Molecular Microbiology (2003)

47

(1), 267–275

© 2003 Blackwell Publishing Ltd

Blackwell Science, LtdOxford, UKMMIMolecular Microbiology0950-382XBlackwell Science, 200347Original Article

Characterization of mutants of Rf1 and RF2L. Mora et al.

Accepted 8 October, 2002. *For correspondence. E-mail Richard.Buckingham@ibpc.fr;

Tel.

(

+

33)

1

58

41

51

20; Fax (

+

33)

1

58 41 50 20.

The essential role of the invariant GGQ motif in the function and stability

in vivo

of bacterial release factors RF1 and RF2

Liliana Mora,

1

Valérie Heurgué-Hamard,

1

Stéphanie Champ,

1

Måns Ehrenberg,

2

Lev L. Kisselev

3

and Richard H. Buckingham

1

*

1

UPR9073 du CNRS, Institut de Biologie Physico-Chimique, 13 rue Pierre et Marie Curie, Paris 75005, France.

2

Department of Cell and Molecular Biology, BMC, Box 596, S-75124 Uppsala, Sweden.

3

Engelhardt Institute of Molecular Biology, Russian Academy of Sciences, 32 Vavilova, Moscow 119991, Russia.

Summary

Release factors RF1 and RF2 are required in bacteriafor the cleavage of peptidyl-tRNA. A single sequencemotif, GGQ, is conserved in all eubacterial, archae-bacterial and eukaryotic release factors and maymimic the CCA end of tRNA, although the position ofthe motif in the crystal structures of human eRF1 and

Escherichia coli

RF2 is strikingly different. Mutationshave been introduced at each of the three conservedpositions. Changing the Gln residue to Ala or Gluallowed the factors to retain about 22% of tetrapeptiderelease activity

in vitro

, but these mutants could notcomplement thermosensitive RF mutants

in vivo

.None of several mutants with altered Gly residuesretained activity

in vivo

or

in vitro

. Many GGQ mutantswere poorly expressed and are presumably unstable;many were also toxic to the cell. The toxic mutantfactors or their degradation products may bind toribosomes inhibiting the action of the normal factor.These data are consistent with a common role for theGGQ motif in bacterial and eukaryotic release factors,despite strong divergence in primary, secondary andtertiary structure, but are difficult to reconcile with thehypothesis that the amide nitrogen of the Gln plays avital role in peptidyl-tRNA hydrolysis.

Introduction

The nascent polypeptide chain is released from peptidyl-tRNA when a termination signal on mRNA is encountered,in a reaction promoted by protein release factors (RFs).In bacterial cells, two class I release factors are requiredto recognize the three stop codons (for a review, seeKisselev and Buckingham, 2000). Thus, RF1 is specificfor UAG and RF2 for UGA, while both factors recognizeUAA. In Eukarya and Archaea, a single class I RF (eRF1and aRF1 respectively) recognizes all three stop codons.Although eRF1 and aRF1 are clearly related, they showno sequence similarity to bacterial RFs with the exceptionof a short motif that contains a tripeptide sequence Gly-Gly-Gln (GGQ), conserved in all three kingdoms in all RFsfor which sequence data are available (Frolova

et al

.,1999). In both RF1 and RF2 from

Escherichia coli

, the Glnresidue of this motif is methylated post-translationally bythe product of the gene

prmC

, previously called

hemK

(Heurgué-Hamard

et al

., 2002; Nakahigashi

et al

., 2002).The methylation has a considerable effect on the activityof RF2

in vitro

, but not of RF1 (Dinçbas-Renqvist

et al

.,2000). A further type of RF (named a class II RF), withribosome-dependent GTPase activity, participates in ter-mination in both bacteria and eukarya (Kisselev andBuckingham, 2000). The bacterial factor, RF3, catalysesthe recycling of class I RFs (Freistroffer

et al

., 1997;Zavialov

et al

., 2001; 2002). The role of the eukaryoticclass II RFs may be similar but is less clearly established.

The role of RFs in codon recognition has encouragedthe idea that these proteins may mimic tRNA moleculesin both size and shape, and possess domains correspond-ing to both the anticodon region and the CCA terminus oftRNA (Cantor, 1979; Nissen

et al

., 1995; Nakamura

et al

.,1996). The crystal structure of human eRF1 supports themimicry hypothesis to some extent (Song

et al

., 2000).The structure shows that the GGQ tripeptide is positionedat the extremity of a stalk protruding from the molecule,and the correspondence proposed between the structuresof eRF1 and tRNA assigns this region to the CCA termi-nus of tRNA, in line with earlier suggestions (Frolova

et al

.,1999). Surprisingly, the crystal structure of RF2 from

E.coli

(Vestergaard

et al

., 2001) appears to be quite differentfrom that of the human factor. The domain containing the

268

L. Mora

et al.

© 2003 Blackwell Publishing Ltd,

Molecular Microbiology

,

47

, 267–275

GGQ motif is positioned differently with respect to the bulkof the molecule, and only 23 Å from a tripeptide region(SPF) that has been implicated as interacting directly withthe stop codon in mRNA on the ribosome (Ito

et al

., 2000).The stop codon and the peptidyl transferase centre areseparated by about 75 Å on the ribosome, like the antic-odon and the CCA terminus in the tRNA molecule. Clearly,the SPF and GGQ regions cannot interact simultaneouslywith their suggested targets on the ribosome if RF2maintains the same structure as in the crystal. This ledVestergaard

et al

. (2001) to propose a different mode ofinteraction, in which the SPF motif docks onto helix 44 of16S rRNA rather than binds directly to the stop codon.Alternatively, it has been suggested that the factor maybecome unfolded on the ribosome (Ehrenberg andTenson, 2002).

