J. h’o2. Biol. (1992) 225, 67-80

Both Forms of Translational Initiation Are Required for Maximal Growth Evidence for Two Translational Christine

Factor IF2 (ar and p) of Escherichia coli

Initiation

Codons for IF20

Sacerdot’j-, Gilles Vachor?, Soumaya Laalami’, Frangoise Yves Cenatiempo’ and Marianne Grunberg-Manago’

Morel-Deville’t

‘Institut de Biologic de Physico-Chimique, URA CNRS 1139 13 rue Pierre et Marie Curie, 75005 Paris, France 21nstitut

de Biologic Mole’culaire et d’Inge’nierie Ge’ne’tique, URA CNRS 40 Avenue du Recteur Pineau, 86022 Poitiers Cedex, France (Received 19 July

1991; accepted 8 January

1172

1991)

The gene infB codes for two forms of translational initiation factor IFZ; IF2a (97,300 Da) and IF2g (79,700 Da). IF2/? arises from an independent translational event on a GCG codon located 471 bases downstream from IF2a start codon. By site-directed mutagenesis we constructed six different mutations of this GUG codon. In all cases, IF2p synthesis was variably affected by the mutations but not abolished. We show that the residual expression of IF2g results from translational initiation on an AUG codon located 21 bases downstream from the mutated GUG. Furthermore, two forms of IF2P have been separated by fast protein liquid chromatography and the determination of their N-terminal sequences indicated that they resulted from two internal initiation events, one occurring on the previously identified GUG start codon, the other on the AUG codon immediately downstream. We conclude that two forms of IF2/? exist in the cell, which differ by seven aminoacid residues at their N terminus. Only by mutating both IF2/? start codons could we construct plasmids that express only IF2a. A plasmid expressing only IF2fi was obtained by deletion of the proximal region of the infB gene. Using a strain that carries a null mutation copy of the same gene on a in the chromosomal copy of infB and a functional thermosensitive lysogenic L phage, we could cure the 1 phage when the plasmids expressing only one form of IF2 were supplied in trans. We found that each one of the two forms of IF2, at near physiological levels, can support growth of Escherichia co& but that growth is retarded at 37°C. This result shows that both forms of IF2 are required for maximal growth of the cell and suggests that they have acquired some specialized but not essential function. Keywords:

translational

initiation

factor

1. Introduction

IF2P;

Escherichia ~01i

The shorter form of the protein, IF2P, is the result of translational initiation at an in-frame GUG codon, valine 158 of IF2a (Morel-Deville et al., 1990; Plumbridge et al., 1985a). The infB gene is located on the E. coli chromosome at 69 minutes and is part of a complex polycistronic operon comprising the genes met Y, 15A, nusA, infB, 15B and 35. The gene met Y encodes a minor form of initiat.or tRNA, nusA encodes a protein involved in transcription-termination and the genes, provisionally identified as 15A, 15B and 35, encode proteins of unknown function with relative molecular mass of 15,000, 15,000 and 35,000 (Ishii et al., 1984; Kurihara & h’akamura, 1983; Plumbridge et al., 1982; Sands et al., 1988).

Initiation of translation in Escherichia coli is promoted by three initiation factors called IFl, IF2 and IF3. IF2 is the largest of the initiation factors and is present in bacterial cells in two forms, IF2a and IF2fi. The two proteins of Mr 97,300 and M, 79,700, respectively, are encoded by the same gene infB (Plumbridge et al., 1982; Sacerdot et al., 1984). TAuthor t,o whom all correspondence should be addressed. fPresent address: Laboratoire de GBn&ique Microbienne, Institut de Biotechnologie, INRA, Domaine de Vilvert, ‘78352 Jouy-en-Josas, France. 0022%2836/9S/090067~14 $03.00/O

2; IF2a;

67

0 1990 Academic Press Limited

C. Sacerdot et ai.

68

**

ACTGTAGCAGGAAGGJ$XAGC ATG ACA GAT GTA ACG ATT IF2a SD Q

ACG CTG

GCC GCA GAG CGA CAG ACC TCC GTG GAA CGC CTG GTA GAG CAA TTT GCT GAT GCA GGT ATC CGG AAG TCT GCT GAC GAC TCT GTG TCT GCA CAA GAG AAA CAG ACT TTG ATT GAC

CAC

CTG

AAT CAG AAdi AAT TCA

GGC CCG GAC ?@A TTG ACG CTG CAA CGT AAA ACA CGC AGC AC@ CTT TCG GTA C

ATC GAA

GTC CGC AAG A?& CGC ACC TTT GTG AAA CGC GAT CCG C

GAG GCT

AAC ATT CCT GGT ACC GGT GGA AAA AGC

GAA CGC CTT GCA GCG GAA GAG CAA GCG CAG CGT GAA GCG GAA GAG CAA GCC CGT CGT GAG GCA GAA GAA TCG GCT AAA CGC GAG GCG CAA CAA A7iA GCT GAA CGT GAG GCC GCA GAA CAA GCT AAG CGT GAA GCT ** * GCT GAA CAA GCG AAA CGT GAA GCT GG SD

Asn Gln AAT cm

Gln cm

Asp Asp Met Thr GAC GAT @ACT

SD

Lys Asn Ala Gin Ala AAA AA~ GCC CAG GCT G

CGC CGT GAG CAG GAA GCT GCA GAG CTC

Ser AGC

WAC -

SF2fh

GCC

543

G CGT

Figure 1. Sequence of the proximal region of ;nfR. The initiator codons of the different forms of JF2 are boxed and the Shine and Dalgarno (SD) sequences are underlined. Val and Met in bold type indicate the Val( 158) and Met( 165) of IFScr. Nueleotides indicated by asterisks were mutated in order to create 2 XhoI sites (CTCGAG): GT (4 to S)-CG and GAAC (457 to 460)+CGAG. The DKA between the XhoI sites was deleted in mut,ant 845. Mutation S changes G(495)-+6, mutation C2: T(515)+C and mutation C3 : G(516)-+C. The double mutants SC2 and SC3 contain the combina,tion of mutations S plus C2 and S plus C3, respectively.

IF2 initiates the formation of the translat’ion-initiation complex by binding with f'Met-tRNAye' and GTP, and interacting with the 30 S ribosomal subunit-mRKA complex (for reviews, see Hershey, 1987; Gualerzi & Pon, 1990). It is not known why IF2 exists in two forms. Tn a coupled in vitro transcription-translation system, IF2p was less active than IF2a (I&kin et al., 1978); while translation of phage R17 RKA was equally efficient with the two forms (Hershey, 1987). In addition to IF!& and IF2P, a third form of IF2, IF2y (65,000 M,) has

been purified. Although apparently an artefact due to proteolysis during preparation of IF2, this Cterminal portion of IF2 exhibits all the same in vitro activities as TF2a and IF2j (Cenatiempo et al., 1987). More recently, Laalami et al. (1991) showed that a genetically engineered fragment of in@ gene expressing the C-terminal 55,000.Mr part of IF2 was able to support growth of E. coli when supplied in excess. However, the presence of the entire in,jB gene clearly enhanced growth of E. eoli.

