Disrupts Interaction between the Ternary Complex and the Ribosome

(1)

JOURNALOFBACTERIOLOGY,Jan. 1993,p.240-250 0021-9193/93/010240-11$02.00/0

Vol. 175,No. 1

A Single Amino Acid Substitution in Elongation Factor Tu Disrupts Interaction between the Ternary

Complex and the Ribosome

IOANNIS TUBULEKASAND DIARMAID HUGHES*

Department of Molecular Biology, Box 590, TheBiomedical Center, Uppsala University, S-751 24 Uppsala, Sweden

Received 30 March 1992/Accepted 2 November 1992

Elongation factor Tu (EF-Tu) GTP has the primary function of promoting the efficient and correct

interaction of aminoacyl-tRNA with the ribosome.Verylittle is known abouttheelements inEF-Tuinvolved in this interaction. We describe amutantform of EF-Tu, isolated in Salmonellatyphimurium, thatcauses a severedefectin the interaction of the ternary complex with the ribosome. The mutationcausesthesubstitution of ValforGly-280in domain^IIofEF-Tu. The in vivo growth and translation phenotypes of strains harboring thismutationareindistinguishable fromthoseof strains in which thesametufgeneisinsertionallyinactivated.

Viable cells are notobtained when the othertufgene is inactivated, showing that the mutant EF-Tualone cannot support cell growth. We have confirmed, by partial protein sequencing, that the mutant EF-Tu is presentin thecells. In vitro analysis of the natural mixture of wild-type andmutantEF-Tu allowsustoidentify the major defect of thismutant. Our datashows that theEF-Tu ishomogeneous andcompetentwithrespect toguanine nucleotide bindingandexchange,stimulation ofnucleotideexchange by EF-Ts, and ternary complex formation with aminoacyl-tRNA. However various measures of translational efficiency show a significant reduction, which is associated with a defective interaction between the ribosome and the mutant EF- Tu. GTP aminoacyl-tRNA complex. In addition, the antibiotic kirromycin, which blocks translation by binding EF-Tuontheribosome, failstodosowith thismutantEF-Tu,although itdoes formacomplexwith EF-Tu. Our resultssuggestthat thisregion ofdomainIIinEF-Tuhasanimportantfunctionandinfluences the bindingoftheternarycomplextothecodon-programmed ribosomeduring protein synthesis.Modelsinvolving eithera directoran indirecteffect ofthe mutationarediscussed.

The prokaryotic translation factor elongation factor Tu (EF-Tu) mediates the productive interaction of aminoacyl- tRNA (aa-tRNA) with the ribosome. The ternary complex

EF-Tu. GTP.

aa-tRNA interacts with the ribosome such that the anticodon region of aa-tRNA is properly positioned in the ribosomal A site on the 30S subunit, while the aa region remains bound to EF-Tu outside the A site (43).

EF-Tu on the ribosome protects bases in the universally conserved a-sarcin loopof 23S rRNAonthe 50S ribosomal subunit, suggesting that this may form at least part of its ribosomalbindingsite(20, 44).This conclusion issupported by data showing that a base substitution mutation in the ot-sarcinloop can affect the binding of ternary complextothe ribosome (62). If the tRNA codon and the mRNA anticodon in the30S Asitematch, there is ahighprobability that the GTPonEF-Tu will be hydrolyzed. Whether theprobability of GTPhydrolysis depends mainlyonthelengthoftime that the ternarycomplex spends on the ribosome or on signals transmittedtoEF-Tuaftercorrectcodon-anticodon interaction is an open question. However, after GTP hydrolysis, EF-Tu- GDP leaves the ribosome and is recycled, via interaction with the nucleotide exchange factor EF-Ts, to

EF-Tu.

GTP and is again capable of forming a ternary complexwith aa-tRNAand participating in translation (30, 41). Accordingtothedata and model of Moazed and Noller (43), it isonlyafter

EF-Tu.

GDP has left the ribosome that aa-tRNA is able tooccupy the A siteon the50S ribosomal subunit. Moazed and Noller(43)haveproposedthat thiscan

*Correspondingauthor.

provideaphysicalbasis forhow kineticproofreadingoccurs (21,46) byshowing that aa-tRNA selection in theAsite is a multistep process. There is experimental support from in vitro translation for the kinetic proofreading of aa-tRNA selection(53, 63).Thus, accordingto the model of Moazed and Noller (43), an important function of EF-Tu is to facilitate the possibility of codon-anticodon interactions while at the same time preventing peptide bond formation until after the GTP on EF-Tu has been hydrolyzed and EF-Tuhas left the ribosome.

One of the aims ofcurrent studies of EF-Tu isto under- stand at the molecular level the relationships between its structureand functions. ThecrystalstructureofEschenchia coli EF-Tu in complex with GDP reveals three distinct domains with theguaninenucleotide bound tothe N-termi- nal domain (13, 33, 35, 47). We have studied Salmonella typhimunumEF-Tu,which is identical in sequencetoE. coli EF-Tuwith the exception of aLeu-*Ile change at residue 189 (65), a residue which is not conserved

(14, 38).

Thus, dataonstructure-functionrelationshipsshouldapply equally toEF-Tusfrom bothspecies.S. typhimunum,besidesbeing closelyrelatedtoE. coli,has the advantagethat each of its twogenesfor EF-Tu is

individually dispensable

forgrowth (23), facilitating mutant analysis. For reasons that arecur- rentlyunclear, theE. coli tufA gene cannot be inactivated (69).

The molecular interactions of EF-Tu with the guanine nucleotides have been studied in most

detail,

but some progress has also been made in

defining

the elements involved in interactions with the nucleotide

exchange

factor EF-Tsand with aa-tRNA. Conserved aminoacid

loops

in the 240

on May 29, 2021 by guesthttp://jb.asm.org/Downloaded from

(2)

TABLE 1. S. typhimuriumstrainsusedin thisstudy

Strain Genotype

TH488...trpE91hisG3720tufA8 TH506...trpE91hisG3720tufA8tufB414

JT506^.. ^...trpE91hisG3720tufA8 tufB414argH1823::TnlO TH195... trpE91tufA8

JT620.. ...trpE91tufA8proB1657::TnlO

JT608^.. ^...trpE91 tufA8 tufB414

JT627^.. ^...trpE91tufA8tufB414proB1657::TnlO JT640... trpE91 tufB414argH1823::TnlO JT642... trpE91 tufB414

JT631... trpE91tufB414proB1657::TnJO

JT863... trpE91proBAl tufB441::MudJ(lacZ950::TnlO) JT848 ^...trpE91proBAl tufA8 tufB414::MudJ(lacZ950::TnlO)

immediate neighborhood

of the

guanine

nucleotide

(10, 15)

were shown bymutagenesis to be involved in

determining

the relative affinity of GDP and GTP for the molecule, its hydrolysis, and thenatureofthenucleotide which is bound (2,25, 26). EF-Tsbindingisinfluenced byamutation close totheguanine nucleotide in domain I(27), but the

primary

binding siteforEF-Tsmay be on domain III(8, 49,56).The binding oftRNA toEF-Tu,studiedby cross-linking

(16,

40,

68)

and chemicalprotection

(5, 28),

indicates that residues in domains I and II of EF-Tuinteractwith aa-tRNA and that the aamoietycanbe crosslinked toHis-66 in domain I. In EF-la (the eukaryotic homolog ofEF-Tu), which lacksthe equivalent of His-66 in domain I, the amino acid has been cross-linkedtotheextremeedge of domainII

(31).