Some studies of the role of each of the three residuesin the conserved GGQ motif have been performed in thecase of both human and yeast eRF1 factors. Thus, Song

et al

. (2000) reported that mutations of the Gln residue inyeast eRF1 to either Leu or Arg completely inactivated thefactor

in vivo

. On the basis of the three-dimensional struc-ture of human eRF1 and of the functional studies of yeasteRF1, it was proposed that the Gln residue plays animportant role in the peptide release reaction by co-ordi-nating a water molecule in the peptidyl transferase centreof the ribosome (Song

et al

., 2000). However, in the caseof the human factor, this appears unlikely, as changing theGln residue to Gly or Arg preserved more than half theactivity of the factor in fMet release

in vitro

(Seit Nebi

et al

., 2000; 2001).Much effort has been devoted in recent years to devel-

oping

in vitro

systems to study peptide release from theribosome that are much closer to the termination of nor-mal cellular proteins than the fMet release assay used inearly experiments. These systems have proved invaluablefor studying the role of both class I and class II releasefactors (Freistroffer

et al

., 1997; Pavlov

et al

., 1997a;Heurgué-Hamard

et al

., 1998; 2000). Here, we have usedboth

in vivo

and

in vitro

approaches to characterizemutants of RF1 and RF2 from

E. coli

.

Results

Complementation of a thermosensitive mutant of

prfA

by wild-type and mutant

prfA

The existence of thermosensitive mutants affecting

prfA

and

prfB

in

E. coli

potentially allows complementationexperiments to be carried out

in vivo

with alleles of thesegenes expressed from plasmids. For RF1, a pBR322-derived plasmid was constructed in which

prfA

wasexpressed from the

trc

promoter, inducible by IPTG. A lowlevel of expression in the absence of induction wasassured by the presence of a compatible plasmid, pAIQ7,

carrying the

lacI

q

gene. In order to study the role of theGGQ region in RF1 and RF2, a series of mutants wasconstructed by site-directed mutagenesis. The ther-mosensitive

prfA

strain US486 was transformed with plas-mids carrying wild-type or mutant

prfA

(and pAIQ7), andthe ability to grow on plates containing IPTG at widelyvarying concentrations was studied at both permissiveand non-permissive temperatures (Fig. 1A and B).

The results are summarized in Table 1. The data in line2 show that, at IPTG concentrations

<

10

-

5

M, expressionof wild-type RF1 complements the thermosensitive muta-tion. At all but low levels, the expression inhibits cellgrowth, a behaviour seen clearly at both 30

∞

C and 42

∞

C.A set of seven mutants affecting one or other or both theGly residues shows widely varying behaviour (Table 1,lines 3–9). None of these mutants is able to complementthe thermosensitive growth of strain US486. However,whereas the expression of some mutants, such as VAQ,MGQ, VGQ and GHQ, is strongly inhibitory for growth, theexpression of others (AGQ, GAQ) is not, and mutant LGQis intermediary in effect. Toxicity may be expected to ariseif mutant RFs retain their affinity for ribosomes paused ata stop signal, but are unable to promote peptide chainrelease. The factors can act as strong competitive inhibi-tors for normal RFs, as removal of a class I release factorby RF3 cannot take place in the presence of a peptide onthe P-site-bound tRNA (Zavialov

et al

., 2001; 2002). Aswill be seen below, the variability in the degree of growthinhibition resulting from the expression of inactive RFmutants may be caused by variations in their intracellularlevels.

Table 1.

Effects on growth at 30

∞

C or 42

∞

C of a thermosensitive RF1mutant resulting from the expression of RF1 mutants altered in theGGQ motif.

Growth

Temperature:

Western

30

∞

C 42

∞

C

IPTG (M): 0 10

-

5

10

-

4

10

-

3

0–10

-

5

pBR

+++ +++ +++ +++

–RF1 (wt) 100

+++ +

– –

+++

RF1(AGQ) ND

+++ +++ +++ +++

–RF1(VAQ) ND

+++

– – – –RF1(GAQ) 11

+++ +++ +++ +++

–RF1(LGQ)

<

3

+++ +++ ++ +

–RF1(MGQ) 109

+++ +

– – –RF1(VGQ) 103

+

– – – –RF1(GHQ) ND

++

– – – –RF1(GGE)

<

3

+

– – – –RF1(GGA)

<

3

+++ ++ ++

– –RF1(GGK)

<

3

+++ +++ ++ +

–RF1(GGE) truncated 84

+++ ++ +

– –

Transformants of strain US4866 were plated as described in thelegend to Fig. 1 on LB agar medium containing the concentration ofIPTG indicated.ND, not determined.

Release factor mutations in the GGQ motif

269

© 2003 Blackwell Publishing Ltd,

Molecular Microbiology

,

47

, 267–275

Three mutants were constructed with changes at thelast position of the GGQ motif, and none complementedthe thermosensitivity of strain US486. A further mutantwas constructed as a result of an erroneous oligonucle-otide or ligation and was found by sequencing to be trun-cated seven residues after the GGE motif. This mutantwas inhibitory to cell growth at 30

∞

C, like the wild-typefactor.

Complementation of a thermosensitive mutant of

prfB

by wild-type and GGQ mutants of RF2

Similar experiments to those described for RF1 were alsoperformed using mutants of RF2(Ala-246) altered in theGGQ sequence. Here, a similar plasmid constructionwas used as for RF1 except that

lacI

q

was carried on thesame plasmid as

prfB

. As in the case of RF1, expressionof wild-type

E. coli

K-12 RF2 is growth inhibitory (Table 2).This is related to the occurrence of Thr-246 in

E. coli

K-12 RF2, in contrast to all other characterized bacterial RFspecies, where an Ala or Ser residue is present. Overpro-duction of RF2(Thr-246Ala) does not inhibit cell growth(Table 2), consistent with previous observations (Uno

et al

., 1996; Dinçbas-Renqvist

et al

., 2000). None of themutants GAQ, GGA or GGE was able to complementthermosensitive growth in the RF2 mutant strain RM737(Fig. 1C and D).

