© 2002 Nature Publishing Group http://www.nature.com/naturestructuralbiology
letters
L-A virus at 3.4 Å resolution
reveals particle architecture
and mRNA decapping
mechanism
Hisashi Naitow1, Jinghua Tang1, Mary Canady1,
Reed B. Wickner2 and John E. Johnson1
1
Department of Molecular Biology, The Scripps Research Institute, 10550
North Torrey Pines Road, La Jolla, California 92037, USA. 2The Laboratory
of Biochemistry and Genetics, NIDDK, National Institutes of Health,
Bethesda, Maryland 20892-0830, USA.
Published online: 16 September 2002, doi:10.1038/nsb844
The structure of the yeast L-A virus was determined by X-ray
crystallography at 3.4 Å resolution. The L-A dsRNA virus is
400 Å in diameter and contains a single protein shell of 60
asymmetric dimers of the coat protein, a feature common
among the inner protein shells of dsRNA viruses and probably related to their unique mode of transcription and replication. The two identical subunits in each dimer are in
non-equivalent environments and show substantially different conformations in specific surface regions. The L-A virus
decaps cellular mRNA to efficiently translate its own
uncapped mRNA. Our structure reveals a trench at the active
site of the decapping reaction and suggests a role for nearby
residues in the reaction.
a
The dsRNA L-A virus (L-A), which infects the yeast Saccharomyces cerevisiae, has a single linear segment of 4.6 kilobases (kb)
encoding a coat protein (Gag) and its RNA-dependent RNA polymerase (Pol)1. Pol is expressed as a Gag–Pol fusion protein formed
by a –1 ribosomal frameshift. The virus is made up of 120 Gag
subunits2, with 2 molecules of Gag–Pol incorporated along with
the dsRNA per particle3. The N-terminal part of Pol is necessary
for the encapsidation of the viral genome, but Gag alone is sufficient to form morphologically normal viral particles4.
The viral mRNA, synthesized by Pol within the virus particles,
lacks both the 5′ cap and 3′ poly(A) structures typical of cellular
mRNAs and is extruded from the virus particles into the cytoplasm, where it can be translated by the cellular ribosomes.
Normally, cellular mRNA is degraded by gradual removal of its
3′ poly(A) structure, followed by decapping by the cellular
Dcp1p enzyme and degradation of the uncapped mRNA by the
Ski1p/Xrn1p exoribonuclease5. The absence of a 5′ cap on L-A
transcripts would be expected to make them immediately subject to degradation by the Ski1p/Xrn1p system. However, the
Gag subunit of the capsid generates uncapped cellular mRNAs
by the enzymatic removal of their 7-methyl-GMP (m7GMP) cap
and attachment to His 154 of Gag6. This decapping activity of
Gag aids in the expression of the viral mRNA, apparently by producing decoys that compete for the attention of the cellular
enzyme Ski1p/Xrn1p that degrades uncapped mRNAs (including the viral mRNA)7. Thus, Gag mutants of His 154 express
viral information poorly, but normal expression and levels of
viral mRNA are restored by a ski1/xrn1 deletion7.
The structure of the L-A virus particle was previously investigated at 16 Å resolution with cryo-EM2. The cryo-EM density
showed that the particles have T = 1 icosahedral symmetry and
b
c
Fig. 1 Structure of L-A virus particle. a, The L-A particle viewed down an icosahedral two-fold axis with the Cα positions traced. The capsomer molecules A and B are purple and red, respectively. The icosahedral symmetry is shown with the lines connecting the five, three and two-fold axes of
symmetry. b, The icosahedral asymmetric unit of the L-A. The path of the polypeptide chains is colored in a gradient from blue at the N-terminus to
red at the C-terminus in each molecule. Icosahedral symmetry elements are indicated by symbols: five-fold is pentagon; three, triangle; and two,
oval. c, The icosahedral asymmetric unit of the VP3 core of bluetongue virus, showing the striking similarity to L-A in the subunit contact regions.
