Modification of the Ribosome and the Translational Machinery during Reduced Growth Due to Environmental Stress

ANTÓN VILA-SANJURJO

[SECTION EDITOR: MICHAEL O’CONNOR]

Posted July 25, 2008

Berkeley Center for Synthetic Biology, University of California, Berkeley, Berkeley, CA 94720-3224

Mailing address: Berkeley Center for Synthetic Biology, University of California, Berkeley, 717 Potter St., Bldg. 977, Room 177, Berkeley, CA 94720-3224. Phone: (510) 486-6839, Fax: (510) 486-4252, E-mail: This e-mail address is being protected from spambots. You need JavaScript enabled to view it .

Understanding the mechanisms used by microbes during the stress response is crucial if one desires to control these organisms within their natural niches. This importance is underscored by the fact that microorganisms reach their most resistant state while living under certain stressful conditions (62). Survival in these situations usually requires a whole reprogramming of gene expression that results in drastic metabolic changes (7, 40, 43). A considerable part of this process appears to involve the adaptation of the translational machinery to the new demand for protein synthesis.

Escherichia coli strains normally used under laboratory conditions have been selected for maximum growth rates and require maximum translation efficiency (64). This means that during logarithmic growth, most ribosomes are engaged in one of the four steps of translation: initiation, elongation, termination, and recycling (see below). Recent studies have shed light on the structural and functional changes undergone by the translational machinery in E. coli during heat and cold shock and upon entry into stationary phase. As we will see, in these situations both the composition and the partitioning of this machinery into the different pools of cellular ribosomes are modified. As a result, the translational capacity of the cell is dramatically altered. The goal of this chapter is to provide a comprehensive account of these modifications, regardless of whether or not their underlying mechanisms and their effects on cellular physiology are known.

The ribosome is an ancient ribonucleoprotein particle whose structural core and fundamental mechanism of action are conserved among all forms of life (for a recent review on translation in E. coli, see reference 111). The ribosome translates the genetic information encoded in the mRNA into the amino acid sequence of a protein. The prokaryotic 70S ribosome consists of a small 30S subunit and a large 50S subunit, which each have three tRNA binding sites: the A (aminoacyl) site that accepts the incoming aminoacylated tRNA, the P (peptidyl) site that binds the tRNA carrying the nascent polypeptide chain, and the E (exit) site that mediates the release of the deacylated tRNA after peptide bond formation. In bacteria, translation initiation generally starts with the binding of the mRNA to the 30S subunit via the interaction between the Shine-Dalgarno and the anti-Shine-Dalgarno sequences (41). This binding is followed by the recognition of the start codon (AUG) at the ribosomal P site by fMet-tRNA^fMet in complex with initiation factor IF2-GTP (16, 39, 113). This process requires the presence of initiation factors IF1 and IF3, which control the fidelity of initiation (41). After the joining of the 50S subunit and the release of the initiation factors, the elongation phase starts with the delivery of the aminoacyl-tRNA, specified by the second codon on the mRNA, at the ribosomal A site (reviewed in reference 38). This process is catalyzed by elongation factor Tu (EF-Tu) and requires the hydrolysis of GTP. Peptide bond formation within the 50S subunit ensues, followed by the translocation of the two tRNAs at the A and P sites to the P and E sites, respectively, by the action of elongation factor G (EF-G) in complex with GTP. The vacant A site then becomes the substrate for another round of aminoacyl-tRNA delivery by EF-Tu. The termination phase of translation follows elongation. At this point, the newly synthesized peptide is released by the action of the release factors RF1, RF2, and RF3. Finally, the ribosomes are recycled into free subunits in a reaction catalyzed by the ribosome-recycling factor and EF-G (55).

Since much of what follows in this chapter deals with the association of ribosomes with different subsets of proteins as a function of the growth conditions, a brief description of the methods and criteria used to identify these proteins is due. Both ribosomal subunits are composed of rRNA and ribosomal proteins (r-proteins), which remain associated with the ribosome after a high-salt wash (typically performed in 0.5 to 1 M NH₄Cl). The two-dimensional (2D) gel electrophoresis developed by Kaltschmidt and Wittmann has been the standard method of separation of r-proteins, which have been named based on their position on 2D gels, i.e., S1 to S21 for 30S subunit proteins and L1 to L34 for 50S subunit proteins (54, 112). Advances in 2D gel electrophoresis introduced by Wada and coworkers in the mid-1980s (104, 105, 107) allowed the identification of additional ribosomal proteins (L35 and L36), as well as additional proteins present on the ribosome during periods of nonlogarithmic growth. More recently, proteomics techniques for the study of the composition of the translational machinery have been developed (49). These methods are particularly powerful for the identification and quantification of ribosome-associated proteins, which are more loosely attached to the ribosomal particles than r-proteins (49).

During the writing of this chapter, a review whose scope partly overlaps with that of this work appeared (110). The reader is referred to this review for a comprehensive account of ribosome-associated proteins known to interact with the bacterial ribosome during protein synthesis.

MODIFICATION OF THE TRANSLATIONAL APPARATUS IN STATIONARY PHASE

Unlike some bacterial species that differentiate into resistant structures such as endospores and myxospores in response to nutrient deprivation, E. coli and Salmonella enterica serovar Typhimurium do not undergo such morphological changes (62). When logarithmically growing E. coli cells exhaust an essential nutrient in the medium, they enter the so-called stationary phase of growth (43), in which the composition, structure, and function of the ribosomal particles become dramatically altered. Stationary-phase cells are highly resistant to a variety of environmental stresses (62, 73). This cross protection is largely the consequence of the binding of the transcription factor σ^S to the RNA polymerase (73). The expression of rpoS, the gene encoding σ^S, is regulated by the alarmone guanosine 3',5'-(bis)diphosphate (ppGpp), which accumulates following a variety of nutritional stresses (73).

