Gateway Role for rRNA Precursors in Ribosome Assembly

Abstract

In Escherichia coli, rRNAs are initially transcribed with precursor sequences, which are subsequently removed through processing reactions. To investigate the role of precursor sequences, we analyzed ribosome assembly in strains containing mutations in the processing RNases. We observed that defects in 23S rRNA processing resulted in an accumulation of ribosomal subunits and caused a significant delay in ribosome assembly. These observations suggest that precursor residues in 23S rRNA control ribosome assembly and could be serving a regulatory role to couple ribosome assembly to rRNA processing. The possible mechanisms through which rRNA processing and ribosome assembly could be linked are discussed.

INTRODUCTION

The assembly of ribosomes from their individual RNA and protein components is an elaborate process that involves dozens of macromolecular components. Despite its complexity, this process proceeds rapidly and is normally completed within minutes in many prokaryotic organisms (7, 12, 19). During the assembly process, rRNAs also undergo processing, with the result that the newly assembled ribosomes generally contain rRNAs with mature 3′ and 5′ ends.

Ribosome assembly and rRNA processing have been extensively studied in Escherichia coli, an organism that has served as a model for defining many aspects of these processes. In E. coli, rRNAs are transcribed as a single transcript that encompasses each of the three rRNAs: 16S, 23S, and 5S (6, 27). The processing of this transcript to generate the final mature products requires the action of multiple RNases. The initial processing steps are carried out by RNase III, a double-strand-specific RNase that cleaves precursor residues between 16S and 23S rRNAs and between 23S and 5S rRNAs, resulting in the separation of regions that correspond to the individual rRNAs (8). The resulting molecules still contain precursor residues at both the 5′ and 3′ ends. Their removal requires several additional RNases, some of which are known and others that still remain to be identified (6).

The processing of rRNA can be retarded under a variety of circumstances that interfere with rapid ribosome assembly, including conditions of slow growth, reduced growth temperature, and defects in either ribosomal factors or in nonribosomal factors that normally promote ribosome assembly (17, 18, 25). How delayed ribosome assembly affects rRNA processing is not completely understood, but presumably it is due to a need for rRNAs to adopt specific conformations during ribosome assembly that are recognized by the processing RNases. To the best of our knowledge, a related question has not been previously addressed: do defects in rRNA processing themselves affect ribosome assembly? With the recent identification of the RNases that process 23S rRNA subsequent to RNase III cleavage (10), we have addressed this question. Here, we show that the presence of as few as three unprocessed residues at either the 3′ or the 5′ end of 23S rRNA interferes with ribosome assembly. The possible mechanisms through which such precursor sequences could regulate ribosome assembly are described.

MATERIALS AND METHODS

Strains.

The wild-type strain used in this study was MG1655*, a derivative of the sequenced MG1655 strain that corrects a frameshift mutation in the rph gene (2). Derivatives of MG1655* containing nonpolar Δrph, Δrnb, Δrnt, and/or Δrng alleles were constructed by transduction of deletion alleles that are interrupted by a kanamycin resistance (KAN^r) marker (1), followed by removal of the KAN^r marker by recombination between the frt sites that flank the gene for the KAN^r marker in each case.

RNA isolation and analysis.

RNA was isolated and analyzed using methods described previously (9, 29). Analysis of 3′-end processing of 23S rRNA was performed using oligonucleotide-directed RNase H cleavage of 23S rRNA 50 nt from the mature 3′ end, as described previously (9). Primer extension of 16S rRNA was performed on total cellular RNA using a 5′-labeled oligonucleotide with the sequence 5′ CGTTCAATCTGAGCATGATC 3′. Primer extension of 23S rRNA was performed using a 5′-labeled oligonucleotide with the sequence 5′ CCTTCATCGCCTCTGACTGCC 3′.

Ribosome analysis.

