An extended Shine–Dalgarno sequence in mRNAfunctionally bypasses a vital defect in initiator tRNASunil Shettya, HimaPriyanka Nadimpallia, Riyaz Ahmad Shaha, Smriti Aroraa, Gautam Dasa, and Umesh Varshneya,b,1

aDepartment of Microbiology and Cell Biology, Indian Institute of Science, Bangalore 560012, India; and bJawaharlal Nehru Centre for AdvancedScientific Research, Bangalore 560064, India

Edited by Dieter Söll, Yale University, New Haven, CT, and approved September 2, 2014 (received for review June 20, 2014)

Initiator tRNAs are special in their direct binding to the ribosomalP-site due to the hallmark occurrence of the three consecutive G-Cbase pairs (3GC pairs) in their anticodon stems. How the 3GC pairsfunction in this role, has remained unsolved. We show that muta-tions in either the mRNA or 16S rRNA leading to extended inter-action between the Shine–Dalgarno (SD) and anti-SD sequencescompensate for the vital need of the 3GC pairs in tRNAfMet forits function in Escherichia coli. In vivo, the 3GC mutant tRNAfMet

occurred less abundantly in 70S ribosomes but normally on 30Ssubunits. However, the extended SD:anti-SD interaction increasedits occurrence in 70S ribosomes. We propose that the 3GC pairsplay a critical role in tRNAfMet retention in ribosome during the con-formational changes that mark the transition of 30S preinitiationcomplex into elongation competent 70S complex. Furthermore, treat-ing cells with kasugamycin, decreasing ribosome recycling factor (RRF)activity or increasing initiation factor 2 (IF2) levels enhanced initiationwith the 3GC mutant tRNAfMet, suggesting that the 70S mode ofinitiation is less dependent on the 3GC pairs in tRNAfMet.

3GC base pairs | 70S initiation | rrsC | G1338 | A1339

Initiation of protein synthesis, assisted by initiation factors, is ahighly regulated process in all life forms. In eubacteria, binding

of both the initiator tRNA (tRNAfMet) and mRNA to the smallribosomal subunit (30S) leads to the formation of a 30S pre-initiation complex primarily with the help of the three initiationfactors (IF1, IF2, and IF3). This stage is then followed by dockingof the large ribosomal subunit (50S) to ultimately produce anelongation competent 70S complex upon the departure of allof the three initiation factors (1). The localization of mRNAonto the 30S subunit is facilitated by a purine rich sequence(Shine–Dalgarno, SD sequence), located upstream of the startcodon, by its pairing with a complementary sequence (anti-SDsequence) at the 3′-terminus of the 16S rRNA (1, 2). ThetRNAfMet binding to ribosome is aided by the unique featuresit possesses. A virtually universal feature of all of the initiatortRNAs, the presence of three consecutive G-C base pairs(G29G30G31:C39C40C41, referred to as 3GC pairs) in the an-ticodon stem is known to be important for their preferentialbinding in the ribosomal P-site (3, 4). Mutations in the 3GCpairs result in poor binding of tRNAfMet to the ribosomalP-site (3, 4).However, the mechanism of how the 3GC pairs help in binding

of tRNAfMet into the ribosome has remained unclear. In the crystalstructure of the initiator tRNA bound 70S ribosome, it was seenthat the universally conserved A1339 and G1338 residues of 16SrRNA establish A-minor interactions with the first GC (G29-C41)and middle GC (G30-C40) pairs, respectively (5). Although theseinteractions were seen as suboptimal, the IF3 induced conforma-tional changes may optimize these interactions (6). Another studyshowed that a major role of IF3 is to uniformly increase the rate ofdissociation of all tRNAs (including tRNAfMet) from the 30Spreinitiation complexes (7). These observations suggest that therole of 3GC pairs in conferring specificity for the ribosomal P-sitebinding is an open question.

Formation of elongation competent 70S complex is a multi-step process wherein both the tRNAfMet and IF2 assist dockingof 50S subunit onto the 30S preinitiation complex. The 70S complexso formed is in a ratcheted state with tRNAfMet bound in the P/Istate wherein the tRNA anticodon stem is placed in the P-site andthe acceptor stem is placed between the P-, and E-, sites. For sucha 70S complex to enter into the elongation phase, the ratchetedstate must unratchet to localize the tRNAfMet in the classical P/Psite and make the A/A site available for the entry of an elongatortRNA. The unratcheting is mediated by the hydrolysis of the IF2bound GTP (8). What specific interactions between the ribo-some and tRNAfMet facilitate retention of the latter into the 70Scomplex during these conformational transitions is not known.Whether the 3GC pairs in the anticodon stem of tRNAfMet playany specific roles during these transitions is also not known.To further our understanding of the role of the 3GC pairs in

the initiation process, we have used a genetic approach to obtainsuppressor strains of Escherichia coli that allow initiation to oc-cur with the tRNAfMet having mutations in the 3GC pairs (9). Inthis study, characterization of one of the isolates, B21, showedthat C1536-to-T1536 mutation in the 3′ terminal region of the16S rDNA, resulting in its extended interaction with the SD re-gion of the reporter mRNA, allowed initiation to occur with themutant tRNAfMet lacking the highly conserved feature of the3GC base pairs in the anticodon stem.

ResultsIn Vivo Assay and Characterization of the B21 Mutation EnablingInitiation with the 3GC Mutant tRNAfMet in E. coli. The plasmid,pCATam1 (AmpR) encodes chloramphenicol acetyltransferase(CAT) reporter mRNA with UAG start codon (as opposed toAUG), which is not translated under normal physiological

Significance

Initiation, a regulatory step in mRNA translation is character-ized by initiator tRNA (tRNAfMet) binding to 30S ribosome to-gether with initiation factors (IFs) to form 30S pre-initiationcomplex (PIC). This step is followed by binding of 50S ribosomeand release of IFs to form elongation competent 70S complex.tRNAfMet is special in possessing a vital feature of three con-secutive G-C (3GC) base pairs in its anticodon stem. However,the role of this feature has remained unclear. We show that the3GC base pairs facilitate tRNAfMet retention in the ribosomeduring the transitions that mark conversion of 30S PIC into 70Scomplex. Furthermore, we show that translation of mRNAshaving an extended Shine–Dalgarno sequence bypasses therequirement of the 3GC pairs in tRNAfMet.

