in the S30 extract (6, 40, 47); the detection of IbpA/IbpB in the pellet fraction of the S135 extract ultracentrifugation product, however, suggests that not all protein aggregates/inclusion bodies are removed by this first centrifugation. The formation of inclusion bodies could possibly explain the altered sedimentation behavior of ribosomes (c.f. Fig. Fig.3A),3A), and we therefore analyzed by Western blotting whether ribosomal subunits were detectable in the inclusion body fraction of FtsY-depleted cells. Both the L2 and the S2 ribosomal subunits were present in purified inclusion bodies of FtsY-depleted cells, while in the corresponding material of FtsY-containing cells, only very weak signals were detectable (Fig. (Fig.3E).3E). The presence of ribosomal subunits in inclusion bodies was further verified by immunogold labeling and electron microscopy (data not shown). Although this analysis confirmed the presence of the L2 subunit within the inclusion bodies, a significant portion of ribosomes was still detectable in the cytosol of FtsY-depleted cells. Therefore, it appeared unlikely that the sequestration of ribosomes in inclusion bodies is the only reason for the in vitro protein synthesis defect.
For determining whether additional mechanisms are involved in changing the sedimentation behavior of ribosomes and potentially interfere with their activity, we analyzed the ribosomal activity directly. Ribosomes were isolated from FtsY-containing and FtsY-depleted N4156 cells by differential centrifugation and tested for their ability to stimulate in vitro protein synthesis of an otherwise inactive FtsY-depleted S135 cell extract. Using Coomassie staining, we did not observe significant differences in the banding patterns of wild-type and FtsY-containing or FtsY-depleted N4156 ribosomes (Fig. (Fig.4A).4A). Nevertheless, purified ribosomes from FtsY-depleted N4156 cells remarkably failed to stimulate in vitro synthesis of the SRP-dependent membrane protein MtlA and the SRP-independent secretory protein OmpA (Fig. (Fig.4B).4B). On the other hand, ribosomes isolated from FtsY-containing N4156 cells and wild-type ribosomes were able to stimulate in vitro protein synthesis (Fig. (Fig.4B4B).
In order to exclude the possibility that FtsY-depleted ribosomes were inactive only in the context of an FtsY-depleted S135 extract, we analyzed ribosomal activity in a nonbiased assay system. The reconstituted in vitro transcription/translation system is a highly purified system which does not contain endogenous ribosomes (31) and is therefore perfectly suited for analyzing ribosomal activity. The reconstituted in vitro system was prepared from wild-type E. coli cells and used for in vitro synthesis of MtlA. Without added ribosomes, the reconstituted in vitro system did not allow synthesis of MtlA (Fig. (Fig.4C).4C). However, upon the addition of increasing amounts of ribosomes from FtsY-containing cells, synthesis of MtlA was easily detectable and reached saturation at about 250 nM ribosomes (Fig. (Fig.4C).4C). In contrast, ribosomes isolated from FtsY-depleted cells exhibited less than 10% of the wild-type ribosome activity. These data demonstrate that FtsY depletion in E. coli not only induces a change in ribosome distribution but also leads to an inactivation of ribosomes.
Ribosome-inactivating proteins have been identified mainly in plants (44), but they exist also in bacteria and seem to be required for survival under stress conditions (18, 28). In E. coli, the RMF has been shown to inactivate ribosomes by inducing the dimerization of 70S ribosomes to 100S ribosomes (29, 50). Because RMF not only inactivates ribosomes but also changes their apparent molecular mass, we analyzed whether RMF was induced in FtsY-depleted cells. In exponentially growing wild-type cells (6 h), we observed by Western blotting only a weak signal for RMF (Fig. (Fig.5A),5A), and in FtsY-containing N4156 cells, there were only slightly elevated concentrations of RMF. In contrast, in FtsY-depleted cells, the RMF concentration was significantly increased (Fig. (Fig.5A).5A). In stationary-phase cells (48 h), comparable amounts of RMF were detectable in wild-type and FtsY-containing/-depleted N4156 cells, which is in agreement with previous data showing that RMF expression is highest during stationary phase (29, 50). In order to directly determine whether RMF was responsible for the inactivation of ribosomes isolated from FtsY-depleted cells, the purified ribosome preparation was analyzed for the presence of RMF. With Western blotting, RMF was strongly detectable in FtsY-depleted ribosomes but only weakly detectable in FtsY-containing ribosomes and almost undetectable in wild-type ribosomes (Fig. (Fig.5B5B).
During stationary phase, E. coli not only induces the expression of RMF but also increases the cardiolipin content of the cytoplasmic membrane (13). We therefore analyzed the phospholipid composition of FtsY-depleted cells. Using TLC, the two major phospholipids phosphatidyl glycerol and phosphatidyl ethanolamine were easily detectable in total-membrane preparations of FtsY-containing cells (Fig. (Fig.5C).5C). Cardiolipin was only weakly detectable because it accounts for only about 5% of the total phospholipid content in wild-type E. coli cells. In total membranes of FtsY-depleted cells, the absolute phospholipid content was not significantly different, but these membranes contained more cardiolipin and slightly less phosphatidyl glycerol (Fig. (Fig.5C).5C). Although these data would in principle support a starvation (σS)-induced stress response (24) upon FtsY depletion, we did not observe a substantial difference in the σS content between FtsY-containing and FtsY-depleted cells (Fig. (Fig.5D5D).
