1. A comparison of membrane-containing bacteriophage and archaeal virus structures by cryo-EM
Sarah Butcher
ERICE 2006
Enveloped virus session

2. Lineage 1 Lineage 2 Lineage 3 irido adeno PRD1 phi6 BTV herpes
Rationale for such a comparison:
Many virus sequences and structures known, shown that there is some similarity between viruses infecting hosts from different domains of life. Heavily weighted towards the tailed bacteriophage, due to the isolation methods.
Only 1 lipid containing virus structure known at high resolution (PRD1). Others, from both eukaryotic and prokaryotic hosts, studied by cryo-em. Are their common folds, principles in virus assembly arising by the restraint of having a membrane? Another way to look at membrane protein structures in a native environment
Jack Johnson will give a talk later in the course on viral protein coat evolution, and Dave Stuart and Nicola Abrescia will give us the state of the X-ray art, so I will address a few points about EM, EM-x-ray synergy, and strategies for learning more from an EM structure.
herpes

3. Membrane-containing viruses: antipasta misto
Membrane required for host cell entry e.g. fusion with endosomes.
Lipids derived from the host.
Bacteriophage:
Cystoviridae – f6 and f8
- fuse with Gram-negative host inner and outer membranes,
releasing polymerase complex into cytoplasm
Tectiviridae – PRD1 and Bam35
- fuse, releasing DNA + terminal proteins into
cytoplasm
Archaeal:
SH not known
Maybe use schematic diagrams here

4. Cryo-em studies
Relatively unstable viruses, so rapid preparation an advantage
Membrane structure preserved in native state
Possibility to look at mutants, protease treated, chemically disrupted
Resolution typically 8-12 Å, but varies within the reconstruction depending on the order.
Capsid proteins mainly peripheral membrane proteins
Integral membrane proteins occasionally revealed
Image reconstruction generally relies on icosahedral symmetry for orientation determination and averaging. Hence symmetry mismatches will not be resolved.

5. f6 entry
Pseudomonas syringae CM OM PG

6. 3D reconstruction of f6 and f8 virion
Central cross section through the phi6 virion reconstruction.
To assign the different layers, we calculated a empty NC reconstruction (next slide)
Difference in spikes due to difference in hosts
Jäälinoja et al. (in preparation)

7. Structure of the f6 nucleocapsid
Three-dimensional reconstruction from cryo-EM data applying icosahedral symmetry
Two protein shells
Hexameric packaging motor at the vertices (symmetry-mismatch)
This shows to you the phi6 nucleocapsid particle which lacks the membrane envelope. We solved this structure to 7Å resolution using cryo-EM and applying icosahedral symmetry. The relatively high resolution allowed us to determine the molecular boundaries between the capsid subunits. This shows to you the polymerase complex particle. PC particle first preassembles in the cell, and after then it packages the three genome segments. We want to understand, how this particle specifically recognizes the three genome segments and packages one copy of each. First I will concentrate on the structure of this hexameric packaging motor, which is the motor responsible for translocating the ssRNA into the particle. Since it is a hexamer located at the five-fold vertex, it was not resolved in this icosahedral reconstruction. For this image they were modeled from an X-ray structure of an related virus.

8. Bacteriophage PRD1
Schematic diagram of an membrane-containing bacteriophage PRD1, could be Bam35, SH1, PM2
PRD1 is a broad host range bacteriophage of the Tectivirus family.
These are dsDNA viruses with a membrane inside an icosahedral shell.
PRD1 is a fairly large and complicated virus.
Its MW is ~66 MDa and its diameter is ~650 A.
It has at least 5 different proteins in the capsid and about 15 proteins interacting with the membrane.
One notable thing about PRD1 is its similarity to the mammalian adenovirus. This similarity runs from the mode of DNA replication (protein primed, sliding back) to the general organization of the capsid (pseudo T=25) to the folding of the major coat protein.
PRD1 is a large and complicated system, and so a variety of structural techniques are being used to study its structure. We have applied the “divide and conquer” technique, where we use cryoEM data at moderate resolution of the whole capsid and then try to use this as a frame to fit the smaller pieces of the virus that are solved by other methods, as Xray crystallography.
This combination of techniques allows us to obtain more information than the sum of the info each one of them gives separately.
Infects Gram negative bacteria-resistant to antibiotics (salmonella, Ecoli?)

