How initiation factors maximize the accuracy of tRNA selection in initiation of bacterial protein synthesis.

Abstract

During initiation of bacterial protein synthesis, messenger RNA and fMet-tRNAfMet bind to the 30S ribosomal subunit together with initiation factors IF1, IF2, and IF3. Docking of the 30S preinitiation complex to the 50S ribosomal subunit results in a peptidyl-transfer competent 70S ribosome. Initiation with an elongator tRNA may lead to frameshift and an aberrant N-terminal sequence in the nascent protein. We show how the occurrence of initiation errors is minimized by (1) recognition of the formyl group by the synergistic action of IF2 and IF1, (2) uniform destabilization of the binding of all tRNAs to the 30S subunit by IF3, and (3) an optimal distance between the Shine-Dalgarno sequence and the initiator codon. We suggest why IF1 is essential for E. coli, discuss the role of the G-C base pairs in the anticodon stem of some tRNAs, and clarify gene expression changes with varying IF3 concentration in the living cell.

Full text

Page 1

Molecular Cell 23, 183–193, July 21, 2006 ª2006 Elsevier Inc. DOI 10.1016/j.molcel.2006.05.030
How Initiation Factors Maximize
the Accuracy of tRNA Selection
in Initiation of Bacterial Protein Synthesis
Ayman Antoun,1,2Michael Y. Pavlov,1Martin Lovmar,1
and Ma ˚ns Ehrenberg1,*
1Department of Cell and Molecular Biology
Biomedical Centre
Uppsala University
Box 596
S-751 24 Uppsala
Sweden
Summary
During initiation of bacterial protein synthesis, mes-
senger RNA and fMet-tRNAfMetbind to the 30S ribo-
somal subunit together with initiation factors IF1,
IF2, and IF3. Docking of the 30S preinitiation complex
to the 50S ribosomal subunit results in a peptidyl-
transfer competent 70S ribosome. Initiation with an
elongator tRNA may lead to frameshift and an aberrant
N-terminal sequence in the nascent protein. We show
how the occurrence ofinitiation errors is minimized by
(1) recognition of the formyl group by the synergistic
action of IF2 and IF1, (2) uniform destabilization of
the binding of all tRNAs to the 30S subunit by IF3,
and (3) an optimal distance between the Shine-Dal-
garno sequence and the initiator codon. We suggest
why IF1 is essential for E. coli, discuss the role of the
G-C base pairs in the anticodon stem of some tRNAs,
and clarify gene expression changes with varying IF3
concentration in the living cell.
Introduction
During initiation of bacterial protein synthesis, messen-
ger RNA (mRNA) associates with the small (30S) ribo-
somal subunit (Gold and Stormo, 1990; Gualerzi and
Pon, 1990), and the mRNA’s Shine-Dalgarno (SD) se-
quence (Shine and Dalgarno, 1974) binds to the anti-
SD sequence at the 50end of the 16S rRNA in the 30S
subunit (Steitz and Jakes, 1975). The SD/anti-SD con-
tact positions the downstream initiator codon (normally
AUG) at the beginning of the open reading frame in the
ribosomal P site, where it is recognized by the initiator
transfer RNA (fMet-tRNAfMet) (Gold and Stormo, 1990;
Gualerzi and Pon, 1990). The efficiency of initiation is
maximal at an optimal distance between the SD se-
quence and the initiator codon (Ringquist et al., 1992).
However, the maximum is broad (Ringquist et al.,
1992), which makes sense codons on either side of the
initiator codon efficient targets for elongator tRNA bind-
ing (Hartz et al., 1989). When a 50S subunit docks to
a 30S subunit in complex with an elongator, rather
than with the initiator, tRNA, this may lead to the synthe-
sis of a protein with an aberrant N-terminal amino acid
