Content uploaded by Jonathan M Grimes
Author content
All content in this area was uploaded by Jonathan M Grimes on Jun 18, 2014
Content may be subject to copyright.
Bluetongue virus VP4 is an
RNA-capping assembly line
Geoff Sutton
1
, Jonathan M Grimes
1
, David I Stuart
1
& Polly Roy
2
Eukaryotic organisms cap the 5¢ ends of their messenger RNAs
by a series of four chemical reactions. Some viruses achieve
this using a single molecule; the crystal structure of such an
enzyme from bluetongue virus reveals an elongated modular
architecture that provides a scaffold for an assemblage of
active sites, two contributed by a domain of novel structure.
The ‘cap’ of eukaryotic mRNAs (methylguanosine connected to the
first nucleoside of the transcript)
1,2
stabilizes the message
1,3
and allows
efficient translation
4
. Many eukaryotic viruses also cap their tran-
scripts, by strategies including encoding their own enzymes. Double-
stranded RNA (dsRNA) viruses use this approach to cap viral mRNA
molecules before release from the transcribing viral core
1,5
. The cap
is made in four steps, and in bluetongue virus (BTV), a dsRNA
orbivirus of the family Reoviridae, all four reactions are catalyzed by a
single protein, VP4 (refs. 6,7). The stepwise process proceeds as follows
(Fig. 1a): hydrolysis of the 5¢-triphosphate to a diphosphate by an
RNA 5¢-triphosphatase (RTPase); addition of GMP via a 5¢-5¢ tripho-
sphate linkage using a guanylyltransferase (GTase); and transfer of a
methyl group to the N7 position by a (guanine-N(7)-)-methyltransfer-
ase (N7MTase) to give ‘cap 0’. Subsequent methylation to form a ‘cap 1’
structure occurs on the 2¢-hydroxyl of the ribose of the first nucleotide,
catalyzed by a (nucleoside-2¢-O-)-methyltransferase (2¢OMTase). The
methyltransferases utilize S-adenosyl-
L-methionine (AdoMet) as the
methyl donor, generating S-adenosyl-
L-homocysteine (AdoHcy). Here
we describe the structure of BTV VP4, assign catalytic activities to the
domains and suggest how capping may be efficiently coordinated.
BTV-10 VP4 was expressed in insect cells and the structure solved
at 2.5-A
˚
resolution using selenomethionine MAD phasing (Supple-
mentary Methods online). The refined model includes 613 of the
644 residues (the N-terminal methionine is missing, and dis-
ordered residues 270–276, 537–549 and 601–610 are omitted) with
good stereochemistry (98% of residues lie in allowed regions of the
Ramachandran plot; Supplementary Table 1 online). The VP4
monomer resembles an eccentric egg-timer, of dimensions 96 A
˚
66 A
˚
52 A
˚
, composed of four domains named according to their
structure and inferred function (see below, Fig. 1b–d and Supple-
mentary Figs. 1 and 2 online). The N-terminal 108 residues form the
KL domain, followed by a 30-residue a-helix and the N-terminal
portion of the N7MT domain. The chain then forms the 2¢OMT
domain, completes the N7MT domain and finally forms the GT
domain. A crystallographic two-fold axis forms a potential dimer
(Supplementary Fig. 3 online), whose 2,130-A
˚
2
interface surface is
contributed by amino acid residues among the N-terminal 114 and C-
terminal 230 residues, in agreement with biochemical data for a dimer
in solution
8
. The domains are detailed below, beginning with the
2¢OMT domain (identification of which anchors the interpretation
of the others).
GTP
m
7
Gpppm
2
Gp-RNA (cap 1)
m
7
GpppGp-RNA (cap 0)
AdoHcy
AdoMet
AdoHcy
AdoMet
PP
i
P
i
GpppGp-RNA
ppGp-RNA
(Nucleoside-2′-O-)-
methyltransferase
(Guanine-N(7))-
methyltransferase
Guanylyl-
transferase
RNA
triphosphatase
1,160 Å
2
1,630 Å
2
924 Å
2
715 Å
2
pppGp-RNA
b
d
ca
Figure 1 Overall structure of VP4. (a) Reaction mechanism for a
type 1 cap. (b) VP4 domain organization (domain names defined in text).
Poorly ordered loops are gray. Contact areas between domains are
indicated. (c) VP4 colored by domain, with active sites and ligands colored
as in a; view rotated 1201 from b. AdoHcy in N7MTase and 2¢OMTase are
azure and orange, respectively. Cap analog is yellow. (d) Active sites mapped
onto electrostatic surface representation of VP4 (colored as in c), and VP4
within BTV core.