Author contributions: S.S. and U.V. designed research; S.S., H.N., R.A.S., S.A., and G.D.performed research; S.S. and U.V. analyzed data; and S.S. and U.V. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.1To whom correspondence should be addressed. Email: varshney@mcbl.iisc.ernet.in.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1411637111/-/DCSupplemental.

E4224–E4233 | PNAS | Published online September 22, 2014 www.pnas.org/cgi/doi/10.1073/pnas.1411637111

Dow

nloa

ded

by g

uest

on

Dec

embe

r 29

, 202

0

conditions. Thus, its (pCATam1) presence does not conferchloramphenicol (Cm) resistance to E. coli (Fig. 1 B, ii, sector 1).However, a simultaneous presence of the mutant metY gene,metYCUA, which encodes tRNAfMet with CUA anticodon (inplace of CAU) in pCATam1metYCUA, results in initiation from theUAG start codon of the reporter mRNA conferring Cm resistance(CmR) to the host (Fig. 1 A, i and B, ii, sector 2) (10). Not un-expectedly, when the metYCUA is further mutated to replace thehighly conserved 3GC pairs in the anticodon stem with U29:A41,C30:G40, A31:Ψ39, the new construct (pCATam1metYCUA/3GC)fails to confer CmR because the encoded tRNAfMet (called the3GC tRNA hereafter) fails to successfully participate in initia-tion (Fig. 1 A, ii, and B, ii, sector 3) despite its efficient amino-acylation and formylation (3, 10, 11).In our earlier work (9), we isolated E. coli B21 strain wherein

a chromosomal mutation rescues initiation defect of the 3GCtRNA to confer CmR (Fig. 1 B, ii, sector 6). As a control, the B21harboring pCATam1 remains Cm sensitive (CmS), and retains CmR

with pCATam1metYCUA (Fig. 1 B, ii, sectors 4 and 5). As yet another

control, all transformants grow on Amp alone plate (Fig. 1 B, i).Genetic mapping using P1 phage mediated transductions(SI Results) localized the site of mutation in B21 at 84.5 min inthe E. coli chromosome (Fig. S1A). At this location, the rrsCgene (encoding for one of the seven 16S rRNA genes) was themost likely candidate. DNA sequence analysis revealed a novelC-to-T mutation in rrsC corresponding to a C to U change atposition 1536 (termed as c1536u or the B21 mutation) in theencoded 16S rRNA (Fig. S1B) in the anti-SD (aSD) region (12),which resulted in an extended interaction with the SD region ofthe reporter mRNA from 6 bp to 8 bp (Fig. 1C, compare i and ii).

Impact of the Extended SD:aSD Interaction on Initiation with the 3GCMutant and the Wild-Type tRNAfMet. To generate a converse systemof extended SD:aSD interaction, we mutated SD sequence ofthe CAT reporter mRNA from UAAGGAAG to UAAGGAGGin pCATam1(a7g) (SDa7g construct) to allow an extended in-teraction with the wild-type 16S rRNA (Fig. 2A). To check theimpact of the extended SD on initiation with the 3GC tRNA, the

A

B

C

Fig. 1. Assay system and characterization of B21 suppressor strain. (A, i) The plasmid pCATam1metYCUA harbors CATam1 and metYCUA genes that encodeCATam1 mRNA with UAG initiation codon (am1) and tRNAfMet having CUA anticodon, respectively. The metYCUA encoded tRNA initiates from the CATam1

mRNA to produce CAT protein and confers CmR to the host. The relevant sequences of anticodon stem loop (box with dashed line) are as shown. (ii) Theplasmid pCATam1metYCUA/3GC is derived from pCATam1metYCUA wherein the metYCUA gene has been further mutated in the three consecutive G-C base pairs.The encoded tRNA (3GC tRNA) fails to initiate from CATam1 mRNA rendering the host CmS. (B) Growth of E. coli KL16 parent strain (sectors 1–3) and thederived suppressor B21 (sectors 4–6) harboring either pCATam1, pCATam1 metYCUA or pCATam1 metYCUA/3GC on LB-agar plates containing, Amp (i) and Ampplus Cm (100 and 30 μg/mL, respectively) (ii). Overnight cultures were streaked and incubated at 37 °C for ∼18 h. (C) Interaction of CATam1 SD sequence withthe wild-type (i) or the B21 mutant (ii) anti-SD sequence of the 16S rRNA (rrsC).

Shetty et al. PNAS | Published online September 22, 2014 | E4225

GEN

ETICS

PNASPL

US

Dow

nloa

ded

by g

uest

on

Dec

embe

r 29

, 202

0

pCATam1metYCUA/3GC or pCATam1(a7g)metYCUA/3GC plasmidswere introduced into the parent strain, KL16, and its derivativeB21. CAT assays showed that compared with the original construct(SDwt), the a7g construct (SDa7g) resulted in an efficient initiationwith the 3GC tRNA even in the parent strain (KL16). In fact, theinitiation in KL16 was higher than that seen in B21 (Fig. 2B). Aplausible reason for this is that although CATam1 mRNA (SDwt)could be preferably used by only a single gene (rrsC) encoded 16SrRNA in B21, the CATam1(a7g) mRNA (SDa7g) could be used by16S rRNA encoded by all of the seven genes in KL16, and by six ofthe seven genes in B21 (the seventh having the B21 mutation).This observation suggests that the extended SD:aSD interactionalone is sufficient to allow efficient initiation by the 3GC tRNA.Similar to earlier studies (13, 14), we observed that compared

with SDwt, the extended SD (SDa7g) caused a decrease in initi-

ation by metYCUA retaining wild-type sequence in its anticodonstem (Fig. 2C, compare bars 1 and 2). However, such an adverseeffect of SDa7g was not seen when, of the three consecutive G-Cbase pairs, we mutated only the first to A-U inmetYCUA/AU (bars 3and 4) or the third to G-U in metYCUA/GU (bars 5 and 6). In fact,the metYCUA/GU construct showed an increase in initiation. Ascontrol, the metYCUA/3GC construct showed an expected increase(bars 7 and 8). In addition, we observed that increased productionof CATam1 mRNA (by introducing another copy of the CATam1gene on another medium copy plasmid, pACDHCATam1, com-patible with the pCATam1) in KL16 did not compensate for the3GC mutations (Fig. 2 D and E) suggesting that the mechanismof rescue of the 3GC defect, in B21 was unlikely due to merelya direct impact of the extended SD on the initial binding of theCATam1 mRNA to the ribosome, or on an early event in initiation.