To analyze the incorporation of newly synthesized labeled rRNA into ribosomes, saturated overnight cultures were separately diluted 50-fold into 1.5 or 120 ml of LB at 37°C and grown to an A₆₀₀ of 0.3 to 0.4. Eight μCi of ³²P-labeled inorganic phosphate was added to the small-scale culture, and incubation was continued for 2 min. The labeled cultures were then rapidly centrifuged and resuspended into 1.5 ml of fresh prewarmed LB. Cells were incubated at 37°C with shaking, and 0.5-ml aliquots were removed at the indicated times. The cells were rapidly centrifuged and cell pellets were frozen on a dry ice-ethanol bath and stored at −80°C. Separately, the large-scale cultures were harvested at mid-log phase to provide a source of unlabeled ribosomes.

For ribosome analysis, labeled and unlabeled cell pellets were resuspended in buffer A (10 mM Tris, pH 7.5, 60 mM KCl, 10 mM MgCl₂) on ice, lysed, and centrifuged in a Microfuge. The resulting lysates were combined and layered onto a 14 to 32% sucrose gradient, followed by ultracentrifugation at 21,000 rpm in a Beckman XL-80 Ultracentrifuge using an SW28 rotor for 19 h. After centrifugation, the gradients were analyzed with a UV monitor (Pharmacia UV-M) to identify the ribosomal peaks. Sixteen ribosomal fractions were collected in each case, and equal volumes of each fraction were filtered through a nitrocellulose membrane (Whatman Protrans BA 85) using a dot blot manifold (Whatman) to trap the labeled rRNA present in ribosomal complexes. The incorporation of the radiolabel in each fraction was determined by using a Molecular Dynamics Storm 840 PhosphorImager. Each set of experiments was performed 2 to 5 times.

For the analysis of RNA from ribosomal and polysomal fractions, cells were grown at 37°C to mid-log phase and chloramphenicol was added to 100 μg/ml immediately prior to harvesting cells in order to stabilize polysomes. Cell extracts derived from the cells collected were layered onto a 14 to 40% sucrose gradient, followed by ultracentrifugation at 21,000 rpm for 10 h in a Beckman XL-80 Ultracentrifuge using an SW28 rotor. Ten to 13 fractions were collected, and RNA was extracted from 50S, 70S, and polysomal fractions by phenol extraction and ethanol precipitation. The RNAs were used for primer extension or for Northern blot analysis following oligonucleotide-directed RNase H cleavage of 23S rRNA (9).

RESULTS

An absence of RNase PH and RNase II causes ribosome assembly defects.

Subsequent to rRNA cleavage by RNase III, the removal of the remaining eight precursor nucleotides at the 3′ end of 23S rRNA involves a combination of 3′ to 5′ exonucleases (Fig. 1A). We have recently shown that RNase PH and RNase II are the major enzymes involved in this process, and an absence of both enzymes results in an accumulation of precursor 23S rRNA (10) (Fig. 1B). To investigate the possible consequences of RNase PH and RNase II absence, we analyzed the ribosome profiles of strains lacking these enzymes using sucrose density gradient ultracentrifugation. In a wild-type strain, most of the ribosomal particles were present as 70S ribosomes with low levels of the 30S and 50S ribosomal subunits observed, a result typically obtained under conditions of rapid ribosomal assembly (Fig. 1C). However, when a Δrph Δrnb strain was analyzed, the abundance of the 30S and 50S subunits was found to be significantly higher. No such accumulation of subunits was observed in strains lacking either RNase PH or RNase II, each of which harbors low levels of unprocessed 23S rRNA (Fig. 1B). These observations suggest that the accumulation of subunits may be related to the increased abundance of rRNA precursors caused by an absence of both RNases.

Fig 1.