Received 17 July 2006; accepted 5 March 2007; published online 8 April 2007; doi:10.1038/nsmb1225
1
Division of Structural Biology, The Wellcome Trust Centre for Human Genetics, University of Oxford, Roosevelt Drive, Headington, Oxford, OX3 7BN, UK.
2
London School of Hygiene and Tropical Medicine, University of London, Keppel Street, London, WC1E 7HT, UK. Correspondence should be addressed to D.I.S.
(dave@strubi.ox.ac.uk) or P.R. (polly.roy@lshtm.ac.uk).
NATURE STRUCTURAL & MOLECULAR BIOLOGY VOLUME 14 NUMBER 5 MAY 2007 449
BRIEF COMMUNICATIONS
© 2007 Nature Publishing Group
http://www.nature.com/nsmb
Residues 175–377 form a seven-stranded b-sheet surrounded by
parallel a-helices, similar to the catalytic domains of class I AdoMet-
dependent methyltransferases (Fig. 2a). Superpositions reveal strong
similarity to RNA 2¢OMTases, in particular vaccinia virus VP39 (r.m.s.
deviation 2.5 A
˚
; 26 of the 170 superposed residues are identical; see
Supplementary Fig. 4 online). The positions of ligands and products
were determined by soaks and cocrystallization (Fig. 2a and electron
density maps in Supplementary Fig. 4). The AdoMet methyl donor
occupies a pocket that largely secludes it from the solvent. Many
residues identified as AdoMet or AdoHcy ligands in VP39 are
structurally conserved in BTV and other orbiviruses. These include
Met241, Val242 and Val266, which define a pocket for the adenine
ring; Gly197 and Asp224, which form hydrogen bonds with the ribose
O2¢ and O3¢; Tyr195 and Asp265, which coordinate the amine
nitrogen of the AdoHcy; and Thr203, which coordinates its carbox-
ylate group. The guanine ring of the cap analog P
1
,P
3
-diguanosine-5¢-
triphosphate (G5¢ppp5¢G) is stacked between Tyr334 (aromatic in all
orbiviruses) and Asn311, and the phosphate moieties are stabilized by
salt bridges with the conserved Arg282 and Arg308. Binding of the cap
and AdoHcy is similar to that observed for VP39 (Fig. 2a). A Lys-Asp-
Lys-Glu (KDKE) tetrad in the active site seems to be characteristic of
RNA 2¢OMTases
9
, with the second lysine directly affecting catalysis
10
.
The VP4 Lys178-Asp265-Lys306-Glu335 tetrad superposes well on the
equivalent residues of VP39 (Fig. 2a), with Lys306 appropriately
aligned with the AdoHcy and the 2¢-O-hydroxyl group of
G5¢ppp5¢G. The 2¢-O-hydroxyl group is 3.6 A
˚
from the methyl
donor (B7.0 A
˚
for the N7). This domain is therefore a 2¢OMTase,
with a catalytic mechanism similar to VP39 (ref. 10).
The N7MT domain also has a class I AdoMet-dependent methyl-
transferase fold (Fig. 2b). Structural similarity to 2¢OMTases is
confined to the b-sheet (the KDKE catalytic tetrad is absent). It is
therefore logical to assign this domain as the N7MTase; indeed, it
is most similar to the Ecm1 mRNA cap N7MTase
11
(3.9 A
˚
r.m.s.
deviation over 161 residues, with nine identical residues; Supplemen-
tary Fig. 5 online). AdoHcy binds in a fissure that runs across the face
of the domain (Fig. 2b) and harbors several residues conserved across
orbiviruses (Supplementary Fig. 5). These include residues 134–139
and 409–414, which define the major binding surface; Tyr411, which
stacks with the adenine ring (analogous to Ecm1 Tyr124); Arg122,
which interacts with the carboxylate (analogous to Ecm1 Lys54); and
Asn417 and Asn396, which interact with nitrogens on the adenine.
a
b
Figure 2 The structure of the methyltransferase domains and ligands. (a) Left, 2¢OMT domain colored from blue (N terminus) to red (C terminus).
Poorly ordered loop is gray. AdoHcy (magenta) and G5¢ppp5¢G (blue) substrates are shown. Right, 2¢OMTase active sites of VP4 (purple) and VP39 (cyan).
Carbon atoms and bonds of significant residues are colored gray. VP4 ligands colored as at left; VP39 AdoHcy and m
7
G5¢ppp5¢G are marine and orange,
respectively. Red and yellow spheres represent cap analog 2¢-O and AdoHcy sulfur, respectively. (b) N7MT domain and AdoHcy, as in a. Right, stereo view
of N7MTase active sites of VP4 and Ecm1. Carbon atoms and bonds of VP4 are gray, with guanine bases purple and AdoHcy pink. Ecm1 is in olive, with
cap analog orange (blue sphere, N7) and AdoHcy lime (yellow spheres, sulfurs).
ac
Size
(kDa)
M
w
1
205
116
84
55
24
97
66
45
36
29
6.5
20
14
2345 1
Lane
2345
db
Figure 3 Structure of the N- and C-terminal domains. (a,b) N-terminal
KL domain (a) and C-terminal GT domain (b) colored as in Figure 2a.
