Báo cáo khoa học: The binding of foot-and-mouth disease virus leader proteinase to eIF4GI involves conserved ionic interactions ppt

The binding of foot-and-mouth disease virus leader
proteinase to eIF4GI involves conserved ionic interactions
Nicole Foeger*, Elisabeth Kuehnel†, Regina Cencic and Tim Skern
Max F. Perutz Laboratories, University Departments at the Vienna Biocenter, Department of Medical Biochemistry, Medical University of
Vienna, Austria
The eukaryotic translation initiation factor (eIF) 4F is
a protein complex that mediates recruitment of ribo-
somes to mRNA [1]. This event is one of the rate-lim-
iting steps for translation and thus an important target
for translational control. The eIF4F complex consists
of several components: eIF4E, a protein recognizing
the 5¢ cap structure of the mRNA; the RNA helicase
eIF4A; and the bridging protein eIF4G, that brings
together mRNA and ribosome via mRNA circulariza-
tion [2]. eIF4G is a central part of this complex as it
provides binding sites not only for the already men-
tioned translation factors eIF4E and eIF4A [3], but
also for the ribosome-associated eIF3 [4], the poly(A)
binding protein [5] and the eIF4E kinases Mnk 1 [6]
and Mnk 2 [7].
Picornavirus infection leads to the so-called host cell
shut-off. Virally encoded picornaviral proteases cleave
eIF4GI and eIF4GII, thereby leading to an inhibition
of cap-dependent cellular protein synthesis [8,9]. Viral
translation is unaffected as it initiates via an internal
Keywords
Foot-and-mouth disease virus; papain-like
proteinase; self-processing; exosite; protein
synthesis inhibition
Correspondence
T. Skern, Max F. Perutz Laboratories,

University Departments at the Vienna
Biocenter, Department of Medical
Biochemistry, Medical University of Vienna,
Dr Bohr-Gasse 9 ⁄ 3, A-1030 Vienna, Austria
Fax: +43 14277 9616
Tel: +43 14277 61620
E-mail:
Website: />medbch
Present addresses
*Division of Cell Biology and †Division of
Tumour Genetics, German Cancer Research
Center, Im Neuheimer Feld 280, D-69120
Heidelberg, Germany
(Received 22 February 2005, revised 20
March 2005, accepted 24 March 2005)
doi:10.1111/j.1742-4658.2005.04689.x
The leader proteinase (L
pro
) of foot-and-mouth disease virus (FMDV) ini-
tially cleaves itself from the polyprotein. Subsequently, L
pro
cleaves the
host proteins eukaryotic initiation factor (eIF) 4GI and 4GII. This prevents
protein synthesis from capped cellular mRNAs; the viral RNA is still trans-
lated, initiating from an internal ribosome entry site. L
pro
cleaves eIF4GI
between residues G674 and R675. We showed previously, however, that
L
pro

binds to residues 640–669 of eIF4GI. Binding was substantially
improved when the eIF4GI fragment contained the eIF4E binding site and
eIF4E was present in the binding assay. L
pro
interacts with eIF4GI via resi-
due C133 and residues 183–195 of the C-terminal extension. This binding
domain lies about 25 A
˚
from the active site. Here, we examined the binding
of L
pro
to eIF4GI fragments generated by in vitro translation to narrow the
binding site down to residues 645–657 of human eIF4GI. Comparison of
these amino acids with those in human eIF4GII as well as with sequences
of eIF4GI from other organisms allowed us to identify two conserved basic
residues (K646 and R650). Mutation of these residues was severely detri-
mental to L
pro
binding. Similarly, comparison of the sequence between resi-
dues 183 and 195 of L
pro
with those of other FMDV serotypes and equine
rhinitis A virus showed that acidic residues D184 and E186 were highly
conserved. Substitution of these residues in L
pro
significantly reduced
eIF4GI binding and cleavage without affecting self-processing. Thus,
FMDV L
pro
has evolved a domain that specifically recognizes a host cell

protein.
Abbreviations
2A
pro
, 2A proteinase; CTE, C-terminal extension; eIF, eukaryotic initiation factor; ERAV, equine rhinitis A virus; ERBV, equine rhinitis B virus;
FMDV, foot-and-mouth disease virus; HRV, human rhinovirus; L
pro
, leader proteinase, containing amino acids 1–201; Lb
pro
, shorter form of
L
pro
containing amino acids 29–201; RRL, rabbit reticulocyte lysate.
2602 FEBS Journal 272 (2005) 2602–2611 ª 2005 FEBS
ribosome entry site (IRES) and therefore is independ-
ent of the interaction between eIF4E and eIF4G
[10,11]. The picornaviral proteinases such as the leader
proteinase (L
pro
) of foot-and-mouth-disease virus
(FMDV) and equine rhinitis A virus (ERAV) or the
2A proteinase of enteroviruses and human rhinoviruses
(HRVs) cleave eIF4GI and eIF4GII so that the eIF4E-
binding part of the eIF4G proteins is separated from
the bulk of the translation complex [4,12,13]. However,
the remainder of the complex is sufficient for viral
translation.
FMDV L
pro
is the most N-terminal protein on the

