The potyviral virus genome-linked protein VPg forms a
ternary complex with the eukaryotic initiation factors
eIF4E and eIF4G and reduces eIF4E affinity for a mRNA
cap analogue
Thierry Michon, Yannick Estevez, Jocelyne Walter, Sylvie German-Retana and Olivier Le Gall
Interactions Plante-Virus, UMR GDPP INRA-Bordeaux 2, Institut de Biologie Ve
´ge
´tale Mole
´culaire, Villenave d’Ornon, France
Lettuce mosaic virus (LMV), is a member of the genus
Potyvirus [1]. Potyviruses are plant viruses with flexible
rod-shaped particles packing a single-stranded, poly-
adenylated, positive-sense genomic RNA [2]. This
RNA, of about 10 kb, is linked at its 5¢end to a viral
protein, VPg (virus protein linked to the genome),
through a tyrosine phosphoester covalent bound [3–5].
The viral particle does not carry the molecular machin-
ery required for its replication in the host cell, and the
viral genome codes for a limited number of proteins.
Thus the infectious cycle needs to recruit various host
factors, including the host translation apparatus.
Eukaryotic mRNAs are capped by the addition of a
7-methylated guanine (m
7
G). This post-transcriptional
modification occurs in the nucleus [6,7]. The 5¢-cap
acts as a flag on the mRNA for cytoplasmic export
and for the ribosomal translation complex assembly.
The recognition of this m
7
G functional group by a
cap-binding protein (eIF4E, the eukaryotic translation
initiation factor 4E, or its isoform eIFiso4E) is the first
step of a complex cascade of molecular events leading
to binding of the 40S ribosomal subunit to the mRNA
[8]. The general structural similarity between the poty-
virus RNA and eukaryotic mRNAs suggests that VPg
may act functionally as a cap-like structure. This hypo-
thesis gained strength when a specific interaction
between eIF4E and VPg was identified in several patho-
systems, such as tomato ⁄tobacco etch virus [9] and
Arabidopsis thaliana ⁄turnip mosaic virus (TuMV) [10].
In the latter case, a single amino acid replacement in
Keywords
eIF4E; eIF4G; fluorescence; interaction; VPg
Correspondence
T. Michon, Virologie Ve
´ge
´tale, GDPP,
IBVM-INRA, BP 81, 33883 Villenave
d’Ornon Cedex, France
Fax: +33 5 57 12 23 84
Tel: +33 5 57 12 23 91
E-mail: michon@bordeaux.inra.fr
Website: http://www.bordeaux.inra.fr/ipv/
(Received 10 October 2005, revised 23
January 2006, accepted 26 January 2006)
doi:10.1111/j.1742-4658.2006.05156.x
The virus protein linked to the genome (VPg) of plant potyviruses is a
25-kDa protein covalently attached to the genomic RNA 5¢end. It was
previously reported that VPg binds specifically to eIF4E, the mRNAcap-
binding protein of the eukaryotic translation initiation complex. We per-
formed a spectroscopic study of the interactions between lettuce eIF4E and
VPg from lettuce mosaic virus (LMV). The cap analogue m
7
GDP and VPg
bind to eIF4E at two distinct sites with similar affinity (K
d
¼0.3 lm). A
deeper examination of the interaction pathway showed that the binding of
one ligand induces a decrease in the affinity for the other by a factor of 15.
GST pull-down experiments from plant extracts revealed that VPg can
specifically trap eIF4G, the central component of the complex required for
the initiation of protein translation. Our data suggest that eIF4G recruit-
ment by VPg is indirectly mediated through VPg–eIF4E association. The
strength of interaction between eIF4E and pep4G, the eIF4E-binding
domain on eIF4G, was increased significantly by VPg. Taken together
these quantitative data show that VPg is an efficient modulator of eIF4E
biochemical functions.
Abbreviations
BN, blue native; eIF4E, eukaryotic translation initiation factor 4E; GST, glutathione S-transferase; LMV, lettuce mosaic virus; PABP, poly(A)-
binding protein; TuMV, turnip mosaic virus; VPg, virus protein linked to the genome.
1312 FEBS Journal 273 (2006) 1312–1322 ª2006 The Authors Journal compilation ª2006 FEBS
VPg abolished its interaction with eIF4E, and this
correlated with a defect in infectivity in Brassica
perviridis [11].