Coloring is the same as in (b).
nature structural biology • volume 9 number 10 • october 2002
725
© 2002 Nature Publishing Group http://www.nature.com/naturestructuralbiology
letters
The short helix from amino acids 106–114, near the
five-fold axis, is shown fit to an electron density map
(Fig. 2), demonstrating the quality of the electron
density map. The core of the structure is mainly
composed of α-helices (Fig. 1b). The C-terminus
(residues 652–680) is disordered and not critical:
residues 671–680 are completely dispensable for
function, and residues 648–680 are unnecessary for
particle formation3. The visible C-terminus (residue
651) is at the interior of the particle (Fig. 3), as
expected because the Pol domain of the Gag–Pol
fusion molecules must face the inside of the particle.
In contrast to the smooth outer surface of the VP3
Fig. 2 A region of the electron density map is shown fit with residues 106–114 particles from BTV, the surface of Gag is uneven.
(Ser-His-Ala-Tyr-Asn-Ile-Thr-Ser-Trp).
This reflects the different roles of the particle surfaces. The VP3 particle from BTV serves as a template for assembly of a T = 13 outer protein shell,
are composed of 12 pentons, each consisting of five sets of two whereas the L-A virus does not have an outer shell and Gag
Gag proteins. This arrangement requires that the two Gag pro- functions as an enzyme.
teins be in non-equivalent environments, which is unusual in
virus structures. This novel virion structure found in L-A is sim- The decapping reaction
ilar to the cores of other dsRNA viruses, bluetongue virus (BTV) The active site of the decapping reaction has been identified as
and reovirus8,9. These viruses are more complex, containing His 154, which is necessary for activity and to which m7GMP
multiple protein shells around the nucleoprotein core, but are becomes attached10. His 154 is at the tip of the loop containing
functionally similar to the L-A virus. Transcription of mRNA residues 144–163 (loop 1), which is found on the surface of the
occurs inside the particles and the nascent mRNAs are extruded capsid in both A and B Gag molecules. This loop is part of a
from holes present in the particles. The unusual arrangement of deep, narrow trench formed along with three other loops
coat proteins was postulated to be an adaptation to the intra- (residues 298–312 (loop 2), 443–458 (loop 3) and 525–562
particulate transcription machinery of the dsRNA viruses. The (loop 4); Fig. 4). This trench is partially disordered (residues 156
structure of this machinery may differ from an ordinary and 444–457) in molecule A of Gag, and overall this region is
arrangement of a coat protein because most virus capsids do not reminiscent of the structure seen in RNA guanylyltransferase, an
participate in transcription. Here, we set out to test this hypo- mRNA capping enzyme11. Similar to the trench observed in the
thesis by determining the structure of the virus particle and its Gag protein of L-A, the guanylyltransferase contains a trench
that can adopt either open or closed conformations during the
method of decapping RNA.
mRNA capping reaction. The secondary structure elements
Structure of Gag
around the trench of guanylyltransferase resemble those in Gag
Using X-ray-based model building of L-A (see Methods), the from L-A but have a different topology, indicating convergent
diameter was determined to be 400 Å. The protein shell is 46 Å evolution. We suggest that cellular mRNAs are bound in this
thick. The particle has 18 Å diameter openings at the icosahedral trench. Two of the four loops are well ordered in the electron
five-fold axes, providing a portal for the entry of nucleotide density, perhaps forming the floor (residues 298–312) and the
triphosphates and exit of viral mRNA. The L-A particle was con- wall (residues 525–562) of the active site, and not directly
structed from 60 copies of pseudo capsomers, each composed of involved in the decapping reaction. However, the other two
two molecules of Gag (A and B; Fig. 1). Although of identical loops are less ordered near His 154. We suggest that the main
sequence, these molecules have significant differences in struc- players in the decapping reaction are residues from these loops,
ture, particularly residues 8–12, 82, 96–99, 111–112
and 387–396, which are all located on the subunit
surface. Although Gag is much smaller than the VP3
subunit of bluetongue virus and λ1 subunit of
reovirus, it has an overall shape that is similar to
coat proteins from the other two viruses and the
quaternary organization of all three particles is
strikingly similar. However, it is impossible to make
a meaningful superposition of either VP3 or λ1 on
Gag. The electron density map shows differences in
side chain density from those expected based on the
Gag sequence at residues 118 and 153 because a different strain of L-A was used for the structural
analysis than that used for the sequence determination (see Methods). These differences are minor and
do not affect interpretations based on the structure.