Expression levels of mutant RF

The results of complementation experiments showed thatnone of the mutants affecting the Gly residues of the GGQmotif led to sufficient RF activity to complement the ther-mosensitive factor expressed from the chromosomal copyof the gene. This could be because the mutant proteinsare inactive or because insufficient mature protein is syn-thesized. The quantity of RF in transformed cells express-ing different mutant alleles of RF1 and RF2 was therefore

Table 2.

Effects on growth at 37

∞

C or 44.5

∞

C of a thermosensitiveRF2 mutant resulting from expression of RF2 (246Ala) mutantsaltered in the GGQ motif.

Growth

Temperature:

Western

37

∞

C 44.5

∞

C

IPTG (M): 0–10

-5 10-4-10-3 0 10-5-10-3

pTrc99C-NdeI +++ +++ – –RF2246T (wt) ND +++ – +++ –RF2246 A (wt) 100 +++ +++ +++ +++RF2(GAQ) 74 +++ – – –RF2(GGA) 60 +++ – – –RF2(GGE) <5 +++ – – –

Transformants of strain RM737 were plated as described in the legendto Fig. 1 on LB agar medium containing the concentration of IPTGindicated. All GGQ mutants are of the Ala-246 form of the factor.ND, not determined.

Fig. 1. Complementation of thermosensitive mutants of prfA and prfB by wild-type and mutant release factors expressed from plasmids.A and B. Transformants of the prfA thermosen-sitive strain US486 by plasmid pBR-prfA and mutant derivatives plated at 30∞C and 42∞C, respectively, in the absence of IPTG. pBR is the control vector without insert. Plasmid pAIQ7 expressing lacIq is present throughout.C and D. Transformants of the prfB thermosen-sitive strain RM737 by pTrc-prfB and mutant derivatives plated at 37∞C and 44.5∞C, respec-tively, in the absence of IPTG. pTrc is the con-trol vector without insert. Growth at the start of some streaks represents reversion to ther-moresistance of the mutant strains. Both prfA and prfB genes are expressed by leakiness of the trc promoter in the absence of IPTG.

270 L. Mora et al.

© 2003 Blackwell Publishing Ltd, Molecular Microbiology, 47, 267–275

studied by Western blotting. Typical results of such exper-iments are shown in Fig. 2, and the amounts of mutantprotein detected as a percentage of wild-type RF areshown in column 2 in Tables 1 and 2. These data showthat the amount of mature mutant protein varies fromundetectable to about the amount of wild-type RF madeunder the same conditions of expression. Thus, the MGQand VGQ mutants of RF1 are made in normal amounts,whereas very little of the GAQ, LGQ, GGA, GGE or GGKmutants is found. We suggest that these variants areunstable and/or fail to fold normally and are rapidlydegraded by cell proteases. Further experiments showedthat more factor was produced in the case of severalmutant RFs when present in the E. coli B strain BL21,lacking the Lon protease, than in K-12 strains. The factthat expression of the C-terminally truncated RF1 GGEmutant is also inhibitory for cell growth raises the possi-bility that unstable RF mutants may give rise to degrada-tion products that inhibit translation.

Activity of mutant RF1 and RF2 in vitro

The results described above suggested that an explana-tion of the in vivo complementation and growth inhibitoryobservations must take into account not only the catalyticproperties of the full-length protein, but also the amountof mature protein produced and putative inhibition bydegradation products in those cases in which little stableprotein was made. It was therefore desirable to isolate the

mutant proteins and determine their catalytic activities ina purified termination assay in vitro. In order to separatemutant factors from endogenous wild-type RF present inthe cells, and to overcome the difficulty of preparing poorlyexpressed mutant RFs, all mutant RFs were purified in aHis-tagged form after induction of the genes in strainsBL21 or BL21(DE3). RF1(GGE)H6 was produced at suchlow levels that purification of the factor for in vitro experi-ments was not possible. The purified proteins were testedfor activity in the release of the tetrapeptide fMet-Phe-Thr-Ile from ribosomal release complexes paused at the stopcodon UAA (Freistroffer et al., 1997).

In the case of both RF1 and RF2, substitution of Glu orAla for the Gln residue of the GGQ motif yielded proteinswith significant release activity. The experimental data forthe release of tetrapeptide by RF2(GGE)H6 are shown inFig. 3, and the results of all measurements of kcat/KM aresummarized in Table 3. The presence of a C-terminal Histag had no significant effect on the release activity of thewild-type proteins. Comparison of the activity of the RF2mutants with the wild-type factor and comparison of in vivoand in vitro experiments are made complicated by theeffects of the post-translational methylation of the Glnresidue on release activity (Dinçbas-Renqvist et al.,2000). Overproduction of RF2 leads to undermethylatedfactor. The unmethylated Ala-246 form of RF2 supportsalmost normal cell growth, in contrast to unmethylatedThr-246 RF2 (Dinçbas-Renqvist et al., 2000; Heurgué-Hamard et al., 2002). Nevertheless, it is clear that bothGGA and GGE mutants of RF2(Ala-246) have significantrelease activity in vitro, although neither is able to com-plement the thermosensitive RF2 mutation in strainRM737. Although the activity for tetrapeptide release isabout 25% that of the control, it should be noted that thecontrol used is the non-methylated form of the factor,which is already less active than the methylated form ofRF2(Thr-246) and much less active than the methylatedform of RF2(Ala-246) (Dinçbas-Renqvist et al., 2000). A

Fig. 2. Western blot of His-tagged wild-type and mutant RF1 and RF2(246Ala). Total protein was extracted from non-induced trans-formed cells or after induction for 2 h with 10-3 M IPTG. RF-carrying plasmids and Western blotting with rabbit anti-RF1 antibodies are described in Experimental procedures. The upper right-hand lane contains RF1 with no His tag, which migrates slightly faster than RF1 with His tag. The lower right-hand lane contains protein from cells transformed with a control plasmid with no RF2 insert and shows endogenous wild-type RF2, which co-migrates with His-tagged mutant RF2.

