NEWS & VIEWS VACCINOLOGY
A sweet cleft in HIV’s armour The structure of an antibody that potently neutralizes a wide range of HIV-1 strains, together with a minimal antigen mimic, is an advance towards the design of vaccines that may elicit protective responses. S A . Q U E N T I N J . S AT T E N TA U
he glycoproteins on the surface envelope of HIV-1 are essential for viral attachment to, and entry into, appropriate host cells. Antibodies that bind tightly to functional envelope glycoproteins neutralize virus infectivity and can protect against infection in vivo1. HIV-1 has, in turn, evolved various defence mechanisms against antibody neutralization, most prominent among which are amino-acid sequence variation and the masking of otherwise vulnerable regions on the envelope glycoproteins with sugars called glycans2. Most such immune-evasion measures have minimal effect on virus infectivity, but there are certain ‘sites of vulnerability’ on the envelope glycoproteins that the virus must keep relatively conserved and exposed to maintain function. In this issue, McLellan et al.3 define a new site of vulnerability on the HIV-1 surface glycoprotein gp120 by solving the atomic structure of a monoclonal antibody, PG9, in complex with a minimal antigenic mimic of gp120 (see page 336). PG9 neutralizes a broad range of HIV-1 strains with high potency4. Therefore, its epitope — the exact part of the gp120 antigen to which it binds — is of particular interest for vaccine design. This epitope resides within a highly variable structure called the V1/V2 domain. This domain is mostly obscured by glycans, and may form the ‘cap’ that brings three gp120 molecules together into a trimer. This, in turn, is associated with the trimer of a second envelope glycoprotein, gp41 (Fig. 1a). McLellan et al. wanted to know where exactly in the V1/V2 domain PG9 binds. But solving the structure of PG9 in complex with its antigen was not a trivial task, because the existence of the epitope depends on the quaternary folding of gp120, to the extent that full antibody binding has been achieved only using intact, membrane-anchored envelope glycoproteins4. As an alternative approach, the authors used molecular modelling of V1/V2 coupled with information about PG9 binding to peptide mimics to select a molecular ‘scaffold’ on which V1/V2 could be restrained in the correct shape. This approach delivered an antigenic mimic of the epitope to which PG9 bound tightly. The mimic consisted
‘Hammerhead’ CDR H3
V1/V2 domain Glycan contact surfaces
Protein contact surface Glycan
CDR H3 contact surfaces
Scaffold Minimal V1/V2 mimic
Figure 1 | Zooming in on the PG9–HIV-1 interaction. a, On the surface of the HIV-1 envelope, three copies of gp120 bind to three copies of gp41 to form a trimeric envelope-glycoprotein complex. The PG9 epitope is located in the V1/V2 domain of gp120, near the tip of this complex and is shrouded by a ‘glycan shield’. b, McLellan et al.3 solve the atomic structure of the PG9 antibody in complex with a scaffolded segment of the V1/V2 domain from two HIV-1 strains (see Fig. 1 of the paper3). To do so, the authors constructed a mimic of the antibody’s minimal natural antigen. The mimic consists of a scaffold that constrains a polypeptide and two glycans such that a miniature canyon is formed. The researchers find that PG9 binds to its epitope mainly through the unusually long CDR H3 loop in its heavy chain, which forms a hammerhead-like structure. CDR H3 inserts between the glycans, contacting both, and interacts with the peptide at its tip. This allows tight binding of the antibody to these conserved antigen surfaces, accounting for its exceptional potency and breadth of HIV-1 neutralization.
of a peptide strand, flanked by two glycans, contained within a four-strand mini-domain. Both PG9 and the closely related antibody PG16 have an unusual structure dominated by an antigen-binding site, which contains an extended region, called CDR H3, of the antibody’s heavy chain. This has a unique hammerhead-like structure5,6. McLellan and colleagues3 report that the hammerhead of the PG9 CDR H3 slots in neatly between the two glycans on gp120, contacting the flanking glycan surfaces and binding to the protein surface at the base of the ‘mini-canyon’ (Fig. 1b). PG9 overcomes variability intrinsic to the V1/V2 protein sequence by binding mainly to the atomic backbone of the protein rather than to the variable amino-acid side chains. This, along with strong evolutionary conservation of the two glycans in gp120, results in its recognizing some 80% of HIV-1 strains. Thus, PG9 is an example of the immune system turning the tables on the viral glycan defences, by binding directly to them.