The contribution of RMF to the in vitro protein synthesis defect was further verified by constructing an RMF knockout strain. We first tried to delete the RMF coding sequence in E. coli N4156 by using the λ-red-based recombination method (11). However, for unknown reasons, we were unable to propagate the required plasmids in this strain. Therefore we deleted the RMF gene in E. coli IY28 (kindly provided by Eitan Bibi), another conditional FtsY depletion strain, carrying FtsY also under the control of the arabinose promoter. IY28 cells grown on fructose were depleted of FtsY (Fig. (Fig.6A),6A), but we did not observe the accumulation of the truncated FtsY-14 isoform (compare Fig. Fig.6A6A with Fig. Fig.1B),1B), suggesting that N-terminal cleavage of FtsY (48) is not observed to the same extent in all E. coli strains. In respect to RMF expression, however, both strains behaved identically, e.g., RMF was strongly upregulated upon FtsY depletion (Fig. (Fig.6A).6A). As expected, independently of the FtsY content, RMF was undetectable by Western blotting in IY28ΔRMF cells. FtsY-depleted IY28 cells also displayed the same morphological features as FtsY-depleted N4156 cells (compare Fig. Fig.6B6B with Fig. Fig.3C).3C). The filamentous and bulged appearance and the presence of inclusion bodies were also observed in IY28ΔRMF cells, indicating that the lack of RMF had no obvious influence on the cell division defect or the formation of inclusion bodies upon FtsY depletion.
Importantly, the S135 cell extract of FtsY-depleted IY28ΔRMF cells contained significantly more ribosomes than the S135 cell extract of FtsY-depleted IY28 cells (Fig. (Fig.6C).6C). In agreement with this, we observed only a weak in vitro protein synthesis defect in IY28ΔRMF S135 extracts (Fig. (Fig.6D).6D). In contrast, protein synthesis in the IY28 S135 extract and the N4156 S135 extract required the addition of wild-type ribosomes (Fig. (Fig.6D6D).
Collectively, these data demonstrate that E. coli responds to the depletion of FtsY by inducing the upregulation of RMF, which concomitantly shuts down protein synthesis. The RMF-induced formation of inactive 100S ribosomes (51) is probably the major reason why ribosomes are pelleted during the preparation of S135 extracts from FtsY-depleted cells, although inclusion body formation most likely also contributes to this effect.
For analyzing whether the protein synthesis defect is specific for FtsY depletion or is also observed when cells are stressed by other means, we analyzed S135 cell extracts from FtsY-containing but heat-shocked IY28 cells. These S135 extracts were also unable to synthesize proteins in vitro, but in contrast to FtsY-depleted S135 extracts, were not reactivated by wild-type ribosomes (Fig. (Fig.7A).7A). This indicates that the protein synthesis defect observed after heat stress is not merely the result of ribosome inactivation and, thus, is different from the effects observed upon FtsY depletion. In agreement with this, Western blotting revealed that the ribosome content was only slightly reduced and the RMF content only slightly elevated (Fig. (Fig.7B).7B). Morphologically, the heat-shocked cells were not significantly different from IY28 cells grown at 37°C, although some cells did show small inclusion bodies (Fig. (Fig.7C).7C). It therefore appears that FtsY depletion provokes a stress response that is similar but not identical to other stress responses.
Finally, we also tried to verify the protein synthesis defect in vivo, e.g., by analyzing the expression of the SRP-independent protein OmpA and the SRP-dependent membrane protein YidC in FtsY-depleted IY28 cells. Although we did observe a synthesis defect for OmpA upon FtsY depletion, we were unable to obtain reliable in vivo data for YidC because expressing YidC under FtsY depletion conditions resulted in severe growth defects and cell lysis (data not shown).
Both Saccharomyces cerevisiae and Streptococcus mutans have been shown to respond to an impairment of SRP-dependent cotranslational protein targeting by reducing the rate of protein synthesis concomitantly with the induction of protein quality control mechanisms like folding and degradation. In particular, the repression of ribosome biogenesis appears to be an important adaptation of both organisms for coping with reduced protein targeting. In S. cerevisiae, the expression of almost all ribosomal proteins is downregulated in response to a loss of the SRβ subunit and in response to an SRP mutation (5, 39). A similar downregulation of ribosomal proteins and upregulation of chaperones was observed in S. mutans (23). Depletion of the SRP components in E. coli also leads to an induction of the σ32-regulated stress response, e.g., the induction of chaperones like DnaK and GroEL and proteases like Lon and ClpQ (9, 25), but the effect on protein synthesis is largely unknown.
