INTRODUCTION

Top Abstract Introduction Materials and Methods Results and Discussion References

In Escherichia coli, the number of ribosomes per cell is proportional to the growth rate to satisfy the cell's demand for protein synthesis (23). At fast doubling times there are as many as 70,000 ribosomes per E. coli cell, while at lower growth rates this number is reduced to 20,000 or less (5). Control of ribosome content is exerted at the level of transcription of the seven rRNA (rrn) operons on the chromosome (18, 29). Expression of rRNA is gene dosage independent; when the number of rRNA operons in the cell is increased by the presence of plasmid-borne operons, total rRNA synthesis rates remain constant (feedback control) (10, 19, 21, 23). Conversely, inactivation of up to four rrn operons on the chromosome leads to a compensatory increase in expression of the remaining intact copies (7). Regulation of rrn expression occurs at the level of transcription initiation (23), and one effector of both growth rate-dependent and feedback control is thought to be the intracellular concentrations of ATP and GTP, the initiating nucleotides of the rrn P1 promoters (14).

Although it is generally assumed that the redundancy of rRNA operons in E. coli has evolved to support the high levels of ribosome production necessary for rapid growth rates (22, 28), there is also evidence suggesting that E. coli requires all of its operons for optimal adaptation to changing physiological conditions (8). rRNA operon multiplicity among the best-studied eubacteria has, however, significantly impeded the genetic study of rRNA structure, function, and evolution in these organisms. Besides the 7 rrn operons present in E. coli (24), Bacillus subtilis (4, 26) and Clostridium perfringens (15) have 10 each and Lactococcus lactis (2, 34) has 6 copies of each of the rRNA genes. As a means of overcoming the multiplicity problem, we sequentially inactivated rrn operons in E. coli until we ultimately succeeded in constructing a strain containing a single exchangeable operon on a plasmid.

We had previously inactivated up to four of the rRNA operons by a deletion-insertion mutagenesis scheme in which each deletion site was filled in with a different antibiotic resistance gene (7). While this technique provided a facile means of operon inactivation, there was an insufficient number of suitable antibiotic resistance genes to inactivate all seven operons, and we were concerned that the accumulation of antibiotic resistance mechanisms would influence the physiology of the cell. In the study reported here, we therefore employed a different approach, in which many of the operons were inactivated without the introduction of antibiotic resistance cassettes, and succeeded in inactivating all seven chromosomal rRNA operons. The survival of this strain is ensured by the presence of a plasmid-encoded rRNA operon. In a separate publication (1), we have demonstrated one important use of this strain by successfully exchanging the wild-type plasmid-borne E. coli rrn operon for operons from Salmonella typhimurium and Proteus vulgaris as well as a hybrid operon in which the GTPase center of the E. coli 23S rRNA had been replaced by the corresponding domain from Saccharomyces cerevisiae. Here we describe in detail the construction of the rrn deletion series and an initial study of their physiological properties, as much to answer questions about the effect of rrn multiplicity on bacterial physiology as to characterize a set of strains we believe will be useful to the scientific community.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results and Discussion References

Bacterial growth conditions. The bacterial growth conditions were described previously (1).

Exchange of alleles with a polA strain and the sacB gene. We have developed an effective method for allele exchange between chromosomal and plasmid-encoded rRNA operons by modifying previously reported techniques (20, 31). DNA fragments containing each of the seven rRNA operons (Fig. 1) and their flanking regions were first cloned into ColE1-type plasmid vectors carrying the ampicillin resistance (Ap^r) gene. Deletion mutations inactivating both the 16S and 23S rRNA genes were then introduced into each operon. A cassette containing the B. subtilis sacB gene and the kanamycin resistance marker (sacB-Km^r) was then prepared from pBIP3 (31) and inserted into the plasmid within the vector sequence (Fig. 2A). Expression of sacB in E. coli is lethal in the presence of sucrose (16). Thus, the cassette allows both positive (Km^r) and negative (sucrose-sensitive [Suc^s]) selection of the resulting plasmid. The plasmid was then electroporated into polA1 (Am) mutant cells in which the corresponding rRNA operon on the chromosome had been inactivated with the chloramphenicol resistance (Cm^r) gene (the cat gene [Fig. 2B]). We took advantage of previous work from our laboratory (9) in which each rRNA operon on the chromosome was inactivated by this gene. Initiation of DNA replication from the ColE1-type origin requires the polA gene product, DNA polymerase I. Thus, polA mutant cells transformed to Ap^r and Km^r are likely to contain the entire plasmid integrated into the chromosome by a single crossover event (Fig. 2C). All integrants showed sucrose sensitivity. Since the rRNA genes encoded in the seven operons have essentially identical primary structures, we relied on flanking sequences to direct recombination with the desired operon, and by Southern blot analysis (rrnB) or P1 transduction (rrnH, rrnG, and rrnA) we confirmed that integration had occurred in the correct operon. In the latter case, P1 lysates were prepared on each integrant and the cotransduction frequency of antibiotic resistance markers was analyzed. If the plasmid integrated into the correct operon, the Cm^r marker of the inactivated chromosomal operon and the Ap^r and Km^r markers introduced by the plasmid cotransduced with a high frequency.

