Mutations at Two Invariant Nucleotides in the 3 (cid:42) -Minor Domain of Escherichia coli 16 S rRNA Affecting Translational Initiation and Initiation Factor 3 Function*

We have investigated the highly conserved GAUCA sequence of small subunit ribosomal RNA. Within this region, the invariant nucleotides G1530 and A1531 of Escherichia coli 16 S rRNA were mutagenized to A1530/ G1531. These base changes caused a lethal phenotype when expressed from a high copy number plasmid. In low copy number plasmids, the mutant ribosomes had limited effects when expressed in vivo but caused significant deficiencies in translation in vitro , affecting enzy- matic tRNA binding, non-enzymatic tRNA binding, subunit association, and initiation factor 3 (IF3) binding. Mutant 30 S ribosomal subunits showed a 10-fold decrease in affinity for IF3 as compared to wild-type sub- units but showed an increased affinity for IF3 when in 70 S ribosomes. Additionally, IF3 did not promote disso- ciation of 70 S ribosomes, which had mutated subunits as monitored by light-scattering experiments. However, extension inhibition experiments (toeprinting) showed that IF3 retained its ability to discriminate between initiator and elongator tRNAs on mutated subunits. The results indicate that the two functions of IF3, tRNA dis- crimination and subunit dissociation, are separable and that the invariant nucleotides are important for correct subunit function during initiation.

We have investigated the highly conserved GAUCA sequence of small subunit ribosomal RNA. Within this region, the invariant nucleotides G1530 and A1531 of Escherichia coli 16 S rRNA were mutagenized to A1530/ G1531. These base changes caused a lethal phenotype when expressed from a high copy number plasmid. In low copy number plasmids, the mutant ribosomes had limited effects when expressed in vivo but caused significant deficiencies in translation in vitro, affecting enzymatic tRNA binding, non-enzymatic tRNA binding, subunit association, and initiation factor 3 (IF3) binding. Mutant 30 S ribosomal subunits showed a 10-fold decrease in affinity for IF3 as compared to wild-type subunits but showed an increased affinity for IF3 when in 70 S ribosomes. Additionally, IF3 did not promote dissociation of 70 S ribosomes, which had mutated subunits as monitored by light-scattering experiments. However, extension inhibition experiments (toeprinting) showed that IF3 retained its ability to discriminate between initiator and elongator tRNAs on mutated subunits. The results indicate that the two functions of IF3, tRNA discrimination and subunit dissociation, are separable and that the invariant nucleotides are important for correct subunit function during initiation.
Ribosomal RNA (rRNA) plays a significant role in the process of translation, and specific rRNA regions have been implicated in several translational functions (1,2). The best evidence for direct involvement of rRNA is the base-paired interaction between the polypurine Shine-Dalgarno (SD) 1 sequence in mRNA and the polypyrimidine anti-SD region at the extreme 3Ј-end of 16 S rRNA during translational initiation (3) and elongation (4,5). The SD interaction does not occur in eucarya, and the anti-SD region is conserved only in archaeal and bacterial rRNA (6). However, the GAUCA sequence (nucleotides 1530-1534 in Escherichia coli 16 S rRNA, see Fig. 1) immediately upstream of the anti-SD region is highly conserved in all three domains (8) and includes invariant nucleotides at positions 1530 and 1531 (9).
The conserved nature of the GAUCA sequence implies a functional significance for these nucleotides. However, little is known about the role of this site. In experiments done with E. coli ribosomes, kethoxal modification of nucleotide G1530 moderately inhibited subunit association (10) while nucleotide A1531 displayed enhanced reactivity toward chemical modification upon 50 S subunit binding (11). Nucleotide A1531 also displayed enhanced modification upon subunit inactivation (12), a reversible conformation change associated with monovalent or divalent cation depletion. The site is protected from nuclease attack by initiation factor 3 (IF3) binding (13), and the adjacent stem structure has been cross-linked to both IF3 (14) and 23 S rRNA (15). Recently, nucleotide G1530 has been cross-linked to mRNA between the Shine-Dalgarno region and the AUG start codon (16). Based partially on the conserved nature of the region, Kössel et al. (17) proposed that an interaction occurs between the GAUCA sequence and the 5Ј-end of 16 S rRNA as a discrete functional state during elongation, whereas Thanaraj and Pandit (18) have proposed that the GAUCA sequence functions as a translational enhancer by base pairing with a complementary sequence in mRNA upstream of the start codon.