In E. coli, a correlation between the number of functional ribosomes and cell viability during stationary phase has been observed (104, 106). To explain this correlation, the following hallmarks are worth mentioning. Entry into stationary phase results in a sharp decrease in the translational capacity of E. coli cells (58, 66, 104). Cell viability decreases to about 50% of the logarithmic phase value during the first 5 to 6 days in stationary phase (104). After this initial decrease in viability, sudden degradation of ribosomes rapidly ensues, accompanied by a dramatic drop in cell viability to less than 10% of the logarithmic phase value (104). This trend is not followed by Salmonella strains, which start degrading their ribosomes immediately after entry into stationary phase (44). Ribosome degradation in E. coli during stationary phase and following starvation for certain nutrients is in contrast to the situation during logarithmic growth, when ribosomes are stable (24). While ribosome degradation during stationary phase is an extreme example of alteration in the translational capacity of the cell, the issue will be just touched upon in this paper. The reader is referred to the following excellent reviews on the topic of ribosome degradation: references 23, 24, and 92.

Modification of the Normal Set of R-Proteins

One of the earliest examples of growth-related alterations in the compositions of ribosomes was the observation that the relative ratio of r-proteins L7 and L12 shifts upon entry into stationary phase. The two proteins have identical amino acid compositions, differing uniquely in the presence of an amino-terminal acetyl group on L7 (68). L7 and L12 constitute a stalk-like feature on the large subunit of the ribosome, where they are present in four copies (120). The stalk has been implicated in the binding of several translation factors to the ribosome and in the stimulation of GTP hydrolysis of these factors (22, 25). Ramagopal and Subramanian observed that the ratio of the large subunit proteins L7 and L12 shifts from 1:6 by mid-logarithmic phase to 3:1 in stationary phase (82). The variation in the L7/L12 ratio during growth appears to occur not through the modification of preexisting ribosomes but through changes in the relative assembly of L7 and L12 species onto ribosomes (81). These results imply that different subpopulations of ribosomes may exist during growth, that is, that ribosomes with both L7 and L12 would coexist with L12-only ribosomes during mid-logarithmic phase and with L7-only ribosomes during stationary phase (81). The increased L7/L12 ratio during stationary phase suggests that the accumulation of L7 on ribosomes may be of adaptive value for E. coli cells under conditions of starvation (80, 82). Although the reason for this accumulation remains obscure, more recent reports claim that it may be related to decreased peptidyltransferase activity during stationary phase (53). This idea must be reconciled, however, with results from structural studies showing that the L7-L12 stalk of the ribosome is neither functionally linked nor structurally proximal to the peptidyltransferase center (9, 42, 70). Therefore, a role for the altered L7/L12 ratio in modulating ribosomal activity awaits confirmation.

S22 is a protein with unknown function that is found in the 30S subunits of stationary-phase ribosomes (61, 104). The levels of S22 increase three- to fivefold in the stationary phase, going from 0.1 to ~0.2 copies per 30S subunit during logarithmic phase to a maximum of 0.6 to ~0.8 copies per 30S subunit in stationary phase and dropping suddenly once ribosome degradation starts (104, 106). Due to its association with ribosomes after isolation in high-salt buffers, S22 was classified as a legitimate r-protein (104). However, recent results showing that S22 is more abundant in 30S fractions than in 70S fractions suggest that S22 may not be a core component of translating ribosomes and that its classification as an r-protein should be revisited (49).

Wada et al. (106) have reported that other r-proteins undergo drastic modifications upon entry into stationary phase. For example, during stationary phase the multiple isoforms of small-subunit protein S3 disappear, L16 is cleaved, and L35 dissociates from the large subunit. Again, the biological significance of these alterations is unknown. What appears clear, however, is that the protein composition of stationary-phase ribosomes must be able to shift to that of their logarithmic-phase counterparts should the growth conditions improve. This shift may be achieved by the release of the stationary-phase forms of the r-proteins and/or by dilution with newly synthesized ribosomes.

Methylation of EF-Tu, Undermodification of tRNAs, and Accuracy of Decoding

Not only is the composition of the ribosome modified upon entry into stationary phase, but the modification of other components of the translational machinery, such as EF-Tu and tRNAs, has also been observed. EF-Tu becomes initially monomethylated at lysine 56 during logarithmic growth, and the same residue undergoes a second methylation when the cells enter stationary phase (102, 119). Since the modification decreases the tRNA-induced GTPase activity of the factor, it has been proposed that the methylation of EF-Tu would render aminoacyl-tRNA selection in the ribosome a more accurate process (102). However, observations of increased levels of frameshifts during entry into stationary phase are in contrast with this idea (10, 31, 89, 109). While the cause for this increase in frameshifts may be related to the undermodification of tRNAs during starvation (11), no direct evidence to prove this point has yet been collected (89). The reader is referred to the chapter by Björk and Hagervall within this work for a discussion on how the growth conditions influence the extent of tRNA modification in E. coli Chapter (Transfer RNA Modification).