Evidence frombiochemical(11) and genetic

(6,

34,57)assaysindicates that EF-Tuinteracts withtheaminoacid, the acceptor stem, andthe

T*C

^helixof aa-tRNA.

Onceformed, the

EF-Tu. GTP.

aa-tRNAcomplexmust interactwiththeribosome. ThatEF-Tuisdirectly involved ininteracting withtheribosome issuggestedby thebinding of

EF-Tu.

GDP to theribosome inthe presenceof

kirromy-

cin (70). EF-Tu, bound to the ribosome in the form EF-

Tu. GDP. aa-tRNA.

kirromycin, protects bases in the ao-sarcin loopon thelarge ribosomalsubunit againstchemi- cal attack(44); thissuggeststhat thisregion of the ribosome maybeatleastpart ofitsbinding site. Inaddition, thereis genetic evidencethat EF-Tu'sinteraction with the ribosome isinfluenced stronglyby protein S12onthesmallribosomal subunit

(7,

60, 62, 64). Thereiscurrentlyverylittle informa- tion on the residues inEF-Tu which are important for the interaction with the ribosome. On the basis of the area of

homology

betweenEF-Tu andEF-G, with which it sharesan overlapping binding siteontheribosome(37, 44),onewould predictthat theribosome interaction should involve domain IofEF-Tu, and thereis evidence supporting this prediction (48, 50). Inthis article,wereportexperiments showing that an amino acid substitution at residue 280, a conserved residue in a loop in domain II, disrupts ternary complex interaction with the ribosome. The effects of the alteration must berelatively specificto the ternarycomplex-ribosome interaction because we do not detect changes in the other molecular interactionsofEF-Tuwhich wehave tested.

MATERIALSANDMETHODS

Bacterial andphage strains. Thebacterial strains used in this study are listed in Table 1. All of the strains are derivatives of the wild-type strain S. typhimurium LT2.

Transductions were made with P22 HT105/1 int-201 (55).

The selection of the mutation studied here,

tufB414,

has

beendescribedpreviously

(1).

Briefly, beginningwithstrain TH488

(which

carries akirromycin-resistant

tufA

allele and a

wild-type kirromycin-sensitive tufB

gene andis thus sen- sitivetotheantibiotic),mutantsresistanttokirromycinwere selected. Among these mutant strains were a number in which the

tufB

gene was no

longer

able to support cell

growth

in the absence ofanactive

tufA

gene.DNA sequenc- ingof thetufb regionsofthese mutantsidentified anumber of amino acid

substitutions, including

thechangeat

position

280 studiedhere.

ThetufB414mutationwasintroducedintodifferent strains

by

cotransduction with the linkedmarker

argH1823::TnlO, selecting

for tetracycline resistance. This marker could be subsequentlyremoved bya transductionselecting for

argi-

nine prototrophy. The cotransduction oftufB414 into the

tufA8

background was confirmed by the appearance of a

kirromycin-resistant

phenotype. The cotransduction of tufB414 into the

tufA'

background was confirmedbothby ourabilityto select kirromycin-resistant colonies at ahigh

frequency

andby the appearance ofa

kirromycin-resistant

phenotype after the subsequent introduction of

tufA8

by transduction. We confirmed that the tufB414 mutation is inviablebyourinabilitytointroduce

tufA::MudJ

alleles(23) by cotransduction with the linked marker zhb-736::TnlO.

ThemarkerproB1657::TnlOwasintroduced by selectionfor tetracyclineresistance and used for thesubsequent selection andmaintenanceof F'factorscarryingtheE.coli lac operon andproAB region byselection forproline prototrophy.

Media. Media and antibiotic concentrations have been described previously (64). Kirromycin (mocimycin) was a gift from Gist-Brocades NV, Delft, TheNetherlands.

Determinations of growth rate and suppression. Growth rates were measured in liquid M9 salts supplemented with glucose (0.4%, wt/vol) and tryptophan(10mM)byinoculat- ing1001.lofanovernightculture into 20 ml of fresh medium in a300-ml flask, aerating the mixture byvigorous shaking, andmeasuringtheincrease inoptical densityof theculture as a function of time. Suppression of the trpE91 and hisG3720 mutations was measured as the time taken to form colonies on M9 minimal medium agar(24). Suppression of nonsense mutations in the lacI part of a lacIZ fusion was determinedby measuring f3-galactosidase activity and nor- malizing it to the activity from a nonmutated fusion inthe samestrain(22). The translationelongation rate invivowas estimated by measuring the step time for

P-galactosidase

synthesis after induction of a wild-type lac operon as describedby Andersson et al. (4).

P-Galactosidase

activitywas measured asdescribed by Miller(42).

In vitro translation assays.EF-Tu was purified according to the method of Leberman et al. (36) with the following modifications. FractionscontainingEF-Tu from the DEAE- Sepharose and Ultrogel AcA44 (AcA) (Sepracor, France) columns were not ammonium sulfate precipitated but were instead pooled and loaded directly onto a small (10-ml) column of Q Sepharose Fast Flow (Pharmacia) previously washed with buffer A(DEAE)orphosphate buffer (AcA) as appropriate. Samples from the DEAEcolumnwere washed from the Q Sepharose by the application of buffer A containing 0.4 M NaCl and loaded directly onto the AcA column. Samples from the AcA column werewashed from theQ Sepharose bythe application of a phosphategradient from 50 to 250 mM. The final EF-Tu-containing fractions from the AcA column (about 20 ml) were concentrated 10-fold with Aquacid and then dialyzed against polymix buffercontainingGDPand stored at

-800C.