The conserved nature of the GAUCA sequence and its proximity to the anti-SD region, the decoding site, and the terminal helix place it in the center of a very important functional region of the 30 S subunit. Here, we describe experiments designed to investigate the function of the GAUCA sequence, especially the invariant G1530 and A1531 residues. We have switched the order of these two purines on a plasmid-borne copy of the rrnB operon and assayed the effects of the mutations in vivo and in vitro. We found that expression of mutant 16 S rRNA affected several subunit functions in vitro, including initiation complex formation, subunit association, and IF3 binding.

MATERIALS AND METHODS
Bacterial Strains and Plasmids-Strain SU1675 ((FЈ lacI q Tn::5 kan r ) recA ⌬(lac-pro) thi ara), a derivative of CSH26 (19) was used routinely as host in this study. Strain MDA6646 (F Ϫ ilvB1202 ilvH2202 rbs221 ara thi ⌬(lac-pro) ⌬gpt pcnB ⌬recA) (a gift from Dr. E. J. Murgola) was used during the initial cloning to reduce plasmid copy number. Strain XL-1 Blue (Stratagene) was used to propagate M13 phage, and the ung dut strain CJ236 (Bio-Rad) was used to prepare uracil containing M13 DNA. The intact rrnB operon was carried in pKK3535, a pBR322-derived high copy plasmid (20), and in pMO11, a derivative of pSC101, a low copy plasmid (21). A single C to U nucleotide change at position 1192 in the 16 S rRNA gene in both pKK3535 and pMO11 confers spectinomycin resistance yielding pKK1192 and pMM1192, respectively. The A1530/G1531 16 S rRNA mutations were carried on plasmids pMFK161 (a derivative of pKK1192) and pMFM161 (a derivative of pMM1192). The rrnB operon in all plasmids was transcribed from the constitutive P 1 P 2 promoters.
Mutagenesis and Ribosome Preparation-The A1530/G1531 mutation was constructed and ribosomes and ribosomal subunits were prepared basically as described (22). Polyribosomes were prepared as de-scribed (23) and separated by sucrose density gradient centrifugation using polysome buffer (20 mM Tris-HCl, pH 7.6, 100 mM KCl, 15 mM MgCl 2 , 1 mM dithiothreitol). Fractions were collected and precipitated with ethanol. The proportion of plasmid-encoded rRNA in total cellular RNA, 30 S subunits, 70 S ribosomes, and polysome preparations was determined by the method of Sigmund et al. (24). A 32 P 5Ј-end-labeled primer complementary to 16 S nucleotides 1193-1216 was used as described in the legend of Fig. 2. Autoradiograms were scanned using an LKB Ultroscan XL laser densitometer.
In Vitro Assays-Aminoacylation and formylation of tRNA f Met (Sigma) were carried out as described (25) except that a commercial synthetase mix was used (Sigma). Binding of f[ 3 H]Met-tRNA f Met to 30 S subunits was carried out as described (26) except that the mRNA used was a random copolymer of adenosine, guanosine, and uracil (poly-(A,G,U), Sigma, 50 pmol), which contained a random distribution of AUG start codons as well as SD sequences. IF1, IF2, and IF3 were gifts from Dr. Claudio Gualerzi.
The ability of 30 S subunits containing mutant rRNA to associate with 50 S subunits to form 70 S ribosomes was analyzed as described (27) except that MgCl 2 concentrations indicated in the Fig. 2 were used. Toeprint analysis was carried out as described (28,29) using free 30 S subunits (93% mutant rRNA) and bacteriophage T4 gene 32 mRNA.
IF3 Binding Experiments-IF3 was labeled with 5Ј-fluorescein isothiocyanate (FITC) (Molecular Probes). IF3 (1 g/l), dissolved in 0.01 N NaHCO 3 , was incubated with a 5-fold molar excess of FITC solution in freshly distilled Me 2 SO for 2 h at 4°C. The total volume was 50 l. The sample was applied to a 1-ml Sephadex G-25 column equilibrated with 50 mM Tris-HCl, pH 7.6, 250 mM KCl, 0.5 M NH 4 Cl, 5% glycerol. The molar ratio was determined as described previously (30). Experiments were performed in 10 mM Tris-HCl, pH 7.8, 50 mM KCl, 6 mM 2-mercaptoethanol, and MgCl 2 as described under "Results." Fluorescence and light-scattering measurements were performed on a Spex Fluorolog 2 fluorometer, which employed a 450-watt mercuryxenon lamp. In steady-state mode, the excitation monochromator entrance and exit slits were 1.4 mm, while the emission monochromator entrance and exit slits were 2.0 mm. A 1.25-mm variable slit in the path of the light was used to diffuse the incoming light into the sample (preventing photobleaching) and reference detectors.