Mutations in RNase PH and RNase II impair 23S rRNA processing and result in elevated levels of ribosomal subunits. (A) Schematic description of the steps involved in the processing of 23S rRNA. The first step involves RNase III cleavage of the 23S precursor just outside the mature 5′ and 3′ ends of this RNA. The primary cleavage sites are located 8 nt downstream of the mature 3′ end and 7 nt upstream of the mature 5′ end. Further processing of the 3′ end involves a combination of exo-RNases, especially RNase PH and RNase II, and, to a lesser degree, polynucleotide phosphorylase (PNPase) (10). The final step of 3′-end maturation, which involves the removal of the last 1 to 4 nt (shown in gray), is carried out by RNase T, an enzyme that is particularly effective at removing residues close to regions of the RNA duplex, as are found in mature 23S rRNA. 5′-End maturation occurs independently and is presumably carried out by an endonuclease that remains to be identified. (B) 3′-End analysis of 23S rRNA. 23S rRNA present in total RNA preparations from a wild-type strain or from strains containing Δrph, Δrnb, or Δrph Δrnb mutations was cleaved 50 nt upstream of the mature 3′ end, followed by fractionation of the resulting products on a denaturing gel and Northern blot analysis using a labeled oligonucleotide probe complementary to the end of 23S rRNA. The positions of the mature rRNA 3′ end (M) and a precursor that contains eight unprocessed nucleotides (+8) are indicated. The ratio of the levels of the +8 precursor to the mature RNA is indicated at the bottom. (C) Ribosome analysis. Cell lysates were prepared from the indicated strains grown to mid-log phase at 37°C. Ribosome profiles were generated at 254 nM following ultracentrifugation of clarified cell extracts on a 14 to 32% sucrose gradient under associative conditions (10 mM Mg²⁺), as described previously (11). The positions of the 70S ribosomes and of the 50S and 30S subunits are indicated. The means and standard errors for the amounts of ribosomal material present in the 70S fraction and in the combined subunit fractions are indicated. A significant increase of subunits was observed in the Δrph Δrnb strain compared to the wild-type or the single-mutant strains (P < 0.005 by one-tailed Student's t test).

Delayed ribosome assembly in strains lacking RNase PH and RNase II.

Increased levels of subunits, as observed in the Δrph Δrnb strain, can be symptomatic of defective or slowed ribosome assembly. To determine directly whether ribosome assembly is affected in strains lacking RNase PH and RNase II, strains containing or lacking these enzymes were pulse labeled during growth with radiolabeled inorganic phosphate. After labeling, cell extracts were layered onto a sucrose gradient, followed by ultracentrifugation. Multiple fractions were collected and analyzed to determine the distribution of the incorporated radiolabel into ribosomal particles. First, we assessed radiolabeled rRNA incorporation in the wild-type strain, in singly deleted Δrph or Δrnb strains, or in a doubly deleted Δrph Δrnb strain at 10 min after labeling. For the wild-type strain and for each of the single-deletion strains, most of the incorporated radiolabel was found in the 70S ribosomal fraction (Fig. 2A). This result indicated that the incorporation of newly synthesized rRNAs into ribosomes is virtually complete within this period of time. In contrast, much smaller amounts of radiolabel were present in the 70S fraction of the Δrph Δrnb strain. Instead, significant amounts were found in fractions that correspond to the 30S and 50S subunits or to precursors of these subunits. Therefore, in a strain lacking both RNase PH and RNase II, a kinetic defect in the formation of 70S ribosomes could be clearly observed.

Fig 2.

Mutations in RNase PH and RNase II result in delayed ribosomal assembly. (A) Strains MG1655*, MG1655* Δrph, MG1655* Δrnb, and MG1655* Δrph Δrnb each were grown at 37°C to mid-log phase and pulse labeled with inorganic phosphate for 2 min. After an additional 10 min of growth, cells were lysed, combined with lysates from unlabeled cells, and ultracentrifuged on a 14 to 32% sucrose gradient. The gradients were analyzed at 254 nM to identify the ribosomal peaks, and 16 fractions were collected in each case. Each fraction was filtered through a nitrocellulose membrane to determine the incorporation of radiolabel into ribosomes. The relative amounts of radiolabel present in each fraction is indicated by vertical bars, with the positions of the 70S, 50S, and 30S particles shown. The standard errors associated with these measurements are also shown. (B) MG1655* and MG1655* Δrph Δrnb strains were grown at 37°C to mid-log phase and analyzed as described for panel A, except that the strains were grown for the indicated times after pulse labeling prior to harvesting cultures for ultracentrifugation.