viral polyprotein, the primary translation product pro-
duced from the viral RNA genome. As a papain-like
cysteine proteinase, L
pro
shows a typical two domain
a-helix ⁄ b-sheet fold of a papain proteinase but is
unique in bearing a so-called C-terminal extension
(CTE) that protrudes from the globular structure [14].
Cleavage of eIF4GI by L
pro
, both in vivo and in vitro,
is highly efficient [15–18]. Nevertheless, the observed
L
pro
proteinase concentration at which eIF4G is
cleaved during viral replication is much lower than
that required in vitro when purified recombinant pro-
teins are employed [16,18,19]. For this reason, it has
been proposed that picornaviral proteinases activate
cellular proteinases, which cleave eIF4G in an indirect
reaction [20,21]. However, such cellular proteinases
have as yet not been identified. Ohlmann et al. have
previously obtained evidence that the substrate for
L
pro
is the eIF4GI–eIF4E complex [22]. In addition,
for the HRV2 2A proteinase (2A
pro
), Haghihat et al.
[23] demonstrated, using purified recombinant proteins,

that the eIF4GI–eIF4E complex was cleaved much
more efficiently than eIF4GI alone. Pertinently, it has
been shown that yeast eIF4GI undergoes an unfolded-
to-folded transition on binding eIF4E [24,25]. This
could be a reason why the eIF4GI–eIF4E complex is
the preferred substrate for picornaviral proteinases.
Recently, we showed that FMDV L
pro
and HRV2
2A
pro
indeed bind directly to the eIF4GI–eIF4E com-
plex, but much less well to eIF4GI alone [26]. Addi-
tionally, we have shown that both of these proteinases
interact with their substrate eIF4GI at a site distant
from their cleavage site [26,27]. For FMDV L
pro
,we
minimized the binding site on eIF4GI to amino acids
640–669; in contrast, the enzyme cleaves eIF4GI
between residues G674 and R675. Here we show that
we can define this region further to the 13 amino acids
between residues 645–657. This region in eIF4GI con-
tains conserved basic residues, which are here demon-
strated to be involved in binding by L
pro
. The L
pro
binding domain for eIF4GI is located 25 A
˚

from the
active site of the enzyme and comprises C133 as well
as residues 183–195 of the CTE. Mutations in these
amino acids significantly decrease eIF4GI cleavage by
L
pro
, without affecting L
pro
processing at the viral
polyprotein sequence. Here we emphasize the role of
two acidic residues (D184 and E186) in the CTE of
FMDV L
pro
which are responsible for eIF4GI binding
and cleavage.
Results
Recently, we have shown that FMDV Lb
pro
can bind
to its substrate eIF4GI between amino acids 640–669
[26]; binding in the presence of eIF4E to eIF4GI frag-
ments containing the eIF4E binding site was more
efficient. However, Lb
pro
cleaves between amino acids
G674 and R675. We wished to further define this
region between amino acids 640–669 and therefore
cloned shorter fragments of eIF4GI, as shown in
Fig. 1A. We started with an eIF4GI fragment contain-
ing amino acids 260–657; RNA from this construct

was translated in vitro in rabbit reticulocyte lysate
AB
DC
Fig. 1. Minimizing the eIF4GI binding domain of Lb
pro
. (A) eIF4GI
fragments used. Sites for eIF4E binding and Lb
pro
cleavage are indi-
cated. (B–D)
35
S-labelled proteins translated in vitro from the indica-
ted cDNA fragments of eIF4GI (input lanes 1, corresponding to a
quarter of that used in each pull-down) were incubated with GST
(lanes 2) or GST-Lb
pro
C51A (lanes 3). Bound proteins were resolved
by SDS ⁄ PAGE and detected by fluorography. All fragments were
reproducibly synthesized as doublets, presumably due to initiation
of translation at two AUG initiating codons in close proximity.
N. Foeger et al. FMDV L proteinase eIF4GI interaction
FEBS Journal 272 (2005) 2602–2611 ª 2005 FEBS 2603
(RRL) in the presence of radiolabelled methionine.
The labelled protein containing eIF4GI residues 260–
657 was then incubated with the fusion protein GST–
Lb
pro
C51A, which had been expressed in bacteria and
purified by binding to glutathione-sepharose beads.
The C51A mutation serves to inactivate the enzyme

during the pull-down assays. Figure 1B shows that
the eIF4GI fragment 260–657 is bound by GST–
Lb
pro
C51A, but not by GST alone. When we used an
eIF4GI fragment of amino acids 260–649 (Fig. 1C),
binding to Lb
pro
could still be detected but was clearly
weaker in comparison to that containing amino acids
260–657. In contrast, a protein containing amino acids
260–645 (Fig. 1D) was essentially not recognized by
the GST–Lb
pro
C51A fusion protein. These results
demonstrate that the 13 amino acids in the region of
645–657 on eIF4GI are important for binding by
FMDV Lb
pro
. Nevertheless, the presence of eIF4E
binding sequences on the eIF4G fragment and the
presence of eIF4E in the binding assay are required
for more efficient Lb
pro
binding.
As a control, we examined the binding of radiola-
belled cortactin, a cellular protein that we have shown
to be cleaved in vitro by Lb
pro
but at much slower

rates than eIF4GI. The radiolabelled cortactin is not
bound by the GST–Lb
pro
C51A fusion protein (data
not shown), indicating that the interaction of the
GST–Lb
pro
C51A complex protein with the eIF4GI is
specific and that it is probably responsible for the
rapid cleavage of Lb
pro
observed on eIF4GI.
When we examined the 13 amino acids of eIF4GI
responsible for binding Lb
pro
as well as those sur-
rounding this sequence more closely, we found that the
sequence of eIF4GI comprised three basic residues
(K643, K646 and R650), which were conserved
between human eIF4GI and eIF4GII (Fig. 2A). Ana-
lysis of protein databases revealed that these residues
were also present in eIF4GI sequences from several
mammalian species such as sheep, cow, hamster, horse,
mouse, pig and rabbit (EMBL accession numbers:
AJ746218–AJ746224 inclusive). This conservation sug-
gested that these basic residues might be recognized by
residues in the Lb
pro
CTE and thus enable the inter-
action to take place.

