Nucleic Acids Research

Construction and expression of penultimate stem mutations

Oligonucleotide-directed mutagenesis was used to introduce four base changes on one side of the penultimate stem (A1472, A1474, U1475 and A1477; Fig. 2b), generating a single-strand mutant referred to as 1472. A second rRNA mutant (designated 1423) was made with base changes complementary to the 1472 mutations (U1423, G1425, U1426 and U1428; Fig. 2c). A third mutant (designated 23/72) was generated with both regions mutated such that base pairing was restored (Fig. 2d). Note that the structure of the double mutant restored a non-canonical U-G pair at 1425/1475 but replaced non-canonical G-U pairs at 1423/1477 and 1426/1474 with U-A pairs. These mutants were initially subcloned into the high copy vector pKK3535. However, in this vector expression of the 1423 and 1472 mutants resulted in an unstable growth phenotype yielding both large and small colonies and reversion to faster growing clones, suggesting that these rRNA mutants had deleterious effects on ribosome function when expressed at high levels. Therefore the mutants were subcloned into pMM1192, a low copy number plasmid, and pL[Delta]H1192, a plasmid with an inducible promoter, for more stable expression.

The effects of the three mutants on growth rates were determined in strain SU1675 using the low copy plasmid. Expression of the two single-strand mutants 1423 and 1472 resulted in an increased doubling time (73 and 71 min respectively, compared with a wild-type doubling time of 57 min), indicating that disruption of the penultimate stem was deleterious to the cell (Fig. 2). However, the double-strand mutant 23/72 showed a wild-type doubling time (57 min). This suggested that secondary structure of the stem, not primary sequence, was the more important element in this region for proper ribosome function.

Effects of penultimate stem mutations on translational fidelity

Since mutations in the base helix and at positions 1469 and 1483 had been reported to have significant effects on translational fidelity (7,9,13), we examined the effect of mutations in the central helix on accuracy of decoding. Using plasmids containing the rrnB operon transcribed from the inducible [lambda] P_L promoter we analyzed the ability of mutant ribosomes to cause frameshifts and read-through of stop codons in a series of lacZ reporter genes (Table 1). A modest increase (50%) in stop codon read-through was observed for the single-strand mutant 1423, while the 1472 mutant had less of an effect on stop codon read-through. Both caused a modest (28%) increase in frameshifting. The double-strand mutant 23/72 displayed levels of frameshifting and read-through virtually identical to wild-type levels. This result, coupled with the recovery in growth rate observed for the double-strand mutant, indicated that the base paired stem structure was important for optimal decoding.

Table 1. Effect of penultimate stem mutations on translational fidelity

lacZ plasmids	[beta]-Galactosidase activity (Miller units)
	1192 (wt)	1472	1423	23/72
Wild-type
pSG25	5500 ± 199	5400 ± 135	5532 ± 224	5153 ± 169
Stop codon mutations
pSG163 (UAG)	56.2 ± 5.6	60.0 ± 2.7	83.5 ± 4.5	55.2 ± 3.7
pSG3/4 (UGA)	138.5 ± 7.9	153.8 ± 7.6	216.0 ± 7.7	134.7 ± 8.1
pSG853 (UAA)	17.5 ± 1.4	22.1 ± 1.4	25.7 ± 1.3	17.4 ± 1.1
Frameshift mutations
pSG12dP (-1)	72.3 ± 1.7	91.9 ± 3.4	90.6 ± 3.6	70.0 ± 1.7
pSGCAUGGA (-1)	10.3 ± 0.5	12.0 ± 0.4	12.9 ± 0.5	10.5 ± 0.9
pSGCCCU (+1)	19.5 ± 0.7	23.4 ± 1.1	25.1 ± 1.4	18.3 ± 2.0
pSGlac7 (+1)	61.0 ± 2.4	77.9 ± 1.1	74.1 ± 1.9	60.5 ± 1.2

Values represent the average of five to ten independent measurements ± 1 SD. [beta]-Galactosidase activities (in Miller units) were measured after induction of transcription of plasmid-encoded rRNA at 42°C for 150 min. The level of enzyme synthesized from the wild-type plasmid is not a good quantitative measure of mutant ribosome activity since the mRNA is not limiting.

