Proc. Nati. Acad. Sci. USA
Vol. 84, pp. 4022-4025, June 1987
AUU-to-AUG mutation in the initiator codon of the translation
initiation factor IF3 abolishes translational autocontrol of its
own gene (infC) in vivo
(protein synthesis control/site-specific mutagenesis)
J. SCOTT BUTLER*, MATHIAS SPRINGERt, AND MARIANNE GRUNBERG-MANAGOt
*Department of Biochemistry, University of Rochester Medical Center, 601 Elmwood Avenue, Rochester, NY 14642; and tInstitut de Biologie
Physico-Chimique, 13, rue Pierre et Marie Curie, Paris 75005, France
Contributed by Marianne Grunberg-Manago, February 20, 1987
translation initiation factor IF3 regulates the expression of its
own gene infC at the translational level in vivo. Here we create
two alterations in the iniC gene and test their effects on
translational autocontrol ofinfC expression in vivo by measur-
ing 13-galactosidase activity expressed from infC-lacZ gene
fusions under conditions ofup to 4-fold derepression or 3-fold
repression of infC expression. Replacement of the infC pro-
moter with the trp promoter deletes 120 nucleotides ofthe infC
mRNA 5' to the translation initiation site without affecting
autogenous translational control. Mutation ofthe unusualAUU
initiator codon of infC to the more common AUG initiator
codon abolishes translation initiation factor IF3-dependent
repression and derepression of infC expression in vivo. These
results establish the AUU initiator codon ofinfC as an essential
cis-acting element in autogenous translational control of trans-
lation initiation factor IF3 expression in vivo.
We previously showed that Escherichia coli
The initiation ofprotein synthesis inEscherichia colirequires
the activity of three protein factors (IF1, IF2, and IF3) (for
reviews, see refs. 1-3). One of these initiation factors, IF3,
binds to the 30S ribosomal subunit and shifts the equilibrium
between 70S ribosomes and their 50S and 30S ribosomal
subunits in favor of the free subunits, thus enhancing the
availability of 30S subunits on which protein synthesis
initiation begins. In addition, IF3 may play a direct role in the
binding of mRNA to the 30S ribosome.
The cellular levels of the protein synthesis initiation fac-
tors, like other protein synthesis components, increase with
increases in cellular growth rate (4), suggesting a shared
system ofgenetic regulation. However, the genes for each of
the factors do not map adjacent to one another, implying that
coordinate control of IF gene expression requires separate,
but not necessarily independent, control elements. The gene
for IF2, infB, lies at 69 min on the E. coli genetic map, and
it appears that the product of the adjacent gene, NusA
protein, regulates both infB and nusA gene expression at the
transcriptional level in vivo (5, 6). The infCgene for IF3 maps
at 38 min on the E. coli chromosome (7), adjacent to other
genes involved in protein synthesis (8). The genetic organi-
zation of infC is very unusual: (i) infC is unique in initiating
translation at an AUU codon (9), (ii) the infC AUU initiator
codon lies only three nucleotides 3' to the stop codon ofthrS,
the gene for threonyl-tRNA ligase (10), and (iii) transcription
of the majority of infC mRNAs initiates at a promoter, pO',
inside the thrS coding sequence and 182 base pairs upstream
of the infC initiator AUU (11).
Our recent experiments demonstrated that IF3 regulates
expression of its own gene at the translational level in vivo
(11). First, infC mutant strains with elevated expression of
structurally altered IF3 do not have corresponding increases
in infC mRNA synthesis rates. Second, infC mutant alleles
cause derepression of infC-lacZ gene fusions (translation
from the translation initiation site of infC) but not infC-lacZ
operon fusions (translation from the translation initiation site
oflacZ). Third, a cellular excess ofIF3 supplied in trans from
a multicopy plasmid represses expression of an infC-lacZ
gene fusion without affecting the rate of synthesis of hybrid
In the present work we tested whether the unusual AUU
initiator codon plays a role in infC translational autoregula-
tion. The results clearly show that a single mutation of
AUU-to-AUG abolishes autogenous translational control of
infC expression. This result is discussed in light of a theo-
retical analysis of the infC system by Gold et al. (12) that
predicted the infC AUU initiator codon to be an essential
element in the autoregulatory system.