Resting Ribosomes in Stationary Phase

Without a doubt, the most dramatic morphological alteration endured by the translational machinery upon entry into stationary phase is the conversion of 70S monomers into dimers, or 100S ribosome particles (63, 107). These particles begin to appear concomitantly with the transition from logarithmic to stationary phase, at which point they reach a level between 40 and 60% of the total ribosome population (99, 107). Stationary-phase dimers are different from those observed at high Mg²⁺ concentrations (reviewed in reference 101), because they are not dissociated by low Mg²⁺ levels and also because their formation requires a protein named ribosome modulation factor (RMF; molecular mass, 6,475 Da) (47, 104, 107, 118). The synthesis of RMF is positively regulated at the transcription level by ppGpp and is independent of the stationary-phase transcription factor σ^S (47, 115). Induction by ppGpp explains why RMF can also be detected during exponential phase at levels that are inversely correlated with the growth rate (47, 115). Moreover, it has been reported that acidic or hyperosmotic conditions during exponential growth further increase the synthesis of RMF (26, 27, 33).

RMF binds to the intersubunit face of the 50S particle, where it covers two vital regions of the ribosome, the peptidyltransferase center and the entrance to the ribosome exit tunnel (104, 117). However, electron micrographs of stationary-phase 100S dimers showed that they consist of two 70S ribosomes linked through the cytoplasmic faces of their 30S subunits (104), thus ruling out a possible role of RMF in bridging the two 70S ribosomes. To explain the apparent discrepancy between RMF’s binding site and the morphology of 100S ribosomes, it has been proposed that the binding of RMF to the 50S subunit induces a conformational change in the 30S subunit that allows the formation of the dimers (104). The details of this putative long-range conformational change remain, nevertheless, obscure.

Two more proteins expressed during stationary phase and related to the formation of the 100S ribosomes are the hibernation-promoting factor (HPF) and its paralog protein Y (PY). These proteins are the products of the yhbH and yfiA genes (encoding HPF and PY, respectively), with molecular masses of 10,750 and 12,653 Da, respectively (61, 99). In contrast to RMF, which is released from the ribosome with high salt, the binding of both HPF and PY to ribosomes is high-salt resistant, requiring subunit dissociation at low magnesium concentrations for their release (61). Despite the fact that HPF and PY show 40% identity in their N-terminal domains (PY has a C-terminal tail of 18 amino acids that is absent in HPF), only HPF is found in 100S dimers (61, 99). In contrast, PY associates solely with the 70S monomer pool, where it is found at almost one copy per 70S ribosome (61, 99). In fact, the two proteins seem to have opposite functions in 100S particle formation. HPF binds to an RMF-bound, immature form of 100S particles (90S ribosomes) and converts them into 100S particles, whereas PY stabilizes 70S monosomes and inhibits their dimerization (99). Interestingly, RMF and HPF bind to stationary-phase ribosomes with different stoichiometries. While two molecules of RMF are present in each dimer, one per 70S (106), the conversion of 90S ribosomes into 100S particles appears to require a single molecule of HPF (99).

The high degree of homology between HPF and PY, together with the fact that the copy number of either one of these proteins relative to the number of ribosomes increases when the other is knocked out, suggests that their binding sites in the ribosome overlap significantly (99). The structures of PY in solution (Fig. 1A) and in complex with the 70S ribosome (Fig. 1B, C, and D) have been determined by nuclear magnetic resonance analysis and X-ray crystallography, respectively (79, 103, 116). The protein is composed of an N-terminal globular domain with two α-helices packed against a four-stranded β-sheet in an overall β-α-β-β-β-α topology and a disordered 23-amino-acid C-terminal extension (Fig. 1A) (79, 116). After the docking of the nuclear magnetic resonance structure of PY into the low-resolution density maps obtained by X-ray crystallography, the globular domain of PY was found to block the binding sites for A- and P-site tRNAs within the 30S subunit of the E. coli 70S ribosome (103) (Fig. 1C and D). The proximity of several of the positively charged and conserved amino acids of PY to 16S rRNA in the crystal structure was consistent with the possible involvement of these residues in ribosome binding (103). Most of these residues are conserved in HPF, also supporting the hypothesis that this protein may bind to the ribosome in a similar fashion (99). In an extension of this idea, it has been proposed that the C-terminal tail of PY, absent in HPF, may be responsible for the inhibition of RMF binding to PY-bound 70S ribosomes (99). Since no density attributable to the C-terminal tail was observed in the crystal structure, this attractive hypothesis could, so far, not be confirmed structurally (103).

Fig. 1(A) Structure of E.coli PY (Protein Data Bank accession no. 1L4S). (B) Structure of the E.coli 70S ribosome in complex with PY (Protein Data Bank accession no. 1VOQ and 1VOR). A surface model of PY (red) bound in the intersubunit space of a 70S ribosome (30S subunit in yellow, 50S subunit in blue) is shown. (C and D) Overlap of the binding sites for PY and A- and P-site tRNAs in the 30S subunit. The structures of the A (cyan)-, P (green)-, and E (grey)-site tRNAs were docked onto the structure of the E.coli 30S subunit (yellow) bound to PY (red; Protein Data Bank accession no. 1VOQ). Shown are two different views of the intersubunit face of the 30S subunit, with the landmark features, direction, and magnitude of rotation indicated. (E) Structure of E.coli Hsp15 (Protein Data Bank accession no. 1DM9). (F) Structure of E.coli CspA (Protein Data Bank accession no. 1MJC). (G) Structure of E.coli IF1 (Protein Data Bank accession no. 1AH9). (H) Structure of E.coli RbfA (Protein Data Bank accession no. 1KKG).