These additional steps with Q Sepharose have the effectofrapidly concen-

(3)

242 TUBULEKAS AND HUGHES

trating the EF-Tu samples and simultaneously giving a significantly improved purificationoverthat achievedbythe traditional method. All other factors, components, and chemicals for invitro translationwerepurifiedandprepared bythe method ofEhrenberget al. (18).

Translationassays in vitroweredone at370C with

100-1.l

volumes in polymix buffer (final ion concentrations as fol- lows: 5 mM magnesium acetate, 0.5 mM

CaCl2,

95 mM

KCl,

5 mM NH4Cl, 8 mM

putrescine,

1 mM

spermidine,

5 mM potassium phosphate [pH7.3], and 1 mM

dithioerythritol).

To assay translation elongation, two separate mixes of components were

prepared

on ice: an initiation mix

(70S

mix) and a factor mix. Each mix was preparedwith appro- priatevolumesofwater,

10x polymix buffer,

20x

potassium

phosphate, and 50x dithioerythritol to balance the final polymix buffer. The 70S mix contained

(per

50

p.l)

10

pmol

of activeribosomes,20

pug

of

poly(U),

and

[3H]NAc (N-acetyl- Phe-tRNAPhe)

in a30% excess overthetotal number of 70S ribosomes. The factor mixcontains(per 50

,ul)

10 plof

A/P (10

mM ATP and 100 mM

phosphoenolpyruvate

inwater at pH 7.0), pyruvate kinase

(5 pg), myokinase (0.3 pg), [14C]Phe (30 nmol),

Phe

synthetase (100 U),

tRNAPhe

(250

pmol), EF-Tu(600

pmol),

EF-G

(250 pmol),

and EF-Ts

(150

pmol). All translation

experiments

were carried out

by

the methods andprinciples detailed

by Ehrenberg

et al.

(18)

and Bohmanetal.

(9),

with the

exception

of the variations noted below. After the

EF-Tu.

GDPconcentrationwasmeasured by a nucleotide

exchange

assay

(18, 19),

different preparations were

routinely compared by gel electrophoresis

and quantified with an LKB-Pharmacia laser densitometer.

Translation assaysas afunctionof

kirromycin

concentration werecarriedout asdescribed

by Hughes (23).

The

ability

of kirromycin to bind EF-Tu on ribosomes was

assayed by

measuringthe translationrate atfixedamountsof EF-Tu

(30 pmol)

and

kirromycin (100 pmol)

and with a titration of active ribosomes from 30 down to 1

pmol (from approxi-

mately 3-fold over to 10-fold under the concentration of

kirromycin-sensitive EF-Tu).

The translation incubation timeswerefrom 15 to120sfor EF-TuA8

B+

and from 20to 25 s for EF-TuA8 B414 to obtain

precipitable poly(Phe)

chains of 20 to 25 amino acids. Other components ^were present in

nonlimiting

amounts. Measurements of the

kcatlKm

of the interaction of the ternary

complex

with the ribosome weremade as described

by

Tubulekaset al.

(64).

Briefly,

wemeasured the rateoftranslation of

poly(U)

^{as a}

function of the amount of EF-Tu

(titrated

from 20 to 600

pmol)

in a 100-pl translation cocktail

containing

¹⁰

pmol

of active ribosomes and an excess ofall other

components

to

achieve a maximal

elongation

rate. Reaction times were

from 4to20s toobtain

precipitable poly(Phe)

chains of 15to 24amino acids. The assayoftranslationrateas afunction of

aa-tRNAPhe

concentrationwasdone as described

by Tapio

et al.

(59). Briefly,

thetranslation rate as a function of the amount of Phe-tRNAPhe (titrated from 100 to

2,000 pmol)

was measured at small and

rate-limiting

amounts of EF-Tu (15

pmol)

and ribosomes

(50 pmol,

of which about25%was

active in

translation).

Thereaction timeswerevariedfrom 10 to 20 s toobtain

precipitable poly(Phe)

chains of about 20 amino acids. Nucleotide

exchange

onEF-Tu

by

EF-Tswas

stimulated asdescribed

by Tapio

etal.

(59).

EF-Tu. GTP. aa-tRNAPMe complex

formation. The

ability

of EF-Tu

preparations

to form a ternary

complex

with aa-tRNAwas

assayed by

^a

nondenaturing polyacrylamide

gel

electrophoresis (PAGE)

system. The details of this method were

developed by

N.

Bilgin,

T.

Gluick,

and M.

Ehrenberg, Department

ofMolecular

Biology, Uppsala

Uni-

versity (7a).

The

electrophoresis

bufferconsisted of 50 mM Tris-HCl

(pH 6.5),

10 mM

Mg

acetate,65 mM

NH4

acetate, 1mMdisodium

EDTA,

10

p.M GTP,

and 1 mM

dithioeryth-

ritol.The vertical slab

gels (5

mmthick

by

20cm

long)

were

madewith

electrophoresis

buffer

containing 5% acrylamide- bisacrylamide (19:1),

1 mg ofammonium

persulfate

per

ml,

and 8.3 mM TEMED

(NNN',N'-tetramethylethylenedi- amine).

The

gels

^wereprerun for 90 minat40 mA and 100 V and after

sample loading

were runforafurther3 hat60 mA and 150 V

(with

fresh buffer and a further buffer

change during

the

run). Electrophoresis

was

performed

in a cold

room

(40C)

with fan

cooling

and buffer circulation. Each

sample

was

prepared

^{as a}

50-plA

translation factor mix

(18) containing

200

pmol

of

EF-Tu,

^{from 0} to 300

pmol

of

tRNAPhe,

andnoEF-TsorEF-G. This mixwasincubatedat

370C

for 10 min

(during

which the tRNAwas

fully charged

and the

ternary complex formed)

andcooledon

ice,

and then 20

jul

^was

immediately applied

^tothe

gel

^afterthe addition of 0.1volumeof50%

glycerol

ⁱⁿ ^water^with ^a^dash^ofbromo-

phenol

blueasamarker.Theremainder of each

sample

was

usedto measuretRNA

charging

levels. The

gels

^werefixed and stained for 20 min in 40% methanol-10% acetic acid- 0.1% Coomassie brilliant blue. The

gels

^were destained in 12%

2-propanol-5%

acetic acid and air dried in a frame between two

cellophane

sheets. For relative

quantification,

the

gels

were scanned withan LKB-Pharmacialaserdensi-

tometer.

Protein

sequencing. Cyanogen

bromide

(CNBr) fragments

ofEF-Tu

(3,000 pmol)

were

electrophoresed

andtransferred

by blotting

onto a

polyvinylidine

difluoride membrane as

described

by

Matsudaira

(39).