The simplest hypothesis is that VPg recruits eIF4E
to initiate the translation of the potyvirus polyprotein
[12]. The involvement of eIF4E in pea seed-borne
mosaic virus cell to cell movement has been suggested
[13]. Some lettuce cultivars display a recessive resist-
ance to LMV. As this resistance is opposed by the
eIF4E isoform, we focused this study on the VPg–
eIF4E interaction [14]. Whatever the biochemical pro-
cess involved, the VPg–eIF4E complex appears to play
a crucial role in the outcome of the plant–potyvirus
interaction. Therefore, we developed a quantitative test
to (a) measure precisely the binding strength between
VPg and eIF4E and (b) assess the pathway of the
interactions between VPg, eIF4E and the cap structure.
A glutathione S-transferase (GST) pull-down test from
plant extracts was used to demonstrate that VPg can
recruit the binary complex eIF4E–eIF4G, a component
of eIF4F, the complex of translation initiation. Finally,
we evaluated the possible effect of VPg on eIF4G
binding to eIF4E using a synthetic peptide that
mimicks the eIF4G-binding domain.
Results and discussion
Interaction between VPg and eIF4E
We initially evaluated the perturbation of the intrinsic
fluorescence of eIF4E upon its interaction with VPg.
As VPg contains no tryptophan, the contribution of its
intrinsic fluorescence was assumed to be negligible with
respect to the nine tryptophans present in eIF4E.
Upon the addition of VPg, a decrease in the overall
fluorescence of eIF4E was observed (Fig. 1). This
seemed to correlate with a specific interaction between
the two proteins, as the fluorescence decrease reached
a plateau at high VPg concentrations. The binding of
VPg only slightly affected eIF4E tryptophan fluores-
cence. The signal-to-noise ratio was poor, implying
low accuracy in the determination of the dissociation
constant (K
d
¼0.3 lm). The binding curve extrapola-
ted to a 1 : 1 stoichiometry for the two proteins
(Fig. 1 inset).
From these data it is likely that VPg binding to
eIF4E is not associated with much modification of the
environment of the eIF4E tryptophans. The two tryp-
tophan residues present in the cap-binding site are
accessible to the surface [15]. Direct access of VPg to
the pocket would probably be associated with greater
modification of the fluorescence of these two residues,
as happens when cap analogues penetrate into the
pocket [16]. The emission maximum of lettuce eIF4E
was found to be at 342 nm, which is rather high. It is
likely that the major contribution to the intrinsic fluor-
escence of the protein comes from these two trypto-
phans in a polar environment, the fluorescence of the
others being partially shielded. This was also found in
previous studies, although not to such an extent [16].
Binding-site characterization
To obtain a better fluorimetric response of eIF4E upon
its interaction with VPg, we attempted to follow indi-
rectly the complex formation using the cap analogue
m
7
GDP. There are considerable discrepancies between
the affinities of the cap analogue published so far. In the
absence of VPg, the value of the dissociation constant
obtained (K
c
¼0.31 ± 0.02 lm) was significantly lower
than in previous reports for this type of analogue
[17,18]. A possible explanation is that, during the proce-
dure used for eIF4E isolation, the elution step from
m
7
GTP–Sepharose 4B is usually performed with free
m
7
GTP. We observed that, even after extensive dialysis,
a significant fraction of active eIF4E retains m
7
GTP
in its binding site (data not shown). In our study, all
eIF4E fractions were previously eluted from m
7
GTP–
Sepharose 4B with 1 mKCl instead of m
7
GTP (see
Experimental Procedures). However, it cannot be exclu-
ded that lettuce eIF4E displays a higher affinity for the
Fig. 1. Fluorescence emission spectra of eIF4E upon VPg addition.
Aliquots of 2.5 lL from a 60-lMVPg stock solution were added
to a 0.5-lMeIF4E solution in buffer M. After each addition, the
mixture was incubated for 5 min at 25 C, and spectra were recor-
ded. (- - - -) No VPg added; (—) from top to bottom increasing
amounts of VPg. Inset: variation in eIF4E fluorescence as a function
of VPg concentration in the medium.