Several factors may explain the observed sharp decrease in the translational capacity of E. coli cells upon entry into stationary phase (58, 86, 104). First, during stationary phase, accumulated ppGpp can compete with GTP for binding to IF2, resulting in a much less active form of IF2 (65). Second, the amount of IF3 in stationary cells has been shown to be drastically reduced relative to that in actively growing cells (85). Third, ppGpp also induces the synthesis of RMF, whose blocking of the peptidyltransferase center and the entrance to the ribosome exit tunnel blocks protein synthesis in a dose-dependent manner (104, 117). Fourth, the blocking of the 30S A and P sites by PY detected in vitro has also been shown to prevent 70S ribosomes from engaging in protein synthesis (3, 103). Finally, HPF has also been shown to inhibit translation in vitro in an initiation independent translation system (98a). Despite all this in vitro evidence implicating RMF, PY, and HPF in the protein synthesis arrest observed during stationary phase, their potential inhibitory role in vivo remains to be elucidated.

The presence of PY and HPF homologs appears to be extended among bacteria, whereas RMF is present only in the gamma proteobacteria group (98a, 99). PY and HPF homologs can be found also in the chloroplasts of higher plants but not in but animals, protists, fungi and archaea (14, 50, 94, 98a). Despite the conservation of both PY and HPF across the bacterial branch of the phylogenetic tree, studies with single- and double-deletion mutants showed no significant growth defects relative to the growth of wild-type cells (8, 13, 99). In contrast, RMF deletion mutants showed decreased viability relative to wild-type cells during stationary phase, indicating that RMF may be essential for survival during periods of decreased growth (6, 115). This finding has been corroborated by Wada and co-workers, who described a close correlation between the loss of 100S dimers in cell lysates and bacterial cell death (104, 106). These studies showed that cell viability is maintained at about 50% during the first few days in stationary phase, at which point 100S dimers are the major species in cell lysates (104, 106). Following this decrease in viability, RMF is abruptly released, the 100S particles dissociate into 70S ribosomes, and cell viability drops to about 10% (104). The subsequent decomposition of ribosomes into RNA and protein results in a further drop in viability (104). Taken together, all these data clearly link RMF to survival in stationary phase. However, it remains to be proven that 100S particles do form in vivo (72), as the formation of 100S ribosomes has been observed only in stationary-phase lysates from cells grown at near-neutral pH (47) but not in those from cells grown under acidic conditions (26, 27) or in those from logarithmic-phase cells growing at low rates (47, 115), despite the presence of RMF in their cytoplasm. These data have prompted some authors to question the biological function of the 100S particles (72).

The claimed essentiality of RMF during stationary phase has also been questioned by a recent report arguing that RMF is dispensable for the growth of E. coli cells during both logarithmic and stationary phases (13). According to the authors of this report, the only advantage of possessing RMF is the ability of wild-type cells to outcompete RMF knockout mutants in mixed cultures (13). This effect is even more apparent when a quadruple mutant devoid of PY, HPF, and SRA (see below), in addition to RMF, is used in the competition experiments (13). While the reasons for the contradictory results regarding the essentiality of RMF remain unclear (6, 13, 104, 106, 115), the outcome of the competition experiments suggest that the roles of the stationary-phase proteins may not always be apparent under the growth conditions normally used in the laboratory. The expression of these proteins may provide a selective advantage that is evident only in natural ecosystems where there is strong competition for common resources. Indeed, the presence of RMF and 100S ribosomes during stationary phase in lysates from natural isolates of E. coli has been corroborated (106).

HPF, PY, and RMF may also be related to the general protection against environmental stress observed in stationary-phase E. coli cells (62, 73), a role that would not be revealed necessarily by the viability assays (6, 13, 106, 115). For example, given the known correlation between the number of functional ribosomes and cell viability during stationary phase (104, 106), these proteins may serve to protect stationary-phase ribosomes against degradation. Biochemical data obtained in vitro for PY and RMF showed that this scenario may be the case, as both proteins protect some of the most important functional areas of resting ribosomes from chemical attack (103, 117). In vivo evidence for such a role for RMF also exists. The factor has been shown to protect stationary-phase ribosomes against degradation during heat shock (72) and acid stress (26) and during spermidine accumulation in spermidine acetyltransferase-deficient mutants (32). Taken together, these observations imply an important role for RMF in cross protecting E. coli cells during stationary phase (62, 73). Moreover, PY is also expressed in the cold, where it is believed to function during the cold shock response (see below) (2, 103). It is possible that as part of their roles in cross protection, PY, RMF, and HPF function also during stationary phase to maintain optimal amounts of both forms of resting ribosomes (PY-bound 70S monomers and HPF- and RMF-bound 100S dimers), so that the cell would be preadapted to additional stress, such as an unexpected temperature drop.

In summary, the expression of the ribosome-associated proteins RMF, PY, and HPF during stationary phase appears to be related to the observed decrease in protein synthesis and to the protection of idle 70S and 100S ribosomes. Therefore, the expression of these proteins may contribute to the cross protection of stationary-phase cells against additional environmental insults. The reversible formation of resting 100S ribosomes during stationary phase has been termed the hibernation stage of the ribosome cycle (118). Clearly, for hibernation to be advantageous to the cell, a mechanism should exist that ensures fast adaptation of the translational machinery to a sudden restoration of the normal growth conditions. Several pieces of evidence suggest that this is indeed the case. First, 100S dimers have been observed to disappear quickly from lysates when stationary-phase cells are transferred into fresh medium (61, 107, 118). The loss of 100S dimers probably occurs in parallel with the release of RMF and HPF from their binding sites. However, no direct experimental evidence exists to prove this point. In addition, the quick release of PY by the action of initiation factors has been shown in vitro, strongly suggesting that PY may dissociate from ribosomes upon resumed translation after transfer into fresh medium (103). A model for the cycle of proteins HPF, PY, and RMF during growth is schematically shown in Fig. 2.

