Alterations in the Two Globular Domains or in the Connecting {alpha}-Helix of Bacterial Ribosomal Protein L9 Induces +1 Frameshifts

The ribosomal 50S subunit protein L9, encoded by the gene rplI,is an elongated protein with an ${alpha}$ -helix connecting the N- andC-terminal globular domains. We isolated rplI mutants that suppressthe +1 frameshift mutation hisC3072 in Salmonella enterica serovarTyphimurium. These mutants have amino acid substitutions inthe N-terminal domain (G24D) or in the C-terminal domain (I94S,A102D, G126V, and F132S) of L9. In addition, different one-basedeletions in rplI altered either the final portion of the Cterminus or removed the C-terminal domain with or without theconnecting ${alpha}$ -helix. An alanine-to-proline substitution at position59 (A59P), which breaks the ${alpha}$ -helix between the globular domains,induced +1 frameshifting, suggesting that the geometrical relationshipbetween the N and C domains is important to maintain the readingframe. Except for the alterations G126V in the C terminus andA59P in the connecting ${alpha}$ -helix, our results confirm earlier resultsobtained by using the phage T4 gene 60-based system to monitorbypassing. The way rplI mutations suppress various frameshiftmutations suggests that bypassing of many codons from severaltakeoff and landing sites occurred instead of a specific frameshiftforward at overlapping codons.

INTRODUCTION

Top

INTRODUCTION

During protein synthesis, mRNA and tRNA move in a coordinatedway through the ribosome. This coordination is dependent onthe formation and maintenance of cognate codon-anticodon pairing.Once established in the ribosomal A-site, the tRNA-mRNA interactionis maintained in the ribosomal P-site and stabilized via extensivecontacts between the ribosome and tRNA (23, 33). However, occasionallythis process fails and an error in the reading frame maintenanceoccurs. Spontaneous frameshift events are very infrequent, probablyfewer than 10^–5 per codon (24). However, the frequencycan be affected by certain sequences in the mRNA, altered tRNAs,and ribosomal mutants. Certain physiological conditions thatmost likely change the balance between the various componentsof the translation machinery also induce errors in the readingframe maintenance (7-9, 11).

Changes in the reading frame can either be one step backward(–1) or one step forward (+1) by the ribosome. However,the ribosome may also detach from one codon (takeoff site) andresume translation downstream at another codon (landing site)and in such a way bypass many codons (15). One of the most intriguingexamples of bypassing occurs during translation of the bacteriophageT4 gene 60 mRNA, which encodes a subunit of T4 topoisomerase,when the ribosome-bound peptidyl-tRNA disengages from the glycinecodon GGA at position 46 and reengages on an identical GGA codon50 nucleotides downstream (20). T4 gene 60 mRNA and translatednascent peptide contain several stimulatory signals facilitatingthis hop. Alteration in these signals reduces or eliminatesbypassing. On the other hand, this effect can be counteractedby mutations in ribosomal protein L9 (15), implying that thewild-type form of L9 prevents initiation of bypassing, and itspresence may be important for reading frame maintenance in general.

Genetic analysis of ribosomal reading frame maintenance is foundedon the analysis of suppression of frameshift mutations, whichare created by either addition or deletion of nucleotides inthe gene sequence. Such changes in the sequence result in anout-of-frame translation of the mRNA and in most cases the productionof a nonfunctional protein. Correction of the phenotype is possibleif the ribosome slips to the original reading frame. One suchframeshift mutation (hisC3072) has been used in this work. Salmonellaenterica serovar Typhimurium carrying this mutation is unableto grow without the addition of histidine (i.e., it has a His^–phenotype). If the ribosome translating hisC3072 mRNA slipsinto the +1 frame, synthesis of a functional HisC protein occurs,and the bacterium becomes His⁺. We have exploited this phenomenonto isolate external suppressors inducing ribosomal slippageto the +1 frame. One class of mutants obtained by such a selectionwas defective in Formula , resulting in either a reduced concentration of the tRNA or a reduced arginylationof it. It was proposed that inefficient decoding of the AGAcodon by the defective Formula stalls the ribosome at the A-site codon, allowing the wild-type formof the peptidyl Formula to slip forward one nucleotide and thereby reestablish the ribosome in the zeroframe (25). Another class of +1 frameshift suppressor mutants,which are analyzed here, shows that the ribosomal 50S subunitprotein L9 plays an important role in reading frame maintenance,in support of earlier observations (15). We describe mutationssuch as amino acid substitutions in the C or N terminus, variousdegrees of truncation, and even the total absence of L9 thatcause the ribosome to slip probably without any other stimulatorysignals. Interestingly, a break in the ${alpha}$ -helix connecting theN- and C-terminal globular domains of L9 induces frameshifting,suggesting that the geometrical relationship between these globulardomains is important for L9 function in maintaining the readingframe.