T. Michon et al.Modulation of plant eIF4E properties by a potyvirus VPg
FEBS Journal 273 (2006) 1312–1322 ª2006 The Authors Journal compilation ª2006 FEBS 1313
cap than its wheatgerm counterpart. It is worth men-
tioning that, in a recent study, values in the nanomolar
range were reported. The authors emphasize that this
could be because the recombinant eIF4E used was
obtained from carefully renatured batches from inclu-
sion bodies [19], thus avoiding affinity purification
involving exposure to cap analogues.
The apparent dissociation constant of m
7
GDP to
eIF4E was higher in the presence of increasing
amounts of VPg (Fig. 2). The plateau value (ligand
saturation) was also affected (Fig. 2, inset). It was sus-
pected that VPg could remain bound to eIF4E even at
high cap analogue concentrations. The fact that
m
7
GDP could not displace VPg argued in favour of
the presence of a VPg-binding site on eIF4E, which is
distinct but structurally related to the cap-binding site.
A possible mixed-type noncompetitive ligand binding
of eIF4E and the cap to the VPg was reported previ-
ously [11]. However, as the amount of free and bound
ligand was not determined, a nonlinear double-recipro-
cal plot was obtained, from which it was difficult to
establish an accurate pathway for the interactions [20].
Complexes between proteins often involve large sur-
face overlaps. It cannot be ruled out that VPg interacts
with eIF4E through several domains, one of them
being structurally linked to the cap site. In such a case,
the presence of VPg might negatively affect the binding
of m
7
GDP. To discriminate between competitive and
noncompetitive interactions, we performed a m
7
GDP–
eIF4E binding test in the presence of a large excess of
VPg (30–60 lm) with respect to eIF4E (2 lm). In such
conditions, the concentration of free VPg ([V]
free
)is
assumed to be close to [V]
total
. The same saturation
behaviour was observed as for lower concentrations of
VPg (Fig. 3A). In the case of strict competition, the
Fig. 2. Effect of VPg on m
7
GDP binding to eIF4E at 25 C. VPg
aliquots were mixed with 2 lMeIF4E in buffer M Before titration
with m
7
GDP. Inset: isotherms of m
7
GDP binding to eIF4E. Solid
lines represent theoretical data calculated using the best fit of the
experimental data to eqn (4) (see Experimental Procedures). (d)No
VPg; (s)0.3lMVPg; (n)1.2lMVPg; (h)2.4lMVPg.
Fig. 3. (A) Binding isotherms of m
7
GDP with eIF4E in the presence
of high VPg concentrations. Experimental conditions were as in
Fig. 2. Lines represent theoretical data calculated using the best fit
of the experimental data to eqn (4) using either two distinct but
dependent sites (solid line) or strict competition at the same site
(dashed line). All measurements were made at least in triplicate.
(d)30lMVPg; (s)60lMVPg. (B) Plots of residuals between the-
oretical and experimental data using either two distinct but depend-
ent sites (solid line) or strict competition at the same site (dashed
line). VPg concentration, 30 lM.
Modulation of plant eIF4E properties by a potyvirus VPg T. Michon et al.
1314 FEBS Journal 273 (2006) 1312–1322 ª2006 The Authors Journal compilation ª2006 FEBS
apparent dissociation constant Kapp
ccan be derived
according to Scheme 1A as:
Kapp
c¼Kc1þ½Vfree
Kv
 ð1Þ
where [V]
free
is the concentration of free VPg in the
medium.
The apparent dissociation constant for binding at
two distinct but dependent sites defined in Scheme 1B
is:
Kapp
c¼Kc
Kvþ½Vfree
Kvþ½Vfree
a
! ð2Þ
where aand K
V
are the ratio of symmetrical dissoci-
ation constants (Scheme 1B) and the VPg–eIF4E disso-
ciation constant, respectively.
In this case, as eIF4E remains partitioned between
the three species eIF4E–m
7
GDP, m
7
GDP–eIF4E–VPg,
and eIF4E–VPg whatever the concentration of
m
7
GDP, the plateau value is also affected and:
Yapp
sat ¼Ysat
Kv
aKvþ½Vfree
ð3Þ
K
c
and Y
sat
were replaced in eqn (4) by the expressions
Kapp
cand Yapp
sat . Assuming [V]
free
to be close to [V]
tot
,
data sets of eIF4E fluorescence as a function of
[m
7
GDP]
total
were fitted to models mimicking either
strictly competitive binding or binding at two distinct
but dependent sites. Plots of residuals obtained from
both fits showed unambiguously that VPg and m
7
GDP
bind to distinct but interdependent sites (Fig. 3B).