One important conclusion of our study is that E. coli does not seem to execute a specific downregulation of ribosome biogenesis upon impairment of the SRP pathway (Fig. (Fig.2A).2A). Instead in E. coli, ribosomes are inactivated by upregulation of RMF in response to FtsY depletion (Fig. (Fig.4,4, ,5,5, and and6).6). RMF belongs to the group of ribosome-inactivating proteins that have been identified in both plants and bacteria (44). Their major role is the inactivation of ribosomes to preserve ribosome integrity and to support cell survival under stress conditions (18). Many of these proteins, e.g., ricin, function as adenine glycosylases, resulting in a depurination of rRNA (19). In E. coli, two proteins have been identified which reversibly inactivate ribosomes by different mechanisms. One is RaiA, which is expressed under cold stress conditions and appears to inactivate the elongation stage of translation (1). The second is RMF, which inactivates ribosomes by covering the peptidyl transferase center (28, 29, 50) and which induces the dimerization of ribosomes into 100S ribosomes (28, 51). This dimerization of ribosomes is probably mainly responsible for the selective removal of ribosomes from the FtsY-depleted S135 cell extracts by centrifugation. So far, RMF expression has been shown to be upregulated during the stationary-growth phase of E. coli (14). Our data now reveal that RMF expression does not seem to be limited to high cell density but is also expressed in response to FtsY depletion. During stationary phase, E. coli increases the cardiolipin concentration of the membrane (13), which we also observe in FtsY-depleted E. coli cells (Fig. (Fig.5C).5C). In addition, our data show the upregulation of YjhC (Fig. (Fig.3B),3B), which has also been linked to the stationary phase (12, 14). Although an exact function for this protein remains to be elucidated, YjhC shows strong homology to glucose-fructose oxidoreductases, which are involved in sorbitol production. Sorbitol belongs to the compatible solutes, a heterogeneous group of small-molecular-weight substances that are involved in stabilizing proteins under stress conditions (10, 30).
Thus, upon FtsY depletion, E. coli induces the σ32 response (9, 25) and a response that is similar to the stationary-phase response, although we did not observe a drastic upregulation of σS. It is important to emphasize that FtsY depletion is normally not encountered by E. coli; however, the bacterial SRP pathway is easily saturated due to the low concentration of SRP (9, 46). The reversible inactivation of protein synthesis by RMF provides an elegant solution to this obstacle because it allows the modulation of protein synthesis in response to the transport capacity and reduces the accumulation of misfolded proteins in the cytoplasm.
In vivo data have suggested that the depletion of FtsY specifically inhibits the synthesis of the SRP-dependent inner membrane proteins SecY and LacY but not the synthesis of secretory or cytosolic proteins (42). Because FtsY is required for the release of SRP from translating ribosomes, one possible interpretation for these observations is that the bacterial SRP is able to induce an elongation arrest by an Alu domain-independent mechanism. However, in our in vitro analyses, we found no indication that FtsY depletion specifically impairs the synthesis of SRP substrates; instead, we observed a general RMF-induced inhibition of protein synthesis. It is important to note that by using in vitro systems, we are able to use more-extensively depleted cell extracts. The level of depletion, however, probably directly influences the subsequent physiological responses and, thus, could in principle explain the difference between the in vivo and the in vitro observations. That the level of FtsY depletion might influence the subsequent protein synthesis defect is also in line with the observation that a specific inhibition of membrane proteins has not been observed in all in vivo studies using FtsY-depleted cells (20, 25, 45). Still, we currently cannot exclude the possibility that the strong effects we observe in our study conceal a more subtle response of E. coli that impairs the synthesis of membrane proteins more severely than the synthesis of soluble proteins.
In summary, our data have revealed that E. coli shuts down protein synthesis in response to an impairment of the essential SRP pathway. However, unlike eukaryotic cells, E. coli employs a mechanism which directly but reversibly inactivates ribosomes by RMF. Considering the growth rate of E. coli in comparison to growth rates of eukaryotic cells, this response might allow a faster adaptation than downregulating ribosome biogenesis. In prokaryotes, a downregulation of ribosome biogenesis has so far only been observed in S. mutans (23). However, S. mutans is, to our knowledge, the only known prokaryote in which the SRP pathway is not essential and, therefore, a slower adaptation might be sufficient in this organism. Still, whether the inactivation of ribosomes by ribosome-inactivating proteins is a general mechanism in bacteria for coping with an inhibition of the essential SRP pathway remains to be analyzed in future studies.
This work was supported by grants from the Deutsche Forschungsgemeinschaft (Forschergruppe 929, Forschergruppe 967, and GRK 1478) to H.-G.K. and the German-French University (DFH) to H.-G.K. and by an F. F. Nord fellowship of the University Freiburg to J.B.
We thank Hideji Yoshida, Osaka University, for providing anti-RMF antibodies and Eitan Bibi, Weizman Institute, for strain IY28.
Published ahead of print on 11 September 2009.