MATERIALS AND METHODS

Top

MATERIALS AND METHODS

Bacteria and growth conditions. The bacterial strains used were derivatives of S. enterica (Table1). As rich media, we used either Luria-Bertani medium (2) orthe complex medium NAA (Difco nutrient broth [0.8%]; Difco Laboratories,Detroit, MI), supplemented with aromatic amino acids, aromaticvitamins, and adenine (6). The minimal solid medium was madefrom the basal medium E (41) with 15 g of agar per liter andsupplemented with 0.2% glucose and required amino acids or vitamins(6). TYS-agar (10 g of Trypticase peptone, 5 g of yeast extract,5 g of NaCl, and 15 g of agar per liter) was used as solid richmedium.

View this table:
[in this window]
[in a new window]

TABLE 1. S. enterica strains used in this study

Genetic procedures. Transduction with phage P22 HT105/1 (int-201) (37) was performedas previously described (6).

DNA sequencing was performed on chromosomal DNA or PCR productsaccording to the manual of Applied Biosystems for ABI Prismcycle sequencing ready reaction kit and BigDye. Mutagenesisof the strain GT6861 (zjf-2531<>cat/pSMP24) was performedby inducing the expression of DinB (42, 43) from plasmid pSMP24(29). DinB is a DNA polymerase (DNA Pol IV) mediating template-directedDNA replication but lacking 3'-to-5' proofreading activity.This leads to untargeted mutagenesis and single-base substitutions,as well as one-base deletions (43). Five overnight culturesof zjf-2531<>cat/pSMP24 were diluted (2 x 10⁶)-fold andinoculated into 500 µl of Luria-Bertani broth plus 100µg of carbenicillin/ml and 0.08% L-arabinose to induceDinB synthesis and initiate mutagenesis. After 24 h of growthat 37°C with agitation, phage P22 was added to make phagelysates of each of the five cultures. The phage lysates wereused to infect the His^– strain GT885 (hisO1242 hisC3072).Chloramphenicol-resistant (Cm^r) transductants were selectedon plates containing TYS plus 12.5 µg of Cm/ml and 10mM EGTA and printed onto plates containing minimal E plus 0.2%glucose and 12.5 µg of Cm/ml. His⁺ clones were collectedevery day for 9 days and saved for further analysis. A totalof 50,000 Cm^r transductants were screened for the His⁺ phenotype.Known suppressors of the frameshift mutation hisC3072 induceframeshifting that is only 1% efficient but can be isolatedas His⁺ clones after 48 h of incubation at 37°C (35). ThehisD3018 allele is suppressed by the sufB2 frameshift suppressor,also operating with an efficiency of only 1% (44), and suchHis⁺ clones are detected after 24 h of incubation at 37°C(data not shown). Thus, the His⁺ selection method used herewill detect suppressor mutants with the suppression efficiencyof only a fraction of a percent.

The deletion mutant rplI11<>cat was constructed by insertinga PCR fragment carrying the Cm^r gene cat between the first andlast sense codons of the rplI gene, fully eliminating the rplIcoding sequence (5). The same method was used to create thezjf-2531<>cat insertion but without deleting any chromosomalsequences. To make the A59P alteration in L9, the tetracyclineresistance (Tc^r) (tetRA) cassette was introduced between thecodons for amino acids D60 and V61 on the chromosome as describedpreviously (5), generating strain GT7216 (hisO1242 hisC3070rplI11<>tetRA). Next, we electroporated a 75-nucleotide-longoligonucleotide homologous to the wild-type sequence of rplIin this area, except for the desired mutation [GCT(Ala)-to-CCT(Pro)codon change at position 59] in the middle of the oligo. Tc^sclones were selected on plates, on which only Tc^s, but not Tc^r,clones can grow, as described earlier (30). The obtained GCT-to-CCTchange in the rplI gene was verified by DNA sequencing.