Two Forms

of IF2

Required for Qrowth

69

Table 1 E. coli strains

and plasmids

A. Bacterial strains Strain SE5000

Genotype

Reference

1337

F-, araD139, A(argF-lac)U169, rpsL150, relA1, $bB5301, de&l, ptsF25, rbsR, reeA56 F-, weAl, uvsA6, phr-1, leuB6, proA&, rgE3, th,i-1, ara-14, lacY1, galK2, ~$5, mtl-I, rpsL31, tsx-33, supE44 F-, thr, leu, minA, m&B; lac Y, gal, rpsL

IBPC5321

Fm, thi-1,

CSR603

IBPC5321R SL598R SLlS-1 SL18A45 SL18SC2 SL18SC3 SL18C2 SL18C3

argG6, rpsL, AlacX74 IBPC5321 recA1, I&PC5321 AinfB IBPC5321 AinfB IBPC5321 AinfB IBPC5321 AinfB IBPC5321 AinfB IBPC5321 AinfB IBPC5321 AinfB

argE3, his-4, ~~1-5, tsx-29, ml :: TnlO :: cat2, recA1 (?.GJ9-2) :: Acat2, recA1, pB181 :: cat&, recA1, pBlSA45 :: cat2, recA1, pB18SC2 :: eat2, recA1, pB18SC3 :: cat2, recA1, pB18C2 :: cat2, recA1, pB18C3

Silhavy Sancar

et al. (1984) et al. (1979)

Derived from P67854, Adler et al. (1967) Plumbridge et al. (19856) Laalami et al. (1991) Laalami et al. (1991) This work This work This work This work This work This work

U. Plasmids Vector

pB18-1 pFY181 pB18A45 pB18SC2 pB18SC3 pB18C2 pB18C3

pCDN4 pNM481 pNMPcat481 pBR322 pRLSlO0 pBR322 pRLSlO0 pBR322 pBR322 pBR322 pBR322 pBR322

Relevant bla, bla, bla, bla, cat, bla, cat, bla, bla, bla, bla, bla,

genotype

tet ‘1acZ

Bleicher et al. (1988) Minton (1984) This work

infB infB AinfB45 infBSC2 infBSC3 infBC2 iqfBC3

Bleicher et al. (1988) Plumbridge and Springer (1983) Morel-Deville et al. (1990) This work This work This work This work This work

‘1acZ

Peat, tet tet nusA, nusA, nusA, nusA, nusA, nusA, nusA,

Reference

Analysis of the N-terminal part of the IF2 primary sequence revealed two adjacent regions that are rich in alanine and charged amino acids and that show striking periodicities in their sequence. Only the second of them exists in IFZP (Sacerdot et al., 1984). These structures could be related to differentiated roles of IF2a and IF2/?. As an approach to elucidate the function of the Nterminal part of IF2, and more specifically the reason why two forms of this factor co-exist in the cell, we constructed mutations that allow the expression of just one form of IF2 to the exclusion of the other. Whereas previous experiments (MorelDeville et al., 1990) described relatively simple methods for obtaining E. coli mutants in which only IF2P was expressed (by deletions in the proximal region of in@), a similar selection for the expression of IF2a alone proved to be more complicated, and in fact revealed a second translation initiation codon for IF!@. We show here that each one of the two forms of IFZ, at near physiological levels, can support growth of E. coli, but is not sufficient for maximal growth at 37 “C.

2. Materials and Methods (a) Strains and media Strains and plasmids used in this work are summarized in Table 1. All strains were routinely grown in Luria Bertani (LB) medium. If required ampicillin was added at 100 pg/ml and chloramphenicol at 10 pg/ml. Growth of liquid cultures was monitored by measuring the absorbance at 650 nm.

(b) Site specilc mutations

mutagenesis and cloning of the in infB expression vectors

Numbering of the sequence starts from nucleotide A located 21 bases upstream from the AUG start codon of IFZa (see Fig. 1). Both pB18-1 and pFY18-1 carry the same nusA&nfB insert, as a BamHI-Hind111 insert, cloned in 2 different vectors: pBR322 and pRLS100, respectively (Table 1). pRLSlO0 is a derivative of pBR322 where the bla gene has been replaced by cat. For some experiments, it was necessary to use one or other series of plasmids: e.g. in order to test the ability of the mutated GUG start codons to promote initiation of translation in vitro using a dipeptide assay (Cenatiempo et al., 1987), it was necessary to use pFY18-1 since it does

70

C. Sacerdot

et ai.

ligat,ed into pFY18I open vector from which the corresponding wild-t,ype fragment had been excised (Fig. 2). Oligonucleotide mutagenesis of the ATG (514 to 516j codon wa,s achieved by the method of Taylor et al. (1985), using thionucleotides. The mutation called C2 changes T (515)X and the mutation called C3 changes G (516)-C. These 2 mutations were constructed on both the BamHI&I wild-type template and on the same template carrying the GTG (493 to 495)-+GTC mutation (mutation 8); the la,tter replacement thus produces the double mutants SC2 and SC3. The same method of Taylor et aZ. (1985) w-as used to create 2 XhoI sites in the proximal region of &f-B. One XhoI site was created 14 nucieotides upstream from the ATG start codon of IFBcc by changing 2 bases: GT (4 and 5)-CG (Fig. 1). The other XhoI site was created 32 nucleotides upstream from GTG (493 to 495) start codon of IF28 by changing 3 bases: GAAC (457 to 460)+CGAG. The DNA between these 2 XhoI sites was deleted on the Ml3 replicative form and the result confirmed by DN’A sequencing (Sanger et al., 1977). This deletion is called A45. The BamHI-SstI DNA fragments carrying the various mutations (C2, C3; SC2, SC3 and A45) were excised from the Ml3 replicative form and replaced into pB18-1 open vector from which the corresponding wild-type fragment had been excised. (c) Miniceil

Figure 2. Cloning

of the mutations of the infB proximal region into iy@ expressing vectors. The plasmids pB18I and pFY 18-l carry the same BamHT-Hind111 DNA insert, containing the nusA and i%fB genes (Table 1). The BamHI-SstI fragment of pB181 was cloned into M13mp19 and submitted to oligonucleotide mutagenesis (see Materials and Methods). In order to place the different mut’ations into the CnfB expressing plasmids, these plasmids were cut wit’h BamHI and S&I and the resulting open vectors were separated from the wild-type RamHI-SstT fragments on agarose gel. before ligation with the purified BarnHI-SstI fragments carrying the different mutations. The asterisks indicate the sites that have undergone mutagenesis. t locates the XhoI site created by oligonucleotide mutagenesis. Relevant restriction sites are indicated as follows: B, RamHI; H; HindIII; S. SstI; K, KpnI.

not carry the Dla gene, which starts with the same dipeptide fMet-Ser as IF@. However. the fact that the 2nd start codon of IF2g produces the same fMet-Thr dipeptide as IFBa, rendered these results difficult to interpret (data not shown). The mutations that allow the expression of just 1 form of IF2 have been cloned into pB181. pFY181 and derivatives were not suitable for transforming SL598R as both these plasmids and this strain are chloramphenicol resistant. A BamHI-S&I DNA fragment of pB18-1 (Plumbridge & Springer, 1983) carrying nusA and the N-terminal region of in@ that extends 69 nucleotides beyond the GTG start codon of IF2a; was cloned into Ml3mpl9 (Yanisch-Perron et al.; 1985). Oligonucleotide mutagenesis by the double-primer technique (Zoller & Smith, 1987) was used to change the GTG (493 to 495) codon into GTN and GNG (i.e. all possible changes of the 3rd and of the The correct mutants were 2nd base, respectively). confirmed by DNA sequencing (Sanger et al.; 1977). The BumHI-SstI fragment containing the mutation was excised from recombinant Ml3 replicative form DNA and

analysis

Minicell analysis of plasmid-encoded proteins was performed as previously described (Morel-Deville et al., 1990). In order to est,imate the relative amount of residual IF2/3 synthesis after mutagenesis of the GUG start codon, the density profiles of several autoradiograms were scanned using a SEBIA system 2-134 scanner. (d) $fazicell Maxicell analysis formed as described

analysis

of plasmid-encoded proteins (Sancar et al., 1979).