In theory, there are two possible explanations for the defects observed in the Δrph Δrnb strain: either the ribosomal particles are unable to progress to 70S ribosomes or 70S formation still takes place but occurs more slowly than in a wild-type strain. To discriminate between these possibilities, pulse-labeling experiments were repeated with the wild-type and Δrph Δrnb strains with the incorporation of radiolabel into ribosomal particles evaluated at different times postlabeling. For the wild-type strain, 5 min after labeling, the incorporated radiolabel was equally distributed between the 70S and the lighter fractions, whereas by 20 min most of it was shifted to the 70S ribosomes, similar to what had been observed at 10 min (Fig. 2A). With the mutant strain, most of the label was present in the precursor fractions at an early time point (5 min). Significantly, by 60 min postlabeling, the radioactivity was found to be shifted to the 70S fraction, as was observed for the wild-type strain at 10 to 20 min postlabeling. These results indicate that similar to the wild-type strain, most of the newly synthesized rRNAs in the mutant strain get incorporated into the ribosomal fraction, but this process occurs more slowly than in the wild-type strain.

Effects of RNase G and RNase T on ribosome assembly.

Three other enzymes that are known to be directly involved in rRNA processing are RNase T, RNase E, and RNase G. RNase T is involved in the final step of 23S and 5S rRNA maturation, in part due to its ability to remove single-stranded precursor sequences close to the base-paired ends of mature RNAs (13, 14) (Fig. 1A). In the absence of this enzyme, 23S rRNA retains 1 to 4 unprocessed residues at the 3′ end (14). Testing whether mutations in RNase T affect ribosomal assembly would be useful to determine whether defects at a late step of 23S rRNA processing can influence ribosome assembly in the same way as the defects at an earlier step caused by an absence of RNase PH and RNase II. The other two enzymes, RNase E and RNase G, have no known role in 23S rRNA maturation, but each promotes 16S rRNA maturation via endonucleolytic cleavage of precursors (15, 30). After RNase III cleavage of the rRNA transcript, the pre-16S precursor contains 115 unprocessed nucleotides at the 5′ end. This precursor is first cleaved by RNase E 66 nt upstream of the mature 5′ end, and thereafter it is cleaved by RNase G to generate a mature 5′ end. RNase E is also involved in an intermediate step of 5S rRNA maturation (5). Testing whether mutations in RNase E or RNase G have an effect on ribosome assembly would allow us to address whether 16S rRNA precursors can influence ribosome assembly. For this reason, the effects of RNase G mutations on ribosome assembly were also investigated. However, RNase E is an essential enzyme and is involved in multiple additional processes, including tRNA maturation and mRNA degradation (4, 21). We decided not to investigate the effects of RNase E mutations on ribosome assembly, since there was a possibility that RNase E mutants might influence ribosome biogenesis indirectly.

To determine the effects of RNase T or RNase G mutations on ribosome assembly, the wild-type, Δrnt, and Δrng strains were pulse labeled, followed by harvesting of cell cultures after 5 or 10 min and subsequent ultracentrifugation of cell extracts to determine the amount of radiolabel present in each ribosomal fraction (Fig. 3A). For each strain, a significant fraction of the radiolabel was found in the 70S fraction 5 min after labeling, whereas at 10 min postlabeling, the incorporation of radiolabel into this fraction was substantially complete. Thus, unlike the situation observed with a lack of RNase PH and RNase II, an absence of either RNase T or RNase G did not strongly influence ribosome assembly. However, the amount of radioactivity in the subunit fractions of the Δrnt strain was observed to be somewhat greater than that in the wild-type strain at each time point, suggesting that there is a small ribosome assembly defect in this strain background.