Plasmid pUST274 was constructed by cloning a DNA fragment containingthe frameshift sequence into the BamHI and EcoRI sites of vectorpGHM57 (16). The DNA fragment was made from two complementaryoligonucleotides (5'-TTTGGATCCCGGGGGAAAGACGCCATTCTCTACTGTCCGAATTCTTTT-3' and 5'-AAAAGAATTCGGACAGTAGAGAATGGCGTCTTTCCCCCGGGATCCAAA-3';BamHI and EcoRI sites are in italics), and the ends were trimmedwith BamHI and EcoRI endonucleases. Ligated plasmids were transformedinto strain DH5 ${alpha}$ , analyzed by sequencing the insert and retransformedby electroporation into different S. enterica strains.

Protein analysis. To monitor ribosomal slippage and to purify the slippage product,a previously described system was used (13, 16). This systememploys a fusion protein consisting of maltose-binding protein(MBP) fused to glutathione S-transferase (GST) at its N terminusand having six histidine residues (His₆) at the carboxy terminus(GST-MBP-His₆). Plasmid pUST274 contained an insert, resemblingthe frameshifting site present in hisC3072, at the fusion pointbetween the gst and malE genes that encode the GST and MBP proteins,respectively. The frameshift window relevant for the experimentis shown in the legend to Fig. 2. Production of complete fusionproteins ( $~$ 70 kDa) required translational slippage to the +1frame, whereas if no slippage occurred translation terminatedafter the gst gene, producing only GST (26 kDa). The two formsof protein were confirmed by Western blot analysis with anti-GST-HRPconjugate and ECL Plus detection reagents from Amersham Biosciences.The full-length GST-MBP-His₆ fusion protein was purified bypassing the protein extract over glutathione-Sepharose (AmershamBiosciences) and then over Ni-NTA-agarose (QIAGEN). To stripthe GST part from the frameshift product, purified GST-MBP-His₆was digested by PreScission protease (Amersham Pharmacia Biotech),cleaving the site directly after the GST moiety. Digestion productswere fractionated by gel electrophoresis and the 43-kDa peptide,corresponding to the slippage junction fused to MBP-His₆, wassubjected to N-terminal sequencing by Edman degradation.

View larger version (20K):
[in this window]
[in a new window]

FIG. 2. pUST274 plasmid construct carrying the +1 frameshift site. (A) The malE gene is in the +1 frame relative to the gst gene. The PreScission protease recognition site in the 3' terminus of gst sequence is indicated as a black box, the first stop codon (TAA) encountered in the 0 frame in malE is marked by an asterisk, and the His₆ tag is indicated as a white box. The total length of the frameshift window generating the GST-MalE fusion is: AAG-TAT-ATA-GCA-TGG-CCT-TTG-CAG-GGC-TGG-CAA-GCC-ACG-TTT-GGT-GGT-GGC-GAC-CAT-CCT-CCA-AAA-TCG-GAT-CTG-GAA-GTT-CTG-TTC-CAG-GGT-CCA-CTC-GGG-ATC-CCG-GGG-GAA-AGA-CGC-CAT-TCT-CTA-CTG- TCC-GAA-TTC-CCA-ACT-GAA-AAT-CGA-AGA-AGG-TAA. The UAG and UAA stop codons in +1 and 0 frame, respectively, are in boldface. The underlined portion is the sequence corresponding to the N-terminal sequence generated following digestion by PreScission protease as also shown in the panel. The BamHI and EcoRI sites used for inserting the desired frameshift site are indicated in italics. (B) Frameshifting (FS) and termination (Term) products from the wild type and rplI11<>Cm^r mutant separated on a sodium dodecyl sulfate gel and detected by anti-GST antibodies by Western blot analysis.