View larger version (9K): [in this window] [in a new window]   FIG. 1.   The common structure of the rRNA operons in E. coli. Open and filled rectangles represent rRNA (16S, 23S, and 5S) and tRNA genes, respectively. rrnB, rrnC, rrnE, and rrnG contain the spacer tRNA gene for Glu-2, and the other operons (rrnA, rrnD, and rrnH) contain the spacer tRNA genes for Ile-1 and Ala-1B (25). Distal tRNA genes are encoded by only three operons: rrnC contains the tRNA genes for Asp-1 and Trp, and rrnD and rrnH contain the tRNA genes for Thr-1 and Asp-1, respectively. The figure also indicates the relative positions of promoters (P1 P2), terminators (ter), and relevant restriction sites.

View larger version (21K): [in this window] [in a new window]   FIG. 2.   The basic strategy for allele exchange. Thick and thin lines represent chromosomal and plasmid sequences, respectively. The hatched rectangles indicate the 16S and 23S rRNA genes. The 5S rRNA and tRNA genes are not shown. Stippled and open rectangles represent the ampicillin and chloramphenicol resistance genes, respectively, and closed rectangles the sacB-Kmr cassette. ori indicates the relative position of the ColE1-type replication origin. Broken lines indicate possible crossover sites for a successful allele exchange. In panels B, C, and D, only a part of the chromosome is shown. See Fig. 1 for definitions of the other symbols.

Construction of rrn deletion strains. As a starting point for rrn operon inactivation, we used Ellwood and Nomura's TX11 strain (12), which carries a large chromosomal deletion spanning the rrnE operon. Before the first allele exchange, the polA1 mutation was introduced into TX11 by P1 transduction by virtue of its linkage to zih::Tn10 (encoding tetracycline resistance), generating TA340 (Fig. 3; Table 1). The presence of the mutation in this strain was verified by its sensitivity to methyl methanethiosulfonate.

View larger version (40K): [in this window] [in a new window]   FIG. 3.   The pedigree of rrn deletion strains. Inactivated rRNA operons are indicated by uppercase letters derived from their specific operon names (for example, A for rrnA). When the inactivation was carried out by a deletion-insertion mutation, an uppercase letter is followed by a lowercase c or z, representing the inserted gene cat+ or lacZ+, respectively (for example, Ac for rrnA::cat+). See Table 1 for precise genotypes. polA mutant strains are resistant to tetracycline since these mutations are linked to Tn10. pTRNA and pHK-rrnC+ contain spectinomycin (Spcr) and Kmr markers, respectively.

(i) Inactivation of rrnB. The rrnB operon in TA340 was first inactivated by introducing the (rrsB-gltT-rrlB)1::kan⁺ mutation (7). (For rrnB, a cat-containing operon was not used in this step.) The resulting strain (TA405) was then transformed to ampicillin resistance (TA406) with pMA101 (Table 2), and the (rrsB-gltT-rrlB)1::kan⁺ allele was removed from the chromosome as described above except that Suc^r cells were screened for sensitivity to ampicillin and kanamycin. The presence of the (rrsB-gltT-rrlB)101 deletion mutation on the chromosome of one of the Ap^s Km^s clones (TA410) was confirmed by PCR amplification of ribosomal DNA followed by agarose gel analysis. The primers used for the PCR were 5'-GGCCTAACACATGCAAGTCGAA-3' and 5'-GCTTACACACCCGGCCTATCAA-3', which hybridize near the 5' end of the 16S gene and the 3' end of the 23S gene, respectively. With these primers, the rrnB operon carrying the (rrsB-gltT-rrlB)101 deletion gives a 2,287-bp PCR fragment whereas the wild-type and kan-containing operons give 4,791- and 4,026-bp fragments, respectively.