Although the cap analogue and VPg displayed com-
parable affinity for eIF4E (K
c
¼0.31 ± 0.02 lmand
K
v
¼0.3 ± 0.03 lm), binding of the first molecule
affected the binding of the second one by a factor 15
(a¼15 ± 3). We examined the effect of disruption of
the cap-binding capacity on eIF4E association with
VPg. To do so, we engineered W123A, an eIF4E
mutant in which W123, one of the two conserved tryp-
tophan residues involved in p-pstacking with the cap
aromatic moiety [15], was substituted with an alanine.
As expected, this substitution abolished m
7
GDP bind-
ing to W123A while retaining its capacity to associate
with VPg (Table 1). The VPg surface defining the zone
of interaction with eIF4E spans at least two distinct
domains, one of which can affect the topology of the
cap-binding pocket. The VPg–eIF4E interaction may
be a mechanism for recruiting the host translation
machinery cut off by several unrelated positive-stran-
ded RNA viruses. In a recent study, an interaction
between VPg and a human enteric calcivirus was dem-
onstrated [21].
GST pull-down assay of plant proteins forming
complexes with VPg
Several studies have reported specific interactions
between the VPg from potyvirus and eIF4E or one of
its isoforms [9]. A recombinant VPgÆGST fusion pro-
tein was used as a bait for trapping molecular species
susceptible to be recruited by VPg in planta. A soluble
protein fraction was recovered after mild detergent
treatment of the plant leaves. This extract was submit-
ted to a VPgÆGST pull-down procedure [22]. To deter-
mine the nature of the protein complexes involved in
specific interactions with VPg, the fraction recovered
was analysed by electrophoresis in native conditions. A
high-molecular-mass (above 200 kDa) species and two
minor species (below 100 kDa) were affinity-purified
from the protein extract (Fig. 4A, lane 2). A western
blot analysis with antibodies to VPg showed that all
Table 1. Equilibrium association constants for various forms of
eIF4E and its ligands. NB, no binding detected.
10
)6
·K
a
(M
)1
) VPg pep4G m7GDP
eIF4E 3.3 ± 0.3 4.1 ± 0.7 3.2 ± 0.3
eIF4E-VPg – 12.1 ± 1.5 0.2 ± 0.07
eIF4E-pep4G 2.9 ± 0.2 – 8.1 ± 0.9
W123A 1.5 ± 0.8 3.9 ± 0.4 NB
Kc
eIF4E eIF4E-m7GDP
VPg
eIF4E-VPg
Kv
m7GDP
A
VPg-eIF4E-m7GDP
α
Kv
α
Kc
B
VPg
Kc
eIF4E eIF4E-m7GDP
m7GDP
VPg
eIF4E-VPg
Kv
m7GDP
Scheme 1. Two plausible models for the interactions between
eIF4E, VPg and the cap analogue
7
mGDP. (A) Strictly competitive
model; (B) binding at two distinct but dependent sites.
T. Michon et al.Modulation of plant eIF4E properties by a potyvirus VPg
FEBS Journal 273 (2006) 1312–1322 ª2006 The Authors Journal compilation ª2006 FEBS 1315
contained this protein (Fig. 4A, lane 3). As the inter-
action of VPg with plant translation initiation factors
was suspected, a western blot was performed with spe-
cific antibodies to eIF4E (lane 4) and eIF4G (lane 5).
Two of the three species (the one >205 kDa and the
intermediate one) contained eIF4E (lane 4). The eIF4G
forms were restricted to the high-molecular-mass spe-
cies (lane 5). When pull-down assays were performed
on extracts from infected leaves, the highest-molecular-
mass species was not detected (lane 6).