(ej PuriJication of hybrid proteins fi-galactosidase activity

was per-

with

Strain SE5000 was transformed with the plasmids carrying the gene fusions iv+la& downstream from promoter Peat. Hybrid IFZP-/?-galactosidase proteins were purified in a l-step method (Ullmann, 1984). Proteins were immediately precipitated with 70% ammonium sulphate, pelleted and kept at -70°C. They were later solubilized in the loading buffer (01 3$-TrisHCl (pH 68): 2% (w/v) SDS, 5% (v/v) glycerol, 0.1 x-2-mercaptoethanol) before SDS/PAGE and subsequent NH,-terminal sequencing. (f) Purilfication

of IF&

and

the 2 species of IlT@

The strain used for purification of IF2 carries a plasmid where the infB gene is placed under the cont’rol of the inducible /zpL promoter and the ~1857 thermosensitive repressor (Dondon et al., 1985). IF&X, IF20, and IF2& have been separated using a fast purification procedure for IF2, according to a method described elsewhere (Mortensen et al., 1991). (g) NH,-terminal

amino

acid sequence determination

The NH, terminal amino acid sequence of the hybrid proteins and of the 2 purified forms of IF2fi was deter-

Two Forms of IF8 Required for Growth mined by step-wise Edman degradation performed at the Service Central d’ilnalyse (CNRS, Vernaison, France), on proteins that had been transferred onto Immobilon (Millipore) membranes after SDS/PAGE, according to Matsudaira (1987). This method necessitates that the prot,ein be at least 60% to 70% pure. (h) Curing

of LinfB strain

transducing SL598R

phage

from

Strain SL598R (Laalami et aZ., 1991) was transformed with the plasmids carrying the mutated IF2a and IF2b initiation codons. Cells were grown overnight in LB Amp (100 pg/ml) with Cm (10 pg/ml), diluted into fresh medium and grown at 30°C to an Asso value of 0.20. Portions (250 ~1) were transferred to 5 ml of LB prewarmed to 42°C and shaken vigorously for 5 min and immediately chilled in ice. Portions (1.5 ml) were concentrated by centrifugation (6000 g, 4°C) and plated on LB plates containing Cm and Amp at 42°C Survivors were purified several times on the same plates at 42°C and the loss of I verified by their sensitivity to ICI- phage. (i) Blotting

and immunodetection IF2u and IF2j

of proteins

The presence of the different forms of IF2, in total extracts of bacteria, was examined by the quantitative immunoblotting procedure of Howe & Hershey (1981), with a few modifications. E. coli cells were grown to late log phase (A,,, = 0.7 to 0.9). Crude cellular extracts were prepared by resuspending 1 A650 unit of cells in 200 ~1 of loading buffer (0.1 iv-TrisHCl (pH 6%), 2% (w/v) SDS, 5% (v/v) glycerol, 0.1 M-2-mercaptoe’chanol) and then incubated for 2 min at 100°C. Protein concentrations were determined as described (Schaffner & Weissmann, 1973). Routinely 10 pg of total protein was analyzed on a 0.1 y0 SDS, 10% (w/v) polyacrylamide gel. Proteins were transferred electrophoretically from the polyacrylamide gel to nitrocellulose membranes in 25 mM-TrisHCl (pH 8.3), 0.2 M-glycine, O.Olo/0 SDS, 20% (v/v) methanol. Then the blot was treated according to Howe & Hershey (1981). Rabbit antiserum against E. coli IF2 was diluted 1 : 4000 in buffer A (Howe & Hershey, 1981). The relative amounts of IF2c( and IF2fi in the cell extracts were evaluated by excising the corresponding bands from the blot and measuring their radioactivity by scintillation counting, compared with that of IFS@ in a similar amount of a wild-type strain.

3. Results (a) Mutagenesis of the GUG initiation codon of IF2p Previous work established that IF2fi results from translation initiation on a GUG start codon, 471 nucleotides downstream from IF2a’s AUG initiation codon (Morel-Deville et al., 1990; Plumbridge et al., 1985a). IF2a’s start codon AUG is preceded by a good SD sequence (AAGGA) while IF‘e/?‘s GUG is preceded by a purine-rich sequence devoid of a strong SD sequence (WAAMA) (Fig. 1). In order to study and try to suppress IF2p expression, we mutated its GUG start codon using site-directed mutagenesis (see Materials and Methods). We constructed two sets of mutants. In one set, the third base of GUG was changed into the three other

71

Table 2 Relative in vivo synthesis of IFZB after mutagefzesis of its GUG start codon Plasmids

%t

(i) GUN mutants GUG+GUA GUG+GUC GUG+GUU

50

(ii) GNG mutants GUG+GAG GUG-tGCG GUG-+GGG

35 66

40 31

71

tAmouts of IF2p synthesized from the mutated plasmids are expressed as a o/O of that synthesized from pPY18-1 (based on scanning of several autoradiograms).

possibilities : GUA, GUC and GUU, all of which conserve the valine 158 of IF2a. The second set of mutants alters the middle base of GUG (GNG mutants), thus permitting three different amino acid substitutions for valine 158 of IF2a: GAG (Ala), GCG (Glu), GGG (Gly). These six mutants were recloned into the pFY181 plasmid that carries a nusA-infB insert in the pLRSlO0 vector (see Materials and Methods, Fig. 2). The resulting plasmids were used to transform a minicell producing strain and protein synthesis directed by plasmid DNA was analysed by SDS/ PAGE followed by autoradiography. The autoradiograms presented in Figure 3 show that IF2P synthesis is variably affected by the mutations but is not abolished by any of them. Scanning of several autoradiograms allowed us to estimate the relative in vivo synthesis of IF2P with the various mutated forms of the GUG codon compared with the wildtype gene (Table 2): the values obtained show that the six mutations allow 31 y. to 71% residual expression. This result, is in contradiction to previous observations where mutations of initiation codons decreased dramatically or abolished translation of the gene (Gold et al., 1981). Either the mutated GUG is not the initiation codon of IFB/? and another initiation codon functions when the GUG is modified, or there exists a second initiation site for IFB/?. (b) Two forms

of IFZP exist in the cell

In order to identify the codon on which translation of IF2b still initiates when the GUG is infB-1acZ fusions mutated, we constructed expressing IF2P-p-galactosidase hybrid proteins corresponding to residual expression of IF2P when the GUG codon was mutated. Figure 4 shows the construction of such fusions. The KpnI-PstI fragment of infB containing the IF2P translation initiation region was isolated from the Ml3 vectors carrying the GUG mutations. These fragments were inserted downstream from the Peat promoter and in phase with the 1acZ gene of the pNMPcat481 vector (Table 1). The IF2B-/?-galactosidase hybrid proteins expressed from the resulting plasmids were purified by affinity chromatography (see Materials and

C. Sacerdot et al

-6F2 a - IF2 B

IF2 a IF2 B

- pNusA

pNusA

CAT -

CAT

(a)

i b)

Figure 3. Minicell analysis of IF28 ex p ression from pFYlS-1 and derivatives carrying the different mutations on the GUG (493 to 495) start codon. Proteins labelled with [35S]methionine were separated on 12,50/b SDS/PAGE and detected by autoradiography. (a) The GUN mutants. Lane 1, GUG-+GUA; fane 2, GUG-+GUC; lane 3, GUG-tGUU; lane 4. control nFYlS-1. ib) The GNG mutants. Lane 1. control nFYlS-1; lane 2, GI;G+GGG; lane 3; GUG+GCG; lane 4: GUG-r GAG. CAT‘is’ the chloramphenicol acetyltransferase: Methods) and their N-terminal sequences were determined. For each of the six mutants the identical sequence was found: Thr-Lys-Asn-Ala-GlnAla-Glu. This amino acid sequence is found in the X-terminal part of IFag, starting eight residues after the GUG initiation codon. Thus after mutagenesis of GUG (493 to 495) translation initiates on the AUG (514 to 516) codon just downstream and permits the synthesis of an IFB/l protein missing seven amino acids compared with the previously identified form of IF2fi. Using low concentration acrylamide SDS/PAGE, we carefully examined the IF2p protein derived from wild-type plasmids (data not shown) and found in all cases that IF2p migrated as polypeptide doublets. These findings strongly suggest that the AUG start codon (514 to 516) also functions even when GUG (493 to 495) is not mutated. Recently two species of IF2P proteins, IFZB, and IFBB,, have been separated by f.p.1.c.t TAbbreviations used: f.p.l.c., fast protein liquid chromatography; RBS, ribosome binding site; Amp, ampicillin; Cm, chloramphenieol; SD, Shine-Dalgarno.