Table 3. Release activity (kcat/Km) for wild-type and mutant RF1 andRF2 proteins retaining significant termination activity.

Release factor kcat/Km (mM-1 s-1)

RF1 49 (± 4)RF1(H6) 46 (± 7)RF1 GGA(H6) 10.2 (± 2.6)RF2Ala 65 (± 4.3)RF2Ala(H6) 62.5 (± 8.5)RF2AlaGGE(H6) 14.7 (± 1.8)RF2AlaGGA(H6) 13.1 (± 2.0)

Catalytic activity is measured by the release of fMFTI peptide fromribosomes paused at a UAA codon and carrying fMFTI-tRNA in theP-site. Values are given with standard deviations of the mean. RF1is the overexpressed protein without His tag. RF2Ala is the overex-pressed protein with an Ala at position 246 and without His tag.Mutant proteins are designated by the sequence of the GGQ motif.

Release factor mutations in the GGQ motif 271

© 2003 Blackwell Publishing Ltd, Molecular Microbiology, 47, 267–275

quantitative description of these differences in activity isnot easily provided, as the kinetic parameters vary withthe length of the peptide used in the release assay(Dinçbas-Renqvist et al., 2000). Furthermore, in vivo, nei-ther GGA nor GGE mutants of RF2 can be methylated,which will amplify the loss in activity compared with themethylatable wild-type factor.

In contrast to the mutations of the Gln residue in theGGQ motif, no mutants affecting the Gly residues werefound that retained significant activity. However, manymutant RFs were too unstable to be prepared in sufficientquantity for precise activity measurements. NeitherRF1(GAQ) nor RF2(GAQ) showed measurable activity invitro. If the populations of mutant RF molecules were

homogeneous, the measurements imply an activity <0.5%those of the wild-type factors. The RF2(GAQ) mutant,which has almost normal stability (Table 2) and could beprepared in larger quantity, shows normal ability to bindto ribosomes and to promote the exchange of guaninenucleotide bound to the GTPase RF3; the lack of peptiderelease activity of the RF2(GAQ) mutant has proved use-ful in studies of the mode of action of RF3 (Zavialov et al.,2002). Although RF1 and RF2 are different in stop codonresponse, recognizing UAG and UGA, respectively, muta-tions of the G residues in the GGQ motif cause verysimilar effects in both factors. This observation is consis-tent with the hypothesis that the motif is involved in thecatalytic step of the termination reaction rather than instop codon recognition (Frolova et al., 1999; Song et al.,2000).

Discussion

The results reported here show that mutations of the Glnresidue in the universally conserved GGQ motif of bacte-rial RF1 and RF2 to either Glu or Ala do not eliminatepeptide release activity in vitro. Nevertheless, these wereunable to complement thermosensitive mutants of RF1 orRF2 in vivo. In contrast, the study of several mutantsaffecting the Gly residues showed that, in all cases,release activity was lost. The results of these in vitroexperiments extend and generalize studies of humaneRF1 (Seit Nebi et al., 2000; 2001) and yeast eRF1 (Songet al., 2000). This is significant because the bacterial andmitochondrial factors, on the one hand, and the eukary-otic/archaeal eRF1/aRF1, on the other hand, clearlybelong to distinct structural families, probably with nocommon ancestor (Frolova et al., 1994; Kisselev et al.,2000; Song et al., 2000). The eukaryotic and bacterialRFs do not cross-react with bacterial and eukaryotic ribo-somes respectively (L. Frolova and M. Ehrenberg, unpub-lished observations). It is therefore not evident a priori thatthe universal GGQ motif is functionally equivalent in thetwo structural families, particularly in view of the fact thatthe currently available crystal structures of human eRF1and E. coli RF2 show strikingly different arrangements ofthe GGQ domain in relation to the rest of the molecule(Song et al., 2000; Vestergaard et al., 2001).

It has been proposed that the Gln residue of the GGQmotif plays a role in co-ordinating a water molecule at thepeptidyl transferase centre of the ribosome (Song et al.,2000). Termination and peptidyl transfer are thought to beessentially similar reactions of the peptidyl transfer centre,except that the nucleophilic attack on the ester bond ofpeptidyl-tRNA by the amino group of aminoacyl-tRNA isreplaced, in the termination reaction, by the nucleophilicattack by a water molecule. Release factors must beinvolved in the positioning of this water molecule. The

Fig. 3. Kinetics of release by wild-type and GGE mutant RF2 of fMFTI from termination complexes paused at a UAA stop codon. Release by wild-type and mutant RF2 is shown by closed circles and open triangles respectively. Values of kcat/Km calculated from this and other experiments with both RF1 and RF2 mutants are shown in Table 3.

