Outline of HIV capsid Chimera demonstration for site visit

Tom Goddard
September 12, 2011

This demonstration will focus on antiviral drugs and vaccines targeting the HIV capsid protein. No drugs targeting the capsid protein have made it to market and only one has been tested in clinical trials. But a recent HIV genetic analysis suggests the capsid protein is the most vulnerable HIV protein. A central problem of HIV treatment is that the virus mutates rapidly and drug resistant strains proliferate when medication is stopped. The broad pool of existing strains makes a vaccine that works against all of them unlikely. Influenza with annual vaccinations presents a similar problem. Analysis of thousands of HIV sequences was used to find groups of amino acids where mutations are negatively correlated. Although all the amino acids in the group are able to mutate and form viable strains, if one mutates, other in the group rarely mutate. In other words, multiple mutations within the group very rarely produce viable strains. Usually drug/immune-response resistance requires multiple mutations. So the aim is to target drugs or the immune system against a group of residues (the epitope contains many residues in such a group) so that the virus cannot escape (ie. mutate to a drug resistant form) because there are few viable escape routes.

The sequence analysis aspects of this problem are excellent Profile Grid and Cytoscape network use cases. Connecting sequence analysis to structure is highly relevant since antibody/HLA/drug targeting is to a spatial region. Important regions are concentrated on capsid protein-protein interfaces making electron microscopy to discern those interfaces valuable. The large size of the capsid (1500 protein copies) and heterogeneity make this a good use case for "composite" model building.

Demonstration Segments

• MAV of reference capsid sequences from different clades. This will be a lead in to the problem of escape mutations. Might show a specific escape mutation against specific HLA allele in sequence and structure views. Need thousands of sequences to find escape mutants -- leads into profile grids and network visualization.

  Crystal structure of a p24 dimer. From "HLA alleles associated with slow progression to AIDS truly prefer to present HIV-1 p24." Entry 1AFV from the protein data bank is shown (PDB, www.rcsb.org/pdb) [44], [45]. (A) One monomer is shown in combined cartoon and surface representation to show the extent of the surface exposed part of the epitopes. For clarity, the other monomer is shown only in cartoon representation. The epitopes located in the dimer interface are highlighted in red (IW9), yellow (TW10) and green (DV9), and are also shown in stick representation. (B) A close-up of the dimer interface. The two monomers are shown in dark blue and cyan, and the TW10 epitopes are shown in green and magenta stick representations. The hydrophobic contact between T110 from one monomer to T110 in the other monomer is indicated by a dashed line.

• Profile grids
  • Can handle alignments of thousands of sequences giving valuable statistical information where showing conventional thousand line alignments is not useful.
  • show Alberto Roca's standalone version, especially how fixing certain residues impacts distributions of other residues at other sites.
  • Strong case to make for connecting this to structure views.
  • Example images showing about 4000 HIV GAG polyprotein sequences.
• Cytoscape residue covariation network. This looks at a mutation which allows HIV to become resistant to a immune response (HLA B*27) raised by some patients, and compensatory viral mutations that retain virus fitness.

  Figure I-B-2: Phylogenetic Dependency Network of p24Gag sites covarying with codon 264. The amino acid sequence of p24Gag is drawn counterclockwise, with the N-terminus at 3 o'clock. HLA-associated polymorphisms with q ≤ 0.2 are identified at their respective positions along the circle's circumference, while covarying amino acids (also q ≤ 0.2) are joined together by arcs within the circle. Colors indicate the statistical strength of the association. This figure highlights the specific codons covarying with codon 264, residue 2 of the well-characterized B*27-restricted KK10 epitope in p24Gag. Codon 264 (yellow dash, 9-o'clock position) is joined to its respective covarying sites (p24Gag codons 136, 173, 215, 256, 260, 268 and 315) via brightly-colored arcs.

  Figure I-B-3: Corresponding three-dimensional model of p24Gag sites covarying with codon 264. The ribbon diagram is a 3D model of the N-terminus of p24Gag. Gag codon 264, residue 2 of the well-characterized B*27-restricted KK10 epitope in p24Gag is shown in pink. Other codons in p24Gag observed to covary with specific codon 264 variants at q < 0.2 (identified in this analysis as Gag codons 136, 173, 215, 256, 260, 268) are shown in green. Codon 315 is not included in the 3D structure. The program used to construct Figures I-B-2 and I-B-3 is described in Carlson et al. [2008] and is available at http://research.microsoft.com/en-us/um/redmond/projects/MSCompBio/PhyloDViewer/

• Cytoscape sequence similarity network could show clade differences for GAG.
• Epitope mapping on structures, from HIV elite controller patients. If you understand the escape mutations perhaps another drug can suppress the primary mutation by targeting the mutated epitope.
• Fitting EM maps to understand protein-protein interfaces. Idea is that compensatory mutations appear randomly located unless you understand the oligmeric capsid structure. This segment can highlight symmetric fitting, iterative fitting, normal mode fitting, energy minimization of linkers between capsid N and C terminal domains. Might fit into 2-D hiv hexagonal grid EM map. Image shows pentamer fitting from Rous sarcoma virus.
• Cage model building to understand needed flexibility to accomodate different hexamer-hexamer joining angles.
• Trim5alpha protein protects old world monkeys from HIV. It binds to the capsid. Single mutation of human trim5alpha prevents HIV infection.
• Tomography solvent flattening to improve image detail. To be able to see trim5alpha or cyclophilin on the core particle need the best possible tomographic reconstruction.