(see NIaterials and Methods). Their N-terminal sequences were det,ermined: IFSfi,, Ser-Asn-Gln-Gln-Asp-Asp-,Meta- Lys-AsnAla, TF2P,, Thr-Lys-Asn-Ala-Gin-Ala. We conclude that two internal initiation events occur on in@ mRWA; one at GUG (493 to 495) to produce IFBP,, the other at AUG (514 to 516) to give IF2P, (Fig. 1). IF2j?, is probably the major species as N-terminal sequencing of a wild-type IF2P-b-galactosidase hybrid protein (which is expected to contain both IF2fi, and IF2p, fusion forms) gives the IF2/3, sequence (Morel-Deville et al., 1990). (c) Construction

of double mutants

thut abolish

IF2P expression In order to investigate the relative roles of IF% and IF28, and in particular to determine whether IFZP is essential to the cell, we were obliged to mutate the initiation codons of both IF2fi, and IFZB, by site-directed mutagenesis under conditions

Two Forms

of IF2

Required for Growth

73

GUG+GUC (493 to 495) and AUG+ACG (514 to 516), and the SC3 mutation refers to the double mutant GUG+GUC (493 to 495) and AUG+AUC (514 to 516). The two double mutants were recloned into the pB181 plasmid, which carries a nusAinfB insert in pBR322 (see Materials and Methods, Fig. 2). The resulting plasmids (pB18SC2 and pBlSSC3) were used to transform the CSR603 strain for maxicell analysis of plasmid DNA encoded proteins. Figure 5(a) shows that, in addition to the NusA protein, both IF2a and IFBB are synthesized from plasmid pB181, whereas just a protein of the size of IF2a is detected in maxicell analysis of the pB18SC2 and pBlSSC3 double mutants. This strongly indicates that both double mutations SC2 and SC3 abolish IFBP, and IF2/3, expression. Finally, we constructed in the same way the control plasmids pB18C2 and pB 18C3, that carry, respectively, the single mutations C2 and C3 for AUG (514 to 516) in the in@ gene of pB18-1, while leaving the GUG of IF2/?, intact. Therefore, pB18C2 and C3 are expected to express both IF2a and IF2/?, (but not IFBP,) proteins, with one amino acid substitution (respectively Met+Thr and Met+Ile) in amino acid 165 of IF2a and in a,mino acid 7 of IFBB,. (d) Constvuction of a deletion of the proximal of infR that abolishes IF2a expression Figure

4. Construction

of infB-la& fusions carrying the mutated GUG (493 to 495) start codon of IF%fi. A PvuIIIPstI fragment of pCDN4 plasmid that carries the promoter of chloramphenicol acetyltransferase (Peat), has been cloned between the XmaI and Pat1 restriction sites of pNM481, upstream from the truncated la& gene (that lacks the 1st 7 codons of P-galactosidase), giving rise to the pNMPcat481 vector. A KpnI-PstI DNA fragment carrying the translation initiation region of IF2P with the different mutations of the GUG (493 to 495) codon, was excised from the mutagenized M13mp19 derivative and cloned between the SmaI and PstI sites of pNM4RlPcat vector. The KpnI restriction site was first treated with T4 DPU’A polymerase in order to create a blunt end, which was necessary for ligation with the SmaI site of pNMPcat481. Relevant restriction sites are indicated as follows : B; BarnHI; P, PstI; Pv, PvuII; S, SstI; Sm, SmaI; K; KpnI.

that permit only IF2a to be expressed. Since each of the six GUG mutants renders this codon inappropriate for translation init,iation, we selected the GUG-+GUC (493 to 495) mutation for two reasons: it most strongly reduces IF2g expression and it conserves the internal Val (158) of IF2cr. This mutat,ion is called S (Fig. 1). On the other hand, changing the AUG start codon of IFBB, inevitably modifies the internal Met (165) of IF2a. We constructed two mutants of this codon. The mutation called C2 changes the second base: AUG-+ACG (Met+Thr). The mutation called C3 changes the third base: AUG+AUC (Met+Ile). So the SC2 mutation refers to the double mutant

region

Previous experiments from our group have described constructions in which IF2a is not synthesized whereas IF28 expression is intact. However, these constructions conserved a large part of IF2a specific DNA. In order to avoid the possibility that any spontaneous mutation in this region could restore partial or complete IF2a synthesis, we constructed an infB gene with a large deletion of the IF2a specific sequence. By site-specific mutagenesis we created two XhoI restriction sites: the first one is located upstream from the IF2a ribosome binding site (RBS) (-14 nucleotides from AUG) and the second one is located 32 nucleotides upstream from the IFB/?, GUG start codon. Therefore, deletion between these two XhoI sites eliminates both the IF2a RBS and most of the IFBa-specific N-terminal coding region of infB (Fig. 1). This deletion (A45) was recloned into pB18-1 to give plasmid pB18A45 (Fig. 2). As expected, maxicell analysis of this plasmid shows that IFB/? (presumably both species IFS/I, and IF2p,) is still synthesized, whereas IF2a is eliminated (Fig. 5(b)).

infB

(e) Complementation of a null mutation of by constructs that express only one of the two fovms of IF2t

In order to elucidate the significance of the existence of the two forms of IF2, we tested our YThe term “the 2 forms of IF2” refers to IF2a and IF28, where IF2p is the shorter form of IF2, which normally consists of a mixture of IF2P, and IF28,.

C. Xacerdot et al

74

IF2 a

IF20

IF2 a pNusA

IF28

0 lactamase

I3 lactamase

Figure 5. (a) Maxicell analysis of IF2v. and IF2p synthesis from plasmids carrying mutations in both start codons of IF@, and IF2&. Proteins labelled with [35S]methionine were separated on 100/b SDS/PAGE and detected by autoradiography. Lane 1, pB18-1; lane 2, pB18SC2; lane 3, pB18SC3; lane 4> pBR322. (b) Xa,xicell analysis of IF20! and IF2b synthesis from the plasmid that carries the A45 deletion in the proximal region of the injB gene. Proteins labelled with [35S] methionine were separated on 10% SDS/PAGE and detected by autoradiography. Lane 1. pBlS-1; lane 2. pBlSA45; lane 3, pBR322.