To investigate this notion, we individually mutated
these three residues to alanine in the plasmid con-
taining the eIF4GI deletion 260–657 to give the cor-
responding three plasmids shown in Fig. 3A,B.
Radiolabelled proteins corresponding to the mutants
were expressed in RRLs and their ability to be bound
by GST–Lb
pro
C51A in the pull-down assays was then
examined. The results are shown in Fig. 3 and Table 1.
The quantitation shows that about 12.5% of the input
material is bound by the wild-type fusion protein in
this assay. The efficiency of the pull-down is probably
limited by the presence of the GST part of the fusion
protein, the small size of the radiolabelled fragment
and the probability that a certain fraction of both
binding partners are incorrectly folded.
When we introduced the mutation K643A into the
eIF4GI construct comprising residues 260–657, we
found that binding by GST–Lb
pro
C51A was reduced
to between 80 and 90% of the wild-type level (Fig. 3A,
Table 1). However, the presence of the mutation
K646A reduced binding by about 75% when com-
pared to the wild type fragment (Fig. 3A, Table 1).
Fig. 2. (A) Sequence alignment of human
eIF4G proteins. The swissprot entries are
I4G1_human (eIF4GI) and I4G3_human
(eIF4GII). (B) Comparison of the C-terminal

extensions of FMDV Lb
pro
, ERAV Lb
pro
and
ERBV Lb
pro
. Asterisks in (A) and (B) indicate
conserved basic residues and acidic resi-
dues, respectively. Amino acids 183–195 in
FMDV Lb
pro
were shown to be required for
eIF4GI binding [26].
FMDV L proteinase eIF4GI interaction N. Foeger et al.
2604 FEBS Journal 272 (2005) 2602–2611 ª 2005 FEBS
Binding to the eIF4GI protein bearing the mutation
R650A was also reduced, but only by between 50 and
70% (Fig. 3A,B, Table 1). The relatively low involve-
ment of K643 for binding was supported by the
behaviour of the double mutant K643AK646A
(Fig. 3A,D); the binding is similar to that of the single
mutant K646A. Finally, no binding to the triple
mutant K643AK646AR650A in which all three con-
served basic residues were substituted with alanine was
observed (Fig. 3E).
The above results show that mutation of any of the
three conserved basic residues of eIF4GI impairs its
interaction with FMDV Lb
pro

but that the effects of
the mutations differ. This implied that conserved acidic
residues should be present in Lb
pro
that represent the
interaction partners of the basic eIF4GI residues. We
have shown previously that residues 183–195 of the
18 amino acid CTE as well as C133 of Lb
pro
were
involved in the interaction of Lb
pro
and eIF4GI [28].
Examination of the amino acids 183–195 of the CTE
revealed two residues, D184 and E186 (Fig. 2B), which
are conserved in all seven serotypes, including those
from Africa [29,30]. ERAV L
pro
is also responsible for
cleavage of eIF4GI and eIF4GII [13]. Analysis of the
CTE of this enzyme revealed that both residues were
also present in its CTE. In contrast, in the CTE of
equine rhinitis B virus (ERBV), L
pro
appears not to
cleave eIF4GI [13]. Fittingly, there is a residue equival-
ent to E186 in ERBV L
pro
but not to D184 (Fig. 2B).
We thus investigated the role of these amino acids in

their interaction with eIF4GI by mutational analysis.
Figure 4A shows that mutations in the amino acid
sequence of this region of Lb
pro
can be readily intro-
duced by using an oligonucleotide cassette spanning
the BsiWI and Bpu10I restriction sites. In this way,
we generated the substitutions D184A, E186K and
Q185RE186K, and investigated the ability of these
Lb
pro
mutants to carry out self-processing and eIF4GI
cleavage when expressed in RRLs. Figure 4B (lanes
1–5) shows that self-processing of wild-type
Fig. 3. Conserved basic residues in eIF4GI
are essential for Lb
pro
binding. (A–E)
35
S lab-
elled proteins translated in vitro from the
indicated cDNA fragments of eIF4GI (input
lanes 1, corresponding to a quarter of that
used in each pull-down) were incubated
with GST (lanes 2) or GST-Lb
pro
C51A (lanes
3). Bound proteins were resolved by
SDS ⁄ PAGE and detected by fluorography.
Table 1. Efficiency of binding of mutated eIF4GI fragments to