RESULTS

Top

RESULTS

Isolation of frameshift suppressor mutations in the rplI gene coding for ribosomal protein L9. The hisC3072 mutant contains a G insertion in a run of G's:GGG GAA AGA (NNN)₂₉ UGA [presented as the zero frame followingthe insertion; the inserted G is underlined, and (NNN)₂₉ standsfor 29 sense codons] and is therefore auxotrophic for histidine(25). External suppressors rendering the hisC3072 mutant ableto grow without addition of histidine were selected. The isolatedsuppressors were expected to induce ribosomal slippage to the+1 frame while translating the hisC3072 mRNA. To obtain suchexternal suppressors, a phage P22 lysate of a random Tn10dTcinsertion pool in the wild-type strain LT2 was treated withthe mutagen hydroxylamine (19) and used to infect strain TR947(GT885; hisO1242 hisC3072). Tc^r clones were first selected and,among these, His⁺ clones were selected. One such Tc^r His⁺ transductantwas selected for further analysis. The location of the Tn10dTcinsertion was determined first by sequencing the chromosomalregions flanking the insertion point, and the sequence showedthat the transposon was in the coding region of the vacB gene,vacB2530::Tn10dTc. Later, transductional mapping indicated thatthe suppressor mutation was only 7% linked to and situated clockwiseof the vacB2530::Tn10dTc insertion. To further map the mutationinducing the His⁺ phenotype, a Cm^r cassette (denoted zjf-2531<>cat)was inserted into the noncoding DNA region between the genessgaE and yjfY, 18 kb from vacB. Since the zjf-2531<>catinsertion appeared to be 99% linked to the mutation inducingHis⁺ phenotype, we explored the neighboring genes. The closestoperon contained genes rpsF, rpsR, and rplI coding for the ribosomalproteins S6, S18, and L9, respectively. This operon was sequenced,and the suppressor of the frameshift mutation hisC3072 was locatedto the gene rplI, resulting in an amino acid substitution G24Din the N domain of L9 (Fig. 1).

View larger version (30K):
[in this window]
[in a new window]

FIG. 1. Alterations in protein L9. The presented protein structure shows L9 from Bacillus (18). The N- and C-terminal globular domains are to the left and right, respectively, with the connecting ${alpha}$ -helix in between. Since protein L9 in Bacillus and Salmonella is conserved, the indicated mutation sites are numbered according to L9 of Salmonella. The amino acid substitutions are in boldface; the frameshift mutations are in italics. An asterisk with the amino acid symbol marks the codon in which it occurred.

Creation of new mutations in the gene rplI. It was of interest to test whether +1 frameshifting on hisC3072mRNA could be induced by other L9 mutants. Mutations in strainGT6861 (zjf-2531::cat/pSMP24) were therefore induced by overexpressionof the dinB gene, and mutations that were closely linked tothe zjf-2531<>cat and that suppressed the His^– phenotypeof TR947(hisO1242 hisC3072) were collected. Nine such cloneswere obtained. The rplI gene was sequenced in all of them, andseveral single-nucleotide substitutions, as well as severalone-nucleotide deletions, were identified. Altogether, the originalrplI2 mutation and later-characterized mutations had amino acidsubstitutions in the N domain (G24D) or the C domain (I94S,A102D, G126V, and F132S) of L9, and they primarily affectedconserved structural elements of the protein (Fig. 1). Differentframeshift mutations in rplI altered either the extreme portionof the C terminus or induced truncations in the C-terminal domainwith or without the connecting ${alpha}$ -helix of L9 (Fig. 1). A deletionmutant of rplI (rplI11<>cat) was also created by replacingthe coding sequence with the Cm^r cassette, proving again thatthe L9 is not an essential protein. Under standard lab conditionsbacteria lacking L9 showed no detectable growth impairment.However, the rplI11<>cat mutation was suppressing His^–phenotype of hisC3072 containing mutant. Since the rplI geneis the last gene in the operon, the induced His⁺ phenotype couldnot be related to a potential polar effect of Cm^r cassette'sinsertion. Thus, a frameshift mutation in hisC3072 could besuppressed by a variety of mutations in ribosomal protein L9,ranging from amino acid substitutions to truncations that alterits function or prevent incorporation into the ribosome, andeven by the absence of L9. All of these mutations suppressedthe hisC3072 mutation, as judged by growth on plates lackinghistidine (Table 2).