Table 3 Analysis

of the cured strains

Relative

derived from SL5981rz

amount of IF2 IF2 (a+B)

Growth on plates

Strains

IF%

IF2p

IBPC 5321R (pI3R322) SL18-1 SL18A45 SL18SC2 SLlSSC3 SLlSC2 SLlSC3

2.2

1

3.2

++++

++++

++++

1.9 0 1.9 1.9 1.9 1.9

1.3 2.2 0 0 0.4 0.55

3.2 2.2 1.9 1.9 2.3 2.45

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

++++ + + + fff +++

+++ i +

44”CT

37 “C-f

30’Cf

Amounts of IF2n and IP2p were determined from liquid cultures at 37”C, bg quantitative immunoblotting assay (see Materials and Methods) and expressed relative to the amount of IF2P in a wild-type strain. + + + +, very good growth; + + +, good growth but slightly smaller colonies; + + , quite small colonies; + , microcolonies: - , no isolated colonies. 7. after 16 h (at 37°C and 42°C). $. after 20 h (at 30°C).

Two Forms of IF2 1

23

45

67

(0) 1

23

45

67

- IF2a -IF20

(b) Figure 6. Western blot analysis of the cured strains carrying pBlS-1 and deriyatives. Cultures were grown in LB at (a) 42°C or (b) 37°C and harvested during late exponential growth (A,,, = @7 to 0.9). Cells were lysed by boiling in SDS sample buffer. The proteins were separated using a 10% SDS/PAGE, transferred to a nitrocellulose membrane and treated with anti-IF2 serum and ‘251-labelled protein A as described in Materials and Methods. Lane 1, wild-type E. coZi strain; lane 2; SL18-1; lane 3: SLlSA45; lane 4, SL18SCZ; lane 5, SL18SC3; lane 6, SL18C2; lane 7, SL18C3.

constructions that synthesize only one form (either IFZa or IFBfl), for their ability to replace the chromosomal infB gene and satisfy the cellular requirement for IF2. We used the strain SL598R, which has been constructed to test the viability of any infB allele (Laalami et al., 1991). This strain carries a chromosomal infB deletion, where most of the infB DNA has been replaced by a chloramphenicol resistance cartridge, while a functional wild-type copy of infB is supplied in trans by a thermosensitive lysogenic 1 phage integrated at att A. After heat induction at 42 “C! , it is possible to cure this 1 infB phage without killing the bacteria, provided that a viable infB allele is supplied in trans (see Materials and Methods). Plasmid pB18-1 and derivatives pBlSA45,

Required for Growth

75

pBl8SC2, pB18SC3, as well as the controls pBl8C2 and pBlSC3, were used to transform SL598R. Plasmid pBl8-1 is a pBR322 derivative multicopy plasmid (Table 1) that carries a nusAinfB insert, which lacks the main promoters of the met Y-infB operon. The amount of IF2 produced by this multicopy plasmid is quite similar to that produced by met Y-infB operon one copy of chromosomal (Plumbridge et aZ., 19856). Operon internal promoters (Plumbridge & Springer, 1983) as well as plasmid promoters may be responsible for expression of the nusA and infB genes from ~1318-1. Strain SL598R transformed with each one of our plasmids was submitted to heat curing of 1 infB (as described in Materials and Methods). Survivors were detected at similar frequencies with all pB18-1 derivatives. The cured strains were shown to be stable through several purifications at 42 “C. Analysis of Western blots of the cured strains confirmed the presence of one form of IF2 (IF2a or IF2/?) to the exclusion of the other in the survivors carrying pB18645, pB18SC2 and pB18SC3 (Fig. 6). This result shows that each one of the two forms of IF2, in near physiological amounts, is sutlicient to support growth of E. coli at 42°C. Surviving strains were named according to the plasmid they carry, by replacing pB by SL. For example, SL18A45 is the cured strain derived from pB18A45 transformed SL598R. We tested the cured strains for growth at various temperatures. Growth on LB plates is summarized in Table 3: it shows that strains carrying a single IF2 form (SLl8A45, SL18SC2 and SLl8SC3) are distinctly retarded at 37”C, compared with SL18-1 and to parental strain IBPC5321R. The single IF2 phenotype is enhanced at 3O”C, where these strains behave as cold Isensitive. In liquid media (LB) cultures at 37°C (Fig. i’), the single IF2 strains (SL18A45 and SL18SC2) display a significantly lower growth rate than the reference two forms of IF2 (SL18-1) strain. The growth rate of the control strain SL18C2, which express both IF2c( and IF2/3, proteins with one amino acid substitution at amino acid 165 of IF2c( and amino acid 7 of IF2/3,, is comparable with. that of SL18-1. At 3O”C, the difference between the reference strain (SL18-1) and the mutants is accentuated. While SL18-1 displays normal exponential growth (doubling time = 40 min), the growth rate of the mutants initially slows down but they do not stop growing. When subcultured for several transfers at 3O”C, their exponential growth rate stabilizes around 175 min (“a only” strains) and 220 min (“fi only” strain). We measured the relative amounts of IF2 in the different cured strains grown at 37”C, using a quantitative immunoblotting technique (see Materials and Methods). The results are summarized in Table 3. They show that in SL18A45 elimination of IF2cr by deletion A45 stimulates IF2/? synthesis by a factor of 1.7. Therefore SL18A45 (the IFS/3 only strain) cells contain 70% of the total amount of IF2 in a wild-type strain. Moreover SL18A45 contains

C. Sacerdot et al

76

only IF2a or IF2b, could be attributed to either t’he mut’ation or a reduced level of totsal IF2, or a combination of the two. It is important t’o state that a strain expressing only one of the two forms of IF2 exhibits defectrive growth compared with a strain with both IF2a and IFS/3 only when the intracellular amounts of IF2 are near physiological levels. Indeed, we constructed plasmids overproducing all our infB constructs (by a factor of 5 to 8, data not shown) and introduced them in SL598R. After heat curing of 2 inf& we observed that all cured strains grew well on plates at all temperatures wit’hout any detectable difference between them. Therefore, each form of IF2 is sufficient to allow maximal growth of E. coli when supplied in excess. 0

50

100

150 200

250 300

350 400

Min

Figure 7. Growth curves of the cured strains at 37 “C in LB medium. Overnight precultures were grown in LB medium at 37°C. These saturated overnight precultures (all at about A,,, = 3) containing probably a higher plasmid copy number than exponentially growing cells, were diluted a 1st time loo-fold with LB. After the cultures had reached A,,, = 0.6 to 07, they were diluted a 2nd time (50-fold) in LB and growth was then monitored by measuring the absorbance at 650 nm, as shown in the Fig. As the growth curves of the 2 pairs SL18SC2 and SLl8SC3, and SL18C2 and SL18C3 superimpose, only 1 member of each pair is shown for clarity. 0, SL18-1; 0, SLlSA45; A, SL18SC2; A, SLlSC2.

amounts of tot’al IF2 that are similar to those in SL18C2 or SLlBC3. Yet these two strains grow much better at 37°C. Thus it is unlikely that t,he growth phenotype of SL18A45 is due to the 30% reduction in total IF2. These data show clearly that the presence of IF2a enhances growth of E. coli. The total intracellular amount of IF2 in the SL18SC2 and SL18SC3 strains (the IF2a only strains) is slightly lower than that of SL18A45: equivalent to about 60% of the total amount of IF2 in a wild-type strain (Table 3). It seems unlikely that this decreased amount of IF2 is responsible for the growth defects of the strains SL18SC2 and SL18SC3 (see Discussion). In order to abolish IFB/? expression, the IF2a protein of these strains has been inevitably mutated at’ one of its amino acids (Met 165) and an effect of the amino acid substitution cannot be positively excluded. The fact that’ both strains behave in the same way at all temperatures in spite of two different substitutions of methionine 165 (for threonine or isoleucine), and the fact that SL18C2 and SL18C3 grow much bett,er than SLlSSC2 and SLlSSC3, argue towards a negligible effect of the amino acid substitutions of methionine 165. However, we note that there is a slight growth defect in the strains SL18C2 and SL18C3 that express the mutated forms of IF2a and IF2P, (but not IFBP,). This effect, which is much less significant than the defect observed in strains expressing