GST–Lb
pro
C51A. The % bound values are expressed relative to the
amount bound by the wild-type, which represents about 12.5% of
the total input. Experiment 2 shows the quantitation of Fig. 3.
eIF4GI fragment
% Bound
Experiment 1 Experiment 2
Wild-type 100 100
K643A 90 80
K646A 25 30
R650A 50 30
N. Foeger et al. FMDV L proteinase eIF4GI interaction
FEBS Journal 272 (2005) 2602–2611 ª 2005 FEBS 2605
Lb
pro
VP4VP2 into Lb
pro
and VP4VP2 takes place
between 4 and 8 min after protein synthesis is initiated.
Similar kinetics of self-processing were observed with
all three mutants tested (Fig. 4B, lanes 6–27). To
examine eIF4GI cleavage by the newly synthesized
Lb
pro
, we took advantage of the presence of eIF4GI in
the RRLs and examined its fate during the synthesis
of Lb
pro
, as reported previously [15]. Accordingly,

aliquots of the translation reactions were subjected to
SDS ⁄ PAGE, the gels blotted onto poly(vinylidene
difluoride) membranes and probed with an antiserum
against the N-terminus of eIF4GI (Fig. 4C, lower pan-
els). eIF4GI itself migrates as a series of bands with a
molecular mass of 220 kDa [31]. The bands have dif-
ferent N-termini, which arise from the use of different
AUG codons during synthesis of eIF4GI [32,33]. Clea-
vage of eIF4GI by Lb
pro
at its single recognition site
between G674R (numbering according to [32]) gener-
ates a series of N-terminal cleavage products that are
detected by the N-terminal antiserum used here (cp
N
,
Fig. 4C).
Figure 4C (lanes 1–5) shows that 50% of eIF4GI is
cleaved after 4 min with wild-type Lb
pro
. However, the
mutation D184A (Fig. 4C, lanes 6–12) showed a delay
in the cleavage of eIF4GI, the time-point of 50%
eIF4GI occurring only after 12–20 min. Thus, we con-
cluded that this residue is involved in eIF4GI recogni-
tion. The E186 mutant (lanes 13–19) also showed a
similar delay in eIF4GI cleavage, with 50% eIF4GI
cleavage being observed only after 12–20 min. To
investigate whether only the charges of the residues
184 and 186 were important for cleavage of eIF4GI or

whether other residues in this region could exert an
influence on the reaction, we constructed the double
mutant Lb
pro
Q185RE186K and investigated its activity
(lanes 20–26). Again, eIF4GI cleavage was impaired as
50% cleavage could only be seen after 20–30 min.
To verify further that these residues were involved
in binding to eIF4GI, we expressed the three mutants
as GST-fusion proteins and examined their ability in
pull-down assays to bind to eIF4GI. In this case, we
used the endogenous eIF4GI present in RRLs as a
source of eIF4GI to ensure the best possible binding
to the GST–Lb
pro
complexes. In Fig. 5A, the purity of
the expressed GST fusion proteins can be seen; Fig. 5B
shows the results of the GST pull-down assays. All
mutant proteins showed weaker binding than that
found in the wild-type (lane 2); furthermore, the extent
of binding correlated with the effect of these mutations
on eIF4GI. The mutants GST–Lb
pro
C51AD184A and
Fig. 4. Substitution of residues D184 and E186 in the CTE of FMDV Lb
pro
affect eIF4GI cleavage. (A) Structure of the expression block of
Lb
pro
VP4VP2 showing the position of restriction sites used to introduce mutations into the CTE of Lb

pro
. (B) Autoradiograms of proteins syn-
thesized from the indicated RNAs. Substituted residues are underlined. Samples were taken at the times indicated and Lb
pro
self-cleavage
from VP4VP2 examined (marked with arrows). (C), Immunoblot with the samples from (B) probed with an anti-eIF4GI antiserum to monitor
the eIF4GI cleavage by Lb
pro
; intact eIF4GI and the cleavage products cp
N
are marked.
FMDV L proteinase eIF4GI interaction N. Foeger et al.
2606 FEBS Journal 272 (2005) 2602–2611 ª 2005 FEBS
GST–Lb
pro
C51AE186K (lanes 3 and 4, respectively)
showed significantly less binding to eIF4GI than
GST–Lb
pro
C51A (lane 2); with the mutant GST–
Lb
pro
C51AQ185RE186K, we observed only very weak
binding (lane 5). Similar results were also obtained
when the
35
S-labelled eIF4GI fragment from 260 to
657 was employed (data not shown).
We showed in Foeger et al. [26] that endogenous
eIF4GII in RRLs could also be bound by the GST–

Lb
pro
complexes. We therefore examined the ability of
the mutated GST–Lb
pro
complexes to bind endogenous
eIF4GII. Figure 5C (lane 2) confirms that the wild-
type GST–Lb
pro
complex binds the endogenous eIF4-
GII. In contrast, binding of the mutant complexes to
eIF4GII is greatly reduced or almost undetectable
(Fig. 5C, lanes 3–5). Furthermore, the binding of the
mutant complexes to endogenous eIF4GII mirrors that
to eIF4GI (Fig. 5B,C). Thus, it seems likely that resi-
dues D184 and E186 are involved in recognizing both
eIF4GI and eF4GII.
Taken together, these results strongly suggest that
there is a specific interaction between the residues
K643, K646 and R650 in eIF4GI, and the amino acids
D184 and E186 in the CTE of FMDV Lb
pro
.Itis
worth noting, however, that none of the mutations
affected the ability of Lb
pro
to carry out the self-pro-
cessing reaction.
Discussion
The cleavage by viral proteinases of the host cell pro-