The

fragment containing

position 280

(29)

wasextractedfrom themembraneand

directly

loaded intothe

sequencing

machine. Automatedaminoacid

sequencing

was

performed

on an

Applied Biosystems

477A Protein

Sequencing System equipped

with an on-line 120A

phenylthiohydantoin analyzer.

RESULTS

We have isolated and

sequenced

^a^{series of}

spontaneous

mutations inS.

typhimurium tuft

whichcause

single

amino acid substitutions

individually

lethaltosomeessentialfunc- tion of EF-Tu

(1).

A mutation was defined as lethal if no

viable cells with an

insertionally

inactivated

tufA

in the presence of the

tuftl

mutation couldbeisolated. Neither

tufA

nor

tuft?

^is

individually

essentialfor

viability

inS.

typhimu- rium,

and either one can beinactivated

by

Mud insertions

(23).

Inthis

study,

we

investigated

thefunctional basisofthe

lethality

ofoneof these

mutations, Gly-280--*Val,

which isa

mutation at aconserved

position

in a

loop

in domain II of EF-Tu

(Fig. 1).

Phenotypes

of

tIuB414

in vivo.The mutation

tufB414

^was

initially

isolatedonthe basis that itconferred^a

kirromycin-

resistant

phenotype

on TH488

(a

strainwhich

already

har- bored a

kirromycin-resistant

and

error-prone

alleleof

tufA

and a

wild-type

copy of

tuff?

and is thus sensitive to the

antibiotic).

Itwasnoted that

tufA8-mediated suppression

of both a nonsense mutation

(hisG3720)

and a frameshift mu-

tation

(trpE91) present

in the strainwas

reduced,

as

judged by colony growth

in the absence of the

appropriate

amino acid. An obvious

possibility

^was ^that

suppression

^{was re-}

ducedbecause ofan increased

efficiency

of termination

by

releasefactors

competing against

areducedamountof active EF-Tu. To

quantify

these effects andextendour

analysis,

we

constructed strains

carrying

lac operon constructs which facilitate the measurement of nonsense

read-through

and J. BACTERIOL.

(4)

222 d

280

cm

FIG. 1. Ribbondiagram of theX-ray crystallographic structure ofamildly trypsinizedform of EF-Tu. GTPlackingamino acids 45 to58,based^onthe model of Clarketal.(13).Thedomains of EF-Tu referred toin the textarelabelled I, II, and III.Theapproximate positionsof the amino acid substitution mutations referredtoin the text(Gly-280-*ValandGly-222-*Asp [Bo]),the site ofcross-linking of the amino acidonaa-tRNA(His-66), and the site of theguanine nucleotide (GDP)areindicated.

translationelongation rate (see Materials andMethods). In addition,wemeasured thegrowthratesof these strains. As

ausefulcomparison,wemeasured each of these parameters for strains carrying a MudJ-inactivated tufB gene. The results of thesemeasurementswerethat all threeparameters (growthrate, step time for P-galactosidase synthesis [elongation rate], and nonsenseread-through) werereduced significantly andtoequivalent degrees by the tufB mutation in eachsetof strains(Table 2). Thus, the tufB414 mutation, in both the tufA+ and tufA8 backgrounds, has an in vivo phenotype which is similar tothat of^an insertionallyinacti- vatedtuft? gene.

The mutant species EF-TuB Val-280 is present. We con-

firmed that the mutant EF-Tu was present in vivo by purifying EF-Tu froma

tufA+

tufB414 strain and sequencing the relevant part of the protein. EF-Tuwas cleaved with

cyanogen bromide and blotted onto polyvinylidine difluoride, and the relevant fragmentwas excised andsequenced

as described in Materials and Methods. The amino acid

sequencewas asexpected for thispartof theprotein (29) and

TABLE 2. Growthrate,translation elongationrate, andnonsense

read-through phenotypesoftufB414 Suppression

(104)atUGA Elongation Growth rate Strain Genotype codona: (amino (doublings/h) acids/s) (duinsh 189 220

JT619 tufA+ tufB+ 18.9 66.0 13.1 1.31

JT620 tufA8 tufB+ 76.7 139.8 13.1 1.18

JT631 tufA+ tufB414 7.3 32.6 10.7 1.08

JT627 tufA8tufB414 35.3 58.1 9.3 0.83 JT863 tufA+ tufB::MudJ 6.4 30.0 10.5 1.20 JT848 tufA8tuJB::MudJ 40.0 80.0 10.2 0.88

aNonsenseread-throughatposition189 or 220 inthe

lacI

partofalacIZ fusionon anF'factor.Suppressionis measuredinarbitrary units,asdefined by Miller(42).

showed the presence of both

Gly

and Val at

position

280 (Fig. 2). This confirms that bothEF-Tuspecieswerepresent in the cells and in the preparations we used for in vitro translation

analysis.

Theratioof

Gly

toValat

position

280 in our

protein preparations,

asestimated from

protein

sequencing, wasapproximately 10:6. This estimate is derived from the increase in the area of each of the amino acid

peaks

compared with the area of the same

peak

in the

(thus,

the

peak

sizes in

Fig.

2 are not

representations of the amount of each amino acid at that

position

in the

protein).

We

performed

control

experiments

toshowthatour

sequencing

result is from apure

fragment

and isnotcontaminated

by

other

peptides

from EF-Tu. The

sequencing

reaction

gives

resultswhichare more

qualitative

than

quantitative,

and our

primary

purpose in this

experi-

ment was to confirm the existence of both types in our

preparation. However,ourestimate that

approximately

one-

third of the EF-Tu is the mutant EF-TuB

species

is in agreementwitha

variety

of data

showing

that theamountof

tufB

product inE.coli and S.

typhimurium

is

normally

close to one-third of thetotalEF-Tu

(3, 23, 51, 67).

In vitro characterization of the tuJB414-encoded EF-TuB Val-280.The

tufB414

mutationcannotsupportcell

growth

in the absence of an active

tufA

gene, so we did all of our assayswith the natural mixtureofAandB EF-Tu

species.

This has the

disadvantage

that in each assaywemeasureda

change

in the

magnitude

ofa

signal against

a

background

of active EF-TuA.

Conversely, however, using

the natural mixtureof

A'

andB414has the

advantage

of

providing

an

internalcontrol for theactivitiesofourEF-Tu

preparations.

EF-Tuconcentration determinationbynucleotideexchange.

The concentration of

EF-Tu.

GDP

preparations

was mea- suredbyanucleotide

exchange

assay

(19)

inwhich

[3H]GDP

released from EF-Tuin the absence ofEF-Tsis

enzymati-

callyconverted into

[3H]GTP.

