HIV Spike Demo Storyboard

Storyboard for HIV spike demonstration for RBVI advisory committee meeting on November 13, 2009.

Aim of demonstration

Aim is to show EM, sequence, and network visualization tools applied to a single biological problem.

Biological Problem

The specific biological problem is to understand interactions of broadly neutralizing antibodies (bNAb) with HIV spike proteins gp120 and gp41. The spikes recognize CD4 receptors on human T-cells to identify and infect those cells. The motivation is to develop an HIV vaccine which works by eliciting production of these antibodies. This month two new bNAbs were reported that are effective against more HIV strains and at lower antibody concentrations than the previously known half-dozen bNAbs.

Broad and Potent Neutralizing Antibodies from an African Donor Reveal a New HIV-1 Vaccine Target.
Walker LM, et al.
Science. 2009 Sep 3 ​PubMed 19729618

Things to show

These have not been ordered in any sensible way yet.

Schematic of HIV virus architecture

EM tomography view of HIV particles (Stephen Fuller lab).

Will show some tomography data and how to get from A to B in above figure.

There are about 15 gp120 crystal structures, all monomeric. The 2 new bNAbs bind only to trimeric gp120 spikes, not to the monomeric form. The trimer conformation can be found from the EM tomography data, averaging and fitting.

HIV-1 BaL spike with CD4 and 17b-FAB. Subramaniam lab. EMDB 5020, PDB 3dnn	HIV-1 BaL spike with b12 FAB. EMDB 5018, PDB 3dno

SIV envelope spike. Roux lab.
EMDB 1246, PDB 2bf1 (gp120), PDB 1tji (gp41), PDB 1tzg (gp41)

There is some conflicting data about what trimeric spikes look like, having a single stalk in the viral envelope, or 3 legs (seen in SIV).

I have SIV spike tomography from Ping Zhu in Kenneth Roux's lab and Florida State University.

Structures aren't available for the two new bNAbs. (None are reported in the article.) So we'll be looking at gp120 structures with other anti-bodies bound and CD4 bound.

Start with PDB 1GC1 from 1998 having gp120 (red), a bound FAB (fragment antibody) (blue) and CD4 (yellow).

Blast against the gp120 sequence (Model Panel / blast protein...). The current Chimera interface doesn't let you specify a chain. We don't want CD4 and other FAB structures, so first delete all but gp120 (chain G) then do the blast. (Elaine: alternatively, "split" beforehand if you would rather keep the other chains)

Multiple alignment of 15 gp120 sequences (through 2QAD in blast dialog) shows mostly identical sequences with a few at the bottom with inserted loops. The gp120 V3 loop which is critical to antibody binding is one of these loops.

Should show more v3 loop sequences in MAV to see conserved regions. Compare those to contacts in a structure having V3 loop.

Loading the top 13 blast hits shows they are all quite similar structures:

Have all the models in the fetch download directory. Point out this nice capability for handling fetched data.

There are other structures showing antibodies bound to just the V3 loop without the gp120 core. Those might be interesting to look at in this demo.

Molecular architecture of native HIV-1 gp120 trimers.
Liu J, Bartesaghi A, Borgnia MJ, Sapiro G, Subramaniam S.
Nature. 2008 Sep 4;455(7209):109-13. ​PubMed 18668044

describes large rotations of the gp120 monomers occurring in antibody and cd4 binding that would be interesting to show since the new bNAbs only bind trimers, not monomers. Here's the caption from figure 3. Can show EM maps of these trimers, EMD 5019, 5020.

"Comparison of the locations of the docked gp120 monomers in the free, b12-bound and CD4-bound states (Fig. 3c-eFigure 3) provides insights into the overall quaternary structural changes that occur in the trimeric spike. The binding of b12 results in a partial opening of the spike, coupled with rotation of each monomer by ~20°–25° around an axis perpendicular to the viral membrane (Fig. 3dFigure 3). However, CD4 binding results in a rotation around this central axis in the same direction that is twice as large, in addition to an out-of-plane rotation (Fig. 3eFigure 3), and slight vertical displacement of gp120. Thus, while the binding sites for CD4 and b12 are on roughly the same face of the gp120 monomer, they result in very different outcomes for the conformation of the Env trimer."

Might show large motions by superimposing inertia ellipsoid depictions.