MATERIALS AND METHODS
The E. coli K-12 strains used here were as described (11).
Genetic techniques andP3-galactosidasemeasurements were
as described by Miller (13). Molecular cloning techniques
were as described by Maniatis et al. (14). Selection and
screening of X monolysogens were as described (15). Mea-
surements ofinfC-lacZmRNA levels were made as described
(11) with details given in the legend to Table 3.
Construction of M13mpl9TSX3. Our first step in in vitro
mutagenesis of infC was to clone the HindIII-Sal I, trpPO-
containing fragment of pDR720 (16) into the same sites in
M13mpl9 to yield the Lac- derivative M13mpl9ptrp4. Next,
the promoterless Sal I-Sst II fragment of pUSX15 (11)
containing the translation initiation site and the first 54
codons of infC was inserted into the same sites between
trpPO and lacZ on M13mpl9ptrp4 to make the Lac' deriv-
Oligonucleotide Directed Site SpecificMutagenesis. We used
the deoxyoligonucleotide 5' TCCGCCTTTCATACCTTA 3'
provided by B. Ehresmann (Institut de Biologie Moleculaire
et Cellulaire du Centre National de laRecherche Scientifique,
Laboratoire de Biochimie, Strasbourg, France) to mutate the
infC AUU initiator codon to AUG using the method of
Kunkel (17). The nucleotide sequence ofthe resulting mutant
M13mpl9TSG2 and the parent M13mp19TSX3 was deter-
mined (18) by sequencing across the M13mpl9 polylinker at
the infC-lacZ fusion all the way to the EcoRI site just 5' to
trpPO. The change of AUU-to-AUG is the only difference
between the two sequences.
Transfer of Wild-Type and Mutant infC-lacZ Fusions to X.
To allow study ofour infC-lacZ gene fusions as single copies
Abbreviation: IF, translational initiation factor.
The publication costs of this article were defrayed in part by page charge
payment. This article must therefore be hereby marked "advertisement"
in accordance with 18 U.S.C. §1734 solely to indicate this fact.
Proc. Nati. Acad. Sci. USA 84 (1987)
in vivo, we constructed a X bacteriophage carrying all of the
lac operon except lacI, the promoter, the translation initia-
tion site, and the first seven amino acids of lacZ. The
EcoRI-Sst II lac fragment of pNM482 (19) was inserted
between the left arm, up to the Sst II site ofXNM540 (20), and
the right arm, from the far right EcoRI site of XSCX16 (11).
After in vitro packaging we characterized by restriction
analysis the c1857, Lac- phage XNNS4.
The EcoRI fragments containing the trpPO-infC fusion of
M13mp19TSX3 and its mutant derivative M13mp19TSG2
were inserted separately into the EcoRI site in front oflac in
XNNS4, creating an in-frame gene fusion between infC and
lacZ. After in vitro packaging and screening ofLac' X phage,
we characterized by restriction analysis the respective de-
rivatives XTSX31 (wild type, AUU) and XTSG25 (mutant,
AUG) (Fig. 1).
Replacement of the inIC Promoter with the tp Promoter
Does Not Affect Translational Autoregulation of iniC Expres-
sion. We replaced the infCpromoterpO' with the controllable
trp promoter because changing the infC initiator AUU to
AUG in the presence pO' resulted in apparent instability of
infC-lacZ gene fusions cloned in X (J.S.B., unpublished
results), possibly due to lethal levels of hybrid IF3-3-
galactosidase expression. The data in Table 1 show that
although IF3-p-galactosidase expression decreases from 519
to 93 units when the infC pO' promoter is replaced with the
trp promoter, the level of derepression from the fusions is
trpPO infC lacZ
gene fusions. The X genomes are abbreviated and
direction opposite to normal. Boxes above the line re
DNA. Open boxes below the line represent X DNA. )
or after a gene means that it is incomplete on that sid
represent mRNA transcripts and are preceded by a
their promoter. The nucleotide triplets represent t
initiation codons and the approximate positions in D
where infC-lacZ protein synthesis initiates. Y, lacY
gene; and J, J gene.