  Missing wedge correction on simulated data -- spherical shell plus noise. Could try running it on actual HIV EM, or clathrin cage EM. Would need masking capability that is not yet developed.

• HTML log with WebGL, for tracking and sharing analysis.
  Example HTML log from a 1 hour session fitting Rous sarcoma capsid EM map and comparing to HIV capsid. Example log was made by hand, while grant proposal is to make it automatically. Links to images, sessions, map and molecule data, numeric results (areas, correlations, rmsds).

Relevant Publications

HIV genetic analysis, residue covariation. Supplemental material discusses resistant patient epitopes.

Coordinate linkage of HIV evolution reveals regions of immunological vulnerability.
Dahirel V, Shekhar K, Pereyra F, Miura T, Artyomov M, Talsania S, Allen TM, Altfeld M, Carrington M, Irvine DJ, Walker BD, Chakraborty AK.
Proc Natl Acad Sci U S A. 2011 Jul 12;108(28):11530-5.
PubMed 21690407

Background material on HIV capsid antiviral drugs.

New approaches for antiviral targeting of HIV assembly.
Prevelige PE Jr.
J Mol Biol. 2011 Jul 22;410(4):634-40.
PubMed 21762804

Escape mutants from immune system pressure with specific examples to p24 capsid protein. Has network showing immune pressure mutations and compensatory mutations that restore viral fitness. Might be intersting to map these mutations onto structure. One figure does that but does not show capsid oligomer interfaces.

Identifying HLA-Associated Polymorphisms in HIV-1 Zabrina L. Brumme, Art F. Y. Poon, Jonathan M. Carlson, Bruce D. Walker HIV Molecular Immunology 2010, p31-44 PDF.

Trim5alpha hiv restriction factor protects Old World monkeys from HIV. Single mutation of human trim5alpha protects humans.

Recent insights into the mechanism and consequences of TRIM5α retroviral restriction.
Sastri J, Campbell EM.
AIDS Res Hum Retroviruses. 2011 Mar;27(3):231-8.
PubMed 21247355

Cyclophilin A (CypA) host protein binds to HIV capsid and is packaged in virions. Structure 1ak4 from 1996 shows the exact interaction with an HIV CA loop. Five of the 9 loop residues are highly conserved. Not sure what effect variants in the other 4 might have. Worth looking more into this and including cyclophilin A in the composite HIV core model. Without CypA incorporated into virions HIV replications is halted at the reverse transcription stage apparently due to an uncoating difficulty. All of the other primate SIV analogs except SIV-chimp do not use CypA. Also the HIV O (outlier) group does not need CypA but it does package it -- the only phenotypic difference of O group from M (main) group clades known in 1996.

Crystal structure of human cyclophilin A bound to the amino-terminal domain of HIV-1 capsid.
Gamble TR, Vajdos FF, Yoo S, Worthylake DK, Houseweart M, Sundquist WI, Hill CP.
Cell. 1996 Dec 27;87(7):1285-94.
PubMed 8980234

Mutations in HIV and MLV that make them susceptible or resistant to human TRIM5alpha restriction.

Homology-based identification of capsid determinants that protect HIV1 from human TRIM5α restriction.
Maillard PV, Zoete V, Michielin O, Trono D.
J Biol Chem. 2011 Mar 11;286(10):8128-40. Epub 2010 Dec 17.
PubMed 21169362

Drug HIV CA assembly/disassembly inhibitor PF-3450074 with crystal structure 2xde and several known drug resistant HIV mutations.

HIV capsid is a tractable target for small molecule therapeutic intervention.
Blair WS, Pickford C, Irving SL, Brown DG, Anderson M, Bazin R, Cao J, Ciaramella G, Isaacson J, Jackson L, Hunt R, Kjerrstrom A, Nieman JA, Patick AK, Perros M, Scott AD, Whitby K, Wu H, Butler SL.
PLoS Pathog. 2010 Dec 9;6(12):e1001220.
PubMed 21170360

More Material

Scooter will get host-pathogen interaction data from Nevan Krogan and see about making a Cytoscape demo segment related to host proteins that interact with the capsid. Cyclophilin A and TRIM5alpha are candidate proteins. Such a segment could replace the residue interaction network or protein similarity network to show clades proposed above.

Investigate what is known about cyclophilin A binding to HIV capsid. It is a host protein necessary for virus replication, and known binder of capsid protein.

May want to include BLAST search of HIV CA in demo.

Another small molecule inhibitor of CA is CAP-1. And a helical peptide inhibitor that interferes with CTD dimerization is CAI with cell permeable derivative NYAD-1.

Bevirimat is a small-molecule drug inhibiting CA-SP1 cleavage by HIV protease. It was tested in phase II clinical trials. Supposedly the only CA targeting drug to reach clinical trials. A single mutation in SP1 confered resistance and spontaneously arose in 3-4 weeks. Other mutations in SP1 and one in CA also confer resistance. Probably not worth including in demo since it targets the SP1 cleavage site rather than CA itself.

Mark Yeager described in a talk a few months ago a human cellular protein that may form large lattices that enmesh HIV capsid and may represent a general host defense against retroviruses. Don't know if that is published. Would be a neat "hypothesis model" example. This is TRIM5alpha and has been published.

There is a great deal of recent published HIV capsid literature I have not yet looked at.

HIV sequences, and resistant patient epitopes are at the Los Alamos HIV database.