Fig 3.

RNase T and RNase G mutations do not significantly influence ribosome assembly. (A) Strains MG1655*, MG1655* Δrnt, or MG1655* Δrng were grown at 37°C and pulse labeled, and the incorporation of labeled inorganic phosphate into ribosomal fractions was evaluated at 5 or 10 min postlabeling, as described in the legend to Fig. 2. The relative amount of radiolabel present in each fraction is indicated by vertical bars, with the standard errors of the measurements indicated. (B) 3′-End analysis of 23S rRNA. 23S rRNA present in total RNA preparations derived from the MG1655* or MG1655* Δrnt strain was analyzed as described in the legend to Fig. 1B. The mature RNA end (M) and a precursor that corresponds to the site of RNase III cleavage (+8) are indicated on the left. The partially processed intermediates that accumulate in a Δrnt strain are indicated by a bracket on the right. (C) 5′-End analysis of 16S rRNA. Total RNA derived from the MG1655* or MG1655* Δrng strain was annealed with a radiolabeled antisense oligonucleotide complementary to 16S rRNA sequences 15 to 35 nt from the 5′ end and extended using Moloney murine leukemia virus reverse transcriptase. The positions of the 5′ ends of mature RNA (M) and a precursor (P) that corresponds to the RNase III cleaved product are indicated. The partially processed 5′ ends that accumulate in the absence of RNase G are indicated by +66.

To confirm the rRNA processing defects expected for these strains, 3′-end analysis of 23S rRNA and 5′-end analysis of 16S rRNA was performed. These analyses showed the expected lack of 23S rRNA maturation in the Δrnt strain, with the RNA containing 1 to 4 unprocessed nucleotides at the 3′ end (Fig. 3B), and defects in 16S rRNA maturation in the Δrng strain, with a large fraction of molecules containing 66 nt of precursor sequence at the 5′ end (Fig. 3C). Overall, these results lead to two conclusions: (i) the presence of shortened precursors at the 3′ end of 23S rRNA, as observed in the Δrnt strain, does not cause significant ribosome assembly defects, and (ii) the presence of precursors at the 5′ end of 16S rRNA does not affect ribosome assembly. Therefore, the consequences of defective rRNA processing on ribosome assembly kinetics appear to be primarily restricted to the 23S rRNA processing defects that occur at an early stage.

Distribution of 3′-end precursor sequences in ribosomal fractions.

To confirm and extend the results described above, Northern blot analysis was used to examine the distribution of the 23S 3′ ends in 50S subunits and in functional ribosomes. Northern blot analysis was performed first on RNAs derived from the 50S, 70S, or polysomal fractions of a wild-type strain (Fig. 4A). In the 50S particles, the majority of the 23S rRNA had a mature end, but small amounts of precursor sequences that were primarily extended by 8 nt were also present. In the 70S particles and polysomal fractions, only mature ends could be visualized and the precursors were no longer visible. These results indicated that rRNAs that contain precursors progress to 70S ribosomes inefficiently, consistent with the kinetic defects in 70S ribosome formation observed under conditions of precursor accumulation (Fig. 2). We next examined 50S and 70S particles in a Δrnt strain. As shown earlier, the RNase III-cleaved precursors undergo partial processing in this strain because RNase T is required mainly to digest the last few 3′-end precursor residues in 23S rRNA (Fig. 3B). It was therefore of interest to determine whether the shortened but still immature rRNAs can be incorporated into ribosomes. In the 50S fraction, precursors ranging from 1 to 8 residues could be observed, as well as small amounts of mature RNA and ends that had been trimmed 1 nt into the mature RNA (Fig. 4B). Although the distribution of 23S rRNA ends in the 50S fractions was somewhat different from those observed in total RNA fractions (Fig. 3B), a ladder of heterogeneous products was clearly evident. Significantly, 23S rRNA precursors were also present in the 70S fractions, suggesting that full processing of the 3′ end is not necessary to permit 70S formation. Moreover, precursors were also present in the polysomal fractions. A careful inspection of the relative amounts of the different-length precursors, however, revealed that rRNAs containing one or two precursor nucleotides were present in the 70S particles and polysomal fractions at levels similar to those in the 50S subunits, whereas those containing three or more nucleotides were present at reduced levels. Thus, there appears to be discrimination against 23S rRNA molecules that retain significant lengths of precursor sequences that interferes with their incorporation into ribosomes. Overall, the results suggest that although full maturation of the 23S rRNA 3′ end is not obligatory, precursors containing more than two precursor residues at the 3′ end are incorporated into ribosomes inefficiently. These results also help to explain the moderate defect in the kinetics of ribosome assembly observed in the Δrnt strain (Fig. 3A), which contains a mixture of longer precursors that are discriminated against for incorporation into 70S ribosomes as well as shorter precursors that are not.