272 L. Mora et al.

© 2003 Blackwell Publishing Ltd, Molecular Microbiology, 47, 267–275

proposed role of the Gln residue was based on the uni-versal occurrence of the motif (Frolova et al., 1999), thecrystal structure of human eRF1 and the study in Saccha-romyces cerevisiae of mutants in eRF1 (Song et al.,2000). The GGQ motif was found to be positioned in thecrystal structure of eRF1 at the tip of a mini-domain. Twomutants in eRF1 affecting the Gln residue of the GGQmotif, GGL and GGR, were found to be unable to supportgrowth and to be neither dominant negative nor dominantlethal. It was concluded that the mutant factors were inac-tive, although no direct evidence was presented to confirmthe supposed presence in the cell of the GGL or GGRmutant proteins. Surprisingly, however, construction of thesame GGR mutant in human eRF1 showed that about50% of the fMet release activity was retained in vitro (SeitNebi et al., 2000; 2001). The reason for the inactivity ofthe yeast factor in vivo therefore needs to be clarified. TheAGQ, GAQ and AAQ mutants of S. cerevisiae eRF1 werealso unable to support growth, but the single mutants wereweakly dominant negative, and the double mutant stronglydominant negative, confirming the presence of mutantprotein in the cell. In the case of bacterial RFs, an inter-action between the GGQ motif and the peptidyl trans-ferase centre is supported by the observation that theinactive GAQ mutants of RF1 and RF2 inhibit the actionof puromycin, which enters the peptidyl transferase centreand acts as acceptor for the nascent peptide (Zavialovet al., 2002). In the experiments that we report here, thesignificant level of activity retained by the GGE mutant ofRF2 might be compatible with the role proposed for theGln residue by Song et al. (2000), as the side-chains ofboth amino acids can make hydrogen bonds to a watermolecule. However, the effect of the change to Ala in bothRF1 and RF2 shows clearly that any such function at thepeptidyl transferase centre, if it occurs, is not essential toRF2 termination activity. The studies that we present here,like those concerning eRF1, illustrate the difficulties inpredicting the behaviour in vivo of mutations purely on thebasis of studies in vitro, as well as of drawing conclusionsabout the mechanism of a reaction purely from studies invivo. The results underline the importance of performingboth types of study.

At least two reasons are likely to contribute to the tox-icity associated with the expression of certain RF mutants.The first is competition with endogenous wild-type factor.In vitro experiments show that GAQ mutants bind to theribosome with the same affinity as the wild-type factor(Zavialov et al., 2002). A second possible mechanism isbased on the observation that overproduction of RF2(Thr-246) leads to an undermethylated protein that is insuffi-ciently active to maintain growth (Dinçbas-Renqvist et al.,2000). Thus, expression of mutant RFs that interact withthe methyltransferase prmC may also inhibit methylation

of RF2. It should also be recalled that RFs recognizecertain sense codons to a limited extent, notably UAU andUGG (Freistroffer et al., 2000). When present in excess,RFs will compete with appropriate tRNAs for sensecodons and may retard protein synthesis and, in somecases, cause premature termination leading to growtharrest. Finally, it cannot be excluded that a change in theconcentration ratio between RF1/2 and RF3 in vivo mayslow down the release of RF1/2 from the ribosome, retardribosome recycling and in this way affect the growth rateof bacteria (Pavlov et al., 1997b).

The fact that the GGQ tripeptide is the only sequenceconserved in all RFs suggests that the motif is functionallyimportant. Furthermore, the observation that variouschanges to the GGQ tripeptide in RFs from man and E.coli have very similar effects, combined with the fact thatGGQ is the only sequence motif common to all class IRFs, is consistent with the interpretation that the regionhas a similar function in both human and bacterial ribo-somes. In this case, it is likely that the crystal structuresof human eRF1 and E. coli RF2, which show quite differ-ent arrangements of the loops containing the GGQ motifsin each of the proteins, are not both representative of thestructures of the factors during the termination reaction onthe ribosome. The GGQ motif is likely to be involved intransmitting to the peptidyl transferase centre the informa-tion that a cognate stop signal interaction has been estab-lished. Clarification of this process would be greatly aidedby information about the precise region of the ribosomewith which the GGQ motif interacts. In view of the enor-mous progress that has been made in the elucidation ofthe ribosome structure, both alone and in complex withtRNA, mRNA and translational factors, it may be antici-pated that precise knowledge about how RFs are posi-tioned on the ribosome will soon be forthcoming.

Experimental procedures

Bacterial strains and plasmids

The thermosensitive prfA and prfB strains US486 (RF1-ts)(Rydén and Isaksson, 1984) and RM737 (RF2-ts) (Kawakamiet al., 1988) were used for cloning, gene modification and invivo experiments. E. coli B BL21 and BL21(DE3) (Studier andMoffatt, 1986) were used for the preparation of His-taggedrelease factors. Luria–Bertani broth (LB) was used as richmedium. Antibiotics were added as necessary at the follow-ing final concentrations: 50 mg ml-1 kanamycin, 200 mg ml-1

ampicillin and 12.5 mg ml-1 tetracycline.Plasmid pFJU335 (Jørgensen et al., 1993) was used as a

source of prfA. The SspI–HindIII fragment containing prfAwas cloned between the NruI and HindIII sites of pBR322.The resulting plasmid, like pFJU335, was found to encodeboth a normal and a minor elongated form of RF1, accountingfor the two induced proteins observed by Jørgensen et al.