sequence or to out-of-frame mRNA reading (Gold and
Stormo, 1990; Hartz et al., 1989). It is therefore essential
that the initiator tRNA, rather than an elongator tRNA,
binds to the mRNA-programmed 30S subunit and even-
tually becomes P site bound in the 70S initiation com-
plex. The problem addressed in this study is how the ini-
tiator tRNA is recognized during initiation of protein
synthesis and how the initiation factors IF1, IF2, and
IF3 contribute to the accuracy by which initiator tRNA
isselectedincompetitionwithelongatortRNAs.Thisac-
curacy problem is conceptually distinct from that in pro-
tein elongation. In initiation, cognate initiator and non-
cognate elongator tRNAs are all able to form matching
codon-anticodon structures with the 30S bound mRNA
(Hartz et al., 1989). In elongation, in contrast, only the
cognate aminoacyl-tRNA can form a matching codon-
anticodon structure, and its selection depends on
base-pairing rules in conjunction with stereochemical
constraints provided by 16S rRNA (Ogle et al., 2002).
30S preinitiation complexes containing various P site
bound elongator tRNAs are easily formed (Hartz et al.,
1989),butthepresenceofinitiationfactorsgreatlyfavors
binding of the canonical initiator fMet-tRNAfMet(Hartz
et al., 1989, 1990). From these and other experiments
(Canonaco et al., 1986; Wintermeyer and Gualerzi,
1983), it was concluded that (1) IF2 helps to select fMet-
tRNAfMeton the 30S subunit by recognizing its formyl
moiety, (2) IF3 recognizes unique motifs in initiator tRNA
andtherebystabilizesitsbindingtothe30Ssubunitinre-
lation to the binding of elongator tRNAs (Gualerzi and
Pon, 1990; Hartz et al., 1990, 1989; Risuleo et al., 1976),
and (3) IF1 contributes little to the accuracy of fMet-
tRNAfMetselectiononthe30Ssubunit(Hartzetal.,1989).
In previous work, we studied how initiation factors
tune the rate of initiation with the initiator fMet-tRNAfMet
(Antoun et al., 2006). In this study, we compare the ef-
fects of combinations of IF1, IF2, and IF3 on the kinetics
of the interaction between fMet-tRNAfMet, Met-tRNAfMet,
and Phe-tRNAPheand the mRNA-programmed 30S sub-
unit and on the rate of 50S subunit docking to 30S pre-
initiation complexes assembled with either one of these
tRNAs. Our results show that the interaction of IF2 with
the formyl group of fMet-tRNAfMetdepends strongly on
the presence of IF1 and is pivotal both for fast formation
of the fMet-tRNAfMet-containing 30S preinitiation com-
plex and for rapid docking of the 50S subunit to this
complex. The present observations do not support the
previous conclusion that IF3 recognizes unique struc-
tural motifs in initiator tRNA (Gualerzi and Pon, 1990;
Hartz et al., 1989, 1990; Risuleo et al., 1976). They sug-
gest, instead, that the main contribution of IF3 to the
accuracy of initiator tRNA selection originates in an in-
crease in the rate of dissociation of all tRNAs from the
30S subunit. Our data show that IF1, acting in synergy
with IF2 and IF3, greatly enhances the accuracy of initi-
ator tRNA selection, which may explain the essential
nature of this factor (Cummings and Hershey, 1994).
The principles behind the effects of initiation factors
on initiator tRNA selection explain how the accuracy
of initiation of protein synthesis can be modulated in
the living cell. They clarify in vivo observations regard-
ing the effects of varying IF3 concentration on gene
*Correspondence: ehrenberg@xray.bmc.uu.se
2Present address: Institute for Cancer Studies, The University of
Birmingham, United Kingdom.