The equilibrium binding constants for IF3/ribosome interactions were determined by monitoring the change in fluorescence intensity as FITC-labeled IF3 (FITC/IF3) bound to ribosomes. IF3 (1 M) was titrated with increasing amounts of ribosomes in buffer 1 containing 1 mM MgCl 2 . Fluorescence excitation was 488 nm, and emission was monitored at 520 nm. The binding of FITC/IF3 to ribosomes was assumed to proceed through the mechanism A ϩ B % C, where A is the FITC/IF3, B is the ribosomal subunit, and C is FITC-IF3-ribosome complex. Equation 1 defines the binding equilibrium constant: where F ϱ is the fluorescence end point and F 0 is the initial fluorescence point. The following conservation equations were used: . By making the appropriate substitutions into Equations 1 and 2 and converting the solved equation into quadratic form, Equation 3 results. Data were fit by ENZFITTER (Elsevier Science Publishers BV) using Equation 3: where K is the equilibrium binding constant, [A] t is the total FITC/IF3, and [B] t is the total ribosome concentration after each addition. The data were fit by allowing K, F ϱ , and [A] t to vary. The large errors in this experiment result partly from allowing the concentration of IF3 to vary. Small differences in [A] t change K eq . As IF3 is a notoriously sticky protein, it is difficult to work with, and the concentration was known to be a relatively imprecise Ϯ5%. Light Scattering-Equation 4 gives the expression for light scattering (31): where R is Rayleigh's ratio (a measure of the intensity of the scattered light), c is the concentration in g/ml, M is the molecular weight, n is the refractive index, and KЈ is an instrumental constant. Equation 4 is valid as long as the diameter of the molecules to be studied is smaller than a tenth of the wavelength of the incident light, the second virial coefficient can be neglected, and no significant depolarization occurs. These assumptions were previously discussed with regard to the ribosome system (31). Equation 4 can therefore be rewritten as follows: where K ϭ KЈ(òn/òc) 2 and is valid as long as no significant depolarization occurs; cЈ represents concentration (moles/liter).
To remove dust, all samples (1.2 ml of a 0.19 M solution of 70 S ribosomes) were filtered through a sterile 0.22-m Millex-GV filter unit (Millipore). Excitation and emission monochromators were set to 560 nm. Intensity of the light was monitored as ribosomes were titrated with increasing amounts of IF3.

RESULTS
Expression of the 16 S rRNA Mutation G1530A/A1531G-Two invariant nucleotides (G1530/A1531) immediately 5Ј to the anti-Shine-Dalgarno sequence in 16 S rRNA were changed to A1530 and G1531 ( Fig. 1) by oligonucleotide-directed mutagenesis. The mutant DNA from M13 was first subcloned into the high copy plasmid pKK3535. Ligations were transformed into both strain SU1675 and strain MDA6646 (pcnB), which reduces the copy number of pBR322-derived plasmids. Transformants were isolated from the pcnB strain, but no clones were obtained in strain SU1675, suggesting that the mutant rRNA was lethal at high gene dosage. Because the mutant plasmid was stably maintained in a low copy system, the mutations were subcloned into a pSC101-derived vector, pMM1192. This low copy plasmid contains the complete E. coli rrnB operon with the wild-type P 1 P 2 promoters. The resulting plasmid carrying the A1530/G1531 mutations was called pMFM161. Both pMFM161 and pMM1192 carried a second mutation at position 1192 (C to U) in 16 S rRNA as a marker for plasmid-encoded rRNA. The doubling time for SU1675 pMFM161 was 54.2 Ϯ 2.3 min compared to 47.7 Ϯ 0.5 min for SU1675 pMM1192 (wild type). Thus, expression of the A1530/ G1531 mutations, which were lethal at high copy, caused a small but detectable deleterious effect on translation at low copy.
The distribution of plasmid-encoded rRNA in 30 S subunits, 70 S ribosomes, and polyribosomes was determined by primer extension analysis (24) utilizing the identity of nucleotide 1192 in 16 S rRNA to distinguish between plasmid-encoded (U1192) and chromosomally (C1192) encoded rRNA. The results of primer extension analysis of rRNA recovered from polysome gradients is shown in Table I. The distribution of rRNA was similar from both wild-type and mutant plasmids. The presence of pMFM161-encoded rRNA in the polysome fraction indicated that mutant rRNA was assembled into 30 S subunits and actively participated in translation. The fact that mutant rRNA was present in polysomes at near wild-type levels and that growth rates were only slightly decreased by the mutant plasmid indicated that mutant ribosomes were not interfering with translation by wild-type ribosomes.