Each Gag molecule is made up of two α-helices
Fig. 3 A stereo view ribbon drawing of the side view of molecule A showing the trench
located near the icosahedral five-fold axis and sev- accessible from the outside of the L-A particle. The labeled N- and C-termini are on the
eral β-sheets near the two- and three-fold axes. inside of the capsid. Coloring is the same as Fig. 1b.
726
nature structural biology • volume 9 number 10 • october 2002
letters
© 2002 Nature Publishing Group http://www.nature.com/naturestructuralbiology
Fig. 4 A stereo view close-up of the trench showing
residue His 154 and its neighboring residues around
the active site. Residues Tyr 150, His 151, Asp 152,
Tyr 452, Tyr 538 and Asp 540 all may contribute to
the mRNA decapping reaction. The four loops corresponding to loop 1, 2, 3 and 4, are colored in red,
green, blue and pink, respectively.
certainly including His 154. Tyr 150, Asp 152 and Tyr 452 are
within a 4 Å radius of a point in the center of the trench, suggesting that they may have direct roles as well (Fig. 4). We predict
that these loops (144–163 and 443–458) open and close the
trench because they are on the surface of the particle (Fig. 4). A
cellular mRNA could be captured by these dynamic and active
loops and positioned in the trench for the decapping reaction to
occur. The cap analog 7-methyl GDP (m7GDP) is an inhibitor6
and can completely block the cap-binding activity. m7GDP is not
sufficient for the decapping reaction to occur, because at least
7-methyl GpppGp is needed6. We propose that the first loop
(residues 144–163), which includes His 154, interacts with the
cap and removes it, while the third loop (residues 443–458)
binds noncovalently to the captured mRNA in the trench. These
active loops are involved in the two step mechanism of the
decapping reaction. Efforts to determine the structure of the L-A
virus with m7GDP are underway.
The 20 Å electron density map showed a single layer of density
inside the protein, interpreted as a portion of the L-A genome.
The density is discontinuous at the five-fold axes. However, this
density gradually disappeared with increasing resolution, and it
is not visible at near-atomic resolution. Because the genome of
Table 1 Structure determination
Number of images
Space group
Cell parameters
a (Å)
b (Å)
c (Å)
α (°)
β (°)
γ (°)
Reflections
Total
Unique
Redundancy
Resolution (Å)1
Rmerge1,2
Completeness (%)1
<I / σ (I)>1
Averaging R-factor3
Correlation coefficient4
596
P21
407.0
403.2
572.0
90.0
90.46
90.0
5,151,654
2,105,482
2.4
89.5–3.4 (3.9–3.4)
0.104 (0.433)
83.8 (70.8)
11.4 (2.7)
0.31
0.78
Number in parentheses is for the highest resolution shell.
Rmerge = ΣhΣj|Ij(h) – <I(h)>| / ΣhΣjIj(h), where Ij(h) is the jth measurement of
reflection indices h and <I(h)> is the mean intensity.
3Averaging R-factor = Σh(|F (h)| – k|F (h)|) / Σh|F (h)|, where k is a scale faco
c
o
tor, Fo is the measured structure amplitudes and Fc is the structure amplitudes calculated from inverse-FFT of the averaged and solvent flattened
map.