A

BC

ED

Fig. 2. Effect of extended SD on translation initiation. (A) Schematics of interaction of SDa7g of CATam1 with anti-SD (aSD) of wild-type 16S rRNA. (B) Ini-tiation with 3GC tRNA as assayed by CAT activity in cell-free extracts of E. coli KL16 or B21 strains harboring CATam1 reporter in the context of SDwt or SDa7g.Fold differences in CAT activities with respect to 3GC tRNA in KL16 are shown. The CAT activity in KL16 for the SDwt context was 110.4 ± 58.8 pmol ofacetylated (Ac) Cm produced per μg of total protein in 20 min at 37 °C. (C) Initiation from CATam1 with metYCUA (bars 1 and 2), metYCUA/AU (bars 3 and 4),metYCUA/GU (bars 5 and 6), and metYCUA/3GC (bars 7 and 8) tRNAs in the context of SDwt or SDa7g. The fold differences in CAT activities with respect tometYCUA

in SDwt context are shown. CAT activity of metYCUA in SDwt context was 5236.4 ± 1204.5 pmol Ac-Cm per μg of total protein. (D) CAT assays from cell-freeextracts of E. coli KL16 harboring pCATam1metYCUA/3GC along with either the empty vector, pACDH (bars 1 and 3) or another construct for CATam1 mRNA,pACDHCATam1 (bars 2 and 4), with (bars 3 and 4) or without (bars 1 and 2) 0.5 mM IPTG. Absolute CAT activity for the reaction represented by bar 1 is 116.1 ±18.3 pmol Ac-Cm per μg total protein. (E) Assay of overexpression of CATam1 mRNA by semiquantitative PCR.

E4226 | www.pnas.org/cgi/doi/10.1073/pnas.1411637111 Shetty et al.

Dow

nloa

ded

by g

uest

on

Dec

embe

r 29

, 202

0

In another experiment, analysis of the ribosome bound fraction ofthe CATam1 mRNA in B21 by Northern blotting and RT-PCR,showed ∼2- to 4-fold increase (compared with KL16). Likewise, theCATam1(a7g) mRNA also increased ∼3.7- to 7-fold in the KL16ribosomes (Fig. 3 A, i and B). We noted that the levels determinedby Northern blotting were lower. More importantly, levels of the3GC tRNA on ribosomes also increased ∼threefold (Fig. 3 A, ii).It may be noted that, immediately upstream of the SD sequence

in the CATam1 reporter, yet another sequence, AGGAG, which,although suboptimally placed with respect to the start codon,could also serve as an SD sequence. However, as shown in Fig. S2,mutations in this sequence did not result in any significant effects.

Distribution of the Wild-Type tRNAfMet and the 3GC tRNA with CAUAnticodon (3GCCAU) in 30S and 70S Ribosomes. Polysome prepara-tions from KL16 strain harboring pmetY3GC encoding 3GCCAUtRNA were fractionated on sucrose density gradients (20–40%)(Fig. 4 A, i) and analyzed for the presence of the wild-type tRNAfMet

and the 3GCCAU tRNAs across the sedimentation profile by dotblot hybridizations (Fig. 4 A, ii). As shown in Fig. 4 A, iii, in thefractional distribution of the tRNAs across the gradient frac-tions 6–17, the free tRNA levels, and those bound to the 30S ribo-somes were represented by nearly the same levels (Fig. 4 A, iii, F6-F7, and F8-F10). However, in 70S fractions (F13–F17), the wild-typetRNAfMet bound more prominently than the 3GCCAU tRNA. Sucha fractional distribution of tRNAs shows that the 3GCCAU tRNA isnot defective in binding to 30S, and suggests that the primary role of

the 3GCpairs may be during the transit of tRNAfMet from 30S to 70Scomplex. To validate these results, in yet another experiment (Fig.4B), we collected the 30S, 50S, and 70S ribosome fractions separately(Fig. 4 B, i) and analyzed them by Northern blotting (Fig. 4 B, ii).In this analysis, whereas the wild-type tRNAfMet was mostly foundin the 70S complex, only a small fraction of the 3GCCAU tRNAlocated in 70S. Ratios of 70S to 30S bound fractions (Fig. 4B, iii) alsorevealed the same. Interestingly, unlike the case with tRNAfMet, weobserved that a significant fraction of the 3GCCAU tRNA was alsofound migrating with the 50S. This migration is most likely due to itspoorer association with 70S resulting in dissociation of a part of itduring centrifugation. It may also be noted that, in E. coli K, wild-type tRNAfMet encoded by metZWV (∼75%) and metY (∼25%)separates into tRNAfMet1 and tRNAfMet2 bands (doublet) whenanalyzed on the native gel shown in Fig. 4 B, ii. Expectedly, bothtRNAfMet species reveal similar distribution.

Extended SD Facilitates the 3GC tRNA to Enter 70S Complex. Theanalysis in Fig. 4 showed that one of the defects in initiation withthe 3GCCAU tRNA is its transition from the 30S to 70S ribo-some. Hence, to understand the mechanism of how an extendedSD rescued the 3GC tRNA defect, we carried out sucrose den-sity gradient analysis (Fig. 5 A, i and ii) to determine the distri-bution of the 3GC tRNA (i.e., with CUA anticodon) and thetRNAfMet. We found that in the case of wild-type SD (SDwt) the3GC tRNA bound well to the 30S but not to the 70S ribosomefractions (Fig. 5B, SDwt). Interestingly, there was more 3GCtRNA on the 70S fractions in the sample with the extended SDcontext (Fig. 5B, SDa7g). Quantitative analysis (Fig. 5C) alsorevealed the same. On the other hand, SDa7g did not seem tomake a change to the distribution of the wild-type tRNAfMet

(Fig. 5 B and C). These results indicate that compensation forthe 3GC pair defect by the SDa7g occurred at a step subsequentto its initial binding to the 30S subunit, most likely at the step ofits transition into the 70S complex.