4. Discussion In order to investigate the observation that IF2 exists primarily in two forms (IF2a, IF28) in E. coE1;, we created strains of E. co& that contained only IF2a or IF2/3. Our attempts to eliminate IF2b expression showed that initiation of IF2g synthesis is quite unusual, consisting of two start codons: GUG (493 to 495) and AUG (514 to 516). Only by mutating both start codons was it possible to eliminate all IF2a synthesis. The determination of the N-terminal amino acid sequence of an IF2/&b-galactosidase hybrid protein expressed from the wild-type IF2g RRS gave the Sterminal sequence of IF2/3, (GUG (493 to 495) start codon: Morel-Deville et al., 1990). This implies t’hat IFSPi is the major species (at least 600/;, to 7Ooi,) of IF2/3 (see Materials and Methods). However, mutating the GUG codon stimulates initiation on BUG (IF2p, start codon) probably by rendering its RBS more accessible. This would explain why the residual t,ranslation of IF2g is 50 ?A, 66 yc and 7 1 yc when the GUG codon is changed to GUU, GGG and GAG, respectively. Since the levels of residual IF%fl expression are different’, it is probable that the nature of the mutations influences the level of initiation on BUG. Schneider et aZ. (1986) have observed that the primary sequence of ribosome binding sites is not random from position -20 to + 13 bases around the st,art codon (the 1st base being taken as 0). According to this numbering the second and third bases of GUG (493 to 495) are located at position - 20 and - 19, respectively, relative to the AUG codon of IF2/I,. The influence of these alterations may be low, but it is interesLing to note that both GUG-tGUC and GUG+GCG mutations decrease IF2P, synthesis to the lowest level (35% and 31%). This is in agreement with Dreyfus (1988) who observed that C bases in the region of RBSs are unfavourable to translation initiation. On the other hand, it is even more difficult to explain why the C2(AUG-+ACG) and CS(ACG+AUC) mutations of the RUG start codon of IF2P, decrease IF2fl, expression to 30 o/0 and 42 o/o, respectively, if initia-

Two Forms

of IF2

tion on AUG does not normally account for more than 40% of IF2/l expression. Mutations C2 and C3 are located at positions +22 and +23 from the GUG start codon, outside the classically defined RBS. However, a consensus sequence found in several efficient translation initiation regions of the E. co& genome and complementary to a 15 base sequence of 16 S rRNA was proposed by Sprengart et al. (1990) to specify a stimulatory interaction between the mRNA and 16 S rRNA during the translation initiation step. This consensus sequence, 5’ UCAUGAAUCACAAAG 3’, called downstream box, was first found at position + 13 to +28 of the highly active translation initiation region of gene 0.3 of phage T7. However, the survey presented by Sprengart and co-workers shows that the distance of the downstream box from the start codon can be quite variable, with 8 to 13 bases complementary within the 15 nucleotide sequence. In the case of IF2P, a similar downstream box can be found (nucleotides 512 to 526), with ten nucleotides complementary to the 16 S rRNA: 16 S rR,NA

3’ AGUACUUAGUGUUU _ C 5’, . . . . . . . . . . . . . . .

Downstream

-G 3’, * C G 3’.

box 5’ UCAUGAAUCACAAA **** * * *** in@ (512 to 526) 5’ AUAUGACU -AAAAA

Mutations C2 and C3 are located within this downstream box and both remove one base-pair between in@ mRNA and the 16 S rRNA. Another factor may be a secondary structure of the mRNA formed between the mutated region and the RBS of IFag. The two IF2fi start sites show interesting features. They are located inside a translated region of the mRNA, which is a situation more common in phage systems than in bacteria. Schottel et al. (1984) observed by gene fusion experiments, that translat>ion initiated upstream prevented detectable initiation of protein synthesis at the cat gene translation start codon. This is obviously not the case here, as IF2/l is a well-expressed protein. A possible explanation may come from the fact that the IF2aspecific part of the infB coding region seems to be less efficiently translated because it contains more codons for rare isoacceptor tRNA species (Sacerdot et al., 1984). Therefore, pausing of ribosomes at rare codons in the proximal part of infB may promote IF2p initiation of translation by rendering its RBS more accessible. This idea is supported by the observation that the A45 deletion, which eliminates the upstream translation, results in an increased synthesis of IF2P by a factor of 1.7 (Table 3). It is very likely that the absence of ribosomes translating IF2a renders IF2P RBS even more accessible. Conversely, abolishing IF2fi expression does not increase IF2c( synthesis (Table 3), indicating that IF2g initiation of translation does not limit the rate of synthesis of IF2cl. The attractive hypothesis of mutual regulatory effects of the two IF2 forms is excluded both by the fact that there is no effect of

Required for Growth

77

IF2 on its own expression (Plumbridge et al., 19856) and by the absence of stimulation or inhibition of IF2a synthesis in the strains lacking IF2fi. Northern blot experiments have shown that ZnfB mRNA undergoes endonucleolytic cleavages that result in considerable heterogenity in length (Regnier & Grunberg-Manago, 1989). In Sl mapping experiments the 5’ end of a minor messenger specific to IF28 was also detected (P. Regnier, personal communication). These observations suggest that a minor part of IF2fl synthesis derives from processed mRNA molecules that are missing the IF2a start site. Perhaps the most unusual feature of IF2/? expression is the fact that IF2P possesses two functional initiation codons, neither of which looks very efficient compared with the consensus sequences established for efficient translational initiation. Nevertheless, IF2B is a highly expressed protein. It is not known what makes the ribosome choose either GUG (493 to 495) or AUG (514 to 516), with a slight preference for the first one. The relative proximity of the two sites means that their RBSs overlap and it is unlikely that initiation occurs simultaneously on the same mRNA. The SD signal of the AUG initiation site (UCAACMA) appears to be less efficient than that of the GUG initiation site (CmAAmA), which could account for IFBB, being the major species of IF2P. Armed with the knowledge of how IF2/? is expressed, we constructed the appropriate tools for the in viva investigation of the role,of the two forms of IF2. Complementation of AinfB :: cat2 of SL598R by plasmids expressing only one form of IF2 shows that IF2a as well as IF2B separately support growth of E. coli. This was to be expected since Laalami et al. (1991) reported that a protein consisting of the C-terminal 55,000-M, fragment of the wild-type IF2 was sufficient to allow growth of E. coli when supplied in excess (at least 5-fold). In this work we demonstrate that each one of the two forms of IF2 is sufficient to allow growth of E. coli when expressed at near physiological intracellular concentrations but that growth is retarded at 37°C compared with a strain with both forms. The effect of the intracellular concentration of IF2 on the growth rate of these strains at 37°C is likely to be a minor effect. Cole et al. (1987) have shown that the intracellular level of IF2 can be decreased by a factor of two without a significant effect on the growth rate at 37°C. In our case, the IF2P strain and both IF2a strains cont>ain, respectively 70% and 60 ye of the total amount of IF2 found in a wild-type strain, which is above the 50% threshold. In spite of this apparent sufficiency in total IF2, they exhibit a clearly visible phenotype of retarded growth both in liquid media and on plates. Moreover, the intracellular level of IF2 in the IF2P only strain (SLlSA45) is similar to that of SLl8C2 and SL18C3, which contain both forms of IF2 and grow much better at 37°C. Therefore, we can conclude that when supplied at near physiological level, IF2P is not sufficient to allow optimal growth