Physical maps oftheXbacteriophage canr
exactly the same: the derepression factor is about four in an
infCl9 background and about two in an infC37 background
whatever promoter is used to express the infC-lacZ gene
fusion. Similarly, the data in Table 2 show that an excess of
IF3 produced from an infC-carrying, multicopy plasmid
represses IF3-l3-galactosidase expression almost 3-fold with
either the trp or the infC pO' promoter. We showed previ-
ously that such repression of infC pO' gene fusions occurs
without changes in the synthesis rate of infC-lacZ mRNA
(11), and the data in Table 3 show that, using the trp
promoter, the derepression seen in an infCJ9 background
occurs without a significant change in the rate of synthesis of
infC-lacZ mRNA. These results demonstrate that transla-
tional autoregulation of infC expression occurs independent
ofthe nature ofthe promoter expressing the gene and thereby
confirm our previous experimental results showing autog-
enous translational control of infC expression in vivo (11).
Changing the infC AUU Initiator Codon to AUG Abolishes
Translational Autoregulation of infC Expression in Vivo. The
unusual usage of an AUU initiator codon in translation of
infC (9) and the fact that IF3 regulates its own gene expres-
sion at the level of translation (11) suggest that the AUU
codon may play a role in the autoregulatory system (12). To
test this hypothesis we changed the infC initiatorAUU to the
more common initiator AUG and tested the ability of IF3 to
control its own gene expression by monitoring the levels of
IF3-f3-galactosidase and infC-lacZ mRNA expressed from
infC-lacZ gene fusions under conditions normally causing
derepression or repression of infC expression in vivo. Mu-
tation ofthe AUU initiator codon ofinfC has two interesting
effects on infC expression in vivo. First, initiation of trans-
lation with AUG (XTSG25) instead of AUU (XTSX31) in-
creases the expression of an infC-lacZ gene fusion 10-fold in
a wild-type background (Table 1) without a comparable
increase in infC-lacZ mRNA levels (Table 3). These results
indicate that the lower level of expression from the fusion
carried by XTSX31 relative to XTSG25 is due to a transla-
tional, instead of transcriptional, phenomenon. Second, and
most dramatically, the mutation ofAUU to AUG completely
abolishes translational autoregulation of infC expression.
Neither infC mutant alleles (Table 1) nor over-production of
IF3 in trans (Table 2) affects AUG-initiated translation ofthe
infC-lacZ gene fusion carried by XTSG25, whereas its AUU
counterpart XTSX31 is normally regulated. We conclude,
therefore, that the infC AUU initiator codon is a necessary
cis-acting element in the translational autoregulation of infC
expression in vivo.
The experimental results presented here identify the unusual
initiatorAUU codon ofinfCas anecessary cis-acting element
in autogenous translational control ofinfCexpression in vivo.
Replacement of the major promoter pO' of infC by the trp
promoter has no effect on autogenous translational control of
infC expression, confirming ourprevious conclusion that IF3
controls its own synthesis at the translational level in vivo
(11). The replacement also deletes the first 120 nucleotides of
the leader of the infC mRNA without affecting regulation of
infC expression, indicating that information required for
translational autocontrol lies in a region 62 nucleotides 5' to
the initiator AUU. Because each of the gene fusions used
here contains the first 54 codons of infC fused to lacZ we
cannot exclude the possibility that control elements lie in the
159 nucleotides 3' to the AUU codon.
The most important result ofour experiments is that a single
nucleotide mutation changing the wild-type AUU initiator
codon to AUG eliminates autogenous translational control of
infC expression. This mutation causes the loss ofderepression
ofinfiC expression by infC mutant alleles (Table 1) and the loss
drawn in the
present E. coli
k prime before
le. The arrows
'NA and RNA
gene; A, lacA
Biochemistry:Butler et al.