(ii) Inactivation of rrnH. TA410 (EB polA) was first transduced to Cm^r (TA415) with (rrsH-ileV-alaV-rrlH)37::cat⁺ (rrnH in reference 9) and then transformed to Ap^r and Km^r (TA418) with pMA103 (Table 2). The (rrsH-ileV-alaV-rrlH)37::cat⁺ allele was removed from TA418 as described above. The presence of the new deletion mutation in one of the Cm^s Ap^s Km^s clones (TA420) was confirmed by PCR with the primers described above. We detected a PCR fragment of the expected size (1,290 bp).

(iii) Inactivation of rrnG. TA420 (EBH polA) was first transduced to Cm^r (TA443) with (rrsG-gltW-rrlG)33::cat⁺ (rrnG in reference 9) and then transformed to Ap^r and Km^r (TA445) with pNY30 (Table 2). This plasmid carries an rrnG allele inactivated by an internal deletion and the concomitant insertion of the lacZ coding region into the site of the deleted operon. The (rrsG-gltW-rrlG)33::cat⁺ allele was removed from TA445 as described above except that 5-bromo-4-chloro-3-indolyl--D-galactopyranoside (X-Gal; 60 µg/ml) was added to sucrose-containing plates. Suc^r, blue colonies were then screened for sensitivity to ampicillin, kanamycin, and chloramphenicol. Cells that were sensitive to these antibiotics should have contained only the (rrsG-gltW-rrlG)30::lacZ⁺ allele on the chromosome and were designated TA447.

(iv) Inactivation of rrnA. The standard inactivation procedure was slightly modified as described below to restore the polA⁺ genetic background. The (rrsA-ileT-alaT-rrlA)1::cat⁺ mutation (rrnA in reference 9) was first introduced into TA410 (EB polA) by P1 transduction. The TA410 transductants were selected for Cm^r and Tc^r and screened for PolA by assaying the transformation efficiency of pBR322. This strain was named TA472. Next, pNY34 (Table 2) was integrated into the chromosome of TA472 near rrnA. Plasmid pNY34 contains an internally deleted rrnA operon. A P1 lysate was prepared on the resulting strain (TA480) and used to transduce integrated pNY34 (zih::pNY34) into TA476. TA476 was generated by transducing the (rrsA-ileT-alaT-rrlA)1::cat⁺ mutation into TA447 (EBHGz polA) and is polA⁺ due to cotransduction of the polA⁺ allele. The TA476 transductants were selected for resistance to ampicillin, kanamycin, and chloramphenicol and screened for sensitivity to tetracycline (Tc^s). The presence of polA⁺ in Ap^r Km^r Cm^r Tc^s cells was confirmed by assaying the transformation efficiency of pACYC184, which also requires DNA polymerase I for replication. The resulting cells (TA485) were grown to saturation, and Suc^r cells were obtained as described above. Finally, the Suc^r cells were screened for sensitivity to ampicillin, kanamycin, and chloramphenicol, thereby obtaining TA488. The presence of the rrnA deletion in this strain was verified by Southern blot analysis.

(v) Inactivation of rrnD. TA500 is the same as TA488 (EBHGzA) but carries the tRNA-containing plasmid pTRNA65 (1). This strain was transduced to Cm^r with (rrsD-ileU-alaU-rrlD)25::cat⁺ (rrnD in reference 9), generating TA516.

(vi) Inactivation of rrnC. pHK-rrnC⁺ (Km^r) (1) was first introduced into TA516 (EBHGzADc/pTRNA65), generating TA520. A spontaneous deletion of the gene for tRNA₂^Glu from pTRNA65 in TA520 resulted in TA520.5 carrying pTRNA66 (1). Restriction mapping indicated that pTRNA65 suffered a spontaneous deletion, likely by recombination between redundant parts of the 23S gene surrounding tRNA₂^Glu. The (rrsC-gltU-rrlC)15::cat⁺ mutation (rrnC in reference 9) was then introduced into TA520.5 by P1 transduction by virtue of its linkage to ilv500::Tn10 in strain TA575. The presence of (rrsC-gltU-rrlC)15::cat⁺ in Tc^r transductants (TA525) was verified by PCR with the following primers: 5'-CTTCCATGTCGGCAGAATGCTT-3' and 5'-GCCTGCATACCGTTGTCGATAG-3'. These primers hybridize near the ends of the cat gene and the rrnC operon, respectively, and amplify an 850-bp fragment. Finally, the ilv⁺ allele was introduced into TA525 (EBHGzADcCc ilv/pTRNA66/pHK-rrnC⁺) by P1 transduction, generating TA527. The lack of intact 16S and 23S rRNA genes on the chromosome of this strain was confirmed by Southern blot analysis (1).