The ratios of the nucleotides as a function of time are then measured after

thin-layer chromatography

and scintillation

counting.

The

experiment

is run in two

halves,

with and without aknown amount of

extracoldGDP. The EF-Tuconcentration is determined as

describedindetail

by Ehrenberg

etal.

(19).

Concentrations determined

by

this method

(for

various

preparations

of

wild-type

andmutant

EF-Tu)

were tested for their

protein

content

by

sodium

dodecyl

sulfate

(SDS)-PAGE

and Coomassieblue

staining

followed

by quantification by

laser

densitometry (see

Materials and

Methods).

We used as a standard an EF-Tu

preparation

whose concentration was

also

independently

measured

by

amino acid

analysis.

For each ofthe

preparations

ofEF-Tu

tested,

there was avery

good

correlation between thetwomethods.

Figure

3shows the results ofa

typical experiment,

inwhichthe concentra-

tions estimated from nucleotide

exchange

assays

(Fig.

3A and

B)

were used to

apply equal

^amounts of EF-Tu for SDS-PAGE

(Fig. 3C),

whichwasfollowed

by

densitometric

quantification,

which revealed no

significant

difference.

These EF-Tu

preparations

are

fully

active in GDP

binding.

The slope of the lines in

Fig.

3A and B is the spontaneous exchange-and-dissociationrateofGDPonEF-Tu (Kd).The

Kd

values of each of these EF-Tu

preparations

areidentical andareveryclosetothe

published wild-type

value of 0.011

S51 (52). Furthermore,

the

slopes

for each EF-Tu prepara- tionare

monophasic, showing

that there isno

heterogeneity

with respecttothe

exchange

rateofGDP. We conclude that EF-TuVal-280 is

fully

competentin GDP

binding

and hasa normal spontaneous dissociation rate.

Stimulation of nucleotideexchangeonEF-Tu

by

EF-Ts.The stimulation of the spontaneous nucleotide

exchange

rate on

(5)

244 TUBULEKAS AND HUGHES AMINO ACID * 20 biganil DataihW umlin

Gly

WOW.~~

1 29 Jun 1991 3:17am I

Va

l

li.8 1s.o 20.0 25.0

Retention Time: Minutes

FIG. 2. Aminoacidsequenceanalysis ofacyanogenbromide-generatedfragment showing thatatthe expectedposition (20min)both the wild-type Gly and themutantValarepresent.Thetwolarge peaksareinternalstandards.

EF-Tu by EF-Ts was measured in a nucleotide exchange

assayasdescribedbyTapioetal. (59). Theconcentration of EF-Ts was 1.2 pmol per 100-1.I assay mixture, and the amount of EF-Tu was varied from 0 to 1,000 pmol. No significant differences between the stimulation of nucleotide exchange on wild-type EF-Tu and that on the wild-type/

mutant

(tufA'

tufB414) species of EF-Tu were observed (Fig. 4). We conclude that EF-Tscanstimulatethe exchange ofguanine nucleotideonbothmutant EF-Tuandwild-type EF-Tu.

EF-TuB Val-280 bindskirromycin. We used the method of Anborgh et al. (3) to test whether EF-TuB414 can bind kirromycin. Briefly, the natural mixture of EF-TuA8 (kirromycin resistant) and EF-TuB414was incubated with kirro-

A wild type EF-Tu

mycin and chromatographedasdescribed previously (3), and thefractions from the columnweretested for thepresenceof EF-Tu andkirromycin. Under the experimental conditions used, EF-Tu- GDP in complex with kirromycin is retarded

on the column and elutes as a separate peak from the kirromycin-resistant EF-Tu. GDP (3). From our EF-Tu mixture, we obtained two separate peaks which clearly indicated that EF-TuB414 bindskirromycin and that itelutes at the same position as wild-type kirromycin-sensitive EF-Tu (Fig. 5). In control experiments, wild-type EF-Tu elutesasasingle peakateitheroftwopositions, depending

onthepresenceorabsence ofkirromycin. Incontrast,pure

EF-TuA8 elutesas asingle peakatoneposition irrespective of whetherkirromycinis present. From thepeakareas,we

B EF-TuA/EF-TuB414 C a. b.

0

C,

0.

a

0 a:

C

0 20 40 60 80 100 120 140

Time (sec) _Time (sec)

FIG. 3. Determination of EF-Tu concentration by nucleotide exchange assayfor EF-Tu isolated frommutant (tufA'

tufBf)

(A) and wild-type (tufA+ tufB414) (B)strains.(C)Twohundredfifty picomolesof each oftheseEF-Tupreparations,asdeterminedbythenucleotide exchangeassay(18,19),waselectrophoresed. Lanes:a,wild-typestrain(tufA+ tufB+); by,mutantstrain(tufA+tufB414). Quantificationof the EF-Tu bandsbylaserdensitometryrevealednodifference in amount.

0.0100 AU

J. BAcrERIOL.

(6)

U

0

I

I-I

u4

1,2 1,0 0,8 0,6 0,4

0,2 0,0

0 2 4 6 8 10 12

[Tu] (10-6 M)

FIG. 4. Stimulationofthe nucleotideexchangerateonEF-Tuby EF-Ts. The dissociationrate constantsfor the release of GDP from EF-Tu intheabsence (K )and inthepresence (K)of EF-Tswere

measuredand calculatedfas describedby Ruusala et al. (52) and Tapioetal.(59).The[Ts]/(K^-Kp)^{ratio is}^plotted^{as a}function of

[Tu].

a

4) ._

94

0

04 c3;6e

Kirromycin(AM)

b

4.)

.3A4

Kirromycin(jiM) were able to measure the relative amounts of each of the

EF-Tuspecies. Measurements ofseveral differentprepara- tions of EF-TuA8 B414 and EF-TuA8 B+ indicated that about30%of eachpreparationisEF-TuB,in agreement with theproteinsequencingdata.