9. 25 Å
14 Å
PRD1 cryo-EM
This is the frame for the puzzle. cryoEm reconstruction of three different PRD1 specimens. Always get a model, but how to interpret the density
Showing here: micrographs, central slice, surface rendering.
Wt: 25 A, icosahedral shell, concentric layers of membrane and DNA
Sus1: mutant that does not package DNA. 14 A. Only shell and membrane present
P3shell: sus1 treated with SDS. 12 A. Membrane and vertices lost.
This way we can see what is the membrane, and further that by detergent treatment it is removed
12 Å
Butcher et al EMBO J.
San Martin et al Structure
San Martin et al NSB

10. Major capsid protein is a peripheral membrane protein
We fitted the 4 trimers of the asymmetric unit of each reconstruction using rigid body refinement in Xplor.
To obtain a meaningful fit, we had to take into account the resolution of each reconstruction, the effect of the CTF, and the uncertainty of the EM scale.
To quantify the goodness of fit we used the crystallographic R factor, which in the end gave us values of between 31% for the P3 shell and 36% for the virion.
This implies that the conformation of is P3 almost the same in the capsid as in the crystal.
R=31%
Benson et al, 1999 Cell; San Martin et al Structure

11. EM/Xray combination
In the case of PRD1 the x-ray structure of the major coat protwin was known
Now there is another very useful thing that we can do with these quasi-atomic models. As I said before, I have built the model by fitting the atomic structure of 240 P3 trimers to the EM reconstruction of the virus. But, the virus is not composed only of P3, it has at least 6 other proteins in the capsid and 15 in the membrane.
So, if I now take my quasi-atomic model, that contains only P3, and subtract it from the EM reconstruction, which contains all the viral components, I can obtain a very detaliled image of these other components.
In doing this, we have learned several interesting points about the way the PRD1 capsid is organized.
Benson et al, 1999 Cell; San Martin et al Structure; San Martin et al NSB

12. Bam35 Infects Gram-positive host Bacillus thuringiensis
Very limited sequence similarity to bacteriophage PRD1
Address structural similarity using cryo-EM and X-ray data from PRD1

13. 7.3 Å icosahedral reconstruction of bacteriophage Bam35
Reconstructions calculated from both full virion and particle lacking DNA
Scalebar 100 nm and
Laurinmäki et al. (2005) Structure

14. Modelling with homologous proteins
N-termini again reaching down to the membrane in Bam35, as peripheral membrane proteins, alpha-helical contact
Laurinmäki et al. (2005) Structure

15. Bam35 membrane proteins
Cluster of 3 or 4 helices surrounded by 6 additional helices. 60 such complexes per virion, no idea what proteins
Membrane curvature- often see that the membrane follows the edge of the capsid – it seems to have the icosahedral symmetry, but then at this resolution start to see local fluctuations in the curvature where the transmembrane proteins are located
Laurinmäki et al. (2005) Structure

16. Effect of curvature and thickness and membrane protein location
Zhang et al. 2003, NSB; Laurinmäki et al. 2005, Structure

17. Archaeal virus SH1
Host is Haloarcula hispanica isolated from salt lakes in Australia
800Å
100Å
Jäälinoja et al. (in preparation)

18. Density distribution and resolution
Capsid
DNA
Spike
Membrane
Radial density distribution from center of the model to outside.
Three different models are plotted. Black line corresponds to the model that is presented here.
Radius 32 nm, excluding the spike.
Resolution highest for the capsid (8.5 Å). Even the model is good to this resolution, other parts are not as icosahedrally ordered.
Previously, small angle x-ray scattiring was used to study density arrangement in the virion. Those assignments match nicely the ones from cryo-EM.
Radial profiles were calculated from 17Å reconstructions of these three specimens, The position of the capsid, lipid bilayer and outermost layer of DNA coincide with values reported by Harrison et al in 1971 from small angle X-ray scattering. The average diameter is 56 nm, The spike density extending the particle out to 78nm
Jäälinoja et al. (in preparation)

19. Archaeal virus SH1
T=28
Jäälinoja et al. (in preparation)

20. Unusual capsomers 2- and 3–fold symmetric capsomers on capsid surface.
Quasi-hexagonal capsomer base.
Slightly skewed at 2f and adjacent capsomers, not elsewhere.
All with similar mass
Jäälinoja et al. (in preparation)

21. Structure at 5-fold axis
Very weak signal from peripheral domain.
Possible symmetry mismatch in external domains.
Transmembrane complex.
Jäälinoja et al. (in preparation)

22. Vertex Reconstruction
Orientation angles and vertex positions derived from output of an icosahedral reconstruction.
Vertices are compared and classified revealing differences in composition and orientation.
Subsets of vertices can be used for reconstruction and any desired symmetry can be applied.

28. SH1 spike reconstruction
Briggs et al. (2005) JSB; Huiskonen et al. (submitted); Jäälinoja et al. (in preparation)

29. Cryo-em studies – conclusions
Capsid proteins mainly peripheral membrane proteins, connected by a-helices
Integral membrane proteins occasionally revealed, so far apparently a-helical
Membranes are icosahedral, locally affected by membrane proteins

30. Acknowledgements Benita Koli Dennis Bamford Hanna Kivelä John Briggs
Stephen Fuller