Subunit Association of A1530/G1531 Ribosomes-To investigate the effect of the A1530/G1531 mutation on subunit association, salt-washed 30 and 50 S subunits were prepared and re-associated by incubating at 37°C in the presence of 1.5, 5, 10, or 15 mM MgCl 2 . Free subunits and 70 S ribosomes were then separated by sucrose density gradient centrifugation in the same buffer conditions. Primer extension analysis of the rRNA from each 30 and 70 S fraction is shown in Fig. 2. The data indicate that the 30 S peaks were enriched with mutant rRNA. Thus, subunits containing the A1530/G1531 mutations were deficient in the ability to form 70 S ribosomes. It is important to note, however, that these subunit re-association assays were carried out in vitro, performed in the absence of translation factors, mRNA, or tRNA. Sucrose density gradients of polysome samples (see Table I) showed mutant rRNA was present in ribosomes and polysomes. Apparently, during initiation of translation in vivo, the subunit association defect of the mutant subunits was suppressed.
In Vitro Initiation Complex Formation on A1530/G1531 Subunits-Given the proximity of the mutant nucleotides to the anti-Shine-Dalgarno region and the IF3 cross-link site (14), it was possible that the A1530/G1531 mutations would affect events of translational initiation. We therefore analyzed the ability of mutant subunits to form the initiation complex in vitro. Activated salt-washed 30 S subunits were used in poly- Met binding assays in the presence of the three initiation factors. The mutant subunit prep-aration used in this experiment contained 65% plasmidencoded (mutant) rRNA. As can be seen in Fig. 3, tRNA binding to mutant subunits was severely reduced relative to wild-type subunits. Indeed, the overall reduction in tRNA binding was almost equal to the amount of mutant rRNA present in the subunit preparation, suggesting that the A1530/G1531 mutations resulted in a complete inability to bind f[ 3 Initiation Factor 3 Binding to A1530/G1531 30 S Subunits and 70 S Ribosomes-Because IF3 has a central role in translational initiation and has been cross-linked very near (nucleotides 1506-1529) to the mutagenized nucleotides (14), the binding of IF3 was examined in more detail. Steady-state experiments were carried out in which the change in fluorescence intensity as FITC-labeled IF3 bound to ribosomes and 30 S subunits was monitored to determine equilibrium binding constants. In these experiments, we took advantage of the propensity of mutant ribosomes to dissociate at lower magnesium ion concentration to obtain a subunit fraction enriched in mutant rRNA. Samples of dissociated mutant ribosomes, which con-   (32). This phenomenon can be visualized by observing the change in light scattering after the addition of increasing amounts of IF3 to 70 S ribosomes. As the 70 S dissociates into 30 and 50 S subunits, the intensity of light decreases. Light-scattering experiments using wild-type and A1530/G1531 mutant 70 S ribosomes are shown in Fig. 4. As expected, at increasing concentrations of IF3 the dissociation of wild-type tight couple 70 S ribosomes into 30 and 50 S components occurs. However, upon addition of IF3 to mutant tight couple 70 S ribosomes (50% mutant rRNA), little change in the intensity of the light was observed, indicating that little or no dissociation of mutant 70 S ribosomes occurred and suggesting that the IF3 dissociative function was compromised. Because of the 30-fold higher affinity of mutant ribosomes for IF3, the IF3 was effectively sequestered, and dissociation of wild-type ribosomes was not observed until much higher concentrations of IF3 were added.