4Correlation coefficient = Σ (|F (h) – <|F (h)|>) (|F (h) – <|F (h)|>) / [Σ (|F (h)| –
h
o
o
c
c
h
o
<Fo(h)>)2 Σh(|Fc(h)| – <|Fc(h)|>)2]1/2.
1
2
nature structural biology • volume 9 number 10 • october 2002
L-A makes a hollow shell of density on the inside of the protein
capsid at low resolution, the Gags have been postulated to attract
the genome to this surface. dsRNA from BTV has similar layered
density12, with three shells of dsRNA found, in agreement with
its higher genome density (38 base pairs (bp) / Å3), which is
twice that of the L-A virus genome2 (19 bp / Å3). Our data indicate that the first genome layer is generated by its interaction
with the coat proteins. The Pol domain is not seen inside the particle because of the 60-fold icosahedral symmetry averaging.
Comparison with coat proteins of dsRNA viruses
Using a program for three-dimensional structural comparison13,
we observe no structural homology between Gag and the coat
proteins of other dsRNA viruses, or any other protein structure
determined so far. Thus, although there is no detectable
structural homology between the major coat proteins of dsRNA
viruses, they share the ability to form the unique T = 1 icosahedral
structure with an asymmetric dimer of Gag, and the capacity to
concentrate the genomic RNA near their inner surface.
Based on sequence analysis of RNA-dependent RNA polymerases (RDRPs), Koonin suggested that dsRNA viruses evolved
from (+) ssRNA viruses, not from each other14. Moreover,
detailed structural comparisons among the core proteins of
dsRNA viruses do not reveal superimposable regions. However,
dsRNA viruses share the T = 1 structure with 120 monomers,
and the close similarity of the quaternary structures of these particles is immediately recognizable. The Gag-like structures have
not been found among naturally occurring capsids of the (+)
ssRNA viruses, suggesting that, if Koonin is correct, the 120
monomer core structure of dsRNA viruses was repeatedly scavenged from a common source — for example, a portion of the
RNA polymerase — and then evolved to the highly varied structures observed today. Alternatively, different gene products may
have been recruited for capsid formation, and these viruses then
converged to the same quaternary structure because of their
common requirements. Even with the variation in the details of
the subunits, it is difficult to believe that their ability to form
such a distinctive T = 1 quaternary structure is a convergent evolutionary event.
All dsRNA viruses synthesize (+) and (–) strands within the
viral (core) particle. The polymerase is fixed to the inside of the
particle wall or at the five-fold axes, so template motion is essential. The T = 1 structure of L-A virions with 120 Gag monomers
per particle demands that Gag be sufficiently flexible to occupy
two different environments in the viral particle. This flexibility
of the coat protein may provide some flexibility of the capsid,
allowing for template motion. Nonetheless, the restricted
intraviral space should limit the rate of elongation; indeed, synthesis of the 4.5 kb L-A (+) strands takes ∼45 min (ref. 15). This
rate (100 nt min–1) is slow compared to elongation rates for DNA
transcriptases, which range from 600 to 24,000 nt min–1 (ref. 16).
727
letters
Poliovirus RDRP elongates at >3,000 nt min–1 (ref. 17).
Rotavirus is similarly rather slow incorporating ∼300 nt min–1
(D. Chen and J. Patton, pers. comm.). Estimates for reovirus
range widely, but elongation is believed to be rate-limiting18.
© 2002 Nature Publishing Group http://www.nature.com/naturestructuralbiology
Methods
Virus crystallization. L-A virus was grown in S. cerevisiae strain
3039 in YPAD broth and crystallized with PEG 8000 as described19.
Although several yeast strains were tested, good crystals were
obtained only from strain 3039. The amino acid sequence of Gag is
available only from strains RE59 (ref. 20) and LO14 (ref. 21), each
related to 3039. The L-A dsRNA sequences from these two strains
differ from each other by <1% (H282R, Y528H, G541S, A604S and
S609R (the first letter is the residue in RE59 at the specificed position and the second is that in LO14)), suggesting that the sequence
of the L-A in 3039 also differs by only a few residues, as indicated by
the electron density.