Distribution of plasmid-encoded rRNA in 30S subunits, 70S ribosomes and polyribosomes

The distribution of plasmid-encoded rRNA in 30S subunits, 70S ribosomes and polysomes was determined in strain SU1675 carrying the low copy pMM1192-derived plasmids. Cells pelleted from exponentially growing cultures were lysed gently with detergent and ribosomes and polysomes were fractionated by sucrose density gradient centrifugation. The results of primer extension analysis of plasmid-encoded 16S rRNA from 30S, 70S and polyribosome fractions are shown in Table 2. Plasmid encoded 16S rRNA was present in polysomes, indicating that all three mutant rRNAs were processed and assembled into functional ribosomes. However, the fraction of all three mutants in 70S ribosomes was reduced, indicating that mutant subunits were less stable as 70S ribosomes. It was surprising that the double mutant also showed this defect, given that the restoration of base pairing restored both growth rate and accuracy to wild-type levels. This suggests that the primary sequence, in addition to secondary structure of the penultimate stem, may have a role in ribosome function.

Table 2. Distribution of plasmid-encoded rRNA in 30S subunits, 70S ribosomes and polysomes

Plasmid	30S	70S	Polysomes
pMM1192	58.6 ± 4.8	58.0 ± 2.9	54.0 ± 1.5
pMF72	61.1 ± 8.4	40.7 ± 0.4	66.7 ± 4.7
pMF23	78.8 ± 4.5	47.1 ± 1.5	68.4 ± 7.9
pMF23/72	67.7 ± 6.2	48.9 ± 2.7	70.6 ± 2.6

Sucrose gradient fractions of 30S subunits, 70S ribosomes and polysomes were analyzed by primer extension as described in Materials and Methods. Numbers represent percentages of plasmid-encoded rRNA in each fraction ± 1 SD and were determined by the formula (plasmid-encoded rRNA/plasmid-encoded rRNA + host-encoded rRNA) × 100. Values are percent mutant ribosomes in each peak and cannot be added to get total percent mutant RNA since most of the total rRNA is in the 70S peak.

Subunit re-association of mutant ribosomes

To investigate the apparent subunit association defect further the ability of dissociated subunits to re-form 70S ribosomes was analyzed. Equimolar amounts of 30S and 50S subunits containing the host- and plasmid-encoded rRNA were incubated in 5, 10 or 15 mM MgCl₂ and subjected to sucrose density gradient centrifugation. The gradient profiles were similar (data not shown), but primer extension analysis of 16S rRNA from each 30S and 70S fraction (Table 3) showed that only small percentages of the 1472 and 1423 single-strand mutant rRNAs were located in the 70S fractions, indicating that subunits containing mutated rRNA could not associate with 50S subunits to form 70S ribosomes. The same was true of the 23/72 double-strand mutant at 5 mM MgCl₂, but at higher MgCl₂ concentrations the wild-type level of association was restored in this mutant. Thus, disruption of base pairing resulted in decreased ability of the mutant 30S subunits to interact with 50S subunits and this defect was corrected upon restoration of base pairing in the helix

Table 3. In vivo subunit re-association: percent of plasmid-encoded 30S subunits incorporated into 70S ribosomes

Plasmid	MgCl2 (mM)
	5	10	15
pMM1192	54.0 ± 4.1	69.5 ± 0.8	86.1 ± 5.5
pMF72	12.5 ± 3.5	6.9 ± 2.6	24.1 ± 5.6
pMF23	6.7 ± 2.4	8.6 ± 3.0	16.1 ± 1.5
pMF23/72	13.7 ± 5.1	82.9 ± 1.6	91.5 ± 6.3

Sucrose gradient fractions of 30S subunits and 70S ribosomes were analyzed by primer extension as described in Materials and Methods and in the legend to Table 2.

30S subunits lacking ribosomal protein S20, which interacts with the penultimate stem, have a subunit association defect (19,20). This defect was shown to be associated with a decrease in methylation of several bases of 16S rRNA which are normally modified. The decrease in methylation was most notable for the dimethyl adenine residues at positions 1518 and 1519. In order to determine if the subunit association defect we observed for subunits containing mutated rRNA was also associated with under-modification we analyzed the methylation state of mutant subunits at the dimethyl adenines (bases 1518 and 1519) by primer extension analysis. In these assays the presence of a methylated nucleotide inhibits extension by reverse transcriptase, resulting in a truncated extension product. We used as controls rRNA isolated from a ksgA strain that lacked methylations at 1518 and 1519. No difference in the length of primer extension products was observed for rRNA isolated from ribosomes containing wild-type or mutated rRNA, indicating that expression of mutated rRNA did not result in hypomethylation (data not shown).