Extended SD Also Facilitates Initiation via 70S Pathway. It had beenshown that the extended SD leads to the formation of 70S ri-bosomal complexes without initiator tRNA (15). Furthermore,the 70S ribosomes are also known to initiate under variousspecialized conditions (16–20). Thus, although extended SDallowed more of the 3GC tRNA to enter the 70S complex in thestandard pathway of 30S mode of initiation (as shown above), itcould also be possible that the initiation by 70S ribosomes in-creased. To check for this possibility, we probed the initiationactivity in the presence of kasugamycin, which inhibits the ca-nonical process of 30S initiation but is less effective against 70Sinitiation (16, 21, 22). Thus, if translation of the CAT reportermRNA by the 3GC tRNA had an advantage of the 70S mode ofinitiation, upon kasugamycin addition to the culture, its trans-lation relative to the rest of the cellular mRNAs would be lessimpacted resulting in a relative increase in CAT. As shown inFig. 6A, compared with the untreated cells, although there wasno significant change in the CAT activity in the SDwt background(Fig. 6 A, i), there was an increase in the CAT activity in theSDa7g background (Fig. 6 A, ii) as a result of initiation by the3GC tRNA. Furthermore, given that the presence of kanamycin(a general inhibitor of protein synthesis) under these conditionsactually led to a small decrease in CAT activity (both in the SDwtand SDa7g backgrounds, Fig. 6A), an enhancing effect of kasu-gamycin in the SDa7g background is significant. Such an effect ofkasugamycin was not seen for initiation with the tRNAfMet (CUAanticodon) possessing the wild-type sequence in the anticodonstem (Fig. S3A). These observations indicate that initiation fromthe SDa7g mRNA particularly by the 3GC tRNA, at least in part,used the 70S mode of initiation.In yet another approach, we analyzed initiation activities in a

strain (MG1655frrts) temperature sensitive (ts) because of a mutation

1 2 3 4

(i) CATam1

(ii) 3GC tRNA

(iii) 5S rRNA

Lanes: 1: KL16/vector 2: KL16/3GC-SDwt 3: B21/3GC-SDwt 4: KL16/3GC-SDa7g

Bars: 1: KL16/3GC-SDwt 2: B21/3GC-SDwt 3: KL16/3GC-SDa7g

Relative - 1 2.2 3.7

Relative - 1 2.6 3.2

A

B

Fig. 3. Binding of mRNA and the 3GC tRNA to ribosomes. (A) Northern blotanalysis for CATam1 mRNA (i), 3GC tRNA (ii), and 5S rRNA (iii) in the ribo-somal pellets. RNA (∼20 μg) were separated on 1.2% agarose gel, transferredto nytran membrane, and probed sequentially for CATam1, 3GC tRNA, and5S rRNA. Signal intensities of CATam1 mRNA, and 3GC tRNA were dividedby those of the corresponding signals of 5S rRNA. The values so obtainedfor the KL16 strain harboring the pCATam1metYCUA (3GC-SDwt) were set as 1,and the remaining values were calculated relative to this reference valueof 1 for both the CATam1 and the 3GC tRNA. (B) Fold increase in ribosomebound CATam1 mRNA as measured by quantitative PCR using 16S rRNA asinternal control.

Shetty et al. PNAS | Published online September 22, 2014 | E4227

GEN

ETICS

PNASPL

US

Dow

nloa

ded

by g

uest

on

Dec

embe

r 29

, 202

0

in the ribosome recycling factor (RRF), an essential factorwhich splits 70S ribosomes into 30S and 50S subunits (23, 24). Adeficiency in the RRF activity is known to increase the level ofempty 70S complexes and thus enhance the 70S mode of initiation(23). In the MG1655frrts strain, increase in initiation by the 3GCtRNA for the SDa7g construct (compared with SDwt construct) bothat the permissive (Fig. 6B, compare bar 3 with 4) and nonpermissivetemperatures (Fig. 6B, compare bar 7 with 8), was higher than inthe strain wild-type for RRF (Fig. 6B, bars 1 and 2 and 5 and 6,respectively). Furthermore, although there was a general decreasein translation at 42 °C, the increase for the SDa7g construct was stillhigher in frrts strain (Fig. 6B, compare bars 6 and 8), supporting thepossibility of 70S mode of initiation in extended SD context.To check for the combined effect of RRFts and kasugamycin

treatment, the MG1655frrts strain was treated with kasugamycinat 30 °C. In this experiment also, the increase in 3GC initiationwas higher for the SDa7g construct than for the SDwt construct

(Fig. 6C, compare bars 2 with 4; and 1 with 3), and it increasedfurther upon treatment with kasugamycin in RRFts background(compare bars 2 with 6 and 4 with 8), whereas it had no effect inSDwt context (compare bars 1 with 5 and 3 with 7). Given that inthe RRFts background, the level of RRF are lower even at thepermissive temperature and it accumulates higher level of 70Sribosomes (23), these observations further support that the 3GCtRNA used the 70S mode of initiation.

Impact of Initiation Factors. The initiation factors IF1, IF2, and IF3are involved in the selection of tRNAfMet in the P-site (7, 25, 26).To investigate the impact of various initiation factors, we overex-pressed them from their plasmid copies. For reference, initiationactivities of pCATam1metYCUA were taken as 100% (individuallyfor each of the strain). All other activities were represented relativeto this activity. As shown in Fig. 7A, the initiation activity frompCATam1metYCUA/3GC (3GC tRNA) in the context of SDwt or

A B

Fig. 4. Binding of 3GCCAU tRNA (wild-type anticodon) to 30S and 70S ribosomes. (A) Analysis of 3GCCAU tRNA across the ribosomal profile. The ribosomalsubunits and polysomes were separated on 20–40% sucrose gradient using SW55 rotor (48000 rpm for 3 h) (i). Dot blot analysis from fraction numbers 6–17was carried out for 3GCCAU tRNA and the tRNAfMet. Dot blots for two biological replicates are shown (ii). The fraction of tRNA in each profile was measured bydividing the individual spot intensities with the total intensity in the profile and the distribution shown as bar diagram (iii), for wild-type tRNA (open bar) and3GCCAU tRNA (black bar). (B) The subunits and polysomes were separated on 20–40% sucrose gradient as in A, and the 30S, 50S and 70S peaks were collectedseparately by monitoring the ribosomal profile (i). The extracted RNA was analyzed on the native PAGE, transferred onto nytran membrane and probed forthe presence of 3GCCAU as well as tRNAfMet (ii). Their fractional distributions between the subunits and 70S ribosome are as shown (iii).

E4228 | www.pnas.org/cgi/doi/10.1073/pnas.1411637111 Shetty et al.