Page 2

expression (Butler et al., 1986; Sacerdot et al., 1996) and
are used to discuss the role of the G-C base pairs in the
anticodon stem of fMet-tRNAfMetand a subset of elon-
gator tRNAs (RajBhandary, 1994).
Results
Kinetics of tRNA Interaction with the 30S Subunit
30SribosomalsubunitswerepreboundtoanmRNAwith
a strong SD sequence and an optimally positioned AUG
(Met) initiation codon followed by a UUU (Phe) codon
(Experimental Procedures) and used to form complexes
with combinations of the initiation factors IF1, IF2, and
IF3. The rate of association of fMet-tRNAfMet, Met-
tRNAfMet, or Phe-tRNAPheto these complexes was
mainly monitored by nitrocellulose (NC) filter binding
with detection of a [3H]-labeled amino acid attached to
the tRNA (typical experiments are shown in Figures 1A
and 1B). The largest tRNA association rates were mea-
sured in stopped-flow experiments with detection of
scattered light (a typical experiment is shown in the in-
sert in Figure 1A). The association rates were of second
order and linear in the concentration of each tRNA (data
not shown), and the results are summarized as associa-
tion rate constants (ka,t, mM21s21) in Table 1.
To determine also the rate constants for tRNA disso-
ciation from the small subunit, we first prepared 30S
complexes containing the same mRNA and combina-
tions of initiation factors as above and, in addition,
[3H]fMet-tRNAfMet,[3H]Met-tRNAfMetor[3H]Phe-tRNAPhe.
Subsequently, each radiolabeled tRNA was chased with
an excess of its nonlabeled counterpart, and the disap-
pearance of
was monitored by NC filtration (a typical experiment is
shown in Figure 1C). The results are summarized as dis-
sociation rate constants (kd,t, s21) in Table 1.
In some cases, NC filtering was used to directly mea-
sure the equilibrium dissociation constants (Kd,t= kd,t/
ka,t) for the binding of tRNA to the 30S subunit in com-
plex with combinations of initiation factors.
3H-labeled tRNA from the 30S complex
Figure 1. Effects of Initiation Factors on the Kinetics of fMet-tRNAfMet, Met-tRNAfMet, and Phe-tRNAPheInteraction with the 30S Ribosomal Sub-
unit
(A) Association of Met-tRNAfMetor Phe-tRNAPheto the 30S$mRNA$IF1$IF2 complex, measured by NC filtration. (Inset) Association of fMet-
tRNAfMetto the same 30S complex, measured by light scattering.
(B) Association of Phe-tRNAPheto mRNA-programmed 30S subunits in the presence of IF1 and IF2, of only IF1, of only IF2, or of no factors, mea-
sured by NC filtration. The association of fMet-tRNAfMetin the absence of initiation factors is shown for comparison.
(C) Dissociation of [3H]Met-tRNAfMetfrom preassembled 30S complexes containing IF3 and different combinations of IF1 and IF2. Dissociation
was measured in a chase experiment by NC filtration.
(D) Dissociation of [3H]fMet-tRNAfMet(filled symbols) or [3H]Phe-tRNAPhe(open symbols) from 30S subunits programmed with xMFTI mRNA or
xFMTI mRNA in the absence of initiation factors. Dissociation was measured in a chase experiment by NC filtration.
Molecular Cell
184