Fig. 2Shifts in the pool of active and inactive ribosomes upon entry of E.coli into stationary phase. (1) During exponential growth, most ribosomal particles are involved in protein synthesis and are found in the polysome pool, shown as strings of ribosomes at different stages of translation. P, A, and E sites on polysome ribosomes are indicated; mRNA is shown in red, with the 5'-AUG-3' start codon indicated; and tRNAs and the nascent polypeptide chain are shown as indicated in the symbol legend on the right. In addition to polysomes, a small pool of free subunits and 70S ribosomes exists in the cell (the sedimentation coefficients are indicated for clarity). (2) In the transition to stationary phase (downward arrow from panel 1 to panel 3), proteins RMF (yellow square), PY (green), and HPF (shown as a red, tailless version of PY) are synthesized. (3) The binding of these proteins to ribosomes decreases the number of ribosomes found as polysomes. RMF-containing 90S particles appear at this stage. (4) The levels of the three proteins increase in stationary phase (downward arrow from panel 3 to panel 5), at which point (5) almost all the existing ribosomes are present as either HPF- and RMF-bound 100S particles or PY-bound 70S ribosomes. After a few days in stationary phase, the ribosomes are gradually degraded (not shown). (6) If optimal growth conditions are restored (upward arrow from panel 3 to panel 1), all three proteins are rapidly ejected from the ribosomes, which then become engaged in protein synthesis; i.e., they are found in polysomes.

SRA

Another protein, termed SRA (stationary-phase induced ribosome associated), that strongly associates with the small subunit of stationary-phase ribosomes has been described previously (48). Should it be confirmed to be an integral r-protein, SRA would be the smallest of all r-proteins (molecular mass, 5,096 Da) (48). The fact that SRA knockout mutants display neither a significant growth difference nor an altered ribosome morphology compared to wild-type cells leaves the question of its functional role unanswered (8, 13, 48).

COLD SHOCK

The ribosome has been suggested to play a central role as the temperature sensor of the cell (100). However, the underlying mechanism for this ribosomal activity remains unknown. What is clear, though, is that the function and the composition of the translational machinery are dramatically affected by temperature, in particular by temperature downshifts. When E. coli cells are shifted from the normal growth temperature of 37°C to temperatures below 20°C but above the minimum temperature of growth of 7.5 to 7.8°C (71, 91), they enter a period of cold adaptation or acclimation. During this period, protein synthesis is inhibited for all but a subset of proteins named cold shock factors (CSFs). It appears that the synthesis of CSFs is essential for the resumption of growth in the cold at the end of acclimation, about 4 to 6 h after the temperature downshift (35, 36, 40). At this point, the situation is reversed and the synthesis of bulk proteins resumes while that of CSFs becomes inhibited (40). While the specific roles of most CSFs remain unclear, it has been proposed that some of them target the translational apparatus (35). CSFs known to directly affect ribosomal function in the cold are PY (2, 103) and the three initiation factors (35, 51). The details of the interaction of initiation factors with the ribosome during general translation have been described elsewhere (references 5 and 69 and references therein). Therefore, only the implications of the increased synthesis of initiation factors during cold shock will be discussed here. Other proteins with proposed but unclear roles during translation in the cold are CspA (cold shock protein A) (75), RbfA (ribosome binding factor A) (51, 114), and DeaD, also named CsdA (cold shock DEAD box protein A) (52).

PY and Initiation Factors

Numerous reports support the idea that protein synthesis during acclimation is arrested by a block of translation initiation (12, 29, 51, 60). Initiation involves the placement of the AUG start codon of the mRNA at the P site of a 30S subunit, the recognition of this codon by the initiator fMet-tRNA^fMet in the presence of the three initiation factors (IF1, IF2, and IF3), and finally, the joining of the 50S subunit with the subsequent release of initiation factors. Until very recently, the only initiation step reported to be cold sensitive was the transition from a 30S to a 70S initiation complex (60). Results from more recent studies showing a temporal coincidence between PY accumulation within the first hour following a temperature downshift (2) and the arrest of translation initiation following cold shock, together with evidence demonstrating an inhibitory effect of PY on P-site tRNA binding (103), are suggestive of a potential role for PY as an inhibitor of translation initiation. Initial evidence establishing PY as a ribosomal inhibitor was obtained by showing that the protein can interfere with the binding of A-site tRNA (1, 3). This finding led to the idea that the main role of the protein was to inhibit ribosomal decoding (3). As a result, the protein was renamed RaiA (ribosome-associated inhibitor A) (3). A structural explanation for this observation was recently provided by results from a crystallographic study showing partial blocking of the 30S subunit A site in the presence of PY (103) (Fig. 1C and D). However, the authors of this study also reported that in the PY-70S ribosome cocrystals, the protein completely blocks the space normally used by the anticodon stem loop of P-site tRNA (103) (Fig. 1C and D), an observation that was biochemically confirmed by showing the inhibition of P-site tRNA binding to the 70S ribosome (103). Since filling of the ribosomal P site precedes the delivery of aminoacyl-tRNA at the A site, it was reasoned that the primary effect of PY is most likely the inhibition of P-site tRNA binding and that the reported inhibition of A-site tRNA binding (3) is an indirect result of PY binding to the P site (103). This proposal appears to be consistent with the absence of PY at the time when most of the cellular ribosomes are involved in translation elongation, that is, when the A site of the ribosome is being loaded with aminoacylated tRNA by EF-Tu. For example, PY is not present in polysomes isolated from exponentially growing cells at the normal growth temperature (61) or in those isolated from cold-shocked cells in the growth phase following acclimation (2).