Construction of an rrn⁺ strain, TA563. An rrn⁺ strain was constructed from TX11 by introducing rrnE⁺ by Hfr mating. The donor and recipient strains were CAG5052 and TA559.5, respectively. TA559.5 was constructed from TX11 by introducing pBEU49 (35). This plasmid was used only to provide convenient counterselection (Ap^r Km^r) of the donor strain. The mating was carried out at 30°C for 20 min, and the cells were plated on LB plates containing tetracycline, ampicillin, and kanamycin. The plates were incubated at 30°C for 16 h, and the exconjugants were screened for Pur⁺ and rrnE⁺. The screening for rrnE⁺ was carried out by PCR. The primers used for the reaction were 5'-GAATTCGACGATACCGGCTTTG-3' and 5'-CCACTCGTCAGCAAAGAAGCAA-3', which hybridize to the purH and 16S sequences, respectively. These primers amplify a 787-bp fragment from the wild-type rrnE region. Finally, pBEU49 was removed from one of the Pur⁺ rrnE⁺ exconjugants (TA560) by using its runaway replication property, generating TA563. Although metA⁺ was most likely introduced into TA560 with pur⁺-rrnE⁺ by Hfr mating, the strain remains Met since the metB1 mutation is located close to btuB::Tn10 in the donor chromosome. The presence of all seven rRNA operons in TA563 was confirmed by Southern blot analysis (1).

Total RNA/total protein and tRNA/rRNA ratios. Total RNA/total protein and tRNA/rRNA ratios were determined as described previously (1).

	RESULTS AND DISCUSSION

Top Abstract Introduction Materials and Methods Results and Discussion References

Construction of rrn deletion strains. The starting material for the construction of our rrn deletion series was TX11, an E. coli strain constructed by Ellwood and Nomura (12), which contains a deletion mutation encompassing the entire rrnE operon (Table 1). The six remaining rrn operons in this strain were sequentially inactivated by deletion mutations spanning the 16S and 23S rRNA genes as summarized in Fig. 3. The details of the inactivation procedures are described in Materials and Methods. Briefly, the rrnB and rrnH operons were inactivated by removal of internal SalI-SalI and SacII-SacII fragments, respectively (Fig. 1). The rrnG operon was inactivated by a deletion-insertion mutation (Gz in Fig. 3) in which the SmaI-HpaI region of the operon (Fig. 1) was replaced by the lacZ coding region. The rrnA operon was initially inactivated with one of the deletion-cat insertion mutations constructed previously in our laboratory (9). This deletion-cat insertion mutation in rrnA (Ac in TA476 [Fig. 3]) was then replaced by a simple deletion mutation (A in TA488) that removed the SacII-SacII region of the operon. The rrnD and rrnC operons were also inactivated with deletion-cat insertion mutations (Dc and Cc, respectively) (9). In the final strain (TA527), carrying no intact rRNA operons on the chromosome (7 prrn), rRNA molecules were expressed from a multicopy plasmid, pHK-rrnC⁺ (prrn). This plasmid is a derivative of pSC101 containing only the wild-type rrnC operon (1).

Cell morphology. We first carried out a microscopic examination of cells from rrn⁺ (TA563) and rrn deletion strains. We found that cells with inactivated rRNA operons showed a pronounced morphological change during exponential growth; the cells became more and more elongated, with this change being very apparent in a 6 strain (TA516 [Fig. 4]). Using a vital stain, we saw no indication that dead cells accumulated in the deletion strains (data not shown). The elongated cell morphology was not completely reversed in a 7 prrn strain (TA527) containing the rRNA and tRNA plasmids (Fig. 4), suggesting that cellular processes other than rrn gene dosage are still perturbed in this strain (see below).

View larger version (32K): [in this window] [in a new window]   FIG. 4.   Microscopic examination. Cells from exponential-phase cultures were stained with BacLight (Molecular Probes Inc., Eugene, Oreg.) and analyzed by fluorescence microscopy.