View this table:
[in this window]
[in a new window]

TABLE 2. Suppression of the hisC3072 mutation by different rplI mutations

The geometrical relationship between the two globular domains of L9 is important to prevent +1 frameshifting. In the intact 70S ribosome, the N-terminal domain of L9 bindsto the ribosome below the L1 stalk, whereas the C-terminal domainextends more than 50 Å from the surface of the 50S subunitdue to the presence of a rigid ${alpha}$ -helix connecting the N- andC-terminal domains (45). The connecting ${alpha}$ -helix of conservedlength was suggested to improve the reading frame maintenanceby determining a proper spacing between ribosomes on the mRNA(16) or looping back to interact with the ribosome, to whichits N-terminal end is bound, as demonstrated by footprintingdata (27) and by cryo-electron microscopy (31). To test therole of the ${alpha}$ -helix in frameshifting, an amino acid substitutionwas made in the middle of it. The Ala59-to-Pro59 change waspredicted to break the a ${alpha}$ -helix and therefore alter the relativeorientation of globular domains flanking the helix. Since insuch a mutant the RNA-binding properties of individual domainremain unchanged, the A59P mutant L9 might still be binding,at least partially, to the ribosome. The L9 with broken ${alpha}$ -helixinduced a +1 frameshift, albeit weaker than the suppressionmediated by the other alterations of L9 (Table 2). These datasuggest that even a small aberration of the conserved connecting ${alpha}$ -helix of L9 induce errors in maintaining the correct readingframe.

Characterization of ribosomal frameshift product in a mutant lacking L9. Plasmid pUST274 was used to analyze ribosomal shifts to the+1 frame. This plasmid codes for a GST-MBP-His₆ fusion proteinwith a coding sequence changing frame, similar to that presentin hisC3072, between the genes for GST and MBP. A complete GST-MBP-His₆fusion protein was only synthesized when the ribosome was shiftedto the +1 frame. Alternatively, translation terminated at theUAA stop codon found downstream of the frameshift sequence,producing only the GST moiety (Fig. 2A). Total protein extractswere prepared from the rplI11<>cat deletion mutant andthe wild type, both containing plasmid pUST274. The terminationproduct and the complete fusion protein were analyzed by Westernblotting (Fig. 2B). The protein extract obtained from the strainlacking L9 had a band corresponding to complete fusion protein,which was absent in the wild type. However, we noticed thatthis protein was not forming a sharp band on the gel, suggestingthat the product was not homogenous. Using the GST and His₆affinity tags, we purified the fusion protein from the hisO1242his3072 rplI11<>cat/pUST274 strain and digested it withPreScission protease. The resulting liberated slippage junctionfused to the MBP-His₆ was subjected to N-terminal sequencing.However, two trials resulted in two different mixed amino acidsequences, with none of them resembling a peptide as deducedfrom the tested frameshift DNA sequence. These results suggestthat if one base frameshift occurs at overlapping codons itmust occur on multiple sites. Alternatively, the results areconsistent with bypassing using various takeoff and landingsites as L9 was earlier shown to induce (15).