teins eIF4GI and eIF4GII is a central event during
picornaviral replication. We have shown recently that
this cleavage is mediated by regions distant from the
active site of two different picornaviral proteinases,
namely the papain-like Lb
pro
of FMDV and the chym-
otrypsin-like 2A
pro
of HRV2 [26,27]. Furthermore,
these regions interact on eIF4GI with amino acids that
are not identical with the cleavage site of the particular
enzyme.
In this paper, we define the binding site recognized
by the Lb
pro
of FMDV to the 13 amino acids from
residues 645–657 of eIF4GI. This sequence is separated
by 17 amino acids from the cleavage site of Lb
pro
between residues 674 and 675. In contrast, the site on
eIF4GI that is bound by HRV2 2A
pro
lies between resi-
dues 600–674 [27], with the C–terminal boundary of
this binding region only seven amino acids away from
its cleavage site. Furthermore, HRV2 2A
pro
only binds
to eIF4GI when the binding site for eIF4GI is present

and eIF4E is included in the binding assay. Although
these two parameters increase Lb
pro
binding, Lb
pro
is
still capable of binding to eIF4GI fragments in their
absence. Thus, the two picornaviral proteinases recog-
nize different sequences on eIF4GI; given their quite
different structures, this is not unexpected.
Investigation of the amino acids both within and
adjacent to the region of eIF4GI to which FMDV
Lb
pro
binds revealed three basic amino acids which
were conserved between human eIF4GI and eIF4GII,
as well as between the eIF4GI proteins of other mam-
mals. In addition, two acidic residues (D642 and
D653) also appeared to be conserved in human eIF4GI
and eIF4GII (Fig. 2A). Furthermore, both of these
aspartic acid residues are found in the seven animal
eIF4GI sequences available in the database. As it
seemed possible that conserved charged residues might
be involved in the interaction, we examined the Lb
pro
CTE sequence from amino acids 183–195. These resi-
dues had been shown previously to be important in
the recognition of eIF4GI by Lb
pro
[26,28]. In this

sequence, we indeed noted a number of acidic and
basic residues. Close investigation of the sequences
of CTEs from other FMDV serotypes revealed that
only residues D184 and E186 were present in all other
serotypes, including the more distant South African
Fig. 5. Specific mutations in the CTE of Lb
pro
reduce binding to
both eIF4GI and eIF4GII. (A) Coomassie Brilliant Blue staining of
purified GST (lane 1) and modified GST-Lb
pro
C51A fusion proteins.
(B and C), 8 lL RRL (input lanes 0, corresponding to 2 lL of RRL)
were incubated with GST (lane 1) or the different GST-Lb
pro
C51A
fusion proteins (lanes 2–5). Bound eIF4GI (B) and eIF4GII (C) were
detected by immunoblotting using anti-eIF4GI and anti-eIF4GII sera.
N. Foeger et al. FMDV L proteinase eIF4GI interaction
FEBS Journal 272 (2005) 2602–2611 ª 2005 FEBS 2607
Territories (SAT) serotypes. In addition, comparison
with the CTE sequence of ERAV L
pro
, which has been
shown to be responsible for eIF4GI cleavage [13],
showed that D184 and E186 were also present in its
CTE. In contrast, only E186 was present in the ERBV
L
pro
sequence [34]; however, this enzyme appears not

to be responsible for eIF4GI cleavage [13]. Thus, the
presence of D184 and E186 in the CTE of all L
pro
shown to be responsible for cleavage of eIF4G pro-
teins suggested to us that these residues might be
involved in interacting with the conserved basic resi-
dues of eIF4GI.
To investigate this, we individually substituted the
three conserved basic residues in eIF4GI with alanine
residues. Replacement of K646 reduced binding by
about 75% whereas replacement of R650 only reduced
binding by between 50 and 75%. The replacement of
K643 had the least effect, with binding only being
reduced by 10–20%. These results strongly suggested
an involvement of residues K646 and R650 in the
interaction with Lb
pro
. This encouraged us to investi-
gate the role of amino acids D184 and E186 in the
cleavage of eIF4GI. The substitution of D184 with
alanine or that of E186 with lysine both severely
delayed eIF4GI cleavage and led to a concomitant
decrease in binding of both eIF4GI and eIF4GII.
Interestingly, the introduction of a second basic resi-
due (arginine in place of glutamine at 185) delayed
cleavage even further, suggesting that the overall
charge of this region is important for eIF4GI recogni-
tion.
Figure 4 shows clearly, however, that eIF4GI clea-
vage still occurs when both D184 and E186 are

replaced by alanine. One reason for this is the presence
of Lb
pro
residue C133. This residue is not part of the
CTE but lies close to it in the three-dimensional struc-
ture [14]; we showed previously that replacement of
this residue affects both the binding to and cleavage of
eIF4GI [26]. However, the results here also do not rule
out further interactions between the CTE of Lb
pro
and
the region 645–657 of eIF4GI involving other sequence
motifs or hydrophobic interactions not considered
here. L188, found in the CTEs of FMDV and ERAV,
may be important in this respect.
How can mutations in the amino acids 184–186
affect eIF4GI cleavage without affecting self-process-
ing? Examination of the structure of the Lb
pro
(Fig. 6)
shows that residues D184 and E186 are at the opposite
side of the molecule from the active site and lie about
12 A
˚
from C133, a residue which has also been shown
to be important for binding eIF4GI [26]. Furthermore,
both residues protrude away from the globular domain
of the enzyme and do not appear to interact with any
residues in the globular domain or in the CTE. Indeed,
they are well positioned to interact with residues from