Kirromycin does notimmobilize EF-TuB Val-280 on ribosomes. EF-Tu purifiedfrom JT642

(tufA+ tufB414)

and the kirromycin-resistant strainJT608 (tufA8

tufB414)

was used

0,5-

0,4-

00,3- a)d

0,2-

0,10

20 30 40 50 60 70 80

Fraction numbers

FIG. 5. Chromatographyon DEAE-Q Sepharose FastFlow of EF-TuA8B414 inthepresence (El)andabsence

(*)

ofkirromycin by the method ofAnborgh et al. (3). The A595 ofthe protein in columnfractionswasmeasuredspectrophotometricallyby theBrad-

c 1,5

1,0

0,5

6

10 20 3b

Ribosomes(pmoles)

FIG. 6. Poly(Phe) synthesis as a function of the amount of kirromycin.(a) Symbols:Ml,EF-TuA' B+;*, EF-TuA'B414. (b) Symbols: El, EF-TuA8 B+;*,EF-TuA8 B414.Inpanelsaandb,the resultsarenormalizedto a0jiM kirromycin concentration for the EF-TuB+ preparationset at100%. (c) Translation at fixed EF-Tu andkirromycin concentrations withatitrationof activeribosomes (Rib.)from 10-fold undertoapproximately3-foldovertheconcen- tration ofkirromycin-sensitiveEF-Tu. Symbols: El,EF-TuA8B+;

*, EF-TuA8B414.

to support translation in vitro in the presence ofdifferent amountsofkirromycin. Translation wascarried outwith an excess of active ribosomes over EF-Tu so that EF-Tu blockedonribosomes bykirromycin(70) wouldnotprevent other EF-Tu molecules fromparticipatingintranslation(23).

As controls, pure wild-type and pure kirromycin-resistant EF-Tu preparations were also used in these experiments.

The results (Fig. 6a and b) show that the EF-TuA' B414 preparation responds to kirromycin as a pure sensitive population, whereas the EF-TuA8 B414 preparation re- spondsas apure resistantpopulation.Thus EF-TuB414 has nophenotypeinthisexperiment excepttoreduce the overall amount ofprotein synthesis. Given that we know, on the basis of protein sequencing and column chromatography, that the EF-TuB414 species ispresent, weconclude that it

9 9~~~m

9 U El~m

U U

U

ford (12) assay. The A325 of kirromycin was measured. In the absence of kirromycin, EF-Tu elutes as a single peak. In the presenceofkirromycin,EF-Tu separates intotwopeaksand kirromycin coelutes with thesecond(smaller) peak, whichcanthusbe identifiedasEF-TuB414.

(7)

246 TUBULEKAS AND HUGHES 9

8 7

C-)4I)

u,

6 5 4 3

0,0 0.2 0,4 0,6 0,8 1,0 1,2 1,4 1,6

V/[Tu] (10-7 M-1 sec-1)

FIG. 7. In vitrotranslationas afunctionoftheconcentrationsof wild-typeEF-TuA+ B+ (El)andthe naturalmixture ofwild-typeand mutant EF-TuA+ B414 (*). Experimental detailsaredescribed in Materials andMethods andinapreviousreportbyTubulekasetal.

(64).

does not detectably participate in translation in vitro, although it does bindkirromycin.

Wetested whetherEF-TuB414 is boundonribosomesby kirromycin, usingtranslation assays asdescribedabovebut with a ribosome titration at fixed EF-Tu and kirromycin concentrations (see Materials and Methods). Under these conditions, wild-type kirromycin-sensitive EF-Tu is bound

onribosomesand doesnotsignificantlycontributetoprotein synthesis. Asthe ribosomeconcentration is lowered below theconcentration of EF-Tu, this sensitive EF-Tupopulation competes with the active kirromycin-resistant EF-Tu for ribosomesandprogressivelypreventsit fromparticipatingin proteinsynthesis.Wecompared the effectsontranslation of mixtures containing the kirromycin-resistant EF-TuA8 and either wild-type EF-TuB ^or the mutant EF-TuB414. The results (Fig. 6c) clearly show that EF-TuB414 does not inhibit translation under conditions of ribosome limitation, indicating that it is not bound on the ribosome by the antibiotic. The small differenceinthetranslation rates sup-

portedby eachpreparationisduetoasmall differenceinthe amountofEF-Tu used.

Kinetics of theinteraction between theternarycomplex and theribosome. We have determinedthe kcatlKm ratioof the interaction between wild-type ribosomes and EF-Tuspecies isolated from

tufA' tufBl3

or tufA+ tufB414 strains. The translation rateinthistypeofexperimentisstrictly limited by the concentration of EF-Tu and the efficiency of its interaction in theternarycomplex with ribosomes (all other components, including aa-tRNA, are present in avast ex-

cess). The resultspresented inFig. 7 show that the mixed EF-Tupopulation

(A'

B414) hasakcat/Km ratio about 30%

lower (average of five experiments) than that of the pure

wild-type EF-Tu. The degree of

kcIt/Km

reduction, taken in thecontextof therelativeamountsofmutantandwild-type EF-Tu in the preparation, indicates complete (or nearly complete) inactivity intranslation associated with EF-TuB Val-280. Areduced kcatlKm ratio for the ternary complex-

30 2,0

1,0 0,0

0 1 2 3 4

V/[t-RNA] (10-6 M-1 sec-1)

FIG. 8. Invitrotranslationas afunction of theconcentrations of

Phe-tRNAPhe

with

wird-type

EF-TuA' B+

(E)

and the natural mixtureof wild-type and mutantEF-TuA'B414

(*).

Experimental detailsare described in Materials and Methods and in a previous report by Tapio et al.(59).

ribosomeinteraction is compatible with the reducedgrowth andelongationrates seenin vivo. Ourinterpretationofthis experiment isthat the mutant ternarycomplex isdefectivein its interaction with the ribosome. However, an alternative possibility is that the formation of the ternary complex is defectivebecause of apoorinteraction between EF-Tu and aa-tRNA(this would give the same result asthatshown in Fig. 7). To test for this

possibility,

we

performed

the experimentsonthe EF-Tuinteractionwithaa-tRNA detailed in the next twosections.

Interaction between EF-Tu and

aa-tRNAPh'

during in vitro translation.Weaskedwhether the mutant EF-Tuwas inactive intranslation becauseof a defect in itsinteractionwith aa-tRNA.To testthis,we

performed

atranslationassaywith 10pmol ofactiveribosomes, rate limited by theamount of EF-Tu (15pmol),and withvaryingamountsof

aa-tRNAPhC

inthe range of 100 to 2,000 pmol. Ifthe mutantEF-Tu is defectiveinforming theternarycomplex,itsactivity should be enhanced at higher aa-tRNA concentrations. Thus, our

expectation

was that the lines would converge at

high

aa-tRNA concentrations if the defectwas

compensated

by excess aa-tRNA. On the basisofthe slopes of the lines in Fig. 8, we conclude that there is no significant difference between the different EF-Tu

species

in their interactions with aa-tRNA. Both the

kctIKm

ratio and the kcat for translation arelower with themutantEF-Tumixture, asin thepreviousexperiment. Ina control

experiment

(datanot shown) using only50% of the normalamount ofwild-type EF-Tu, weplotted alineparallelto and

slightly

below that seenwiththenatural

mutant/wild-type

mixture.This

again

is consistentwith the mutant EF-Tumakinguponly 30%ofthe EF-Tu mixture. We conclude that excessaa-tRNAcannot compensatefor the mutantEF-Tu defect intranslation and thus thatthisdefect isunlikelytobe duetolowefficiencyin formingacomplexbetweenaa-tRNA andEF-Tu.However, this conclusion would notbe valid if the defect in ternary J. BACTEMRIOL.