Fig 4.

Analysis of 23S rRNA 3′-end precursors in ribosomal fractions. Cell extracts were ultracentrifuged on a 14 to 40% sucrose gradient, and RNA was analyzed from fractions corresponding to 50S subunits, 70S ribosomes, or polysomes. (A) Analysis of strain MG1655*; (B) analysis of strain MG1655* Δrnt. On the left are ribosomal profiles. The bars at the bottom demarcate the fractions from which RNA was isolated for analysis. On the right are the results obtained from Northern blot analysis of 23S rRNA from the respective fractions following RNase H cleavage, as shown in Fig. 1B. The positions of the mature 3′ rRNA end (M) and precursors containing the indicated number of unprocessed nucleotides are indicated. The assignment of the bands in panel B is based on comparison to the mature 23S rRNA, which was analyzed in parallel (not shown).

Distribution of 5′-end precursor sequences in ribosomal fractions.

Because the analysis described above indicated that rRNA molecules containing 3′-end extensions are inefficiently incorporated into ribosomes, it was of interest to investigate whether precursors at the 5′ end also could interfere with ribosome assembly. As depicted in Fig. 1A, cleavage of the 23S rRNA operon by RNase III primarily occurs 7 nt upstream of the mature 5′ end. Additional precursors containing shorter extensions have also been observed; these are derived either by alternative RNase III cleavage or by further cleavage of the +7 precursor via unknown mechanisms (28). To analyze the consequences of the 5′-end precursors on ribosome assembly, we examined, by using primer extension, the distribution of precursors in total RNA derived from a wild-type strain as well in the 50S, 70S, and polysomal fractions (Fig. 5A). In a wild-type strain, the predominant precursors in total RNA were found to contain extensions of 1, 3, and 7 nt. The same precursors were also present in the 50S fractions, although there were some qualitative differences in the relative amounts of the precursors present. We next examined the levels of precursors present in 70S particles. These particles were found to contain levels of the +1 precursor equivalent to those in the 50S subunit fraction, but the levels of the +3 and +7 precursors were reduced to trace levels. Thus, similar to what was observed for the 3′-end precursors, the retention of three or seven precursor nucleotides caused a bias against their incorporation into ribosomes. Also, as was observed earlier, the profiles of the 70S fractions were indistinguishable from those of the polysomal fractions, suggesting that once 70S ribosomes are assembled, there is no further discrimination against the particles that contain precursor residues.

Fig 5.

Analysis of 23S rRNA 5′ end precursors in ribosomal fractions. Primer extension reactions were performed on total RNA (T) or on RNA isolated from 50S, 70S, or polysomal fractions (P), as indicated. The RNAs were derived from MG1655* (A) or the MG1655* Δrph Δrnb strain (B). The positions of the mature end (M) as well as precursors containing three or seven precursor residues are shown. Other precursors that are also present are not explicitly indicated.