another protein. Thus, it seems that Lb
pro
, despite
being one of the smallest papain-like enzymes, has
been able to evolve a site which can significantly accel-
erate cleavage of a host cell molecule without reducing
the self-processing reaction. Of the two reactions, the
eIF4GI cleavage reaction appears to be more sensitive
to mutation than the self-processing reaction. This
emphasizes the importance of the interaction of Lb
pro
with the eIF4G proteins for the successful replication
of FMDV.
In summary, we have defined closely the regions on
eIF4GI and Lb
pro
that enable them to interact with
each other. A minimal binding site on eIF4GI between
residues 645 and 657 has been identified, although
binding is more efficient when eIF4G fragments con-
tain the eIF4E binding site and eIF4E is present in the
binding assay. Once again, the versatility of viral pro-
teins is amply illustrated. Although viral proteins must
remain small in order to limit genome size, they are
still able to evolve domains away from the canonical
active site which can interact with a second substrate
and contribute to the efficiency of viral replication.
Experimental procedures
Reagents
The FMDV L

pro
is the most N-terminal protein on the
FMDV polyprotein. L
pro
frees itself by cleavage between its
own C-terminus and the N-terminus of VP4. As the initi-
ation of protein synthesis on the FMDV RNA can occur at
one of two AUG codons lying 84 nucleotides apart, two
forms of L
pro
(designated Lab
pro
and Lb
pro
) are synthesized
in the infected cell. The reason for this is not clear, as both
Fig. 6. Arrangement of Lb
pro
residues involved in recognizing
eIF4GI. Stereo diagram of Lb
pro
(green, a -helices; purple, b-sheets;
yellow, coils) showing C133, D184, Q185 and E186 as balls-and-
sticks. The catalytic residues C51 (alanine in the crystal structure
[14]) and H148 are also shown. The drawing was produced using
the program
MOLSCRIPT [37,38] and rendered with RASTER3D [39].
The PDB coordinates used for Lb
pro
were 1QOL, molecule G.

FMDV L proteinase eIF4GI interaction N. Foeger et al.
2608 FEBS Journal 272 (2005) 2602–2611 ª 2005 FEBS
forms appear to have the same enzymatic properties [35].
All work described here was carried out with the Lb
pro
form.
Plasmid pCITELb
pro
VP4VP2, which encodes FMDV
amino acids 29–364 corresponding to the Lb
pro
form (29–
201), VP4 (202–286) and part of VP2 (287–364) has been des-
cribed previously [28]. Fragments of Lb
pro
to be expressed as
GST fusions were introduced as EcoRI ⁄ XhoI fragments into
the plasmid pGEX5X (Amersham Biosciences, Little Chal-
font, Buckinghamshire, UK) as required [26]. Fragments of
eIF4GI for in vitro translation were amplified from plasmid
pSKHC1, which contains the human eIF4GI cDNA from
amino acid 197–1600 [36], and cloned as EcoRI ⁄ HincII frag-
ments into pBluescriptKS (Stratagene, La Jolla, CA, USA).
Mutations were introduced into the cDNAs for Lb
pro
and
eIF4GI using standard PCR mutagenesis except for the
amino acid substitutions described in Fig. 4 which were
introduced by replacing the 36 bp BsiWI and BpuI0I frag-
ment [28] of Lb

pro
VP4VP2 with the appropriate synthetic
oligonucleotides.
The following antibodies were used. Rabbit polyclonal
antiserum raised against the N-terminus of eIF4GI (kindly
provided by R. Rhoads, Shreveport, LA, USA) was diluted
1 : 8000. Rabbit polyclonal antiserum raised against the
C-terminus of eIF4GII (kindly provided by N. Sonenberg,
Montreal, Quebec, Canada) was diluted 1 : 2000. Secon-
dary horse radish peroxidase (HRP)-conjugated antibodies
were diluted 1 : 10000 (BioRad, Hercules, CA, USA), and
second alkaline peroxidase (AP)-conjugated antibodies were
diluted 1 : 5000 (Sigma, St Louis, MO, USA).
Purification of GST fusion proteins
E. coli JM101 cells were transformed with plasmids enco-
ding the GST-Lb
pro
fusion proteins or GST alone. To
express GST-Lb
pro
, an overnight culture was diluted 1 : 10
in 50 mL medium, isopropyl thio-b-d-galactoside added to
a final concentration of 2 mm and the cells incubated at
30 °C for 3 h. The proteins were purified on glutathione-
agarose resin (Amersham Biosciences) using standard tech-
niques.
GST pull-down assays
Glutathione-sepharose beads coated with GST fusion pro-
teins were incubated in binding buffer (50 mm Tris ⁄ HCl
pH 7.4, 10 mm EDTA, 150 mm NaCl) with either an ali-

quot (8 lL) of RRL or with radiolabelled in vitro translated
proteins for 2 h at 4 °C. The amount of radiolabelled pro-
tein was adjusted so that the same amount was added in
each set of binding experiments. After three washes with
binding buffer, bound proteins were eluted by boiling in
SDS ⁄ PAGE loading buffer, resolved by SDS ⁄ PAGE and
visualized by western blotting and using the enhanced
chemiluminescence system (Pierce, Rockford, IL, USA) for
detection, or fluorography. Quantitation of binding of
radiolabelled fragments was done with a BioRad Fluor-
S
TM
MultiImager using quantity one 4.4.0 (Basic) soft-
ware.
In vitro translation
In vitro expression of radiolabelled proteins for GST pull-
down assays was performed in RRLs (Quick Coupled
Transcription ⁄ Translation system; Promega, Madison,
WI, USA) in the presence of [
35
S]methionine (20 lCi per
reaction; Hartmann Analytic, Braunschweig, Germany).
Labelled proteins were resolved by SDS ⁄ PAGE and gels
were dried and exposed to X-ray films. In vitro translations
in RRLs (Promega) to examine Lb
pro
self-processing and
eIF4GI cleavage were performed using in vitro transcribed
RNAs as described previously [15,28].
Acknowledgements