The possibility that PY may interfere with translation initiation in the cold was confirmed by in vitro experiments showing that, in the presence of the initiation factors, PY is able to compete with the initiator tRNA for binding to the ribosome at low temperatures but not at 37°C (103). Should this ability of PY be confirmed in vivo, the protein would be demonstrated to have a direct role in the general arrest of translation observed during acclimation by blocking the formation of initiation complexes. In addition, the reported subunit association function of PY (4) stabilizes a substantial number of ribosomes, up to one-third of the entire pool (2, 4), as resting 70S ribosomes. This function of PY indirectly contributes to the general arrest of protein synthesis by storing a substantial fraction of 70S ribosomes in an inactive form.

The accumulation of resting 70S ribosomes in the cold has been reported on numerous occasions (references 12, 19, 51, and 98, but see also references 28 and 30). This accumulation is due only in part to PY’s subunit association activity, as low temperatures are known to strongly shift the subunit-70S ribosome equilibrium towards 70S monosome formation (34, 37, 121). Indeed, some authors believe that these two effects combined may make the amount of ribosomal subunits that are available for initiation too low to support the translation of CSF mRNAs (37). Recent evidence demonstrating a special role for initiation factors during acclimation explains how E. coli cells can surmount the strong bias towards 70S ribosome formation in the cold, thus allowing the translation of CSF mRNAs (34, 35, 37). The increased expression of initiation factors shortly after cold shock, which result in a transient imbalance of the constant ratio of initiation factors to ribosomes (~0.15 during normal growth versus ≥0.3 during acclimation) (35), facilitates the translation of CSF over non-CSF mRNAs via two mechanisms: (i) indirectly by counteracting the accumulation of 70S monosomes in the cold and (ii) directly by discriminating against non-CSF mRNAs. The elevated levels of synthesis of IF1 and IF3 are required to overcome the strong shift of the subunit-70S ribosome equilibrium in the cold and, as a consequence, to release ribosomal subunits for the translation of CSF mRNAs (34, 37). While this effect is related mostly to the well-known ribosome-dissociating activity of IF3 (39), the contribution of IF1 appears to be particularly important at low temperatures (34, 37). Interestingly, the subunit association activity of PY in vitro can effectively compete with ribosome dissociation by IF3 (103), raising the interesting question of why the cell would synthesize two proteins with opposing functions as part of the cold shock response. Solving this riddle would certainly require a comparison of the accumulation rates of the two proteins following cold shock, and to assess the performance of ribosomes when both initiation factors and PY are present at biologically relevant ratios.

As for the second mechanism, that of the discrimination in favor of CSF mRNAs during acclimation, IF3 appears to bear most of the responsibility (35, 37). For many years, the only element that could be linked to the observed discrimination in favor of CSF mRNAs was their 5' untranslated region, believed to stabilize the interaction of these transcripts with the ribosome (35). New evidence shows that IF3 can directly recognize CSF and non-CSF mRNAs and induce the formation of nonproductive 70S initiation complexes on the latter transcripts, thus directly discriminating against their translation (37). Notably, this cold-related function of IF3 reaches its maximum efficiency at stoichiometric levels similar to those observed during acclimation (37). How CSF and non-CSF mRNAs are recognized by IF3 in the ribosome remains obscure, however. What is clear, though, is that as a result of the increased levels of initiation factors and, in particular, of the new activities of IF3 in the cold, CSF mRNAs are preferentially translated during acclimation (37). The translation of non-CSF mRNAs occurs in cold-adapted cells undergoing postacclimation growth (51), raising the question of how the non-CSF mRNA transcripts are able to overcome the discriminatory effect of IF3. A possible explanation is given by the observation that the pre-cold shock ratio of ribosomes to initiation factors is restored shortly after the end of acclimation (34) and the discriminatory ability of IF3 against non-CSF mRNAs should no longer be relevant at the restored ratio (37).

The release of PY from 70S ribosomes has also been detected at a time that likely coincides with the end of acclimation (2). While the identity of the factor(s) that triggers the release of PY remains unknown, this release likely contributes to the increase in postacclimation polysomes (51). It has been reported that the amount of ribosomes that are recruited into these polysomes is proportional to the magnitude of the temperature downshift (98). This relationship may mean that upon growth resumption in the cold, not all resting 70S ribosomes will become active in protein synthesis and that a fraction of PY-bound 70S ribosomes will remain. The proven ability of the initiation factors to compete with PY at 37°C in vitro (103) suggests that PY-sequestered ribosomes would be made immediately available for translation should the normal growth temperature be restored (103). Thus, while PY can block protein synthesis in the cold, directly by inhibiting initiation complex formation and indirectly by stabilizing 70S monomers, it also provides a mechanism to ensure that the cell can rapidly respond to a sudden improvement of the environmental conditions. Similarly, the discriminatory function of IF3 in the cold has not been detected at 37°C (37), suggesting that if a culture undergoing cold acclimation is rapidly shifted back to 37°C, excess IF3 should not impair growth by preventing the translation of regular mRNAs. A model for the cycle of PY and initiation factors during growth is schematically shown in Fig. 3.