(8)

complexformation were so severe as to almostcompletely prevent the interaction. In other words, if it had been necessary to titrate aa-tRNA to muchhigher concentrations than we hadtitratedittobefore the defect was overcome,we mighthaveobtainedthe result shown in Fig. 8. Totestthis possibility, ^weperformed aphysical assay of the abilityof the mutant EF-Tutoform the ternarycomplex with GTP and aa-tRNAasdetailed in the next section.

Interaction between EF-Tu and aa-tRNAPhe by native PAGE. Weperformed aphysical assay of the ability of the mutant EF-Tu to form a ternary complex with aa-tRNA (using amethod developed by N. Bilgin, T. Gluick, and M.

Ehrenberg [7a] [see Materials and Methods]). Complexes were formed in solution with a constant amount ofEF-Tu andatitrationof aa-tRNAappliedtogels andelectrophore- sed under nondenaturing conditions, and we analyzed the resultsbydensitometry.Thegelsshown in Fig. 9A and B are typical, and thedensitometric data from these gels plotted in Fig.9Crevealnosignificant difference in the abilities of the wild-type EF-Tu and the mutant/wild-type EF-Tu mixture to form ternarycomplexes. Both curves plateau at the same level,correspondingto200pmolof EF-Tu, indicating thatall of the inputEF-Tuis inacomplexed form. Controlexperi- mentsshow thatcomplexformation is absolutelydependent on aa-tRNA and on GTP. Thus, the translation defect associated with EF-TuB Val-280 in vitro is not due to an inability to form a complex with aa-tRNA. Furthermore, complex formation in this assay proceeds with the same kinetics forwild-type and wild-type/mutantEF-Tu species.

WeconcludefromthisexperimentthatourEF-Tuprepara- tions are fully active in binding aa-tRNA. These results, showingno obvious defect in the EF-Tu-aa-tRNA interaction, allow us to make aclear interpretation ofthekcat/Km reduction in Fig. 7 as being due to a reduced efficiencyof interaction between the mutant ternary complex and the ribosome.

EF-TuB Val-280 does not influencetranslational errors in vitro. We havemeasured missense error in vitro (data not shown) forthe naturalmixtures ofwild-type

EF-TuA'

and EF-TuB Val-280 and oferror-prone EF-TuA8 and EF-TuB Val-280inparallel withpurewild-type and pureerror-prone EF-Tu.Translationalaccuracyinthisassayis verysensitive to

perturbations,

yetwe measured in each natural mixture only the errorleveltypical of the wild-type orerror-prone EF-TuAspecies alone(23). Thus, EF-TuB Val-280 does not influence the errorlevel of the other EF-Tu.

GTP

hydrolysis

duringternary complex-ribosome interaction. The datapresented in the previous sections show that the mutant EF-TuB Val-280 is competent and apparently normal with respect to its interactions with the guanine nucleotides, EF-Ts, and aa-tRNA. By elimination, and as shownbythe assays ofthe ternarycomplex-ribosome interaction(reduced

kcat/Km

ratioandfailure to block translation when complexed with kirromycin), this suggests that the defect of EF-TuB Val-280 in translation is related to its inability to interact productively with ribosomes. We performed anotherassayoftheinteractionof the mutant EF-Tu species with the ribosome by measuring the hydrolysis of GTPonEF-Tuassociatedwith this interaction. We used the method developed byEhrenberg et al. (18, 19) to measure theamountof GTPhydrolyzed on EF-Tu per peptide bond formed.This assay uses a translation elongation system that is complete except for the absence of EF-Ts. Thus, upon mixing, preformedternarycomplexes take part in an initial rapidburstof translation and GTP hydrolysis followed by a veryslowtranslation phase rate limited by the spontaneous

A.

EF-Tu

EF-Tu/t-RNA

B.

EF-Tu

EF-Tu/t-RNA

C. 300- z

m

46

I-

iSc-

w us.

0 20 40 60 80 100 150 200 300

0 100 200 300 400

aa-tRNA (pmol)

FIG. 9. Formation of a complex between EF-Tu and Phe-

tRNAPbe

visualized by nondenaturing PAGE. (A) Wild-type EF- TuA+ B+; (B) natural

wild-type/mutant

mixture of EF-TuA' B414.

The amount of EF-Tu in each lane was 200 pmol, andPhe-tRNAPhe wastitrated from 0 to 300 pmols (amounts shown below lanes). (C) Plot ofdensitometric data for formation of ternary complex derived byscanning gelsAand B.

nucleotide exchange rate on EF-Tu in the absence of EF-Ts.

The intrinsic GTPase level of EF-Tu (which is enhanced in some EF-Tu mutants although not detectably in EF-TuB Val-280) is taken into account and subtracted as a background level. In the first few seconds of this assay, there is aninitial burst of proteinsynthesis during which virtually all the GTP bound to the

EF-Tu.

aa-tRNA complex during preincubation is hydrolyzed. Our results with wild-type EF-Tuindicate that almost all of the GTP on EF-Tu (approximately 500 pmol) is hydrolyzed in this burst, with the concomitant synthesis of 300 pmol of poly(Phe). In contrast, an equivalent amount of the natural mixture of mutant and wild-type EF-Tu hydrolyzed approximately 400 pmol of GTP, and 250pmol ofpoly(Phe) wassynthesized. Thus, the levels of both GTP hydrolysis and poly(Phe) synthesis associated with the ternary complex-ribosome interaction

(9)

248 TUBULEKAS AND HUGHES

areapproximately 20% lower for the wild-type/mutant mixture than for the wild type. This difference, although small, is reproducible. The maximum difference we expect if the mutantEF-TuBVal-280is completely inactive in interacting withribosomesis30%onthebasis of the amount of EF-TuB and the magnitude of the R-factorreductionfor the ternary complex-ribosome interaction. We conclude that under the conditions of this assay, the ternary complex with the mutant EF-TuB Val-280fails inmost cases to interact with ribosomes, resultinginGTPhydrolysis. This result, showing reduced GTP hydrolysis, the kcatlKm reduction, and the failure of kirromycin-bound mutant EF-Tuto inhibit translation, is consistent with the translation defect caused by EF-TuB Val-280 being in the interaction of the ternary complex with ribosomes.