Dow

nloa

ded

by g

uest

on

Dec

embe

r 29

, 202

0

SDa7g remained nearly unchanged upon overexpression of IF1(compare activity blocks of IF1, bars 5–8, with those of vector, bars1–4). Interestingly, overexpression of IF2 caused a significant in-crease in initiation from pCATam1metYCUA/3GC in the context ofSDa7g (Fig. 7A, compare bars 2 and 10) but not in the SDwt context(bars 1 and 9). The fact that IF2 functions as association factor

between 30S and 50S and drives the reaction to form 70S, the ob-servation further supports the view that initiation by 3GC tRNAused 70S ribosomes.Overexpression of IF3 had a small negative effect (27) on the

activity from pCATam1metYCUA/3GC in the SDwt context (Fig. 7A,compare bars 1 and 13) but it had no effect in SDa7g context

A

B

C

Fig. 5. Effect of SDa7g on 3GC tRNA binding. (A) Sucrose density gradients (20–40%) analysis of cell-lysates from KL16 strain harboring pCATam1metY3GC/CUA

with SDwt (i) or SDa7g (ii). (B) Northern blot analysis for 3GC tRNA and the tRNAfMet across the profile. (C) The fractional distribution of the tRNAs across thefractions, measured by dividing the individual intensities with the total intensity in all fractions.

Shetty et al. PNAS | Published online September 22, 2014 | E4229

GEN

ETICS

PNASPL

US

Dow

nloa

ded

by g

uest

on

Dec

embe

r 29

, 202

0

(compare bars 2 and 14). However, we observed that compro-mising IF3 function in infC135 background resulted in increasedactivity from pCATam1metYCUA/3GC in SDa7g construct (Fig. 7B,compare wild-type infC with infC135, bars 2 and 6) but the foldincrease is less than the SDwt (Fig. 7B, bars 1 and 5).

DiscussionCharacterization of an E. coli isolate, B21, in this study hasrevealed that a mutation in one of the rRNA genes (rrsC) cor-responding to a C to U change at position 1536 of the mature16S rRNA, resulting in an extended interaction with the SDsequence of the reporter mRNA, allowed initiation to occur witha tRNAfMet deficient in the highly conserved feature of the threeconsecutive G-C base pairs in its anticodon stem. This observa-tion was further validated by making a directed mutation in SDsequence of the reporter mRNA to extend its interaction withaSD sequence in the wild-type 16S rRNA. In contrast to theincreased initiation by the 3GC tRNA, an extended SD:aSD in-teraction was suboptimal for initiation by the wild-type tRNAfMet.Suboptimal initiation by the wild-type tRNAfMet may be a con-sequence of lowered efficiency of ribosome clearance from thetranslation initiation region of the mRNA (14).Polysome profiling and northern analysis of the fractions

showed that the 3GC tRNA binding to the 30S subunit was notthe rate-limiting step for its participation in initiation. Thus, it wasnot unexpected that overexpression of CATam1 mRNA (Fig. 2D)did not rescue the initiation defect of the 3GC tRNA. However,compared with the tRNAfMet, the 3GC tRNA occurrence in the70S fractions was very low (Fig. 4), suggesting that during transi-tion of the 30S preinitiation complex to 70S complex, the 3GCtRNA was rejected. This observation suggests that the 3GC basepairs play an important role in stabilizing tRNAfMet during thetransition of the 30S preinitiation complex to the 70S complex

(Fig. 8A). In yeast, the 3GC tRNAs also showed stable binding tothe 40S (28, 29) but not the 80S ribosomes (28), supporting ourproposal of the role of the 3GC pairs, a feature of initiator tRNAsconserved in all domains of life. In vitro studies using 70S ribo-somes also showed that the 3GC pairs are required for tRNAfMet

binding to the ribosome (4).How does an extended SD:aSD allow initiation with the 3GC

tRNA? Using single-molecule studies, it was shown that theSD:aSD interaction stabilizes the 70S and it requires more energyto separate the subunits (30). The presence of extended SD allowsmore 3GC tRNAs to enter the 70S complex (Fig. 5), suggestingthat a more stable 70S complex may compensate for the bindingenergy contributed by the 3GC base pairs. The SD:aSD helixcomprising nucleotides 1534–1540 in 16S rRNA interacts withh28 (forms neck of 30S), h23a (forms platform of 30S) and h26.Using translation libration screw (TLS) analyses of the structureof 70S complex containing SD sequence with those lacking it, itwas suggested that the presence of the SD:aSD helix reduces themobility of the head and platform (31). Thus, positioning of theSD:aSD helix may fix the orientation of the mobile head ofthe 30S subunit for optimal interaction with tRNAfMet in the 30SP site. It was also shown that in comparison with the vacant 70Sribosome, h26 and h28 move apart by >2 Å to house the SD:aSDhelix (31). Such a displacement near the hinge of the neck resultsin ∼13 Å change in the positions of G1338 and A1339 in the headregion. This change may allow 3GC tRNA accommodation inthe 70S and lead to initiation. Interestingly, in another crystalstructure, which used an extended SD in the mRNA, interactionof G1338 and A1339 with 3GC base pair of initiator tRNA wasfound to be suboptimal (5). This observation suggests that thediscriminatory role of the 3GC pairs is diminished in the contextof extended SD:aSD interaction.

RRFwt RRFts RRFwt RRFts

Kasugamycin

RRFwt RRFts RRFwt RRFts

30 oC 42 oC

SDa7g SDwt

(i) SDwt (ii) SDa7g

Fold

Cha

nge

in C

AT a

ctiv

ity

Fold

Cha

nge

in C

AT a

ctiv

ity

Fold

Cha

nge

in C

AT a

ctiv

ity

Fold

Cha

nge

in C

AT a

ctiv

ity

SDa7g SDwt

Time (h) Time (h)

1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8

A

B C

Fig. 6. Effect of SDa7g on initiation with 70S ribosome. (A) Initiation with the 3GC tRNA in the context of SDwt (i) and SDa7g (ii) of the CATam1 reporter uponkasugamycin or kanamycin treatment. Cultures at OD600 of 0.5, were split into three parts and CAT activities were determined at various times from theuntreated controls or upon treatment with kasugamycin (500 μg/mL), or kanamycin (25 μg/mL). The experiment was carried for three biological replicates foreach construct. Fold differences in CAT activity with respect to untreated control at the time of splitting the original culture into three are shown. Forreference, the average CAT activities for 3GC tRNA in the context of SDwt and SDa7g in untreated KL16 were 92.9 ± 4.0 and 2225.1 ± 656.3 pmol Ac-Cmproduced per μg of total protein, respectively. (B and C) Activity assays in MG1655 (RRFwt) and MG1655 frrts (RRFts) in the context of SDwt and SDa7g CATam1

reporters. Cultures were grown to OD600 of 0.5 at 30 °C and then either kept at 42 °C (B) or treated with 500 μg/mL kasugamycin (C) and incubated for 2 hbefore determination of CAT. Fold differences in CAT activity with respect to 3GC in MG1655 (6.2 ± 2.2 pmol Ac-Cm produced per μg total protein) are shown.