To visualize the unprocessed RNAs in the 70S polysomal fractions more clearly, the above-described analyses were also performed in the Δrph Δrnb strain, which is expected to contain higher levels of not only 3′-end precursors (Fig. 1B) but also 5′-end precursors (9). As expected, enhanced levels of precursors were observed in the 50S fractions compared to those in a wild-type strain (Fig. 5B). Again, in the 70S and polysomal fractions, the majority of the rRNAs contained mature ends. However, low levels of the +3 and +7 precursors also could be observed somewhat more clearly than in a wild-type strain but to a reduced extent compared to that in the 50S fraction. Overall, the results of these analyses indicate that 23S rRNA containing 1 or 2 nt of precursor sequences at either end can be assembled into ribosomes effectively, but precursors containing three or more nucleotides are not.

DISCUSSION

The assembly of ribosomes comprises a complex set of processes that involves not only the association of multiple ribosomal proteins with rRNA but also the processing and modification of rRNAs. It has been shown previously that diverse ribosome assembly defects result in an accumulation of rRNA precursors (6, 12, 27). It is believed that rRNA processing occurs more efficiently when these molecules adopt specific conformations that occur during the assembly process, and when these conformations are not attained due to assembly defects, processing of rRNAs becomes inefficient. For example, for RNase III cleavage to initiate rRNA processing, complementary precursor sequences at the 5′ and 3′ ends of each of the three rRNAs need to base pair to form duplexes that can be cleaved by this double-strand-specific endonuclease (3, 16). Similarly, the action of RNase T on the 23S rRNA precursor is more efficient if the enzyme acts on ribonucleolytic particles rather than on naked RNA (30).

One specific aspect of ribosome assembly that has not been explored in detail is whether rRNA processing defects can themselves influence ribosome assembly. To investigate this question, we characterized ribosome assembly in strains lacking the RNase genes known to be involved at defined steps of rRNA processing. We found that an absence of RNase G, which is involved in the final step of 16S rRNA 5′-end maturation, had no discernible effect on ribosome assembly, even though a significant fraction of the 16S rRNA in RNase G-deficient strains contains a 66-nt extension (Fig. 3C). Consistent with this interpretation, high levels of 16S precursors have been observed in 70S ribosomes derived from rng strains (23). It is possible, however, that the precursors confer a small defect in assembly that is diluted by the formation of mature-like ends for a population of the 16S rRNA molecules through an RNase G-independent mechanism (Fig. 3C). On the other hand, defects in 23S rRNA 3′-end maturation conferred a significant defect in ribosome assembly. In particular, in a strain lacking RNase PH and RNase II, ribosome assembly was observed to be significantly slower than that in a wild-type strain (Fig. 2). Interestingly, an absence of RNase T, which mediates the final step of 23S rRNA maturation, conferred a defect in ribosome assembly, but one that was considerably less dramatic. Thus, ribosomal assembly appears to be influenced mainly by the defects that occur at an early stage of 23S rRNA processing. A nucleotide-level analysis indicated that precursors containing more than two unprocessed 3′ or 5′ residues are inefficiently incorporated into 70S ribosomes and polysomes, whereas those that contain 2 nt or fewer suffer no such discrimination (Fig. 4 and 5). The delayed kinetics of ribosome assembly in the Δrph Δrnb strain therefore are consistent with the notion that the assembly delays are caused by precursor sequences, and it requires their removal for assembly to proceed further. These results suggest that precursor residues in 23S rRNA serve an important role to ensure that only rRNAs that are fully processed, or nearly so, are incorporated into ribosomes efficiently.