Release factor mutations in the GGQ motif 273

© 2003 Blackwell Publishing Ltd, Molecular Microbiology, 47, 267–275

(1993). The elongated form carried a 25-amino-acid N-terminal extension originally encoded by pTrc99A, in phasewith prfA. For later preparations of RF1, the anomaly wasremoved by digestion of the plasmid with NcoI and SacI andreligation in the presence of the oligonucleotide 5¢-CATGAGCT-3¢, which disrupted the reading phase before theRF1 start codon, yielding the ampicillin-resistant plasmidpBR-prfA. This maintains prfA under the control of the IPTG-inducible trc promoter present in pFJU335 and derived pre-viously from pTrc99A (Amersham Pharmacia Biotech). Theplasmid was always co-transformed with the tetracycline-resistant plasmid pAIQ7, which carries lacIq to reduce expres-sion of prfA in the absence of inducer. pAIQ7, a kind gift fromM. Springer, is derived from pACYC184 and is thereforecompatible with pBR-prfA. To prepare mutants in the GGQsequence, mutagenic oligonucleotides covering the HpaI sitejust downstream of the sequence encoding GGQ in prfA(for example, 5¢-GGTGGTGTTAACGTGAGCACCACCC-3¢for GGA mutation) were used in conjunction with a secondoligonucleotide (5¢-CCCGTTCTGGATAATGTTTTTTGC-3¢)upstream of the BsrGI site for polymerase chain reaction(PCR). The product was digested with HpaI and BsrGI andused to replace the corresponding fragment in pBR-prfA. AC-terminal His-6 tag was added to RF1 encoded by pBR-prfA, and mutant derivatives of this plasmid, by an analogousprocedure using the oligonucleotides 5¢-GGGGGATCCTTAGTGGTGGTGGTGGTGGTGTGATTCCTGCTCGGACAACGCCGC-3¢ and 5¢-ACTTTCTGTTCTCGGTGCTCGCATC-3¢. The PCR product and the recipient plasmids weredigested and religated using the unique SgfI and HindIIIsites.

The expression vector pET-prfB for preparation of RF2,based on plasmid pET11A (Stratagene) and described pre-viously by Dinçbas-Renqvist et al. (2000), was used as asource of prfB. The frameshift site (Craigen and Caskey,1986) has been removed from prfB in this vector, and residue246 is Ala, as in all strains of E. coli characterized so farexcept for K-12 (Dinçbas-Renqvist et al., 2000). For the con-struction of mutants, the SalI–EcoO109 fragment containingpart of prfB was first cloned between the EcoRV andEcoO109 sites of pUC19 (Yanisch-Perron et al., 1985). TheHpaI and upstream BspEI sites were used for recloningmutant fragments generated in a similar way to thatdescribed above for prfA, and a C-terminal His tag was intro-duced. Preparative expression vectors were made for eachmutant by recloning the SalI–BamHI fragment into pET-prfB.For in vivo experiments, a variant (pTrc99C-Nde) of pTrc99Cwas constructed with an NdeI site in place of the NcoI site inthe polylinker. First, the unique NdeI in pTrc99C was removedby digestion with NdeI, the ends filled with Klenow DNApolymerase and religated. A mutagenic oligonucleotide 5¢-GCTCGAATTCCCCATATGCTGTTT-3¢ was used for PCRamplification with another oligonucleotide upstream of theEcoRV site in pTrc99C. The product was then digested withEcoRV and EcoRI and ligated between the same sites inpTrc99C, yielding pTrc99C-Nde. The NdeI–BamHI fragmentsfrom pET-prfB and mutant His-tagged derivatives were thenrecloned into pTrc99C-Nde, yielding pTrc-prfB and mutantderivatives. All cloning steps were verified by sequencing ofplasmids on both strands by the dideoxy method (Sangeret al., 1977).

Recombinant DNA techniques and genetic manipulations

General procedures for recombinant DNA techniques, plas-mid extraction, agarose gel electrophoresis, etc., were per-formed as described by Sambrook et al. (1989). DNAfragments from agarose gels were extracted using Jetsorbgel (Bioprobe) according to the manufacturer’s instructions.Transformations were performed as described by Miller(1992).

Purification of His-tagged RF1 and RF2 proteins

For in vitro study of the mutant RF1 and RF2 factors, His-tagged variants of the proteins were prepared by nickel affin-ity chromatography. E. coli BL21 cells transformed with plas-mid pBR-prfA or mutant variants and pAIQ7 were grown in1 l of LB medium supplemented with ampicillin and tetracy-cline. E. coli BL21(DE3) cells containing pET11a encodingthe RF2 mutants were grown in 1 l of LB medium supple-mented with ampicillin and kanamycin. Cultures were grownto an absorbance of 0.5 at 600 nm, and RF expression wasthen induced by the addition of IPTG at 1 mM final concen-tration, and incubation continued for 2 h. The cells werepelleted, resuspended in 20 ml of 30 mM Tris-HCl, 1 mMdithiothreitol (DTT), 100 mg of DNase, 0.5 mM phenylmethyl-sulphonyl fluoride (PMSF) and broken by passage througha French pressure cell at 7000 p.s.i. The supernatant wasloaded on to a 1 ml nickel column (resin Ni-NTA superflow;Qiagen) previously equilibrated with the same buffer. Thecolumn was washed, and the His-tagged protein was elutedby an imidazole step gradient from 10 mM to 100 mM. Puri-fied protein was visualized by SDS-PAGE followed by Coo-massie staining. Fractions containing His-tagged RF wereprecipitated by dialysis overnight at 4∞C against the samebuffer saturated to 80% with ammonium sulphate. The pre-cipitated proteins were pelleted, dissolved in polymix buffer(Ehrenberg et al., 1990), dialysed against the same bufferand stored in aliquots at -80∞C.

Determination of kcat/Km values for RF1 and RF2

Absolute values were determined as described byDinçbas et al. (1999) using ribosomal release complexesprogrammed with the mRNA XR7 (Freistroffer et al., 1997)containing the stop signal UAAU and encoding the tet-rapeptide MFTI. In vitro components apart from releasefactors were purified as described by Dinçbas et al.(1999), and 70S ribosome tight couples from strainMRE600 were isolated by zonal centrifugation (Noll et al.,1973; Rodnina and Wintermeyer, 1995).