DISCUSSION

In this article, we have presented data showing that a single amino acid substitution in domain II ofEF-Tu, Gly-

280-WVal,

abolishes or severely reduces the activity of EF-Tu in translation. The severe effect of this mutation,

tufB414,

invivo is similar tothat ofinsertional inactivation of the samegene. Because of theinviability of strains with

tuJB414

as theonlyactivetufgene,we have done all ofour invitroexperiments with EF-Tupurified fromstrainswhich also carry an active tufA gene. Doing in vitro

experiments

with the natural mixture of EF-TuA and EF-TuB414

(our

measurements show that about 30% of the mixture is EF- TuB414)meansthatwe canidentifythe gross defects caused by thismutant but cannot exclude the

possibility

that there areadditionaldefects of smallermagnitudein other interactions. We have shown that the

disruptive

effect of the mutation is associated with the ternary

complex-ribosome

interaction in vitro. However, mutant EF-TuB Val-280 is active andindistinguishable fromwild-typeEF-Tuin

binding

andexchangingguanine

nucleotides,

instimulationof nucleotide exchange by EF-Ts, and in

forming complexes

with aa-tRNA. Results of three different in vitro assays are consistent with the mutant EF-Tu

having

a defect which severely reduces or abolishes

productive

ternary

complex-

ribosome interactions. The assay of the ternary

complex-

ribosome interaction,underconditions where theamountof EF-Tustrictly limitstherateoftranslation, showslittleor no translation activity associated withthemutant. The assay of translation in the presence of

kirromycin

shows that the mutantEF-Tuisnotboundonthe ribosomeby

kirromycin,

althoughit doesbindtheantibiotic.

Finally,

mostof the GTP onthemutantEF-Turemains

unhydrolyzed during

anassay ofpeptidebond formation.

Other datasuggestthe involvementofdomain II of EF-Tu in the ternary

complex-ribosome

interaction. A mutation isolated in the E. colitufB gene,

tufBO (Gly-222--*Asp) (17,

66) probablyhasamajoreffectonthecorrectinteractionof the ternarycomplex with ribosomes

(58).

Invitrotranslation experiments suggest that this mutant also is inactive in translation (61)exceptat high,

nonphysiological magnesium

concentrations, atwhichadefect inbindingtoribosomes is overcome (58). Together, the data on the

Gly-222

and

Gly-280

mutants support the

suggestion

that domain II of EF-Tuhas a role in the ternary

complex-ribosome

interaction.

Itis worthpointingoutthat thenormal interaction isone of the ternarycomplex

(rather

than

EF-Tu)

with ribosomes and that when EF-Tu dissociates from aa-tRNA it also leavesthe ribosome. There isno a

priori

reason to assume a

direct EF-Tu-ribosome interaction. In

principle,

the interactions

might

all be dictated

by

the bound aa-tRNA molecule. An indication that in the ternary

complex

EF-Tuitself is

directly

involved in

interacting

with ribosomes is that it hasa

region

of

homology (in

domain

I)

with

EF-G,

afactor which interacts

directly

with ribosomes. The GTPase activ-

ity

of the isolated domain I of E.

coli

EF-Tu is activated

by

70S ribosomes

(48),

while

tryptic cleavage

or

enzymatic

modificationstothis

region

of either factor

(EF-Tu/EF-la

or

EF-G/EF-2)

^can interferewith the ribosome interaction in eachcase

(32, 45, 50, 54),

and each

factor,

when stabilized

on ribosomes

by

an

antibiotic, provides

an

overlapping

pattern of

protection

of 23S rRNA

against

chemical and

enzymatic

attack

(44).

In summary, these

data, although

often basedon grossalterations to the

protein,

supportthe

expectation,

based on

homology,

that domain I of EF-Tu contains a

region

involved in

interacting

with ribosomes.

Our data and those of Swart et al.

(58)

suggest

that,

in addition to a domain

I-ribosome interaction,

domain II is

involved,

either

directly

or

indirectly,

in the ternary com-

plex-ribosome

interaction.We shall present here alternative models which could describe in

general

terms this interaction and how it

might

be

disrupted by

the domain II mutations in EF-Tu.

Directinteractions. EF-Tu may have

temporally

and spa-

tially

differentinteractions with ribosomes. For

example,

an

initial interaction

(possibly involving

domain

II) leading

to

codon-anticodon interaction

might

be followed

by

a second interaction

(involving

domain Iand the

a-sarcin region

ofthe 50S

subunit) leading

toGTPase and the

subsequent

dissociation of

EF-Tu.

GDP from the ribosome.

Arguing against multiple

sitesonribosomesatwhich EF-Tu-ribosome interactions occur are rRNA

protection experiments (43, 44)

which donotdetect any EF-Tu-mediated

protection

outside of the a-sarcin

region. However,

these

experiments depend

on

freezing

EF-Tu on ribosomes

by using kirromycin

and hence

binding

theGDPform of

EF-Tu,

and thus

they

donot

provide

informationon the initial interaction of theternary

complex

with ribosomes.

Furthermore,

these

experiments

wouldnot

necessarily

havedetected

protein-protein

interactions. An alternative model for a direct interaction is that EF-Tu in the ternary

complex

may have a conformation which

brings regions

ofdomains I and II close

together

^so

that

they form,

in essence, one

hybrid

EF-Tu site for interactionwith the ribosome.

Suggestive

of this

possibility

is the

cross-linking

of theaa

moiety

of aa-tRNAtodomain I in EF-Tu

(16)

andto domain II in EF-la

(31).

Areasonable

prediction

from each of the direct-interaction models is that it should be

possible

to select

compensating

mutations with alterationsin the ribosome.

Indirect effects. The domain II mutation may exert an

indirect effect

by altering

the

tertiary

structure of EF-Tu such that the interaction of domain I with the ribosome is

severely

affected. Given the

position

of the mutation in a

loop

^atthe

edge

ofdomainII

(like

the

Gly-222 mutation)

this

seems

unlikely,

but itcannotbe ruled outwith the

present

data. A more

likely

alternative for indirect effects is sug-

gested by

the recentresults of

Kinzy

^et al.

(31)

who made extensivecross-links betweenEF-laandseveralaa-tRNAs andused the data to model the interaction between the EF and the aa-tRNA.

According

to their

model,

thedomain II mutations areeachvery close to the 5' and 3' ends of the tRNAmolecule. If this model iscorrect,itseemsreasonable thatlocal

changes

indomain II

might

haveadirect influence eitheronthe conformation of the bound

aa-tRNA, affecting

its

ability

tointeract with the

codon-programmed ribosome,

J. BACTERIOL.