This work was supported by the Austrian Science
Foundation (grants P-16189 and P-17988) to T.S.
We thank Bob Rhoads and Nahum Sonenberg for
reagents.
References
1 Gingras AC, Raught B & Sonenberg N (1999) eIF4
initiation factors: effectors of mRNA recruitment to
ribosomes and regulators of translation. Annu Rev Bio-
chem 68, 913–963.
2 Morley SJ, Curtis PS & Pain VM (1997) eIF4G: transla-
tion’s mystery factor begins to yield its secrets. RNA 3,
1085–1104.
3 Imataka H, Sonenberg N & Olsen HS (1997) Human
eukaryotic translation initiation factor 4G (eIF4G) pos-
sesses two separate and independent binding sites for
eIF4A. Mol Cell Biol 17, 6940–6947.
4 Lamphear BJ, Kirchweger R, Skern T & Rhoads RE
(1995) Mapping of functional domains in eukaryotic
protein synthesis initiation factor 4G (eIF4G) with
picornaviral proteases – Implications for cap-dependent
and cap- independent translational initiation. J Biol
Chem 270, 21975–21983.
5 Imataka H, Gradi A & Sonenberg N (1998) A newly
identified N-terminal amino acid sequence of human
eIF4G binds poly(A)-binding protein and functions in
poly(A)-dependent translation. EMBO J 17, 7480–7489.
6 Pyronnet S, Imataka H, Gingras AC, Fukunaga R,
Hunter T & Sonenberg N (1999) Human eukaryotic
translation initiation factor 4G (eIF4G) recruits mnk1
to phosphorylate eIF4E. EMBO J 18, 270–279.

7 Scheper GC, Morrice NA, Kleijn M & Proud CG
(2001) The mitogen-activated protein kinase signal-
N. Foeger et al. FMDV L proteinase eIF4GI interaction
FEBS Journal 272 (2005) 2602–2611 ª 2005 FEBS 2609
integrating kinase Mnk2 is a eukaryotic initiation factor
4E kinase with high levels of basal activity in mamma-
lian cells. Mol Cell Biol 21, 743–754.
8 Kra
¨
usslich HG, Nicklin MJ, Toyoda H, Etchison D &
Wimmer E (1987) Poliovirus proteinase 2A induces clea-
vage of eucaryotic initiation factor 4F polypeptide p220.
J Virol 61, 2711–2718.
9 Gradi A, Svitkin YV, Imataka H & Sonenberg N
(1998) Proteolysis of human eukaryotic translation
initiation factor eIF4GII, but not eIF4GI, coincides
with the shutoff of host protein synthesis after polio-
virus infection. Proc Natl Acad Sci USA 95, 11089–
11094.
10 Jackson R, Howell M & Kaminski A (1990) The novel
mechanism of initiation of picornavirus RNA transla-
tion. Trends Biochem Sci 15, 477–483.
11 Belsham GJ & Jackson RR (2000) Translation initiation
on picornavirus RNA. In Translational Control of Gene
Expression (Sonenberg, N, Hershey, J W B & Mathews,
M B, eds), pp. 869–900. Cold Spring Harbor Press,
Cold Spring Harbor, NY.
12 Borman AM, Kirchweger R, Ziegler E, Rhoads RE,
Skern T & Kean KM (1997) elF4G and its proteolytic
cleavage products: effect on initiation of protein synth-

esis from capped, uncapped, and IRES-containing
mRNAs. RNA 3, 186–196.
13 Hinton TM, Ross-Smith N, Warner S, Belsham GJ &
Crabb BS (2002) Conservation of L and 3C proteinase
activities across distantly related aphthoviruses. J Gen-
eral Virol 83, 3111–3121.
14 Guarne
´
A, Tormo J, Kirchweger K, Pfistermueller D,
Fita I & Skern T (1998) Structure of the foot-and-
mouth disease virus leader protease: a papain-like fold
adapted for self-processing and eIF4G recognition.
EMBO J 17, 7469–7479.
15 Glaser W & Skern T (2000) Extremely efficient cleavage
of eIF4G by picornaviral proteinases L and 2A in vitro.
FEBS Lett 480, 151–155.
16 Kirchweger R, Ziegler E, Lamphear BJ, Waters D, Lie-
big HD, Sommergruber W, Sobrino F, Hohenadl C,
Blaas D, Rhoads RE & Skern T (1994) Foot-and-mouth
disease virus leader proteinase: Purification of the Lb
form and determination of its cleavage site on eIF-4
gamma. J Virol 68, 5677–5684.
17 Thomas AAM, Scheper GC, Kleijn M, Deboer M &
Voorma HO (1992) Dependence of the adenovirus tri-
partite leader on the p220 subunit of eukaryotic initia-
tion factor-4F during in vitro translation – effect of
p220 cleavage by foot-and-mouth-disease-virus L-pro-
tease on in vitro translation. Eur J Biochem 207,
471–477.
18 Belsham GJ, McInerney GM & Ross-Smith N (2000)