The +1 frameshifts mediated by defective ribosomal protein L9 require a long frameshift window. Since the amino acid sequence of the slippage junction peptideindicated a lack of specificity for ribosomal frameshiftingwhen L9 was absent, we tested whether those mutant ribosomessuppressed other frameshift mutations. The rplI2 (G24D) mutationwas therefore introduced into strains carrying various frameshiftmutations in different his genes, rendering them His^–,and the double mutants were monitored for the ability to growwithout the addition of histidine. Of 15 his mutations, onlyfour were suppressed: hisB6480, hisD2780, hisF2439, and hisC3072(Table 3). The hisB6480 and hisD2780 mutations have a characteristicrun of C's and are suppressed by sufA, which induces frameshiftingat such CCC sites (36). The hisF2439 and hisC3072 mutationshave a characteristic run of G's and are suppressed by sufD,which induces frameshifts at GGG sites (34). However, the otherfive CCC-containing sites and three GGG-containing sites werestill His^– in the presence of the rplI2 mutation. Theseresults indicated the absence of specificity for suppressionof runs of C's or G's mediated by an altered L9. Nevertheless,the suppressed frameshift mutants had one feature in common,since in all of them the frameshift window was at least 31 codonslong. A long frameshifting window increases the probabilityfor the ribosome to shift frame at multiple places or to bypassseveral codons before the stop codon is reached. Of course,this requires that produced protein tolerates short portionsof aberrantly inserted amino acids without losing its function.The frameshift window in the hisD3068 mutant is of 42 codonsbut was not suppressed by the rplI2 mutation. This could beexplained because the third codon downstream the G insertionpoint is coding for conserved His326 involved in the catalysisreaction performed by HisD (40). The mutated HisD produced bythe double hisD3068 rplI2 mutant most likely has Thr326 instead,and therefore the strain was His^–. Similarly, the hisG3037frameshift mutant (containing a 26-codon-long frameshiftingwindow) was not suppressed by rplI2 because the G insertionpoint is located adjacent to and 3' of the sequence coding forthe conserved 13-amino-acid-long binding site for 5-phosphoribosyl-1-pyrophosphate,which is one of the substrates for HisG (28). HisG may not tolerateamino acid changes at this site, and the mutated HisG producedafter rplI2 suppression may not be active. The hisD3018, hisD6610,hisD3749, hisD6580, and hisF6527 mutants have short frameshiftwindows of four to eight codons, and none of them was suppressedby rplI2 (Table 3). The hisD3749 mutation is located only 13codons from the N-terminal end, which can sustain various aminoacid substitutions without affecting the activity of HisD (12,21), and still it was not suppressed by the rplI2 mutation.We conclude that mutations in rplI suppressed all tested hismutants that had long frameshifting windows located in the sequencescoding for nonessential parts of the proteins. The requirementfor a long frameshift window is consistent with ribosomal bypassing,where more than one takeoff and landing site could be used,as has been shown earlier for ribosomes with defective L9 (15).Apparently, an altered L9 can induce ribosomal frameshiftingwithout any other stimulatory signals present on the mRNA.

View this table:
[in this window]
[in a new window]

TABLE 3. Suppressor specificity of different frameshift sites monitored by growth on medium lacking histidine

DISCUSSION

Top

DISCUSSION

Here we describe alterations in the ribosomal protein L9 thatinduce frameshifting to suppress a chromosomal +1 frameshiftmutation in the hisC gene. Most of the alterations obtainedwere located in the C terminus of L9, and only one was in itsN-terminal part. A break in the ${alpha}$ -helix was introduced by changingan Ala-to-Pro change at position 59 within this helix, whichalso induced frameshifting, suggesting that the geometric relationshipbetween the two globular domains is important for L9 to improvereading frame maintenance.

All mutants isolated here were required to synthesize a functionalHisC protein for growth. This was possible only via shifts tothe +1 frame during the translation of hisC3072 mRNA. However,among the His⁺ mutants it was not possible to differentiatewhether the functional HisC protein was synthesized via +1 frameshiftingto an overlapping codon or via bypassing to a downstream codonin the +1 reading frame. The efforts to identify the point offrameshift or bypassing by N-terminal sequencing of the frameshiftproduct (Fig. 2) gave no clear answer. The requirement for longframeshift windows (Table 3), the synthesis of several frameshiftproducts as revealed by a fuzzy band on the gel (Fig. 2), andno definitive sequence over the frameshift site is consistentwith either a shift within overlapping codons at multiple placesor bypassing from several take-off sites over many codons occursproducing different frameshift peptides. An increase in bypassingand the diversity of landing sites have been previously observedin L9-deficient ribosomes (17).