Fig. 3Inhibition of translation initiation during cold shock. (1.1) Normal initiation occurs at 37°C (the normal growth temperature is shown as a yellow background). (1.2) The 30S subunit (with the A-, P-, and E-site tRNA binding sites indicated) binds the mRNA (with the 5'-AUG-3' start codon indicated); the three initiation factors IF1, IF2-GTP in complex with fMet-tRNA^fMet, and IF3; and the 50S subunit to form a 70S initiation complex (the previous formation of a 30S initiation complex is not shown). (1.3) Polysomes form during elongation when several ribosomes translate the same mRNA (the peptide chain is shown as a string of white ovals). (2.1) Cold shock acclimation (the low temperature is indicated by a cyan background). Upon cold shock, the subunit/70S equilibrium shifts towards the formation of 70S ribosomes or monosomes (sedimentation coefficients are indicated for clarity). Shortly after cold shock, during acclimation, initiation factors (2.2) and PY (3.1) are expressed. Increased levels of initiation factors are required to dissociate cold-induced 70S ribosomes into the 30S and 50S subunits (2.3a and b) needed to form 30S initiation complexes both on CSF mRNAs (2.4) and on non-CSF mRNAs (2.7). 30S initiation complexes formed on CSF mRNAs (2.4) will result in productive 70S initiation complexes that can proceed to the elongation phase of translation, thus allowing the formation of polysome structures (2.5). These polysomes synthesize CSFs (2.6). This situation is in contrast to that for 30S initiation complexes formed on non-CSF mRNAs (2.7), which due to the discriminatory ability of IF3 in the cold, give rise to nonproductive 70S initiation complexes unable to proceed to the elongation phase of translation (indicated by a stop sign in panel 2.8). In the PY cycle, cold-induced 70S ribosomes (2.1) and free ribosomal subunits (2.1 and 2.3b) are substrates for the 70S ribosome-stabilizing function of the newly synthesized PY (3.1). As a consequence, PY-bound resting 70S ribosomes start accumulating in the cell (3.2). PY is known to compete in the cold with initiation factors and initiator tRNA for binding to ribosomes. As a result, PY may reduce the accumulation of stalled 70S initiation complexes formed on non-CSF mRNAS (3.3) by displacing initiation factors and initiator tRNAs (3.5). Since, in principle, PY could disrupt the process of initiation complex formation on any mRNA, this possibility is noted in panel 3.4, where red mRNAs represent both CSF and non-CSF transcripts being recruited into initiation complexes. The disruption of initiation complexes by PY in the cold would increase the fraction of PY-bound resting 70S ribosomes (3.2). (4.1) Post-acclimation phase (the end of acclimation is indicated by a broken line). Upon growth resumption, at the end of acclimation, PY is released from resting 70S ribosomes and the pre-cold shock 70S ribosome-initiation factor balance is restored (not shown). The factor(s) triggering PY release remains unknown (as indicated by red question marks). (4.2) At the same time, the decreased levels of initiation factors allow polysome formation on non-CSF mRNAs, thus permitting cell growth in the cold. Also shown in panel 4.2 is the contribution of some CSFs to the translation of bulk mRNAs by acting as chaperones and/or helicases. (5.1) Return to 37°C during acclimation. Should the normal growth temperature be suddenly restored during acclimation , initiation factors can rapidly eject PY from the PY-bound, resting 70S ribosomes. (5.2) Because the discriminatory activity of IF3 against non-CSF mRNAs is no longer relevant at 37°C, initiation complexes can rapidly form on regular mRNAs, which are now competent to proceed to the elongation phase of translation.

Association of Other CSFs with the Ribosome

It has been shown previously that ribosomes from cold-shocked cells show certain preferences for cold shock mRNAs even when stripped of all loosely associated proteins (35, 40). Evidence showing that the composition of cold-shocked ribosomes is altered relative to that of their 37°C counterparts has been gathered recently by mass spectrometry studies. These studies showed that ribosomes obtained from cells grown at 16°C lack large-subunit proteins L1, L2, L3, L4, L22, and L23 (49). In addition, lower levels of free 50S subunits than of 30S subunits in cold-shocked cells have been reported (17, 18, 49). How these alterations in the composition of the translational machinery relate to the observed preference for cold shock mRNAs in the cold remains, however, unclear.

A few CSFs, besides PY and initiation factors, have been proposed to associate with the ribosome during translation in the cold. One group of CSFs contains the major cold shock protein CspA (Fig. 1F) and some of its structural paralogs, named CspB to CspI in E. coli (75). CspA has been shown to have a favorable effect on the translation of cold shock mRNAs during acclimation (35). However, this result is likely an indirect effect due to the ability of the Csp family of proteins to bind and melt nucleic acids, thus aiding the translational machinery by opening an extra stable mRNA secondary structure in the cold (40, 75). This RNA-chaperoning activity of Csp family proteins is also used for transcription antitermination (75), which is likely the main function of these proteins, as it is the only activity of the cold-induced Csp homologs known to be essential during cold shock (76).

The common fold of CspA homologs is very similar to the S1 domain structure (87, 88), which is also present in other nucleic acid binding proteins, such as IF1 (Fig. 1G) (90). As mentioned above, IF1 levels are increased upon cold shock (34, 35), and some mutations in the infA gene coding for IF1 result in cold sensitivity (20), further strengthening the possible role of S1 domain proteins in cold acclimation. An intriguing connection between IF1 and CspA homologs has been uncovered in studies performed with Bacillus subtilis (108), in which E. coli IF1 functionally replaced the Bacillus CspA homologs in the cold (108). While these results would be consistent with a possible role of the Csp family of proteins in translation, more recent data showing that IF1 can also function as an RNA chaperone and promotes antitermination (77) provide a more likely explanation for the replacement studies (108).