78

c. Sacerdotet al.

of E. coli and that under these conditions IF2a is required for maximal growth rates. Concerning the two strains that express only IF2a protein, it is necessary to point out that the two amino acid substitutions of methionine 165 by threonine or isoleucine, which eliminate IFS@, may affect’ TF2a activity and conseinitiation, quently the gr0wt.h rate of the cells. This is the reason why we analyzed two such mutants. We observe that both strains SL18SC2 and SL18SC3 behave identically under our experimental conditions. Furthermore the corresponding controls (SL18C2 and SL18C3), that carry respectively the same muta.tion introducing one amino acid substitution in both forms of IF2, grow much better than the IF2a only strains. This suggests that t,he mutations Met(l65)-tThr and Met(l65)-+Ile are not important for IF2 activity. However, the two control strains (SL18C2 and SLl8C3) show a slightly retarded growth compared with the reference strain SL18-1, and this is why we cannot formally exclude an effect of our mutations on methionine 165. The fact that these strains contain a distinctly decreased amount of TF2P (30% for SL18C2 and 42% for SLl8C3) is more likely the reason why they behave differently from 51,18-l. Therefore all these observations argue toward a specific contribution of TF2B to optimal growth of the cell at a near physiological level of TF2. To summarize, the growth phenotype of the single form of IF2 strains at 37°C allows us to conclude that both IF2c( and IF2P are required for maximal growth rates of E. co&. The phenotype of the mutants at 30°C is accentuated and they behave as cold sensitive. However, at this temperature the intracellular level of IF2 is likely to contribute significantly to the cold sensitivity of the cell. Western blot analysis of the cured strains grown at 30°C indicate that the single form of IF2 strains contain less than 40% of total IF2 present in a wild-type strain grown at the same temperature. The reference cured strain with both forms contains 66% of the wild-type level (data not shown). Conversely this reference strain grown at 42 “C displays a higher intracellular concentration of IF2 than a wild-type strain grown at the same temperature (visible in Fig. 6(a)). The expression of IF2 and the cellular requirement for IF2 at different temperatures are under investigation. It should be noted that IF2 (both forms), as well as pNusA, are cold shock proteins of E. eoli (Jones et al., 1987). The fact that IF2 is a cold shock protein is in good agreement with the hypothesis that the intracellular amount of IF2 is crucial at low temperatures. An unanswered question is whether different roles exist for IF2a and IFSB. Reside the function of IF2 in protein translation initiation, this factor has been reported to be implicated in transcription: Travers et aZ. (1980) have observed that IF2 was capable of affecting transcription in vitro and specifically that it increased the production of rRNA. They found that IFSP preparat.ions had a, similar effect on transcriptional selectivity as IF20(. Nomura et al. (1986)

reported that fMet tRSA~“’ enhanced transeript’ion from a group of promoters, only when it was charged with formylated methionine. More recently the same group (Ishihama, 1988) showed tha,t only the IF2g form of IF2 was able to stably associate with RNA polymerase. They proposed a regulatory role for IF2/3, fMet tRINAye’ and also EF-Tu, in the coupling of transcription to translation. One of the two forms of IF2 may have a role in secretion since cold sensitive mutations of IF2 have been isolated as temperature resistant revedants to the thermosensitive secY24 mutation (Shiba eb a!.. 1986). The ability of IF2cr or IF2/? alone to support growth might have been predicted since it was already shown that the 55,000-1V, C-t,erminal fragment of IF2 is capable of supporting growth when supplied in excess, although growt,h was somewhat slow compared with fully complemented strains (Laalami et al.: 1991). The data of the present, paper of rF2 show that the full length versions (X = 97,300 M, and p = 79,700 &‘,) are not equivalent and that the two forms are necessary for optimal growth when present in physiological amounts. This implies that both forms have acquired specialized functions. This function could acts as a heterodimer be simply that IF2 normally (or other multimeric forms) of IF2or plus TF2g. Although TF2 has always been considered to act as a monomer, we know of no experimental data that eliminates this idea. The conclusion that TF2a and lF2/? have independent roles is support,ed by the observation that two forms of IF2 have been detected in other Enterobacteriaceae (Howe & Ilershey, 1984), and in the Gram-positive bacteria Racillus subtilis (Shazand et al., 1990). It thus seems probable that. same specialized but, non-essential function has been conserved t,hrough evolution. We are grateful sequencing

to B. Derijard

experiments

for his help in protein

and t.o B. Savelli

ings, to J. 8. Plumbridge,

for the draw-

J. W. B. Hershey and D. Levin

for crit,ical reading of the manuscript and to J. VT. B. Hershey and H. U. Petersen for providing us with an& bodies against E. coli TF2. We also t,hank B. Savelli and P. Foreman for t,echnical assistance in the purification of the 2 species of IFZP, and the 2 referees for their constructive comments. This work was supported by grants from the following : C.N.R.S. (URA 1139 and URA 1172), eontrat C.E.E. no. i%l*/0194-C(AM), Contrat~ de Recherche Externe I.N.S.E.R.M. no. 891017. Fondation pour la Recherche Medicale to M. Grunberg-Manago and S. Laalami, Universite Paris 7 (Chapitre 66-71, Soutien de

Programme)

and Universite

de Poitiers.

References Adler;

H. I.. Fisher, W. D., Cohen, A. & Hardigree, A. -4. (1967). Miniature Escheriehia Coli cells deficient in DNA. Proc. Sat. Acad. Sci., U.S.A. 57. 321-326. Bleicher. F., Deville, F.: Cort’ay, J. C.. h’egre; D.; Cozzone, A. cJ. & Cenat’iempo, Y. (1988). Promoter vectors with restriction-site banks. Gene, 63, 135-139. Cenatiempo, P., Deville. F., Dondon. J.. Grunberg-

Two Forms

qf IF2

Manago, M., Sacerdot, C., Hershey, J. W. B., Hansen, H. F., Petersen, H. U., Clark, B. F. C., Kjeldgaard, M., la Cour, T. F. M., Mortensen, K. K. & Nyborg, J. (1987). The protein synthesis initiation factor 2 G-domain. Study of a functionally active Cterminal 65.kilodalton fragment of IF2 from Escherichia, coli. Biochemistry, 26, 5070-5076. Cole, J. R., Olsson, C. L.; Hershey, J. W. B., GrunbergManago, M. & Nomura, M. (1987). Feedback regulation of rRNA synthesis in Escherichia coli. Requirement for initiation factor IF2. J. Mol. Biol. 198, 383-392. Dondon, J.; Plumbridge, J. A., Hershey, J. W. B. & Grunberg-Manago, M. (1985). Overproduction and purification of initiation factor IF2 and pNUSA proteins from a recombinant plasmid bearing strain. Biochimie, 67, 6433649. Dreyfus, M. (1988). What constitutes the signal for the initiation of protein synthesis on Escherichia coli mRNAs? J. Mol. Biol. 204, 79-94. Eskin, B., Treadwell, B., Redfield, B., Spears, C., Kung, H.-F. & Weissbach, H. (1978). Activity of different forms of initiation factor 2 in the in vitro synthesis of 189, Arch. Biochem. Biophys. P-galactosidase. 531-534, Gold, L., Pribnow, D., Schneider, T., Shinedling, S., Singer, B. S. & Stormo, G. (1981). Translational initiation in prokaryotes. Annu. Rev. Microbial. 35, 3655403. Gualerzi, C. 0. & Pon, C. L. (1990). Initiation of mRNA translation in prokaryotes. Biochemistry, 29; 5881-5889. Hershey, J. W. B. (1987). In Protein Biosynthesis (Neidhardt, F. C., Ingraham, J. L., Low, K. B., Magasanik, B., Schaechter, M. &. Umbarger, H. E., Society for 613-647, American eds), PP. Microbiology, Washington: DC. Howe, J. G. & Hershey, J. W. B. (1981). A sensitive method for measuring protein immunoblotting synthesis initiation factor levels in lysates of Escherichia coli. J. Biol. Chem.; 256, 1283612839. Howe, J. G. & Hershey, J. W. B. (1984). The rate of evolutionary divergence of initiation factors IF2 and IF3 in various bacterial species determined quantitatively by immunoblotting. Arch. Microbial. 140, 187-192. Ishihama, A. (1988). Promoter selectivity of prokaryotic RNA polymerases. Trends Genet. 4, 282-286. Ishii, S., Kuroki, K. 8: Imamoto, F. (1984). tRNAf2met gene in the leader region of the nusA operon in 81, Escherichia coli. Proc. Nat. Acad. Sci., U.S.A.