Protein L9 is part of the 50S ribosomal subunit and is a highlyelongated protein consisting of two globular domains (N domainand C domain) separated by a connecting nine-turn ${alpha}$ -helix. Eachdomain has conserved aromatic and positively charged amino acidresidues on their surface, allowing them to bind to rRNA (18).The binding of the N domain was localized to domain V of 23SrRNA close to the base of the L1 stalk (27, 31, 38, 45). TheL1 stalk is actively involved in the translocation movementof tRNA from the P site to the E site (32). The C domain andthe connecting helix of L9 undergo significant movement duringribosomal cycling (10). The amino acids involved in RNA bindingare clustered together in the N domain, whereas amino acidspredicted to be involved in RNA binding in the C terminus aremore scattered (18). Therefore, even though the changes in theC-terminal domain are close to the putative RNA-binding sites,there still may be enough sites left for interaction with theRNA and allow such altered L9 to assemble into the ribosome.

The functional aspect of L9 has been studied only by analyzinghow its various alterations influence the translational bypassingof an mRNA encoded from a multicopy plasmid containing the bacteriophageT4 gene 60 site (15). Although the first mutant (S93F) isolatedwas a genomic mutant (14), all other characterized mutants wereobtained after mutagenesis of the rplI gene located on a plasmid(1). Such an approach forces the mutant form to compete withthe wild-type protein encoded by the chromosomal allele rplI⁺.Thus, to manifest a phenotype the mutant must be dominant. Herewe have only isolated mutants with altered L9 encoded on thechromosome, which allows the detection of both dominant andrecessive rplI mutations. Since L9 is not an essential protein,some mutant forms of L9 may not be incorporated into ribosomesand phenotypically would be equivalent to an rpsI deletion.The unbiased selection for frameshift suppressors in the rplIarea on the chromosome allowed isolation of rplI mutants withunchanged gene expression levels and producing a homogeneousribosome population containing only mutated or no protein L9.Hence, the approach taken in the present study is distinctlydifferent from that used by Adamski et al. (1). Mutant analysisin the present study relied on the ability of the altered L9to induce frameshifting on a chromosomally located +1 frameshiftmutation (hisC3072), a very sensitive system to detect low levelsof frameshifting.

Although the number of mutants is rather low, all (four aminoacid substitutions and all frameshift mutations, see Fig. 1)except one resulted in an altered or deleted C-terminal globulardomain. The exception is the mutant G24D (rplI2) in the N-terminaldomain. This amino acid substitution is at the same positionas the mutations G24A and G24P isolated earlier (1). Althoughthe latter two alterations did not induce bypassing, purifiedL9 protein containing G24P bound poorly to a 23S rRNA fragment,which suggests that such altered L9 is not assembled into ribosomes.Since the absence of L9 induces bypassing, it seems likely thatG24P is a recessive mutation allowing the wild-type L9 encodedfrom the chromosome to outcompete the mutant form in the ribosomeassembly process resulting in a wild-type ribosome. The G24D-alteredL9 isolated here induced frameshifting that could also dependon impaired rRNA binding, which hindered G24D-containing L9to be incorporated into the ribosome. In addition, the N-terminaldomain is also the dominating RNA-binding domain and, in fact,in the three-dimensional structure of the 70S ribosome the Ndomain is buried in the 50S subunit, and the ${alpha}$ -helix and C domainare extended beyond the surface of the ribosome (45). All thistogether suggests that the discrepancy between the phenotypesinduced by the various amino acid substitutions at position24 of L9 is likely explained by the presence or absence of awild-type form of L9.

Five isolated frameshift suppressors had frameshift mutationsin the rplI gene (Fig. 1). One of them, rplI1, caused a truncationof L9, leaving only a small portion of the N-terminal domain,which was probably too small to be incorporated into the ribosome.Two other suppressors, rplI3 and rplI4, encoded a L9 proteinwith an intact N domain, but lacking the ${alpha}$ -helix and the C domain.Mutation rplI5 omitted only the C domain of L9, while rplI10mutation changed the extreme C terminus of L9 and extended itby 16 amino acids. It is possible that differently truncatedforms of L9 with the intact N domain and with or without partsof the C domain, as in the rplI3, rplI4, rplI5, and rplI10 mutants,could be assembled into the ribosome. However, we cannot ruleout that these frameshift mutations render L9 unable to bindto the ribosome and therefore act like a deletion of rplI.