Proc. Natl. Acad. Sci. USA 84 (1987)
Effect of infC alleles on infC-lacZ expression from various X bacteriophage
3-Galactosidase expression, units
(wild-type infC promoter,
AUU initiator codon)
519 ± 20
AUU initiator codon)
mutant AUG codon)
1037 ± 56
Measurements were made on monolysogenic, AWac strains IBPC5311 (wild type), IBPC5231 (infCl9), and IBPC5251
(infC37) (11) growing in exponential phase at 30'C in 4-morpholinepropanesulfonic acid (Mops)-glucose culture media (21)
supplemented with arginine at 0.05 mg/ml. The doubling times for the strains are 120 min each. The values are the average
of five or six measurements ± SD and are expressed per A650 unit of bacteria as described by Miller (13).
1942 ± 239
1020 ± 20
of repression of infC expression by increased cellular levels of
IF3 (Table 2) without a significant increase in infC-lacZmRNA
levels (Table 3). It is not likely that the increased translational
efficiency caused by mutation of AUU-to-AUG can, alone,
account forloss oftranslational autocontrol because excess IF3
represses, 2.5-fold, comparable levels of IF3-f3-galactosidase
expression from the wild-type fusion on XSCX16 (Table 2).
Instead, theAUU initiatorcodon must itselfspecify recognition
by the system governing translational autocontrol of infC
expression, and its mutation to AUG results, most likely, in a
combination ofderepression oftranslation initiation due to loss
of autocontrol and an increase in efficiency of translation
initiation from AUG instead ofAUU.
Our earlier experimental results showing autogenous trans-
lational control ofinfC expression (11) and the present results
showing its dependency on an AUU initiator codon confirm
two predictions of a theoretically based model proposed by
Gold et al. (12). In their model infCmRNA is translated in an
IF3-independent manner due to the abnormal AUU initiator
codon and other unusual features ofthe translation initiation
site ofinfC. The model predicts thatwhen the IF3 level is low
relative to the number of 30S ribosomal subunits, IF3-
independent translation of infC mRNA will be preferred to
IF3-dependent translation ofother cellular mRNAs-causing
a relative increase in IF3 levels. The Gold et al. model also
predicts twenty-nucleotide base pairing interactions between
the translation initiation site of infC mRNA and various
domains of the 16S RNA that specify infC mRNA-30S
ribosomal subunit interaction in the proposed IF3-indepen-
dent mode. In this respect, it is somewhat surprising that a
single mutation of U-to-G, changing only a single base pair,
should result in complete loss oftranslational autoregulation.
It is, however, possible that the AUU-to-AUG change is
enough to favor translation ofthe mutated infC mRNA in the
IF3-dependent mode, which should, if the model is correct,
cause loss of autoregulation of IF3 synthesis.
The negative autoregulation of infC translation can be
explained alternately by a more conventional model whereby
IF3 binds to an operator site covering the translation initia-
tion site on the infC mRNA. In this case, the third base ofthe
AUU initiator codon should be essential to the binding ofIF3
to the translational operator. Chemical crosslinking experi-
ments (22) and assays showing binding ofIF3 to the 3' end of
the 16S RNA (23) suggest that IF3 may interact directly with
sites on the 16S RNA during translation initiation. Some
nucleotide sequence homology exists between these sites and
the translation initiation site of infC (9, 22, 23)-yet these
homologies do not include the infC initiator AUU codon.
Such nucleotide sequence homologies might participate in
secondary structures recognized by IF3 and destabilized by
mutation of AUU-to-AUG. However, examination of the
nucleotide sequence of the portion of infC shown here to
specify autocontrol does not reveal significant secondary
structure involving the translation initiation site. Lack of
apparent homology between the 16S RNA and infC mRNA
does not rule out the existence of a functional competition
between the RNAs for IF3. We see little reason at this time,
however, to favor a classical IF3-mRNA interaction model
over the more indirect model of Gold et al.
A growing number of proteins control their own gene
expression at the translational level in vivo. These include
genes for ribosomal proteins (24, 25), the gene 32 protein (26,
Effect of a cellular excess of IF3 on infC-lacZ expression from various X bacteriophage
XSCX16 (wild-type infC
promoter, wild-type infC
AUU initiator codon)
981 + 86
XTSX31 (trp promoter,
wild-type infC AUU
135 ± 10
XTSG25 (trp promoter,
mutant infC AUG
790 ± 83
The strains and techniques are described in the legend of Table 1 except that 4-morpholinepropane-
sulfonic (Mops)-glucose culture media was supplemented with all the amino acids (21) except
tryptophan; ampicillin was added to 0.1 mg/ml every 2 hr during growth; and the doubling time for the
strains was 90 min.