Page 3

Combining all these results, a complete set of asso-
ciation and dissociation rate constants as well as equi-
librium dissociation constants for the interaction of
fMet-tRNAfMet, Met-tRNAfMet, or Phe-tRNAPhewith the
mRNA-programmed30Ssubunitwasobtained(Table1).
Effects of the Joint Action of All Three Initiation
Factors
The presence of all three initiation factors on an mRNA-
programmed 30S subunit increased ka,tby a factor of
400 from 0.031 to 12.5 mM21s21for formylated initiator
fMet-tRNAfMetbut only 10-fold for nonformylated Met-
tRNAfMetand 5-fold for elongator Phe-tRNAPhe(Table
1). At the same time, the dissociation rate constant in-
creased 300-fold for fMet-tRNAfMetand Phe-tRNAPhe
and more than 1000-fold for Met-tRNAfMet. As a result,
the addition of all three initiation factors to the 30S sub-
unit left the Kd,tvalue for fMet-tRNAfMetvirtually un-
changed but increased the Kd,tvalues for Met-tRNAfMet
and Phe-tRNAPheby two orders of magnitude (Table 1).
This means that the presence of all three initiation
factors strongly favored the canonical initiator fMet-
tRNAfMetin relation to Met-tRNAfMetand Phe-tRNAPhe
for binding in the 30S preinitiation complex.
IF2 Effects on Initiator tRNA Binding to the 30S
Subunit Depend on Formylation
The interaction between the initiator fMet-tRNAfMetand
the 30S subunit was strongly affected by the presence
of IF2, while the interactions between the nonformylated
Met-tRNAfMetor the elongator Phe-tRNAPheand the 30S
subunit were not (Table 1). Remarkably, in the absence
of IF2 the rate constants ka,tand kd,tfor fMet-tRNAfMet
and Met-tRNAfMetwere very similar, but addition of IF2
greatly decreased Kd,t, mainly by increasing the rate
constant (ka,t) of fMet-tRNAfMetassociation to the 30S
subunit (Table 1).
IF3 Destabilizes the Binding of All Types of tRNA
to the 30S Subunit
In the absence of IF1 and independent of the presence
of IF2, addition of IF3 to the 30S subunit led to
a 10- or 20-fold increase in Kd,tfor Phe-tRNAPheand
Met-tRNAfMet, due to one order of magnitude increase
in the association and two orders of magnitude increase
in the dissociation rate constant for these tRNAs. Fur-
ther addition of IF1 led to an additional 5-fold increase
in Kd,tfor Phe-tRNAPheand Met-tRNAfMetdue to a corre-
sponding increase in the dissociation rate constant
(Table 1). In the absence of IF2, the effects of IF3 and
IF1 on fMet-tRNAfMetand Met-tRNAfMetbinding kinetics
were very similar (Table 1), highlighting the primary role
of IF2 in formyl-group recognition.
IF1 Amplifies the Effects of IF2 and IF3 on tRNA
Interaction with the 30S Subunit
Removal of IF1 from the full set of initiation factors
greatly decreased the rate constants kd,tand ka,tfor
fMet-tRNAfMet, leaving their ratio, Kd,t, unaltered while
it decreased kd,tas well as Kd,tfor the other two tRNAs
(Table 1). In the absence of IF2, addition of IF1 to an
IF3-containing 30S subunit amplified the destabilizing
effect of IF3 on the binding of all tRNA types (see also
the previous section). In the absence of IF3, addition of
IF1 to an IF2-containing 30S subunit amplified the
stabilizing effect of IF2 on the binding of fMet-tRNAfMet
(Table 1).
Effects of the Distance between the SD Sequence
and the Initiation Codon
The rate constants for the interactions of Met-tRNAfMet
and Phe-tRNAPhewith the 30S subunit were similar for
allcombinationsofinitiationfactors,apartfromaconsis-
tently larger dissociation rate constant for Phe-tRNAPhe
in all cases (Table 1). This difference was due to the dif-
ferent positioning of the AUG (Met) and UUU (Phe) co-
dons in relation to the upstream SD sequence, as con-
cluded from experiments with two types of mRNA on
the 30S subunit. In these mRNAs, denoted xMFTI and
xFMTI and containing either A(+1)UGUUC or U(+1)UCAUG
sequences, the AUG and UUC (Phe) codons were swap-
ped, such that the +1 nucleotide in each case was at the
same optimal distance from the strong SD sequence.
The kinetic interactions between 30S subunits pro-
grammed with either one of these mRNAs and fMet-
tRNAfMetor Phe-tRNAPhewere studied in NC filter bind-
ing experiments. The association and dissociation rate
constants were virtually identical for the two types of
tRNA when their respective codons had the same dis-
tance to the SD sequence. Furthermore, when the AUG
or UUC codon was moved three bases downstream of
its optimal distance to the SD sequence, there was
an w10-fold increase in the dissociation rate constant
and no change in the association rate constant for the
correspondingcognatetRNA(Figure1D,Table2).These
results suggest that the dissociation rate constants for
Met-tRNAfMetwere smaller than those for Phe-tRNAPhe
in Table 1 because the AUG codon, but not the UUU
codon, was positioned at an optimal distance from the
Table 1. Kinetic and Equilibrium Parameters for tRNA Interaction with the mRNA-Programmed 30S Subunit in the Presence of Different
Combinations of Initiation Factors
IF1IF2IF3
fMet-tRNAfMet
Met-tRNAfMet
Phe-tRNAPhe
ka,t
kd,t
Kd,t
ka,t
kd,t
Kd,t
ka,t
kd,t
Kd,t
+
+
2
2
+
+
2
2
+
2
+
2
+
2
+
2
+
+
+
+
2
2
2
2
12.5
0.24
1.65
0.15
1.35
0.035
0.34
0.031
0.034
0.11
0.0065
0.013
1.5 3 1024
3.5 3 1024
1.8 3 1024
1.1 3 1024
2.70.27
0.22
0.19
0.24
0.022
0.024
0.024
0.026
0.15
0.16
0.027
0.025
2.1 3 1024
3.5 3 1024
1.2 3 1024
1.2 3 1024
560
730
140
100
0.22
0.21
0.23
0.22
0.022
0.025
0.036
0.045
0.42
0.34
0.10
0.079
7.5 3 1024
21 3 1024
13 3 1024
16 3 1024
1900
1600
450
360
34
84
36
35
460
3.8
87
0.11
10
0.54
3.5
9.5
14
5.0
4.6
Association rate constants (ka,t), dissociation rate constants (kd,t), and dissociation constants (Kd,t) are given in units of mM21s21, s21, and nM,
respectively. The rate constants and their standard deviations were calculated using the nonlinear regression program Origin 7 (OriginLab). The
relative standard deviations (s/mean) were normally around 5% and never exceeded 15%.
Accuracy of Initiation of Protein Synthesis
185