E4230 | www.pnas.org/cgi/doi/10.1073/pnas.1411637111 Shetty et al.

Dow

nloa

ded

by g

uest

on

Dec

embe

r 29

, 202

0

IF3 has long been suggested to scrutinize the 3GC base pairs(6, 26, 32). However, this scrutiny may be indirect and negotiatedthrough G1338- and A1339-mediated A-minor interactions (6).In our studies, overexpression of IF3 led to a small decrease ininitiation with the 3GC tRNA in the SDwt context, whereas it hadno effect in the SDa7g context (Fig. 7A). However, an inefficientlyfunctioning IF3 (infC135) led to a 6- to 8-fold increase in initia-tion with 3GC tRNA in the context of SDwt, and a 2- to 3-foldincrease in the context of the SDa7g (Fig. 7B). The diminution ofIF3 action by the presence of SD sequence was also observed inthe context of reinitiation by the 70S ribosomes (33). Theseobservations lend further support to the view that the displace-ment of G1338 and A1339 (by the extended SD interaction) maymitigate IF3 mediated scrutiny of the 3GC sequence.Kasugamycin treatment resulted in increased initiation with

the 3GC tRNA in the context of SDa7g, suggesting the possibilityof 70S mode of initiation, which is relatively insensitive tokasugamycin (i.e., compared with the 30S mode). DecreasingRRF (RRFts) as well as combining the RRFts background withkasugamycin treatment further supported the 70S mode of ini-tiation with the extended SD. The 70S mode of initiation is thepredominant mode of initiation in the leaderless mRNAs. Im-portantly, the presence of initiator tRNA is necessary for theformation of 70S complex on the leaderless mRNAs (34). Also, itwas shown that the kasugamycin-treated 70S complex had accessto bind to SD sequences (23). Taken together, these observationssuggest that leadered mRNAs with strong SD sequences may (inconverse) facilitate initiation with the tRNAs lacking the 3GC

pairs. In addition, the 70S mode of initiation was found to betemperature sensitive (17). Consistent with this observation, wenoted that in the strain wild-type for RRF, an increase in thegrowth temperature resulted in a decrease in initiation from theSDa7g construct by the 3GC tRNA (Fig. 6B, compare bar 2 withbar 6; see also Fig. S3B). Although maximum pairing of 12 or 13nucleotides in the SD:aSD helix has been reported (35), most ofthe mRNAs consist of 4- to 6-base-long SD interactions (36).The most optimal SD:aSD interaction in E. coli (at 37 °C) is 6nucleotides long, with shorter interactions being optimal at coldtemperatures (36).IF2 acts as association factor and it helps in docking of the 50S

subunit onto the 30S preinitiation complex. Overexpression ofIF2 has a synergistic effect on initiation with the 3GC tRNA inthe extended SD context, most likely by increasing the pop-ulation of 70S ribosomes. Overexpression of IF2 was also shownto increase the 70S mode of initiation in bicistronic constructs(33). Thus, the conditions that favor 70S formation also lead toincreased initiation by the 3GC tRNA especially in the context ofan extended SD. In fact, in the classical in vitro experimentsdesigned to decipher the genetic code (37) initiation occurredeven with the elongator tRNAs by increasing salt concentrationto stabilize the 70S complex, suggesting yet again that 70S modeof initiation may dispense with the requirement of the 3GC pairsin the tRNA.Thus, our results suggest that the 3GC pairs in the anticodon

stem of initiator tRNA play a major role in its retention on theribosome during transitions to generate an elongation competent

A

B

Fig. 7. Effect of initiation factors on initiation with 3GC tRNA. (A) Initiation from CATam1 containing SDwt or SDa7g by 3GC or tRNAfMet (metYCUA) tRNAs uponoverexpression of IF1 (pACDHinfA), IF2 (pACDHinfB), IF3 (pACDHinfC) or the vector control (pACDH, V). CAT activities are shown with reference to metYCUA

(SDwt), which was set as 100%, individually for each strain. For reference, CAT activities formetYCUA tRNA in the context of SDwt were 6619.2 ± 989.1, 14459.3 ±1892.4, 29456.6 ± 1760.1 and 7521.2 ± 2312.5 pmol Ac-Cm produced per μg total protein for the strains harboring vector control (V), IF1, IF2 or IF3 over-expression constructs, respectively. (B) Effect of compromising IF3 activity on initiation by 3GC tRNA. CAT assays in KL16 (infC) or infC135. CAT activities (%) withrespect to metYCUA in SDwt for each strain are as shown. For reference, 100% CAT activities for metYCUA (SDwt) in infC and infC135 were 5236.4 ± 1204.5 and13054.9 ± 2409.3 pmol Ac-Cm produced per μg of total protein, respectively.

Shetty et al. PNAS | Published online September 22, 2014 | E4231

GEN

ETICS

PNASPL

US

Dow

nloa

ded

by g

uest

on

Dec

embe

r 29

, 202

0

70S complex (Fig. 8 A, i). Presence of the extended SD in mRNApartly compensates for this critical requirement, possibly bystabilizing the 70S complex during these transitions and pre-venting rejection of the 3GC tRNA. In the absence of the 3GCpairs in the initiator tRNA and an extended SD, the 70S com-plexes decompose and fail to enter into the elongation step (Fig. 8A, ii). The current study also suggests that for the 70S modeof initiation, which may do away with the transitional changesneeded during the 30S mode of initiation, the requirement of the3GC pairs is not vital (Fig. 8B). Importantly, although the twomechanisms (Fig. 8 A and B) are not mutually exclusive, they bothsupport that the 3GC pairs play a critical role in tRNAfMet re-tention in ribosome during the conformational changes that markthe transition of 30S preinitiation complex into 70S complex.Finally, the initiation by 70S ribosomes by tRNAs lacking the

3GC pairs may be relevant from an evolutionary consideration.As this mode of initiation does not necessarily require 3GC basepairs, a primitive type of initiation may have been feasible solelyby the elongator tRNAs. Initiation process must have evolvedsubsequent to the process of elongation to direct protein syn-thesis from a definite point in an mRNA with the coevolution ofthe 3GC pairs in the tRNA, and a subsequent specialization toinitiate with the 30S ribosomes. Interestingly, some species ofprimitive bacteria, such as mycoplasma, alpha-proteobacteriaand also the yeast mitochondria have been observed to lack thefull complement of the 3GC pairs in initiator tRNAs.