Data collection and processing. X-ray diffraction data (λ = 1.0 Å)
were collected at BioCARS beamline 14BM-C at the Advanced
Photon Source (APS), Argonne, Illinois, using ADSC 2 X 2 arrayed
CCD detector (Area Detector Systems). Data were collected by oscillating an L-A crystal through 0.25° at 100 K. The lifetime of the
frozen crystal within the X-ray beam was enough to permit the collection of one whole data set per crystal, consisting of 596 frames
with 40 s exposure per frame. The detector was placed 300 mm
from the crystal. Data were processed using the HKL package22.
Statistics are shown in Table 1. The crystal belongs to the space
group P21, with unit cell dimensions a = 407.0 Å, b = 403.2 Å, c =
572.0 Å and β = 90.46° and containing one particle in the asymmetric unit. The latter resulted in 60-fold noncrystallographic symmetry.
The position and orientation of the L-A particle in the asymmetric
unit was determined according to virus packing and the result of a
locked self-rotation function23.
Map calculation and model building. The L-A structure was
solved by molecular replacement, noncrystallographic symmetry
averaging and solvent flattening. The inner core VP3 of BTV has a
larger size (901 amino acids) than the Gag protein from L-A (681
amino acids). To use the structure of VP3 as a starting model, the
coordinates were multiplied by a reduction factor so the subunits
occupied a volume consistent with that of the asymmetric unit of
L-A. The initial map was generated at 29 Å resolution. The phases
were refined by sequential averaging and solvent flattening of the
2Fo – Fc map in real space (Fc originated either from the phasing
model or from inverse-FFT of a revised map). Phase extension was
carried out in steps of one reciprocal lattice point to 3.4 Å. The Fo
data were sharpened with a B-factor of –38 Å2. The final R-factor
and correlation coefficient were 0.31 and 0.78 between sharpened
Fo and Fc derived from the inverse-FFT sequential-averaged and solvent-flattened map. The density averaging, solvent flattening and
phase extension were performed with the program AVE in RAVE24
and the CCP4 suite25. All unmeasured reflections were exchanged
for weighted Fc, and the weights were calculated with SIGMAA26 in
the CCP4 program suite. Cα models of the two molecules A and B of
Gag were constructed based on the 3.8 Å 2Fo – Fc map using O27. The
2Fo – Fc map was of sufficient quality for the chain tracing and
assignment of amino acid sequence for both molecules of Gag.
Crystallographic refinement. The particle position and orientation were improved further by a locked self-rotation function and
R-factor search by X-PLOR28. Positional and grouped B-factor refinements were carried out by X-PLOR with strict 60-fold noncrystallo-
728
graphic symmetry. The crystallographic refinements led to the final
R-factor of 32.3% (Rfree = 32.4%) for Fo data between 8.0 and 3.4 Å.
The r.m.s. deviation from ideal bond lengths is 0.014 Å and that
from ideal angles is 1.627°. The percentage of all residues in the
model that fell into either the most favored or allowed regions of
the Ramachandran plot was 99.3%, and only 0.1% in the disallowed
regions29.
Low-resolution map calculation. A low-resolution map at 20 Å
was calculated to visualize viral genome density. Averaging was carried out using phases from the final model. All figures were prepared using O27, RasMol30 and MolScript31.
Coordinates. The atomic coordinates were deposited in the
Protein Data Bank (accession code 1M1C).
Acknowledgments
We thank T. Lin and G. Cingolani for their help in the data collection. Use of the
Advanced Photon Source was supported by the U.S. Department of Energy, Basic
Energy Sciences, Office of Science. Use of the BioCARS Sector 14 was supported
by the National Institutes of Health, National Center for Research Resources. The
crystallographic studies were supported by a grant from the National Institutes of
Health to J.E.J.
Competing interests statement
The authors declare that they have no competing financial interests.