Page 4

SD sequence. The reason why the choice of distance
between initiation codon and SD sequence affected the
dissociation rate constants, but not the association rate
constants, for the tRNAs is clarified in Figure 3C and its
legend.
Kinetics of 70S Ribosome Formation from
Ribosomal Subunits
Stopped-flow technique with detection of scattered
light (Antoun et al., 2003, 2004) was used to monitor
the association of 50S subunits with mRNA-pro-
grammed 30S subunits in complex with either one of
fMet-tRNAfMet, Met-tRNAfMet, or Phe-tRNAPheand either
one of the eight possible combinations of initiation fac-
tors. All results from the subunit docking experiments
are summarized in Table 3 as first-order rate constants,
kc(s21).
Subunit Docking in the Absence of IF3
In the absence of initiation factors, the kcvalues differed
but little with the type of tRNA on the 30S subunit
(Figure 2A). When IF1 was added to the 30S subunit,
the kcvalues remained virtually unchanged for all three
tRNA types (Table 3). When, in contrast, IF2 was added
to the 30S subunit, kcincreased about 100-fold to 36 s21
with fMet-tRNAfMetbut only about 4-fold with Met-
tRNAfMetor Phe-tRNAPhe(Figure 2B, Table 3). Further
addition of IF1 to IF2-containing 30S subunits increased
kc by only 15% to 42 s21for fMet-tRNAfMetbut by
about 10-fold to 5.2 s21for Met-tRNAfMetand about
Table2. Effects oftheDistancebetweentheSDSequenceandthe
Initiation Codon on the Kinetics of Initiator and Elongator tRNA
Interaction with the 30S Subunit
mRNAtRNA
ka
(mM21s21)
kd
(s21)
Kd
(nM)
First position
binding
xMFTI
xFMTI
Second position
binding
xMFTI
xFMTI
fMet-tRNAfMet
Phe-tRNAPhe
0.025
0.026
1.6 3 1024
2.0 3 1024
6.4
7.7
Phe-tRNAPhe
fMet-tRNAfMet
0.029
0.021
12 3 1024
13 3 1024
41
62
Rate constants and their standard deviations were calculated as in
Table 1. The relative standard deviations were around 7% in all
cases.
Figure 2. Docking of the 50S Subunit to 30S Preinitiation Complexes Assembled with Different tRNAs and Combinations of Initiation Factors
Measured by Light Scattering
(A) 30S preinitiation complexes contained no initiation factors.
(B) 30S preinitiation complexes contained IF1 and IF2 (closed symbols) or only IF2 (open symbols). (Inset) The same curves in a short time range.
(C) 30S preinitiation complexes contained IF1 and IF3 (closed symbols) or IF3 (open symbols).
(D) 30S preinitiation complexes contained IF1, IF2, and IF3 (closed symbols) or IF2 and IF3 (open symbols). (Inset) The same curves in a short
time range.
Molecular Cell
186