(c) Coomassie-stained gel and radiograph of VP4 radiolabeled with
[a-
32
P]GTP and subjected to proteolytic digestion. Lanes 1–3, digested
with trypsin for 2, 30 and 120 min, respectively; lane 4, digested with
endoproteinase Asp-N for 120 min; lane 5, incubated for 120 min with no
protease. (d) Surface depression showing conserved histidine and cysteine
residues. Arrow in b indicates view. Arrows in d indicate direction of labeled
a-helices. Residues conserved in orbiviruses are shown in red; His554 and
Cys518 are in blue and yellow, respectively.
BRIEF COMMUNICATIONS
450 VOLUME 14 NUMBER 5 MAY 2007 NATURE STRUCTURAL & MOLECULAR BIOLOGY
© 2007 Nature Publishing Group
http://www.nature.com/nsmb
The conserved loop 134–136 locks the AdoHcy in place, aided by
specific interactions with a carbonyl oxygen (Gly134) and water
molecule, which interact with the NH
3
group of AdoHcy, recapitulat-
ing interactions seen in Ecm1. Cocrystallization and soaking experi-
ments did not locate further ligands in the N7MT active site. However,
the AdoHcy map has additional electron density above the aromatic
rings of Phe387 and Phe410. Overlaying the Ecm1–AdoHcy–cap
complex on VP4 places the G0 base (the base added by GTase) in
the electron density above Phe410 (Fig. 2b). Several conserved
residues form a pocket for G0 analogous to that in Ecm1 (ref. 11),
with Phe410 in VP4 fulfilling the role of Phe141 in Ecm1 and VP4
Arg122 positioned to coordinate the ribose hydroxyl (analogous to
Ecm1 Lys54). The density observed above Phe387 may correspond to
the G1 base of the cap (Fig. 2b). The structure of the VP4 N7MT
domain supports the suggestion
11
that N7MTase activity proceeds by
proximity-facilitated catalysis.
The N-terminal 108 residues of VP4 form an a/b-domain with a
kinase fold (the KL domain, Fig. 3a). However, this domain lacks the
catalytic apparatus and essential P-loop of kinases (see Supplementary
Fig. 6 online) and is the least conserved domain of VP4, with 13%
identical residues (N7MT, 22%; 2¢OMT, 19%; GT, 27%). This domain
is therefore unlikely to have catalytic activity. An alternative function
might be in protein-protein interactions. The domain is well posi-
tioned for such a role, being on the side of the otherwise linear
molecule. Noncatalytic kinase-like domains, in particular the guany-
late kinase domain of membrane-associated guanylate kinases, act as
protein-binding domains in higher eukaryotes
12
. Likely binding part-
ners would be the viral polymerase or VP3
13
, which forms the inner
layer of the core.
The C-terminal 135 residues of VP4 (the GT domain) form a
compact stack of six a-helices (five antiparallel to one another;
Fig. 3b) atop the N7MT domain. These close interactions stabilize
the connecting loops (Fig. 1b). The GT and KL domains abut each
other, with interactions involving loops and the antiparallel GT helix.
The helical stack is crowned by poorly ordered loops. The GT domain
is the most strongly conserved domain (the N-terminal portion shows
36% identity; see Supplementary Fig. 6); however, no similar struc-
tures could be found in the PDB. By process of elimination, this
domain might have RTPase or GTase activity, or both. GTase activity
is usually achieved via a covalent phosphoramidate bond between
GMP and a protein side chain—either lysine, arginine or histidine. To
obtain biochemical evidence for such a mechanism, VP4 was incu-
bated with [a-
32
P]GTP to allow formation of the covalent adduct and
then subjected to controlled proteolysis (Fig. 3c). N-terminal sequen-
cing indicated that the labeled B10-kDa fragment starts at residue 547
(consistent with trypsin cleavage after arginine), whereas the unlabeled
B60-kDa fragment is identical to the N terminus of the protein
minus the initial methionine. The labeled catalytic residue therefore
lies in the C-terminal 98 residues. A number of conserved lysine and
histidine residues in this region could form the guanine adduct
required for GTase activity. Finally, conserved residues (SLCRFxGL/
IR, where x is any residue) toward the N terminus of this domain
include a cysteine residue (Cys518) at the base of a deep depression
(Fig. 3d). By analogy with the HCxxxxxR motif of cysteine phospha-
tases
14
, this could form part of the RTPase catalytic apparatus. Indeed,
the presence of a conserved histidine (His554) makes it conceivable
that the RTPase and GTase activities are colocalized in this depression.