Correspondence should be addressed to J.E.J. email: [email protected]
Received 24 June, 2002; accepted 27 August, 2002.
1. Wickner, R.B. in Fields Virology 4th edn (eds Knipe, D.M. & Howley, P.M.) 629–658
(Lippincott, Philadelphia; 2001).
2. Caston, J.R. et al. J. Cell. Biol. 138, 975–985 (1997).
3. Ribas, J.C. & Wickner, R.B. J. Biol. Chem. 273, 9306–9311 (1998).
4. Fujimura, T., Ribas, J.C., Makhov, A.M. & Wickner, R.B. Nature 359, 746–749
(1992).
5. Muhlrad, D., Decker, C.J. & Parker, R. Mol. Cell. Biol. 15, 2145–2156 (1995).
6. Blanc, A., Goyer, C. & Sonenberg, N. Mol. Cell. Biol. 12, 3390–3398 (1992).
7. Masison, D.C. et al. Mol. Cell. Biol. 15, 2763–2771 (1995).
8. Grimes, J.M. et al. Nature 395, 470–478 (1998).
9. Reinisch, K.M., Nibert, M.L. & Harrison, S.C. Nature 404, 960–967 (2000).
10. Blanc, A., Ribas, J.C., Wickner, R.B. & Sonenberg, N. Mol. Cell. Biol. 14, 2664–2674
(1994).
11. Hakansson, K., Doherty, A.J., Shuman, S. & Wigley, D.B. Cell 89, 545–553 (1997).
12. Gouet, P. et al. Cell 97, 481–490 (1999).
13. Holm, L. & Sander, C. Science 273, 595–603 (1996).
14. Koonin, E.V. Semin. Virol. 3, 327–339 (1992).
15. Fujimura, T., Esteban, R. & Wickner, R.B. Proc. Natl. Acad. Sci. USA 83, 4433–4437
(1986).
16. Uptain, S.M., Kane, C.M. & Chamberlin, M.J. Annu. Rev. Biochem 66, 117–172
(1997).
17. Lundquist, R.E., Ehrenfeld, E. & Maizel, J.V. Jr Proc. Natl. Acad. Sci. USA 71,
4773–4477 (1974).
18. Zarbl, H. & Millward, S. in The Reoviridae (ed. Joklik, W.K.) 107–196 (Plenum;
Dordrecht; 1983).
19. Naitow, H., Canady, M.A., Lin, T., Wickner, R.B. & Johnson, J.E. J. Struct. Biol. 135,
1–7 (2001).
20. Icho, T. & Wickner, R.B. J. Biol. Chem. 264, 6716–6723 (1989).
21. Diamond, M.E. et al. J. Virol. 63, 3983–3990 (1989).
22. Otwinowski, Z. & Minor, W. Methods Enzymol. 277, 505–524 (1997).
23. Tong, L. & Rossmann, M.G. Methods Enzymol. 276, 594–611 (1997).
24. Kleywegt, G.J. & Read, R.J. Structure 5, 1557–1569 (1997).
25. Collaborative Computational Project, Number 4 Acta Crystallogr. D 50, 760–763
(1994).
26. Read, R.J. Acta Crystallogr. A 42, 140–149 (1986).
27. Jones, T.A., Zou, J.Y., Cowan, S.W. & Kjeldgaard, M. Acta Crystallogr. A 47,
110–119 (1991).
28. Brünger, A.T. X-PLOR, version 3.1. (Yale University Press, New Haven; 1992).
29. Laskowski, R.A., MacArthur, M.W., Moss, D.S. & Thornton, J.M. J. Appl.
Crystallogr. 26, 283–291 (1993).
30. Sayle, R.A. & Milner-White, E.J. Trends Biochem. Sci. 20, 374 (1995).
31. Kraulis, P.J. J. Appl. Crystallogr. 24, 946–950 (1991).
nature structural biology • volume 9 number 10 • october 2002