How can one explain why 23S rRNA precursors affect ribosome assembly? One possibility is that cells harbor a signaling mechanism that detects the accumulation of unprocessed 23S rRNA to adjust the kinetics of ribosome assembly based on the levels of precursors detected. In this scenario, we expect that a mechanism exists to measure precursor length and to modulate the ribosome assembly rate based on the length of precursors measured. In this context it is worth pointing out that E. coli possesses a related sensing mechanism that detects the nutritional environment and the growth rate of cells to adjust rRNA synthesis accordingly (20, 22). A more direct mechanism, however, is suggested in a model presented below. The ends of the 23S rRNA are located on the periphery of the ribosome and point outwards (26), therefore the presence of precursor sequences is unlikely to cause any interference in the mature ribosome (Fig. 6A). However, it is conceivable that during the assembly process, the ends of mature 23S rRNA transiently pass close to other residues in the nascent ribosome, so that precursor sequences, if they were present, would cause a steric clash (Fig. 6B). In that event, further assembly would be blocked, leading to ribosome assembly defects, as are observed in the Δrph Δrnb strain. In this sense, the residues that clash with the unprocessed rRNA residues can be considered to function as a molecular gate. We propose that the gate allows mature rRNAs or those containing one or two precursor nucleotides to pass through without any hindrance but that when the precursors are longer, passage through the gate is impeded, with negative consequences for ribosome assembly. Additionally, the finding of precursors containing three or more unprocessed nucleotides in the ribosomal fractions, albeit at low levels, suggests either that the gate displays some flexibility, that the precursor sequences can transiently attain compacted conformations that allow transit through the gate, or that assembly can proceed through a pathway that bypasses the gateway mechanism, though inefficiently.

Fig 6.

Structure of the 23S rRNA ends and a model for regulation of ribosome assembly by precursor sequences. (A) The ends of 23S rRNA in the E. coli 50S subunit, derived from the published structure (26), are shown on the left using PyMOL. The RNA component of the subunit is colored light blue, whereas the ribosomal proteins are depicted by different colors. The region that corresponds to the ends of the 23S rRNA is boxed and shown enlarged on the right. The penultimate 5′ and 3′ residues that can be visualized in the crystal structure (residues 2 and 2903, respectively) are indicated; in each case the 3′- and 5′-most residues cannot be located in the structure. (B) Model for role of the rRNA ends in ribosomal assembly. The base-paired ends of the 23S rRNA are shown as parallel lines, with precursor sequences indicated with dashes. In this depiction, the terminal residues of the 23S rRNA need to pass under a set of ribosomal residues, depicted by an arch, for ribosome assembly to proceed further. The presence of precursor residues beyond a certain length, however, causes a steric clash that interferes with this process. As a consequence, rRNA molecules that contain substantially processed ends selectively progress to form fully assembled ribosomes, whereas those that contain precursors do so inefficiently.

Irrespective of the mechanisms involved, the question arises of why it is necessary for ribosomes to be substantially free of 23S precursor sequences. One possibility is that the presence of precursors in ribosomes interferes with some aspects of translation. For example, although precursors on 16S rRNA appear to cause little interference with the assembly process, their presence has recently been found to decrease the translational fidelity of ribosomes (23). Whether precursors on 23S rRNA also affect some aspects of translation is a question that remains to be addressed. In addition, it is possible that precursor sequences, by regulating the assembly process, provide a checkpoint to control the fidelity of ribosome assembly inside the cell. We suggest that this aspect of ribosome assembly represents a quality-control mechanism that ensures that only those rRNAs that are properly folded and have undergone a normal processing pathway will be incorporated into ribosomes efficiently. Mutations in precursor sequences have been previously shown to confer a number of defects, including a cold-sensitive growth phenotype (24). We have also shown recently that processing of the 23S rRNA 3′-end precursors occurs in a regulated manner and have proposed that this represents a regulatory mechanism to ensure that digestion by RNases does not proceed into the mature rRNA sequences (10). The identification of a gatekeeper mechanism that controls ribosome assembly based on precursor length further highlights the importance of rRNA precursor sequences in ribosomal biogenesis. Future work on this topic should help to define the molecular basis by which rRNA precursors modulate ribosome assembly and function to channel the significant resources used by the cell to yield ribosomes that are optimally poised to engage in translation.