The family Reoviridae is composed of two branches, turreted and
nonturreted, characterized by differing capping logistics. The non-
turreted viruses, including orbi- and rotaviruses, hold their capping
enzyme in the core, close to the polymerase
15
, whereas turreted virus
(such as orthoreovirus) cores bear projections, where five capping
enzymes surround the transcript
16
. Nonturreted viruses’ capping
enzymes are smaller and fewer copies are required (probably two at
each vertex). This reduced capping enzyme concentration and its high
efficiency in solution suggest coordination of the four chemical
reactions
6
. The BTV VP4 structure provides evidence for the basis
of several of these catalytic activities. The active sites have an overall
linear layout (Fig. 1d), and the nature of the molecular surfaces
between them is consistent with partial substrate channeling and tight
protein-protein interactions with the polymerase enzyme
13
,which
may allow immediate capping of the emerging chain (as shown for a
closely related virus
17
). Coupling via the KL domain would place the
polymerase adjacent to the GT domain, with the next (N7MTase)
active site 30–40 A
˚
along the molecule and the final 2¢OMTase
40 A
˚
beyond that. The fundamental differences between VP4 and
the capping cage of orthoreovirus suggest that the split between the
two branches of the family Reoviridae may be very ancient and
provides further evidence for the diversity of viral solutions to the
capping problem.
Accession codes. Protein Data Bank: Coordinates and structure
factors have been deposited with accession codes 2JHP (VP4–
AdoHcy), 2JH8 (VP4–m
7
GDP), 2JH9 (VP4–GTP), 2JHA (VP4–
G5¢ppp5¢G) and 2JHC (apo-VP4).
Note: Supplementary information is available on the Nature Structural & Molecular
Biology website.
ACKNOWLEDGMENTS
We thank L. Lyne, W. Lu and the staff of BM14 and ID14 beamlines at the
European Synchrotron Radiation Facility for technical support. This work was
supported by the Medical Research Council, UK, EC SPINE grant QLG2-CT-
2002-00988 (D.I.S.), and by the Royal Society, UK (J.M.G.).
COMPETING INTERESTS STATEMENT
The authors declare no competing financial interests.
Published online at http://www.nature.com/nsmb/
Reprints and permissions information is available online at http://npg.nature.com/
reprintsandpermissions
1. Furuichi, Y. & Miura, K. Nature 253, 374–375 (1975).
2. Shatkin, A.J. Cell 9, 645–653 (1976).
3. Furuichi, Y., LaFiandra, A. & Shatkin, A.J. Nature 266, 235–239
(1977).
4. Muthukrishnan, S., Both, G.W., Furuichi, Y. & Shatkin, A.J. Nature 255,33–37
(1975).
5. Mertens, P. Virus Res. 101,3–13(2004).
6. Ramadevi, N., Burroughs, N.J., Mertens, P.P., Jones, I.M. & Roy, P. Proc. Natl. Acad.
Sci. USA 95, 13537–13542 (1998).
7. Ramadevi, N. & Roy, P. J. Gen. Virol. 79, 2475–2480 (1998).
8. Ramadevi, N., Rodriguez, J. & Roy, P. J. Virol. 72, 2983–2990 (1998).
9. Bujnicki, J.M. & Rychlewski, L. Genome Biol. 2, RESEARCH0038 (2001).
10. Li, C., Xia, Y., Gao, X. & Gershon, P.D. Biochemistry 43, 5680–5687 (2004).
11. Fabrega, C., Hausmann, S., Shen, V., Shuman, S. & Lima, C.D. Mol. Cell 13,77–89
(2004).
12. Funke, L., Dakoji, S. & Bredt, D.S. Annu. Rev. Biochem. 74, 219–245 (2005).
13. Nason, E.L. et al. J. Virol. 78, 8059–8067 (2004).
14. Takagi, T., Moore, C.R., Diehn, F. & Buratowski, S. Cell 89, 867–873 (1997).
15. Gouet, P. et al. Cell 97, 481–490 (1999).
16. Reinisch, K.M., Nibert, M.L. & Harrison, S.C. Nature 404, 960–967 (2000).
17. Lawton, J.A., Estes, M.K. & Prasad, B.V. J. Virol. 75, 1632–1642 (2001).
BRIEF COMMUNICATIONS
NATURE STRUCTURAL & MOLECULAR BIO LOG Y VOLUME 14 NUMBER 5 MAY 2007 451
© 2007 Nature Publishing Group
http://www.nature.com/nsmb