Growth rate. We next measured the growth rates of rrn⁺ and rrn deletion strains in a rich nutrient medium (1), in which a large number of ribosomes are needed for short cell division times and in which the effects of rrn inactivation should be most pronounced (7, 8). For this and the other physiological studies described below, we constructed new 1 to 4 strains (TA566, -567, -568, and -430, respectively) (Fig. 3; Table 1) that contained the polA⁺ allele and the deletion-cat insertion mutation in rrnA) (Ac). The 5 strain used for physiological studies was TA476, which also contained polA⁺ and Ac. For a 6 strain, we used TA516, which contained the deletion-cat insertion mutation in rrnD but not in rrnA. Unlike these strains, the 7 prrn strain, TA527, contains two deletion-cat insertion mutations on the chromosome, within rrnD and rrnC.

View larger version (25K): [in this window] [in a new window]   FIG. 5.   Physiological effects of rrn inactivation. (A) Relative growth rates, rRNA/protein ratios, and ribosome efficiencies. Growth rates (doublings per hour) were determined by monitoring the turbidity of each culture with a Klett-Summerson photoelectric colorimeter and are presented relative to the rrn+ strain values. The maximum standard error of growth rate measurements was 0.07. The actual growth rate of the rrn+ strain, TA563, was 2.0 doublings/h. rRNA/protein ratios were determined from the data presented in panel B. Ribosome efficiencies were calculated from growth rates and rRNA/protein ratios as described by Bremer and Dennis (5). (B) Relative total RNA/total protein and tRNA/rRNA ratios. The total RNA/total protein and tRNA/rRNA ratios in each total RNA sample were determined as described previously (1). These parameters were normalized to the total amount of RNA in the rrn+ strain. Closed and open bars represent the amounts of rRNA and tRNA, respectively. The maximum standard error of RNA measurements was 0.05. Note that the 6 strain contains only the tRNA plasmid while the 7 strain contains both the tRNA and rRNA plasmids.

RNA/protein and tRNA/rRNA ratios. Ninety-eight percent of the total RNA is stable RNA in wild-type E. coli (5). We therefore expected that the total RNA/total protein ratio would be one of the physiological parameters most profoundly affected in the rrn deletion strains. We measured total RNA and total protein from rapidly growing cells and determined the RNA/protein ratio. Contrary to our expectations, the ratio remained relatively constant, with the 6 strain's RNA/protein ratio being approximately 84% of that of the rrn⁺ strain (Fig. 5B).

Ribosome efficiency. The rRNA/protein ratio is proportional to the number of ribosomes per protein, i.e., the ribosome concentration (5). In the 2 strain, the percent decrease in the growth rate was similar to the decrease in the rRNA/protein ratio (Fig. 5A), suggesting that the reduced growth rate of this strain was caused by a reduced ribosome concentration. When the number of rrn deletions was extended from three to six, however, the decrease in the growth rate was greater than that of the rRNA/protein ratio. This difference between the decreases in the growth rate and the rRNA/protein ratio was most pronounced in the 7 prrn strain, in which the rRNA/protein ratio was completely restored to wild-type levels but the growth rate was not. These results suggest that the reduced growth rates of the multiply deleted strains are not simply due to reduced ribosome concentrations. It has been proposed that ribosome concentration and efficiency are growth limiting in any living cell whose protein turnover is negligible and that the growth rate of an exponential-phase culture is proportional to the multiplication product of these two values (5). According to this theory, the reduced growth rates of the 3 to 7 strains can be attributed, at least in part, to reduced ribosome efficiencies. We have calculated the theoretical ribosome efficiency in each of the rrn deletion strains by using the equation presented in reference 5. By these calculations, the ribosome efficiency of 7 prrn strains falls to below 80% of wild-type levels (Fig. 5A).

Applications of rrn deletion strains. Using the 7 prrn strain, we have successfully constructed strains containing an rrn operon from a foreign microorganism, such as Salmonella typhimurium and P. vulgaris (1). Characterization of these strains with hybrid ribosomes emphasizes the usefulness of this system for evolutionary studies of the translation machinery. The deletion strains can also be used to examine current models of rrn regulation and to answer questions about the evolution of bacteria with multiple rrn operons. The 7 prrn strain should be especially useful for in vitro analysis of ribosome functions, since pure mutant ribosome populations from this strain are available (1). In addition, 7 prrn provides a powerful method for the isolation of new rRNA mutations, including conditionally lethal mutations for examining essential functions of rRNA. Questions concerning specific rRNA domains or sequences, modified bases, particular structures, long-range interactions, changes leading to drug resistance, and interaction with other components of the translation apparatus should all be more readily addressed by using the deletion strains described here.