Page 5

5-fold to 4.0 s21for Phe-tRNAPheon the 30S subunit
(Figure 2B, Table 3). This shows that aberrant 30S pre-
initiation complexes with IF1 and IF2 containing either
Met-tRNAfMetor Phe-tRNAPheassociate rapidly with
50S subunits.
Subunit Docking in the Presence of IF3
In the presence of IF3 as the only initiation factor, the kc
value was small with fMet-tRNAfMet(0.04 s21), smaller
with Met-tRNAfMet(0.01 s21), and smallest with Phe-
tRNAPhe(0.003 s21) on the 30S subunit (Figure 2C,
Table 3). Addition of IF1 to IF3-containing 30S subunits
reduced considerably the kcvalues with fMet-tRNAfMet
and Met-tRNAfMetand effectively blocked the formation
of 70S initiation complexes from Phe-tRNAPhe-contain-
ing 30S subunits (Figure 2C, Table 3). When, instead,
IF2 was added to IF3-containing 30S subunits, the kc
values increased about 200-fold to 8.7 s21with fMet-
tRNAfMet, about 9-fold to 0.09 s21with Met-tRNAfMet,
and about 5-fold to 0.015 s21with Phe-tRNAPheon the
30S subunit (Figure 2C, Table 3).
Figure 3. Cartoon of Initiation of Protein Synthesis with Definitions of Elemental Rate Constants
(A) Different tRNAs associate to mRNA-programmed 30S subunits equipped with three initiation factors with the association rate constant ka,t
and dissociate with the dissociation rate constant kd,t. IF3 dissociates from the 30S preinitiation complex with the rate kd,3, after which the 50S
subunit can dock to an IF3-free 30S preinitiation complex with the association rate constant ka,r. Alternatively, IF3 can rebind to the 30S preini-
tiation complex with the association rate constant ka,3. Rate constants ka,t, kd,t, kd,3, and ka,rare different for fMet-tRNAfMet, Met-tRNAfMet, and
Phe-tRNAPhe.
(B) The same as (A) but in the absence of IF3. In this case, the rate of tRNA dissociation (kd,t) is much smaller than the rate ka,r$[50S] of 50S dock-
ing. Thus, any tRNA bound to the 30S preinitiation complex has very little chance to dissociate and therefore always ends up in the 70S initiation
complex.
(C) Cartoon illustrating how a varying distance between the SD sequence and an initiation codon affects the kinetics of tRNA interaction with the
30S subunit. The 30S subunit has four states: 30S0or T:30S0, where the initiator codon is off the P site with or without tRNA, respectively, and
30Smor T:30Sm, where the initiator codon is in the P site with or without tRNA, respectively. 30S0and T:30S0are converted to 30Smand T:30Sm
with rate constants ksand kts, respectively. 30Smand T:30Smare converted to 30S0with rate constants k0
with 30S0and 30Smwith the same rate constant kaand dissociates from T:30S0and T:30Smwith rate constants k0
T:30S0and T:30Smare in rapid equilibrium (ks=k0
dissociation rate constant koff=k0
mizes koff, and leaves konunaltered.
sand k0
ts, respectively.tRNA associates
dand kd, respectively. When
s? 1 and kts=k0
ts[1), then the measured association rate constant kon= kaand the measured
ts=ktsÞ. Thus, optimizing the SD-initiator codon distance under these conditions maximizes kts=k0
dðk0
ts, mini-
Table 3. Effects of Initiation Factors and tRNA Type on the Rate Constant (kc) of 50S Subunit Docking to 30S Preinitiation Complexes and on
the kcat/KMParameters for the Whole Initiation Process
IF1 IF2IF3
kc(s21)kcat/KM(mM21s21)
fMet-tRNAfMet
Met-tRNAfMet
Phe-tRNAPhe
fMet-tRNAfMet
Met-tRNAfMet
Phe-tRNAPhe
+
+
2
2
+
+
2
2
+
2
+
2
+
2
+
2
+
+
+
+
2
2
2
2
2.9
0.0098
8.7
0.038
42
0.24
36
0.30
0.056
0.0016
0.090
0.011
5.2
0.11
0.51
0.13
0.0071
z 0
0.015
0.0032
4.0
0.16
0.79
0.19
12.4
0.020
1.65
0.11
1.35
0.035
0.34
0.031
0.072
0.0022
0.15
0.078
0.022
0.024
0.024
0.026
0.0037
—
0.029
0.0087
0.022
0.025
0.036
0.045
Thefirstorderrateconstantkccorresponds totherateof70Sribosomeformationmeasuredat0.3mMinitialconcentrationof30Scomplexesand
50S subunits in the reaction mixture. Rate constants and standard deviations were calculated from light scattering data using the nonlinear re-
gression program Origin 7 (OriginLab Corp.). The relative standard deviations (s/mean) for the kcestimates were in most cases around 3% and
never exceeded 10%. The kcat/KMvalues were calculated as the tRNA association rate constant ka,t(Table 1) times the probability that a bound
tRNA ends up in the initiation complex PInas defined in Equation 2 in Experimental Procedures.
Accuracy of Initiation of Protein Synthesis
187