363 ± 80
762 ± 65
Biochemistry:Butler et al.
Proc. Natl. Acad. Sci. USA 84 (1987)
infC-lacZ gene fusions
Cellular levels of infC-lacZ mRNA expressed from
% x 102
cpm x l0-4
gene fusion with
gene fusion with
gene fusion with
The strains were grown as described in the legend ofTable 1. The
[3H]uridine-labeled RNA was isolated and hybridized to an excess of
the lacZ-specific probe M13mp81acl4 as described (11). Background
hybridization to single-stranded M13mp8 was <25% ofthe values in
column 3 and has been subtracted from them.
10391 ± 170.9 ± 0.2
58 90 ± 241.5 ± 0.4
166 170 ±
1.0 ± 0.01
27) and the RegA protein (28) of bacteriophage T4, and
threonyl-tRNA ligase (15, 29). In all ofthese examples (RegA
being a possible exception) the gene products made by and
controlling each gene interact functionally, in a well-defined
manner, with nucleic acids. The ability of the protein to
control translation is thought to depend upon its binding to a
region on its own mRNA that shares some homology with its
primary nucleic acid substrate. For example, mutations on
the thrS mRNA disrupting translational repression by thre-
onyl-tRNA ligase lie in an area ofthe mRNA adjacent to the
ribosome binding site showing striking homologies in se-
quence and structure to threonine isoacceptor tRNAs (30).
IF3 differs from these examples because, although it interacts
with a variety of RNAs, it does so without any apparent
nucleotide sequence specificity (23, 31, 32). Nevertheless,
IF3 may control its own translation simply by binding to its
AUU codon, thereby blocking the access of 30S ribosomes.
Such a model, together with the results presented here,
suggests that IF3 should have a measurable difference in
binding affinity for infCmRNAs containing an AUU initiator
codon or a AUG initiator codon.
Translation initiation ofE. coli mRNAs occurs at specific
sites characterized by a certain number of determinants
whose importance depends upon their spatial organization
(33). One of these determinants is the translation initiation
codon to which initiator fMet-tRNA binds on the 30S ribo-
somal subunit. The translation initiation codon is generally
AUG, but E. coli initiatortRNA can also recognizeGUG and
exceptionally AUA and UUG (33). Surprisingly, IF3, which
is essential for efficient translation initiation complex forma-
tion in vitro, has a unique AUU initiator codon. The results
reported here show that this exception is not fortuitous, but
is necessary for the controlled synthesis of IF3 itself.
The authors are grateful to Terry Platt, for reading the manuscript
and for generously providing laboratory facilities where this work
was completed. J.S.B. was supported by the National Institutes of
Health-French Centre National de la Recherche Scientifique Pro-
gram for Scientific Collaboration. This workwas supported by grants
from Centre National de la Recherche Scientifique, from the Fonda-
tion pourlaRecherche Medicale, andfrom E. I. du Pontde Nemours.
Grunberg-Manago, M. (1979) in Ribosomes: Structure, Func-
tion, and Genetics, eds. Chambliss, G., Craven, G. R., Da-
vies, J., Davis, K., Kahan, L. & Nomura, M. (University Park
Press, Baltimore), pp. 445-447.
Hershey, J. W. B. (1980) in Cell Biology: A Comprehensive
Treatise, eds. Prescott, D. M. & Goldstein, L. (Academic,
New York), Vol. 4, pp. 1-68.
Kozak, M. (1983) Microbiol. Rev. 47, 1-45.
Howe, J. G. & Hershey, J. W. B. (1983) J. Biol. Chem. 258,
Plumbridge, J. A., Dondon, J., Nakamura, Y. & Grunberg-
Manago, M. (1985) Nucleic Acids Res. 13, 3371-3388.
Nakamura, Y., Plumbridge, J. A., Dondon, J. & Grunberg-
Manago, M. (1985) Gene 36, 189-193.