Critical aspect of FAB binding is that gp120 has extremely high glycosylation. N-glycosylation sites are at amino acid triplets: Asparagine - X - Serine/Threonine called the glycosylation sequon. Do a prosite search (pattern N-x-[ST]) of the gp120 sequence in 1GC1 finds 18 sites. Full length gp120 has 25 sites. There are also O-glycoslyation sites on serines and threonines. More than 50% of the mass of gp160 (gp120 + gp41 before cleavage) is glycans.

For clarity you could hide the non-search-result regions with Tools→Region Browser

Only one gp120 crystal structure (SIV) has glycosylation, 13 sites, only some of the sugars at each site are resolved. PDB 3fus.

Would be interesting to show in a demonstration adding glycans to all N-glycosylation sites to get an idea of appearance of gp120 glycan shield. Maybe they could be added far from molecule and minimization could drag them in close. Would be very interesting to see the dense coverage by sugars.

Glycosylation alternative (Elaine)

Instead of showing glycosylation consensus sites on the single sequence, you could use Matchmaker to superimpose HIV1 gp120 and glycosylated SIV gp120 and show the corresponding pairwise sequence alignment. Regions that superimpose well are automatically highlighted with orange boxes, and consensus glycosylation sites could then be highlighted on both sequences.

Say 1gc1 chain G (HIV gp120) is already open as model 0, 3fus (glycosylated SIV gp120) as model 1. You could focus on the SIV structure and show the glycosylation:

focus #1

alias glyco #1 & ~ protein

show glyco

Then discuss glycosylation. May want show attached Asn residues (disp #1:asn & glyco z<4) or fill sugar rings (fillring glyco). Then superimpose the structures and show the pairwise sequence alignment:

mm #1 #0 show true

If the display is too busy, could hide the atoms at some point (~disp).

The orange boxes on the sequence alignment are the parts used in the final fit iteration and coincide with low values of the RMSD header, which in the pairwise case is simply CA-CA distance per column. Clicking the orange region selects the corresponding parts of the structures.

Multalign Viewer menu Edit... Find PROSITE Pattern N-x-[ST] reveals commonalities and differences.

Alternatively, instead of finding the pattern on the sequence, just select it on the structures directly with main menu Select... Sequence, PROSITE Pattern, which is then reflected in the green selection region on the sequences (could "color yellow :asn & sel" or something like that). Another possibility here is calculating percent ID (in Multalign Viewer menu, Tools... Percent Identity).

Conservation alternative (Elaine)

Could color the gp120 structure by conservation, showing that the antibody and CD4 in structure 1gc1 are interacting with fairly well-conserved regions of gp120.

Open 1gc1 and the alignment pdb1gc1.aln (attached). Ahead of time, use Multalign Viewer Preferences... Headers to change Conservation style to AL2CO (otherwise default). Then use Structure... Render by Conservation and adjust settings as shown in the Render dialog (lowest or most variable red, highest or most conserved blue, no-value residues gray).

The gray segment in gp120 is where the alignment column had too few sequences (too many gaps) to calculate a reliable conservation value. The redder, more variable parts of the gp120 core structure are not interacting with the other chains.

Another possibility is using (main menu) Select... Sequence... gag to show where longer loops were replaced with gly-ala-gly to create this crystallizable gp120 core construct. Some other sequences in the alignment have the full-length loops.

Cytoscape

Would like to demonstrate network visualization of HIV sequence strains clustered into clades. Associate gp120 structures with sequences and be able to see what clades the structures belong too. (All clade B, US/Europe?)

Thousands of HIV complete rna sequences available. Not sure how to measure similarity for cytoscape clustering. Also technical problem to associate PDB gp120 structures with those sequences.

SAXS

SAXS profile of gp120 / FAB / CD4 complex could be used to compare conformational states. Bound gp120 and CD4 was studied by SAXS in

Binding of full-length HIV-1 gp120 to CD4 induces structural reorientation around the gp120 core.
Ashish, Garg R, Anguita J, Krueger JK.
Biophys J. 2006 Sep 15;91(6):L69-71. ​PubMed 16844755

We could show SAXS profile calculation of different conformations. Would be best to get experimental SAXS data. Did not see any database with HIV SAXS on web. Could just make some example target conformation profile and demonstrate testing alternatives against it.

Plastic trimer model

Make plastic model of gp120 trimers in 5 parts with bound CD4 and FAB connected by magnets, pieces colored distinctly. Maybe include gp41 stalk.