Western blot

Western blot experiments were performed using rabbit anti-RF1 or anti-RF2 antibodies commercially prepared from purerelease factors. Cultures were centrifuged, and the cells werelysed for 10 min at 100∞C in lysis buffer (50 mM Tris-HCl,pH 6.8, 100 mM DTT, 2% SDS, 0.1% bromophenol blue).

274 L. Mora et al.

© 2003 Blackwell Publishing Ltd, Molecular Microbiology, 47, 267–275

Proteins were separated by electrophoresis on 10% poly-acrylamide gels as described by Laemmli (1970). Transferto nitrocellulose membranes and Western blotting withantibodies (diluted ¥5000) were performed as described bySambrook et al. (1989), using 125I-labelled protein A (Amer-sham) to reveal the antibody. The radioactivity in individualbands was determined using a phosphorimager (MolecularDynamics).

Acknowledgements

We thank Monica Rydén and Yoshi Nakamura for the kindgift of thermosensitive RF strains. This work was supportedby the Centre National de la Recherche Scientifique(UPR9073), l’Association pour la Recherche sur le Cancer,the Fondation pour la Recherche Medicale, the SwedishResearch Council for Engineering Sciences, the SwedishNatural Science Research Council, the Russian Foundationfor Basic Research and the Programme of Support forRussian Scientific Schools. V.H.-H. and L.K. thank theHuman Frontier Science Programme for support. L.K. waspartly supported by INTAS (grant 00–041).

References

Cantor, C.R. (1979) tRNA–ribosome interactions. In TransferRNA: Structure, Properties and Recognition. Schimmel,P.R., Soll, D., and Abelson, J.N. (eds). Cold Spring Harbor,NY: Cold Spring Harbor Laboratory Press, pp. 363–392.

Craigen, W.J., and Caskey, C.T. (1986) Expression of pep-tide chain release factor 2 requires high-efficiency frame-shift. Nature 322: 273–275.

Dinçbas, V., Heurgué-Hamard, V., Buckingham, R.H.,Karimi, R., and Ehrenberg, M. (1999) Shutdown in proteinsynthesis due to the expression of mini-genes in bacteria.J Mol Biol 291: 745–759.

Dinçbas-Renqvist, V., Engström, Å., Mora, L., Heurgué-Hamard, V., Buckingham, R.H., and Ehrenberg, M. (2000)A post-translational modification in the GGQ motif of RF2from Escherichia coli stimulates termination of translation.EMBO J 19: 6900–6907.

Ehrenberg, M., and Tenson, T. (2002) A new beginning ofthe end of translation. Nature Struct Biol 9: 85–87.

Ehrenberg, M., Bilgin, N., and Kurland, C.G. (1990) Designand use of a fast and accurate in vitro translation system.In Ribosomes and Protein Synthesis. Spedding, G. (ed.).Oxford and New York: IRL Press, Oxford University Press,pp. 101–129.

Freistroffer, D.V., Pavlov, M.Y., MacDougall, J., Buckingham,R.H., and Ehrenberg, M. (1997) Release factor RF3 in E.coli accelerates the dissociation of release factors RF1 andRF2 from the ribosome in a GTP dependent manner.EMBO J 16: 4126–4133.

Freistroffer, D.V., Kwiatkowski, M., Buckingham, R.H., andEhrenberg, M. (2000) The accuracy of codon recognitionby ribosome release factors. Proc Natl Acad Sci USA 97:2046–2051.

Frolova, L., Le Goff, X., Rasmussen, H.H., Cheperegin, S.,Drugeon, G., Kress, M., et al. (1994) A highly conserved

eukaryotic protein family possessing properties of polypep-tide chain release factor. Nature 372: 701–703.

Frolova, L.Y., Tsivkovskii, R.Y., Sivolobova, G.F., Oparina,N.Y., Serpinsky, O.I., Blinov, V.M., et al. (1999) Mutationsin the highly conserved GGQ motif of class 1 polypeptiderelease factors abolish ability of human eRF1 to triggerpeptidyl-tRNA hydrolysis. RNA 5: 1014–1020.

Heurgué-Hamard, V., Karimi, R., Mora, L., MacDougall, J.,Leboeuf, C., Grentzmann, G., et al. (1998) Ribosomerelease factor RF4 and termination factor RF3 are involvedin dissociation of peptidyl-tRNA from the ribosome. EMBOJ 17: 808–816.

Heurgué-Hamard, V., Dincbas, V., Buckingham, R.H., andEhrenberg, M. (2000) Origins of minigene-dependentgrowth inhibition in bacterial cells. EMBO J 19: 2701–2709.

Heurgué-Hamard, V., Champ, S., Engstöm, Å., Ehrenberg,M., and Buckingham, R.H. (2002) The hemK gene inEscherichia coli encodes the N(5)-glutamine methyltrans-ferase that modifies peptide release factors. EMBO J 21:769–778.

Ito, K., Uno, M., and Nakamura, Y. (2000) A tripepeptideanticodon deciphers stop codons in messenger RNA.Nature 403: 680–684.

Jørgensen, F., Adamski, F.M., Tate, W.P., and Kurland, C.G.(1993) Release factor-dependent false stops are infre-quent in Escherichia coli. J Mol Biol 230: 41–50.

Kawakami, K., Inada, T., and Nakamura, Y. (1988) Condi-tionally lethal and recessive UGA-suppressor mutations inthe prfB gene encoding peptide chain release factor 2 ofEscherichia coli. J Bacteriol 170: 5378–5381.