Another function of IF3 is to discriminate between an elongator tRNA and an initiator tRNA binding on the 30 S subunit during initiation complex formation. We used extension inhibition analysis (toeprinting) to analyze this function of IF3 on mutant subunits. In this system, the complex between T4 gene 32 mRNA, tRNA, and 30 S subunits halts the extension of a primer annealed downstream on the mRNA and results in a characteristic band (the toeprint) when the assay is examined by gel electrophoresis (28,29). Under the conditions used in this experiment, the toeprint was strictly dependent on bound tRNA. In the presence of tRNA f Met and tRNA Phe (which decodes the second codon), two stops were observed. A toeprint band corresponding to position ϩ16 of T4 gene 32 mRNA (numbering the A of the AUG start codon as ϩ1) was seen corresponding to tRNA f Met bound in the P-site, and a second stop was at ϩ19 corresponding to tRNA Phe bound at the P-site. As increasing amounts of IF3 were added to mutant subunits in the presence of both tRNAs, the tRNA Phe toeprint band decreased, leaving only the tRNA f Met toeprint (Fig. 5). This demonstrates that the tRNA discrimination function of IF3 was not disrupted by mutant subunits. DISCUSSION Expression of the A1530/G1531 mutant rRNA produced severe effects on translation. It was lethal when expressed from a high copy number plasmid, and even at low copy it had significant, yet subtle, effects on growth rate. In vitro, the Met were bound to a constant amount of activated 30 S subunits from cells containing wild-type (å) or mutant (ϩ) rRNA plasmids in the presence of excess IF1, IF2, IF3, and poly(A,G,U) mRNA. Reaction mixtures were filtered through nitrocellulose, and the radioactivity retained on the filters was determined by scintillation counting. Approximately 65% of the subunits from cells containing the mutant plasmid were the A1530/G1531 mutant subunits as determined by primer extension. mutant subunits displayed explicit functional defects, the most dramatic involving IF3 binding and function. IF3 is bifunctional. It promotes proper selection of the initiator tRNA by recognizing specific determinants on tRNA f Met and promotes subunit dissociation by preferential binding to free 30 S subunits (32)(33)(34). IF3 is an elongated protein consisting of two separate domains (35) and has been shown by footprinting (13,36) and cross-linking (14) studies to interact with the 30 S subunit at both the central and the 3Ј-minor domains of 16 S rRNA (see Ref. 2 and references therein). The central domain binding site includes the 700 region, the 790 loop, and the 840 stem while the 3Ј-binding site includes the 1500 region and the 3Ј-terminal helix. These sites of IF3 interaction with 16 S rRNA correlate well with IF3 function as the 700 region, the 790 loop, and the terminal helix have all been implicated in subunit association and are clustered in the vicinity of the P-site (2). Additionally, a mutation at 791 not only resulted in a 10-fold decrease in IF3 binding, an effect remarkably similar to the effect seen in the A1530/G1531 mutant, but also resulted in decreased subunit association (27), further linking the two functions to the central domain. The 3Ј-end of 16 S rRNA has also been implicated in subunit association (2,9,14,37,38). In the present study, the toeprint data (Fig. 5) indicate that the A1530/G1531 mutation does not alter IF3's ability to discriminate between initiator and elongator tRNAs. However, subunit dissociation is affected. The binding of IF3 does not cause dissociation of mutant 70 S ribosomes, a particularly interesting result since mutant ribosomes have a propensity for dissociation. Thus, it appears that the mutations at 1530 and 1531 uncouple the two functions of IF3 on the 30 S subunit.
Tapprich and co-workers (27) showed that the 10-fold decrease in IF3 binding seen for the mutation at 791 was entirely due to a 10-fold increase in off-rate (k Ϫ1 ), with on-rate (k 1 ) remaining unchanged. This led to their proposal that IF3 can interact with either binding site but requires a specific set of interactions, including G791, to form a stable complex. This model is supported by studies in which IF3 was bound to 30 S subunits lacking the 3Ј-colicin E3 fragment (nucleotides 1493-1542) in which only weak IF3 interactions were seen (39). Our data are consistent with a model in which IF3 first interacts with the 3Ј-domain binding site and is then stabilized by cooperative interactions with the central domain binding site. The loss of the cooperative interactions in subunits containing the A1530/G1531 mutant rRNA could account for the 10-fold decrease in IF3 binding. This model also provides a mechanism for IF3 release upon 50 S binding to the 30 S initiation complex. After proper initiation occurs, 50 S subunits can bind the central domain, eliminating the cooperative interactions and reversing the process of IF3 binding.
The decrease in enzymatic poly(A,G,U) f[ 3 H]Met-tRNA f Met binding may be due, in part, to the inability of mutant subunits to reform the active conformation. Activation, defined by the ability of the subunit to bind N-acetyl-Phe-tRNA Phe in the presence of poly(U) (40), is associated with a conformational change in the 3Ј-minor domain (12). In fact, one of the mutated sites, A1531, has been shown to have enhanced reactivity in inactive subunits (12). Finally, the conserved nucleotides between 1529-1534 have been proposed to base pair with mRNA and function as a translational enhancer (the TP interaction, Ref. 18). Using compensatory base change analysis, we have compared the in vivo expression of a lacZ reporter gene containing a TP site upstream of the SD sequence, which was complementary to either the wild-type or A1530/G1531 mutant 16 S rRNA. No significant change in expression was observed (data not shown). Furthermore, if the region was functioning as a translational enhancer, we might expect differences in spot intensities on two-dimensional protein gels corresponding to increased or decreased complementarity to specific mRNAs, much as was seen in experiments involving Shine-Dalgarno mutants (41). The fact that we see no differences in relative intensities for proteins in two-dimensional gels (data not shown) together with the results of compensatory base change analysis indicates that expression of different proteins was not altered and does not support the translational enhancer model.