Materials and MethodsStrains and Plasmids. Strains and plasmids used in this study have been listedin Table S1. E. coli KL16 and its derivatives were grown in LB and LB–agarplates (Difco). Unless mentioned otherwise, media were supplemented withampicillin (Amp, 100 μg/mL), chloramphenicol (Cm, 30 μg/mL), kanamycin(Kan, 25 μg/mL), or tetracycline (Tet, 7.5 μg/mL).

Isolation, Characterization, and Genetic Mapping of B21. The isolation andpreliminary characterization of E. coli B21 has been described (9). Mapping ofthe suppressor mutation was carried out by the methods described in ref. 9.

Site-Directed Mutagenesis. The pCATam1metYCUA/3GC plasmid was used astemplate to modify the upstream of CATam1 reporter. Using oligomers withdesired mutation, inverse PCR was carried out, the amplification productwas digested with DpnI and transformed into E. coli TG1, and the desiredmutants were confirmed by DNA sequencing. The details of primers areprovided in Table S2.

Polysome Profiling. The total ribosomal preparations from E. coliwere carriedout as described (15) with few changes. Translation was inhibited by 0.3 μMtetracycline before harvesting the cells, and chilled on ice–salt mix. Thebuffer 1 contained 10% sucrose. Approximately 10–20 OD260 of total RNAwas used for profile analysis on 20–40% (wt/vol) (15–35% in some cases)sucrose gradients (buffered in 20 mM Hepes-KOH, pH 7.5, 50 mM NH4Cl,10 mM MgCl2, 4 mM β-mercaptoethanol) were prepared using a BioCompgradient master (BioComp Instruments) in polyclear tubes (SETON, catalogno. 7022). Samples were loaded on gradients and spun in an ultracentrifugeusing an SW55 rotor at 45,000 rpm for 2–3 h at 4 °C as described in the figurelegends. Gradients were fractionated using a BioComp Gradient Fraction-ator with the flow rate set to 0.3 mm per second. Subunits were eithermanually collected by following the absorption trace at 254 nm using a BIO-RAD Econo UV monitor or equal fractions of 200–300 μL volume were col-lected using fraction collector (BIO RAD Model 2110 Fraction Collector).

RNA Analysis. Analysis of fractions for the presence of mRNA and tRNA by dotblotting was carried as earlier (15). For analysis by native gel, the RNA frompolysome fractions were extracted using hot phenol, ethanol precipitatedand separated on native PAGE, transferred on Nytran membrane. The mem-branes were probed for 16S rRNA, wild-type tRNAfMet, and the 3GC tRNA,as required.

Preparation of Cell-Free Extracts and CAT Assays. The midlog phase grownE. coli strains were used for extract preparation as described (9, 38).

A

B

Fig. 8. Role of 3GC base pairs in translation initiation. (A) Translation initiation begins with the binding of initiator tRNA and the initiation factors IF1 (1), IF2(2), and IF3 (3) to the 30S subunit to generate 30S initiation complex (30S IC) followed by binding of the 50S subunit resulting in 70S complex with tRNAfMet

bound in P/I state (shown by tilted initiator tRNA). An unratcheting movement assisted by the hydrolysis of GTP in IF2 together with the release of initiatorfactors brings the tRNA into classical state (i). Our results suggest that the 3GC base pairs in the anticodon stem of initiator tRNA play a major role in itsretention on the ribosome during these transitions to generate an elongation competent 70S complex. The initial binding of the tRNAfMet to 30S subunit isindependent of the 3GC base pairs. Presence of the extended SD in mRNA partly compensates for this critical requirement. In the absence of the 3GC pairs inthe initiator tRNA or an extended SD in mRNA the 70S complex decomposes (ii). (B) The extended SD may also allow bypass of the 3GC base pair requirementby the 70S mode of initiation.

E4232 | www.pnas.org/cgi/doi/10.1073/pnas.1411637111 Shetty et al.

Dow

nloa

ded

by g

uest

on

Dec

embe

r 29

, 202

0

ACKNOWLEDGMENTS. We thank our laboratory colleagues for theirsuggestions on the manuscript. This work was supported by the De-partment of Science and Technology (DST) and the Department of

Biotechnology. U.V. is a J. C. Bose fellow of DST. S.S. is supported bya Shyama Prasad Mukherjee senior research fellowship of the Council ofScientific and Industrial Research.

1. Gold L (1988) Posttranscriptional regulatory mechanisms in Escherichia coli. Annu RevBiochem 57:199–233.

2. Shine J, Dalgarno L (1974) The 3′-terminal sequence of Escherichia coli 16S ribosomalRNA: complementarity to nonsense triplets and ribosome binding sites. Proc NatlAcad Sci USA 71(4):1342–1346.

3. Mandal N, Mangroo D, Dalluge JJ, McCloskey JA, Rajbhandary UL (1996) Role of thethree consecutive G:C base pairs conserved in the anticodon stem of initiator tRNAs ininitiation of protein synthesis in Escherichia coli. RNA 2(5):473–482.

4. Seong BL, RajBhandary UL (1987) Escherichia coli formylmethionine tRNA: Mutationsin GGGCCC sequence conserved in anticodon stem of initiator tRNAs affect initiationof protein synthesis and conformation of anticodon loop. Proc Natl Acad Sci USA84(2):334–338.

5. Selmer M, et al. (2006) Structure of the 70S ribosome complexed with mRNA andtRNA. Science 313(5795):1935–1942.

6. Lancaster L, Noller HF (2005) Involvement of 16S rRNA nucleotides G1338 and A1339in discrimination of initiator tRNA. Mol Cell 20(4):623–632.

7. Antoun A, Pavlov MY, Lovmar M, Ehrenberg M (2006) How initiation factors maxi-mize the accuracy of tRNA selection in initiation of bacterial protein synthesis. MolCell 23(2):183–193.

8. Marshall RA, Aitken CE, Puglisi JD (2009) GTP hydrolysis by IF2 guides progression ofthe ribosome into elongation. Mol Cell 35(1):37–47.

9. Das G, et al. (2008) Role of 16S ribosomal RNA methylations in translation initiation inEscherichia coli. EMBO J 27(6):840–851.