Kisselev, L.L., and Buckingham, R.H. (2000) Translationaltermination comes of age. Trends Biochem Sci 25: 561–566.

Kisselev, L.L., Oparina, N.Y., and Frolova, L.Y. (2000)Class-1 translation termination factors are structurally andfunctionally similar to suppressor tRNAs and belong todistinct structural/functional families (prokaryotes/mito-chondria and eukaryotes/archaea). Mol Biol (Moscow) 34:427–442.

Laemmli, U.K. (1970) Cleavage of structural proteins duringthe assembly of the head of bacteriophage T4. Nature 227:680–685.

Miller, J.H. (1992) A Short Course in Bacterial Genetics. ColdSpring Harbor, NY: Cold Spring Harbor Laboratory Press.

Nakahigashi, K., Kubo, N., Narita, S., Shimaoka, T., Goto,S., Oshima, T., et al. (2002) HemK, a class of proteinmethyl transferase with similarity to DNA methyl trans-ferases, methylates polypeptide chain release factors, andhemK knockout induces defects in translational termina-tion. Proc Natl Acad Sci USA 99: 1473–1478.

Nakamura, Y., Ito, K., and Isaksson, L.A. (1996) Emergingunderstanding of translation termination. Cell 87: 147–150.

Nissen, P., Kjeldgaard, M., Thirup, S., Polekhina, G.,Reshetnikova, L., Clark, B.F.C., and Nyborg, J. (1995)Crystal structure of the ternary complex of Phe-tRNAPhe,EF-Tu and a GTP analogue. Science 270: 1464–1472.

Noll, M., Hapke, B., and Noll, H. (1973) Structural dynamicsof bacterial ribosomes. II. Preparation and characterizationof ribosomes and subunits active in the translation ofnatural messenger RNA. J Mol Biol 80: 519–529.

Release factor mutations in the GGQ motif 275

© 2003 Blackwell Publishing Ltd, Molecular Microbiology, 47, 267–275

Pavlov, M.Y., Freistroffer, D., MacDougall, J., Buckingham,R.H., and Ehrenberg, M. (1997a) Fast recycling of E. coliribosomes requires both ribosome recycling factor (RRF)and release factor RF3. EMBO J 16: 4134–4141.

Pavlov, M.Y., Freistroffer, D.V., Heurgué-Hamard, V.,Buckingham, R.H., and Ehrenberg, M. (1997b) Releasefactor RF3 abolishes competition between release factorRF1 and ribosome recycling factor (RRF) for a ribosomebinding site. J Mol Biol 273: 389–401.

Rodnina, M.V., and Wintermeyer, W. (1995) GTP consump-tion of elongation factor Tu during translation of heteropoly-meric mRNAs. Proc Natl Acad Sci USA 92: 1945–1949.

Rydén, S.M., and Isaksson, L.A. (1984) A temperature-sensitive mutant of Escherichia coli that shows enhancedmisreading of UAG/A and increased efficiency for sometRNA suppressors. Mol Gen Genet 193: 38–45.

Sambrook, J., Fritsch, E.F., and Maniatis, T. (1989) Molecu-lar Cloning: A Laboratory Manual. Cold Spring Harbor, NY:Cold Spring Harbor Laboratory Press.

Sanger, F., Nicklen, S., and Coulson, A.R. (1977) DNAsequencing with chain-terminating inhibitors. Proc NatlAcad Sci USA 74: 5463–5467.

Seit Nebi, A., Frolova, L.Y., Ivanova, N., Poltarous, A., andKisselev, L.L. (2000) Substitutions of the glutamine residuein the ubiquitous GGQ tripeptide in human eRF1 do notentirely abolish the release factor activity. Mol Biol (Mos-cow) 34: 899–900.

Seit-Nebi, A., Frolova, L., Justesen, J., and Kisselev, L.(2001) Class-1 translation termination factors: invariantGGQ minidomain is essential for release activity and ribo-

some binding but not for stop codon recognition. NucleicAcids Res 29: 3982–3987.

Song, H., Mugnier, P., Das, A.K., Webb, H.M., Evans, D.R.,Tuite, M.F., et al. (2000) The crystal structure of humaneukaryotic release factor eRF1 – mechanism of stop codonrecognition and peptidyl-tRNA hydrolysis. Cell 100: 311–321.

Studier, F.W., and Moffatt, B.A. (1986) Use of bacteriophageT7 RNA polymerase to direct selective high-level expres-sion of cloned genes. J Mol Biol 189: 113–130.

Uno, M., Ito, K., and Nakamura, Y. (1996) Functional speci-ficity of amino acid at position 246 in the tRNA mimicrydomain of bacterial release factor 2. Biochimie 78: 935–944.

Vestergaard, B., Van Bich, L., Andersen, G.R., Nyborg, J.,Buckingham, R.H., and Kjeldgaard, M. (2001) Bacterialpolypeptide release factor RF2 is structurally distinct fromeukaryotic eRF1. Mol Cell 8: 1375–1382.

Yanisch-Perron, C., Vieira, J., and Messing, J. (1985)Improved M13 phage cloning vectors and host strains:nucleotide sequences of the M13mp18 and pUC19 vec-tors. Gene 33: 103–119.

Zavialov, A.V., Buckingham, R.H., and Ehrenberg, M. (2001)A post-termination ribosomal complex is the guaninenucleotide exchange factor for peptide release factor RF3.Cell 107: 115–124.

Zavialov, A.V., Mora, L., Buckingham, R.H., and Ehrenberg,M. (2002) Release of peptide promoted by the GGQ-motifof class 1 release factors regulates the GTPase activity ofRF3. Mol Cell 10: 789–798.