Springer, M., Graffe, M. & Hennecke, H. (1977) Proc. Natl.
Acad. Sci. USA 74, 3970-3974.
Plumbridge, J. A., Springer, M., Graffe, M., Goursot, R. &
Grunberg-Manago, M. (1980) Gene 11, 33-42.
Sacerdot, C., Fayat, G., Dessen, P., Springer, M., Plum-
bridge, J., Grunberg-Manago, M. & Blanquet, S. (1982)EMBO
J. 1, 311-315.
Mayaux, J.-F., Fayat, G., Fromant, M., Springer, M., Grun-
berg-Manago, M. & Blanquet, S. (1983) Proc. Natl. Acad. Sci.
USA 80, 6152-6156.
Butler, J. S., Springer, M., Dondon, J., Graffe, M. & Grun-
berg-Manago, M. (1986) J. Mol. Biol. 192, 767-780.
Gold, L., Stormo, G. & Saunders, R. (1984) Proc. Natl. Acad.
Sci. USA 81, 7061-7065.
Miller, J. H. (1972) Experiments in Molecular Genetics (Cold
Spring Harbor Laboratory, Cold Spring Harbor, NY).
Maniatis, T., Fritsch, E. F. & Sambrook, J. (1982) Molecular
Cloning: A Laboratory Manual (Cold Spring Harbor Labora-
tory, Cold Spring Harbor, NY).
Springer, M., Plumbridge, J. A., Butler, J. S., Graffe, M.,
Dondon, J., Mayaux, J. F., Lestienne, P., Blanquet, S. &
Grunberg-Manago, M. (1985) J. Mol. Biol. 185, 93-104.
Russel, D. R., Auger, E. A., Vermersch, P. S. & Bennett,
G. N. (1984) Gene 32, 337-348.
Kunkel, T. (1985) Proc. Natl. Acad. Sci. USA 82, 488-492.
Sanger, F., Coulson, A. R., Barrel, B. G., Smith, A. J. H. &
Roe, B. (1980) J. Mol. Biol. 143, 161-178.
Minton, N. P. (1984) Gene 31, 269-273.
Borck, K., Beggs, J. D., Brammar, W. J., Hopkins, A. S. &
Murray, N. E. (1976) Mol. Gen. Genet. 146, 199-207.
Neidhardt, F. C., Bloch, P. L. & Smith, D. F. (1974) J.
Bacteriol. 119, 736-747.
Ehresmann, C., Moine, H., Mougel, M., Dondon, J., Grun-
berg-Manago, M., Ebel, J.-P. & Ehresmann, B. (1986) Nucleic
Acids Res. 14, 4803-4821.
Wickstrom, E. (1983) Nucleic Acids Res. 7, 2035-2052.
Lindahl, L. & Zengel, J. (1986) Annu. Rev. Genet. 20, 297-326.
Nomura, M., Gourse, R. & Baughman, G. (1984) Annu. Rev.
Biochem. 53, 75-177.
Russel, M. L., Gold, L., Morrissett, H. & O'Farrell, P. Z.
(1976) J. Biol. Chem. 251, 7263-7270.
Lamaire, G., Gold, L. & Yarus, M. (1978) J. Mol. Biol. 129,
Karam', J., Gold, L., Singer, B. & Dawson, M. (1981) Proc.
Natl. Acad. Sci. USA 78, 75-117.
Butlet', J. S., Springer, M., Dondon, J. & Grunberg-Manago,
M. (1986) J. Bacteriol. 165, 198-203.
Springer, M., Graffe, M., Butler, J. S. & Grunberg-Manago,
M. (1986) Proc. NatIl. Acad. Sci. USA 83, 4384-4388.
Schleigh, T., Wickstrom, E., Twombly, K., Schmidt, B. &
Tyson, R. W. (1980) Biochemistry 19, 4486-4492.
Johnson, B. & Szekely, M. (1977) Nature (London) 267,
Gold, L., Pribnow, D., Schneider, T., Shinedling, S., Singer,
B. S. & Stormo, G. (1981) Annu. Rev. Microbiol. 35, 365-403.
Biochemistry:Butler et aL