10. Varshney U, RajBhandary UL (1990) Initiation of protein synthesis from a terminationcodon. Proc Natl Acad Sci USA 87(4):1586–1590.

11. Mayer C, Stortchevoi A, Köhrer C, Varshney U, RajBhandary UL (2001) Initiator tRNAand its role in initiation of protein synthesis. Cold Spring Harb Symp Quant Biol 66:195–206.

12. Jacob WF, Santer M, Dahlberg AE (1987) A single base change in the Shine-Dalgarnoregion of 16S rRNA of Escherichia coli affects translation of many proteins. Proc NatlAcad Sci USA 84(14):4757–4761.

13. Takahashi S, Furusawa H, Ueda T, Okahata Y (2013) Translation enhancer improves theribosome liberation from translation initiation. J Am Chem Soc 135(35):13096–13106.

14. Komarova AV, Tchufistova LS, Supina EV, Boni IV (2002) Protein S1 counteracts theinhibitory effect of the extended Shine-Dalgarno sequence on translation. RNA 8(9):1137–1147.

15. Mawn MV, Fournier MJ, Tirrell DA, Mason TL (2002) Depletion of free 30S ribosomalsubunits in Escherichia coli by expression of RNA containing Shine-Dalgarno-like se-quences. J Bacteriol 184(2):494–502.

16. Moll I, Bläsi U (2002) Differential inhibition of 30S and 70S translation initiationcomplexes on leaderless mRNA by kasugamycin. Biochem Biophys Res Commun297(4):1021–1026.

17. Grill S, Moll I, Giuliodori AM, Gualerzi CO, Bläsi U (2002) Temperature-dependenttranslation of leaderless and canonical mRNAs in Escherichia coli. FEMS Microbiol Lett211(2):161–167.

18. Moll I, Grill S, Gründling A, Bläsi U (2002) Effects of ribosomal proteins S1, S2 and theDeaD/CsdA DEAD-box helicase on translation of leaderless and canonical mRNAs inEscherichia coli. Mol Microbiol 44(5):1387–1396.

19. Moll I, Grill S, Gualerzi CO, Bläsi U (2002) Leaderless mRNAs in bacteria: Surprises inribosomal recruitment and translational control. Mol Microbiol 43(1):239–246.

20. Vesper O, et al. (2011) Selective translation of leaderless mRNAs by specialized ribo-somes generated by MazF in Escherichia coli. Cell 147(1):147–157.

21. Poldermans B, Van Buul CP, Van Knippenberg PH (1979) Studies on the function oftwo adjacent N6,N6-dimethyladenosines near the 3′ end of 16 S ribosomal RNAof Escherichia coli. II. The effect of the absence of the methyl groups on initiation ofprotein biosynthesis. J Biol Chem 254(18):9090–9093.

22. Kaberdina AC, Szaflarski W, Nierhaus KH, Moll I (2009) An unexpected type of ribo-somes induced by kasugamycin: A look into ancestral times of protein synthesis? MolCell 33(2):227–236.

23. Moll I, Hirokawa G, Kiel MC, Kaji A, Bläsi U (2004) Translation initiation with 70S ri-bosomes: An alternative pathway for leaderless mRNAs. Nucleic Acids Res 32(11):3354–3363.

24. Seshadri A, Dubey B, Weber MH, Varshney U (2009) Impact of rRNA methylations onribosome recycling and fidelity of initiation in Escherichia coli. Mol Microbiol 72(3):795–808.

25. Antoun A, Pavlov MY, Lovmar M, Ehrenberg M (2006) How initiation factors tune therate of initiation of protein synthesis in bacteria. EMBO J 25(11):2539–2550.

26. Hartz D, McPheeters DS, Gold L (1989) Selection of the initiator tRNA by Escherichiacoli initiation factors. Genes Dev 3(12A):1899–1912.

27. Mangroo D, RajBhandary UL (1995) Mutants of Escherichia coli initiator tRNA de-fective in initiation. Effects of overproduction of methionyl-tRNA transformylase andthe initiation factors IF2 and IF3. J Biol Chem 270(20):12203–12209.

28. Kapp LD, Kolitz SE, Lorsch JR (2006) Yeast initiator tRNA identity elements cooperateto influence multiple steps of translation initiation. RNA 12(5):751–764.

29. Dong J, et al. (2014) Conserved residues in yeast initiator tRNA calibrate initiationaccuracy by regulating preinitiation complex stability at the start codon. Genes Dev28(5):502–520.

30. Masuda T, et al. (2012) Initiation factor 2, tRNA, and 50S subunits cooperativelystabilize mRNAs on the ribosome during initiation. Proc Natl Acad Sci USA 109(13):4881–4885.

31. Korostelev A, et al. (2007) Interactions and dynamics of the Shine Dalgarno helix inthe 70S ribosome. Proc Natl Acad Sci USA 104(43):16840–16843.

32. Hartz D, Binkley J, Hollingsworth T, Gold L (1990) Domains of initiator tRNA andinitiation codon crucial for initiator tRNA selection by Escherichia coli IF3. Genes Dev4(10):1790–1800.

33. Yoo JH, RajBhandary UL (2008) Requirements for translation re-initiation in Esche-richia coli: Roles of initiator tRNA and initiation factors IF2 and IF3. Mol Microbiol67(5):1012–1026.

34. O’Donnell SM, Janssen GR (2002) Leaderless mRNAs bind 70S ribosomes more stronglythan 30S ribosomal subunits in Escherichia coli. J Bacteriol 184(23):6730–6733.

35. Yusupova G, Jenner L, Rees B, Moras D, Yusupov M (2006) Structural basis for mes-senger RNA movement on the ribosome. Nature 444(7117):391–394.

36. Vimberg V, Tats A, Remm M, Tenson T (2007) Translation initiation region sequencepreferences in Escherichia coli. BMC Mol Biol 8:100.

37. Nirenberg MW, Matthaei JH, Jones OW, Martin RG, Barondes SH (1963) Approxi-mation of genetic code via cell-free protein synthesis directed by template RNA. FedProc 22:55–61.

38. Kapoor S, Das G, Varshney U (2011) Crucial contribution of the multiple copies of theinitiator tRNA genes in the fidelity of tRNA(fMet) selection on the ribosomal P-site inEscherichia coli. Nucleic Acids Res 39(1):202–212.

Shetty et al. PNAS | Published online September 22, 2014 | E4233

GEN

ETICS

PNASPL

US

Dow

nloa

ded

by g

uest

on

Dec

embe

r 29

, 202

0