The purified fraction was also analysed by
SDS ⁄PAGE to determine the composition of the com-
plexes. They mainly contained three groups of poly-
peptide chains according to their relative molecular
masses: a group with bands in the range of 180 kDa, a
54-kDa band, and a single chain of 26 kDa (Fig. 4B,
lane 4). The proteins were identified by western blot-
ting. Antibodies to eIF4G revealed many polypeptides
in the 70–200 kDa region (Fig. 4B, lane 5). It has pre-
viously been reported that, in several organisms,
eIF4G is highly susceptible to proteolysis [23]. It is
likely that, during the extraction step, cleavage
occurred along the polypeptide chain. Whereas blue
native (BN)-PAGE native conditions revealed a single
molecular species probably because of conformational
locking, SDS ⁄PAGE denaturing conditions revealed
the proteolysis.
The VPg antibodies reacted with a single 54-kDa
polypeptide corresponding to the GSTÆVPg fusion
(Fig. 4B, lane 6). The eIF4E antibodies revealed only a
26-kDa polypeptide in accordance with the expected
plant eIF4E (Fig. 4B, lane 7). Our GST pull-down
experiment showed that VPg can recruit eIF4E
(26 kDa [14]) and eIF4G (180 kDa according to
A. Thaliana eIF4G [24]). From the SDS ⁄PAGE
pattern we could determine precisely the nature of the
three complexes revealed by BN-PAGE analysis.
The highest-molecular-mass band (>205 kDa) corres-
ponds to the heterotrimer eIF4G–eIF4E–VPgÆGST
(260 kDa; Fig. 4A, compare lanes 3, 4 and 5).
The intermediate band (80 kDa) observed in native
conditions contained VPg and eIF4E but not eIF4G
(Fig. 4A, compare lanes 3 and 4 with 5). It was
attributed to the binary complex GSTÆVPg–eIF4E.
The band of lowest molecular mass, 54 kDa, on
BN-PAGE was unambiguously identified as the
GSTÆVPg fusion on SDS ⁄PAGE (Fig. 4B, lane 6; see
also lane 2 for comparison with pure GSTÆVPg).
We demonstrate here that the VPg–eIF4E interac-
tion previously described in other pathosystems such
as TuMV ⁄B. perviridis is also found in the LMV ⁄let-
tuce system. However, this is the first report of a phys-
ical interaction between VPg and eIF4G. At this stage,
we found no evidence for a direct interaction between
VPg and eIF4G. The recruitment of eIF4G by VPg is
probably indirectly mediated by eIF4E as the central
element of the complex. This should display at least
two distinct binding sites, one for VPg and the other
for eIF4G. In control experiments, the protein extract
was incubated with unfused GST before the affinity
chromatography. GST alone failed to pull down any
A
B
Fig. 4. Analysis of the plant soluble proteins complexed with VPg.
(A) BN-PAGE of the protein fraction retained on glutathione–Seph-
arose 4b. Lane 2, the plant soluble extract was incubated with
recombinant VPgÆGst fusion and mixed with glutathione–Seph-
arose 4b beads. After extensive washing, the proteins specifically
retained on the resin were eluted with glutathione (see Experimen-
tal procedures for details). The fractions obtained were loaded on
to a 6–18% polyacrylamide slab gel. After migration, the complexes
were Coomassie stained. Proteins from lane 2 were transferred to
a nitrocellulose membrane. Complexes were revealed with poly-
clonal antibodies raised against VPg (lane 3), eIF4E (lane 4) and
eIF4G (lane 5). Lane 6, same as lane 2 except that protein extracts
were from LMV-Infected plants (40 lg VPgÆGst added). (B)
Sds ⁄PAGE of the proteins retained on glutathione–Sepharose 4b.
Affinity chromatography was as in (A). Lane 4, electrophoretic pat-
tern of the protein fraction under denaturing conditions; Coomassie
blue staining. Lanes 5, 6 and 7: Western blot analysis of the pro-
teins retained using polyclonal antibodies raised against eIF4G, VPg
and eIF4E, respectively. Lane 1, molecular mass markers. Lane 2:
recombinant GSTÆVPg fusion extracted from E. coli and affinity-
purified on glutathione–Sepharose 4b. Lane 3, as a control, the plant
soluble extract was incubated with nonfused recombinant GST
and mixed with glutathione–Sepharose 4b. After being washed,
the fraction eluted with glutathione was analysed By SDS ⁄PAGE.
Modulation of plant eIF4E properties by a potyvirus VPg T. Michon et al.
1316 FEBS Journal 273 (2006) 1312–1322 ª2006 The Authors Journal compilation ª2006 FEBS