Mutant L9 proteins isolated earlier were shown to be dominant,suggesting that they are assembled into the ribosome and stimulatebypassing (1). The amino acid substitutions in the C domainof L9, isolated here as inducing suppression of +1 frameshiftmutations, are located close to already-described amino acidsubstitutions (1). Moreover, based on the fact that our andthe earlier-described mutations alter the same area of L9, weassume that L9 harboring I94S, A102D, or F132S could also beassembled into the ribosome. The alteration at position 126(G126V) mediated frameshifting (Table 2), although the G126Wand G126V (i.e., the same alteration as ours) isolated earlier,did not induce bypassing (1). The explanation of this discrepancyis either the presence of dominating wild-type copy of L9 inthe analysis of bypassing by Adamski et al. (1) or differencesin sensitivity of the assays used. In conclusion, by using differentsystems we show that the alterations in the C-globular domainresult in less restrictive reading frame maintenance, supportingthe role of L9 in translational bypassing as seen in bacteriophageT4 gene 60-derived mRNA.

The length of the rigid connecting ${alpha}$ -helix of L9, but not itsamino acid sequence, is conserved. However, inserting four alaninesor removing four amino acids of the ${alpha}$ -helix does not increasebypassing, whereas altering the length by one or two alaninesdoes (1). Such mutations alter the length of the ${alpha}$ -helix andthe relative positions of the globular domains. A four-amino-acidinsertion or removal changes the alignment less than an insertionof one or two amino acids, suggesting that the geometric relationshipbetween the two domains is important (1). The Ala-to-Pro substitutionat position 65, as constructed by Adamski et al. (1), or atposition 59, as constructed here, should both break the ${alpha}$ -helixand misalign the globular domains. Whereas A65P does not inducebypassing (1), the A59P alteration induced weak frameshifting(Table 2). A suppression level induced by the A58P mutationlower than that observed in the mutants lacking L9 hints thatthe A59P mutant protein is incorporated into the ribosome butdoes not function to full extent. The difference in the A65Pversus A59P mutant results could depend on the differences inthe assay systems. The A65P-altered L9 was synthesized froma plasmid and, because the wild-type L9 was produced from thechromosome, the ribosome population was heterogeneous. On theother hand, the A59P protein was synthesized from the chromosome,and no other form of L9 was present; therefore, only the mutantform of ribosomes was assembled. Moreover, the assay monitoringframeshift suppression of the hisC3072 mutation is most likelymore sensitive than the method used by Adamski et al. (1) (seeMaterials and Methods). We conclude that the geometric orientationof two globular domains is important to correctly position theC domain to the same ribosome (looping back) or, if the functionof L9 is to act like a "strut," to a neighboring ribosome onthe mRNA.

ACKNOWLEDGMENTS

This study was supported by grants from the Swedish Cancer Foundation(project 680) and the Swedish Science Research Council (projectBU-2930).

We thank Gunilla Jäger, Kristina Nilsson, and Kerstin Jacobssonfor technical assistance and Per-Ingvar Ohlsson at the Departmentof Plant Physiology, Umeå University, and the ProteinAnalysis Center at the Department of Medical Biochemistry andBiophysics, Karolinska Institute, for protein sequencing. Wealso thank Norma M. Wills and John F. Atkins for providing thevector pGHM57; Dan Andersson, Uppsala, Sweden, for providingthe plasmid pSMP24 before its publication; and John Roth, Universityof California at Davis, for the generous gift of numerous strainscontaining his mutations. We also thank John Atkins, Tord Hagervall,Joakim Näsvall, and Jaunius Urbonavi $c$ ius for critical readingof the manuscript.

FOOTNOTES

* Corresponding author. Mailing address: Department of Molecular Biology, Umeå University, S-90187 Umeå, Sweden. Phone: 46-90-7856756. Fax: 46-90-772630. E-mail: glenn.bjork{at}molbiol.umu.se

Published ahead of print on 27 July 2007.

Present address: Children's Hospital Oakland Research Institute,5700 Martin Luther King, Jr. Way, Oakland, CA 94609.

REFERENCES

Top