The cases of the CSFs RbfA and DeaD (or CsdA) are similar in that both have been proposed to function in translation in the cold but definitive proof for such a role has not yet been obtained. The rbfA gene encodes a 133-amino-acid protein (~15 kDa) with strong fold homology to K homology domains (45) (Fig. 1H). Its original isolation as a suppressor of a dominant cold-sensitive 16S rRNA (C23U) mutation (21) is consistent with its proven role in 16S rRNA maturation (15). While the protein has been shown to bind directly to 30S subunits (114), the available data do not so far support a role for RbfA during translation in the cold. For example, the observed constitutive induction of the cold shock response in both the C23U mutant and an RbfA knockout mutant strongly suggests that the role of RbfA in the cold is merely to aid in the processing of 16S rRNA (51). This conclusion is supported by the finding that the removal of the 25-amino-acid C-terminal extension of RbfA inhibits its association with the 30S subunit but not its ability to function in the maturation of 16S rRNA or to allow E. coli cells to survive in the cold (45, 114).

The deaD/csdA gene was initially identified as a suppressor of mutations in the gene encoding r-protein S2 (rpsB) (96). CsdA was established as a CSF due to its expression in the cold (52) and to the fact that the deletion of its gene results in a cold-sensitive phenotype (17). CsdA is a 64-kDa protein with a high degree of similarity to the members of the DEAD family of RNA helicases (96). Recent mutational analyses have confirmed the function of CsdA as an RNA helicase in the cold both in vitro and in vivo (97). CsdA has been proposed to have a role in the biogenesis of the small subunit (67), which would be consistent with its identification as a suppressor of rpsB mutations (96). However, more recent data have challenged this idea by showing that 30S subunits from deaD null mutants do not reveal any processing defects (17). The authors of this study showed that the protein can be found associated with precursors of the 50S subunit during cold shock (17), thus implicating the protein in the maturation of the large subunit. Other functions proposed for CsdA are mRNA stabilization (46), interaction with poly(A) polymerase (83), and participation in the cold shock "degradosome" (78). A role during translation for CsdA, in which the protein would assist the formation of initiation complexes on templates with high levels of secondary structures, has also been proposed (59). As for the possible association of CsdA with mature ribosomes, however, the data are contradictory (40, 52), thus leaving open the question of whether or not this protein can be considered a legitimate ribosome-associated factor.

HEAT SHOCK

Early studies found that the thermal stability of ribosomes from different bacterial species correlates positively with the maximal growth temperatures of the organisms (74). Moreover, ribosome degradation has been proposed as the cause of cell death in heat-stressed bacteria (72, 95). Therefore, one would expect that high temperatures could interfere with normal ribosomal function and that mechanisms to prevent this interference should have been evolved. It has been mentioned already that the induction of RMF in stationary phase protects ribosomes against degradation by heat shock (72). However, this protective effect of RMF induction is part of the enhanced resistance to environmental stress acquired during stationary phase and unrelated to the heat shock response, as heat shock induction of RMF has not been observed.

E. coli responds to temperature upshifts within the temperature range of 30 to 42°C with the rapid induction of a number of protein chaperones and proteases globally known as heat shock proteins (HSPs) (7). The induction of HSPs is positively regulated by the product of the rpoH gene, which encodes the σ³² subunit of the RNA polymerase (7). So far, the only HSP which can be considered a ribosome-associated factor is Hsp15. This protein has general affinity for nucleic acids and, in particular, for the 50S ribosomal subunit (56, 57). Hsp15 is induced 43-fold upon heat shock, thus making it one of the most highly heat-inducible genes in E. coli (84). The structure of Hsp15 is composed mainly of the so-called αL motif, shared by a family of proteins with more than 500 members that include r-protein S4 (Fig. 1E) (93). Interestingly, Hsp15 was found to be associated with 50S subunits carrying a nascent peptide chain, a ribosomal species whose accumulation is favored by heat shock conditions (56). The authors of the study proposed that these 50S subunits would be the result of erroneously dissociated, elongating ribosomes and that the protein would be important for the recognition and repair of these subunits (56). Hsp15 homologs were found in both gram-negative and gram-positive bacteria, indicating that the function of Hsp15 is conserved in eubacteria (57).

CONCLUSIONS

Learning about the mechanisms used by microbes during the stress response is crucial if one desires to control these organisms within their natural niches and in their most resistant state. The last few years have witnessed great progress towards expanding our understanding of how E. coli modulates protein synthesis as part of its response to sudden changes in the environment. In particular, the involvement of small RNA binding proteins as ribosome-associated factors during stress appears to be a powerful way to induce drastic changes in the translational capacity of the cell. In addition, these proteins may have a role in protecting the ribosome against environmental insults, thus contributing to the acquisition of a higher resistance state during stress. It is likely that the known number of such factors will increase in the near future, as new candidates are already emerging (48).

Despite this progress, many unanswered questions still exist. For example, while it is clear that certain stressful conditions result in the modification of the ribosome itself (e.g., a change in the L7/L12 ratio or alterations in the primary structures and the abundance of r-proteins) or of general translation factors (e.g., the methylation of EF-Tu, increased synthesis of initiation factors, or the undermodification of tRNAs), the reasons behind these changes are, in most cases, completely unknown. Even for the best-characterized ribosome-associated factors induced under stress (RMF, PY, and initiation factors), we are far from a complete understanding of their modes of action. For example, their role in vivo and how their presence results in an advantage to the cell remain to be elucidated.

ACKNOWLEDGMENTS

I acknowledge Anna Maria Giuliodori and Claudio O. Gualerzi for sharing their unpublished results, Alma Leticia Avila-Juarez, Gary R. Janssen, Michael O’Connor, Raj Pai, and Barbara S. Schuwirth for interesting discussions, and Fidedigna Vila-Sanjurjo for assisting with balance and coherence.