409-413. Jones, P. G., van Bogelen, R. A. & Neidhardt, F. C. (1987). Induction of proteins in response to low temperature in E. coli. J. Bacterial. 169, 2092-2095. Kurihara, T. & Nakamura, Y. (1983). Cloning of the nusA gene of Escheriehia coli. Mol. Gen. Genet. 190, 189-195. S., Putzer, H., Plumbridge, J. A. & Laalami, Grunberg-Manago, M. (1991). A severely truncated form of translational initiation factor 2 supports growth of Eseherichia coli. J. Mol. Biol. 220, 335-349. Matsudaira, P. (1987). Sequence from picomole quantities polyvinylidene electroblotted onto of proteins membranes. J. Biol. Chem. 262, difluoride 10035-10038. Minton, N. P. (1984). Improved plasmid vectors for the isolation of translational lac gene fusions. Gene, 31, 269273.

Required .for Growth

79

Morel-Deville, F., Vachon, G., Sacerdot, C., Cozzone, A., Grunberg-Manago; M. & Cenatiempo, Y. (1990). Characterisation of the translational start site for IF2, a short form of Escherichia coli initiation factor IF2. Eur. J. Biochem. 188, 605-614. Mortensen, K. K., Nyengard, N. R., Hershey, J. W. B., Laalami, S. & Sperling-Petersen, H. C. (1991). Superexpression and fast purification of E. coli initiation factor IF2. Biochimie, 73, 983-989. Nomura, T., Fujita, N. & Ishihama, A. (1986). Promoter selectivity of Escherichia coli RNA polymerase: alteration by fMet-ARNt)““‘. Nuel. Acids Res. 14, 6857-6870. Plumbridge, J. A. & Springer, M. (1983). Organisation of the Eseherichia coli chromosome around the genes for translation initiation factor IF2 (infB) and a transcription termination factor (nusA). J. Mol. Biol.

167, 227-243. Plumbridge, J. A., Howe, J. G., Springer, M.: TouatiSchwartz, D., Hershey, J. W. B. & GrunbergManago, M. (1982). Cloning and mapping of a gene for translational initiation factor IF2 in Escherichia co&. Proc. Nat. Acad. Sci., U.S.A. 79, 503335037. Plumbridge, J. A., Deville, F., Sacerdot, C., Petersen, P., Cenatiempo, Y., Cozzone, A.; Grunberg-Manago, M. & Hershey, J. W. B. (1985a). Two translational initiation sites in the infB gene are used to express initiation factor IF2/J and IF2 in Escherichia coli. EMBO J. 4; 223-229. Plumbridge, J. A., Dondon, J., Nakamura, Y. & Grunberg-Manago, M. (1985b). Effect of NusA protein expression of the nusA, infB operon in E. coli. Nucl. Acids Res. 13, 3371-3388. Regnier, P. & Grunberg-Manago, M. (1989). Cleavage by RNase III in the transcripts of the met I’-nusA-infB operon of Escherichia coli releases the tRNA and initiates the decay of the downstream mR.NA. J. Mol. Biol. 210, 293-302. Sacerdot, C., Dessen, P., Hershey, J. W. B.: Plumbridge, J. A. & Grunberg-Manago, M. (1984). Sequence of the initiation factor IF2 gene: unusual protein features and homologies with elongation factors. Proc. Nat. Acad. Sci., U.S.A. 81, 778777791. Sancar, A., Hack, A. M. & Rupp, W. D. (1979). Simple method for the identification of plasmid coded proteins. J. Bacterial. 137, 692-693. Sands, J. F., Regnier, P., Cummings, H. S.; GrunbergManago, M. & Hershey, J. W. B. (1988). The existence of two genes between infB and rps0 in the Escherichia coli genome: DNA sequencing and Sl nuclease mapping. Nucl. Acids Res. 16, 10803-10816. Sanger, F., Nicklin, S. & Coulson, A. R. (1977). DNA sequencing with chain-terminating inhibitors. Proc. Nat. Acad. Sci., U.S.A. 74, 5463-5467. Schaffner, W. & Weissmann, C. (1973). A rapid, sensitive and specific method for the determination of protein in diluted solution. An&. Biochem. 56; 502-514. Schneider, T. D., Stormo, G. D. & Gold, L. (1986). Information content of binding sites on nucleotide sequences. J. Mol. Biol. 188, 415431. Schottel, J. L., Sninsky, J. J. & Cohen, S. N. (1984). Effects of alterations in the translation control region use of cat gene on bacterial gene expression: constructs transcribed from the lnc promoter as a model system. Gene, 28, 177-193. Shazand, K., Tucker; J., Chiang, R., Stansmore, K., Sperling-Petersen, H. U., Grunberg-Manago, M., Rabinowitz, J. C. & Leighton, T. (1990). Isolation and molecular genetic characterization of the Bacil-

80

C. Xacerdot et al.

Pus subtilis gene (infB) encoding protein synthesis initiation factor IF2. J. Bacterial. 172, 2675-2687. Shiba, K.: Ito, K., Nakamura; Y., Dondon, J. & Grunberg-Manago, M. (1986). Altered translational initiation factor 2 in the cold sensitive ssyG mutant affects protein export in Escherichia coli. EMBO J. 5, 3001-3006. Silhavy, T. J., Berman, M. L. & Enquist, L. W. (1984). Experiments with Gene Fusions. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY. Sprengart, M. L.: Fatscher, H. P. & Fuchs, E. (1990). The initiation of translation in E. coli: apparent base pairing between the 16 S rRNA and down&ream sequences of the mRNA. Nucl. Acids Res. 18, 1719-1723. Taylor, J. W., Ott, J. & Eckstein, F. (1985). The rapid generation of oligonucleotide-directed mutations at

high frequency using phosphorothioate-modified DNA. Nucl. Acids Res. 13, 8’765-87% Travers, A. A., Debenham, P. G. & Pongs, 0. (1980). Translation initiation factor 2 alters transcriptional selectivity of Escherichia coli ribonucieic a.cid polymerase. Biochemistry, 19. 1651-1656. Ullmann; A. (1984). One-st,ep purification of hybrid proteins which have ,8-galactosidase activity. Gene, 29, 27-31. C.: Vieira, J. & Messing, J. (1985). Yanish-Perron, Improved Ml3 phage cloning vectors and host strains : nucleotide sequences of the M13mp18 and pUCl9 vectors. Gene: 33, 103-l 19. Zoller, M. J. & Smith. M. (1987). Oligonucleotide-directed mutagenesis: a simple method using t,wo oligonucleotide primers a,nd a single-stranded DNA template.

Meth,ods Enzymol. 154. 329-350.

Edited by 42. Yaniv