Foot-and-mouth disease virus 3C protease induces clea-
vage of translation initiation factors eIF4A and eIF4G
within infected cells. J Virol 74, 272–280.
19 Bovee ML, Lamphear BJ, Rhoads RE & Lloyd RE
(1998) Direct cleavage of elF4G by poliovirus 2A pro-
tease is inefficient in vitro. Virology 245, 241–249.
20 Lloyd RE, Toyoda H, Etchison D, Wimmer E & Ehren-
feld E (1986) Cleavage of the cap binding protein com-
plex polypeptide p220 is not effected by the second
poliovirus protease 2A. Virology 150, 299–303.
21 Wyckoff EE, Lloyd RE & Ehrenfeld E (1992) Relation-
ship of eukaryotic initiation factor 3 to poliovirus-
induced p220 cleavage activity. J Virol 66, 2943–2951.
22 Ohlmann T, Pain VM, Wood W, Rau M & Morley SJ
(1997) The proteolytic cleavage of eukaryotic initiation
factor (eIF) 4G is prevented by eIF4E binding protein
(PHAS-I; 4E-BP1) in the reticulocyte lysate. EMBO J
16, 844–855.
23 Haghighat A, Svitkin Y, Novoa I, Kuechler E, Skern T
& Sonenberg N (1996) The eIF4G-eIF4E complex is the
target for direct cleavage by the rhinovirus 2A protei-
nase. J Virol 70, 8444–8450.
24 Hershey PE, McWhirter SM, Gross JD, Wagner G,
Alber T & Sachs AB (1999) The Cap-binding protein
eIF4E promotes folding of a functional domain of yeast
translation initiation factor eIF4G1. J Biol Chem 274,
21297–21304.
25 Gross JD, Moerke NJ, von der Haar T, Lugovskoy
AA, Sachs AB, McCarthy JE & Wagner G (2003) Ribo-
some loading onto the mRNA cap is driven by confor-

mational coupling between eIF4G and eIF4E. Cell 115,
739–750.
26 Foeger N, Glaser W & Skern T (2002) Recognition of
eukaryotic initiation factor 4G isoforms by picornaviral
proteinases. J Biol Chem 277, 44300–44309.
27 Foeger N, Schmid EM & Skern T (2003) Human rhino-
virus 2, 2A
pro
recognition of eukaryotic initiation factor
4GI: Involvement of an exosite. J Biol Chem 278,
33200–33207.
28 Glaser W, Cencic R & Skern T (2001) Foot-and-mouth
disease Leader proteinase: involvement of C-terminal
residues in self-processing and cleavage of eIF4GI.
J Biol Chem 276, 35473–35481.
29 George M, Venkataramanan R, Gurumurthy CB &
Hemadri D (2001) The non-structural leader protein
gene of foot-and-mouth disease virus is highly variable
between serotypes. Virus Genes 22, 271–278.
30 van Rensburg H, Haydon D, Joubert F, Bastos A,
Heath L & Nel L (2002) Genetic heterogeneity in the
foot-and-mouth disease virus Leader and 3C protei-
nases. Gene 289, 19–29.
31 Etchison D, Milburn SC, Edery I, Sonenberg N &
Hershey JWB (1982) Inhibition of HeLa cell protein
synthesis following poliovirus infection correlates with
the proteolysis of a 220,000-dalton polypeptide asso-
ciated with eucaryotic initiation factor 3 and a cap
binding protein complex. J Biol Chem 257, 14806–
14810.

FMDV L proteinase eIF4GI interaction N. Foeger et al.
2610 FEBS Journal 272 (2005) 2602–2611 ª 2005 FEBS
32 Byrd MP, Zamora M & Lloyd RE (2002) Generation of
multiple isoforms of eukaryotic translation initiation
factor 4GI by use of alternate translation initiation
codons. Mol Cell Biol 22, 4499–4511.
33 Bradley CA, Padovan JC, Thompson TL, Benoit CA,
Chait BT & Rhoads RE (2002) Mass spectrometric ana-
lysis of the N terminus of translational initiation factor
eIF4G-1 reveals novel isoforms. J Biol Chem 277,
12559–12571.
34 Skern T, Fita I & Guarne A (1998) A structural model
of picornavirus leader proteinases based on papain and
bleomycin hydrolase. J General Virol 79, 301–307.
35 Cao X, Bergmann IE, Fullkrug R & Beck E (1995)
Functional analysis of the two alternative translation
initiation sites of foot-and-mouth disease virus. J Virol
69, 560–563.
36 Yan RQ, Rychlik W, Etchison D & Rhoads RE (1992)
Amino acid sequence of the human protein synthesis
initiation factor-eIF-4gamma. J Biol Chem 267, 23226–
23231.
37 Kraulis PJ (1991) MOLSCRIPT: a program to produce
both detailed and schematic plots of protein structures.
J Appl Crystallogr 24 , 946–950.
38 Esnouf R (1997) An extensively modified version of
Molscript that includes greatly enhanced colouring
capacities. J Mol Graphics 15, 133–138.
39 Merrit E & Murphy M (1994) Raster3d, Version 2.0. A
program for photorealistic molecular graphics. Acta

Crystallogr D50, 869–873.
N. Foeger et al. FMDV L proteinase eIF4GI interaction
FEBS Journal 272 (2005) 2602–2611 ª 2005 FEBS 2611