3 2 4 | NAT U R E | VO L 4 8 0 | 1 5 D E C E M B E R 2 0 1 1
Another site of vulnerability was previously identified7 in gp120. This site, which binds the CD4 receptor on the host-cell surface, was subsequently used to isolate antibodies that neutralize a broad range of HIV-1 strains8. Although a valid template for vaccine design, the CD4-binding surface has intrinsic complexities that make its use for eliciting antibodies difficult. The surface is highly convoluted, and, until a couple of months ago9, attempts to reproduce it as a polypeptide mimic were largely unsuccessful. Moreover, immunization with gp120 has so far failed to induce neutralizing antibodies of required breadth and potency, most likely because the elicited antibodies could not bind tightly to the assembled trimer of envelope glycoproteins10. The PG9 epitope is highly accessible on the envelope glycoprotein trimer, and so is not beset by the difficulties encountered in attempting to raise broadly neutralizing antibodies to the partly occluded CD4-binding surface on gp120. Moreover, by virtue of its
NEWS & VIEWS RESEARCH structure in a different region of gp120 called the V3 loop11. This suggests that the production of glycopeptide-reactive, broadly neutralizing antibodies may not be a particularly rare event during HIV-1 infection. The convergent evolution in antibody recognition also suggests that the immune system is rising to the challenge of generating antibodies with appropriate shapes to fit the structural constraints required to pierce the gp120 glycan shield and interact with protein surfaces beneath. Vaccination relies on triggering the immune system’s B cells to produce antibodies. Will vaccination with appropriately designed antigen mimics such as that described in this paper3 elicit equivalent antibody responses to those seen during HIV-1 infection? This cannot be answered yet. What we do know is that B cells from some individuals infected with HIV-1 recognize this type of glycopeptide epitope after stringent selection and a substantial degree of mutational evolution to
S OF T M ATERI AL S
Marginal matters Most soft materials, such as sand, can be in either a solid-like or a liquid-like state. New experiments probe the surprisingly rich nonlinear physics that can occur in between these two states. S L . V I N C E N Z O V I T E L L I & M A R T I N VA N H E C K E
ll around us, things are falling apart. The foam on our cappuccinos looks solid, but gentle stirring irreversibly changes its shape. Sand, a granular material, mimics a solid when we walk on the beach but a liquid when we pour it out of our shoes. Such examples suggest that we can think of the mechanics of soft disordered materials as either jammed (solid-like) or unjammed (freely flowing). On page 355 of this issue, however, Bi et al.1 describe experiments showing that such materials come in more than just these two flavours. The hybrid behaviour of sand and foams has fascinated physicists for decades, and lies at the heart of Liu and Nagel’s jamming diagram2 (see Fig. 1a of the paper1). The idea behind this diagram is that the mechanical state of a whole range of soft materials — such as granular media, pastes, foams and emulsions — is controlled by how densely their constituents (grains, bubbles or droplets) are packed. Dense packings are jammed, loose packings are unjammed, and when the constituent particles just touch, the material is said to be marginal. Think of mayonnaise, an emulsion of oil droplets in water: only when a sufficient amount of oil is added to the mixture do the oil droplets start to touch and the
mayonnaise acquires its solid-like consistency. Bi and colleagues’ experiments1 suggest that the jamming diagram should be revisited to describe a broader variety of states than the strictly jammed and unjammed ones. The authors placed a granular material consisting of small plastic discs in a box whose side walls could be moved in order to compress or shear (deform without compression) the material. The discs were photoelastic and so permitted direct visualization of the forces operating a
increase antibody binding strength for antigen. Whether vaccination can select the appropriate B-cell specificities and drive sufficient antibody evolution remains to be seen, but this is likely to be the major remaining hurdle on the way to an antibody-based HIV-1 vaccine. ■ Quentin J. Sattentau is in the Sir William Dunn School of Pathology, University of Oxford, Oxford OX1 3RE, UK. e-mail: email@example.com 1. Burton, D. R. et al. Proc. Natl Acad. Sci. USA 108, 11181–11186 (2011). 2. Wei, X. et al. Nature 422, 307–312 (2003). 3. McLellan, J. S. et al. Nature 480, 336–343 (2011). 4. Walker, L. M. et al. Science 326, 285–289 (2009). 5. Pejchal, R. et al. Proc. Natl Acad. Sci. USA 107, 11483–11488 (2010). 6. Pancera, M. et al. J. Virol. 84, 8098–8110 (2010). 7. Zhou, T. et al. Nature 445, 732–737 (2007). 8. Wu, X. et al. Science 329, 856–861 (2010). 9. Azoitei, M. L. et al. Science 334, 373–376 (2011). 10. Chen, L. et al. Science 326, 1123–1127 (2009). 11. Pejchal, R. et al. Science 334, 1097–1103 (2011).
between them when the material was subjected to deformation: the more incident light that went through the material, the larger the forces3. The first question Bi et al. addressed was, what happens when such granular material is sheared? Typically, when we spread mayonnaise on a sandwich, smear foam on our skin or kick a sandcastle, we unjam these materials. By contrast, the authors1 find that, when sheared under constant volume, collections of loose, unjammed discs build up pressure and resist further deformation — they become rigid. This phenomenon is reminiscent of Reynolds’ dilatancy, in which granular materials expand when sheared under constant pressure. A familiar manifestation of this is how footprints on a wet beach tend to become dry as the deformed sand expands and sucks in water. The second question Bi et al. tackled was, what is the nature of the novel states in which mechanical rigidity is generated by b
Figure 1 | Fragile and shear-jammed states. Bi and colleagues1 prepared a granular material consisting of a mixture of plastic discs of diameter 0.74 and 0.86 centimetres and subjected it to shear deformation. a, For small applied shear deformation, the material becomes fragile: the discs come into contact with each other mostly in one direction (here, vertical). b, For large applied shear deformation, the material is said to be shear-jammed: the discs come into contact with each other in more than one direction. The intensity of light seen on each grain is proportional to the amount of force it experiences. 1 5 D E C E M B E R 2 0 1 1 | VO L 4 8 0 | NAT U R E | 3 2 5
J. ZHANG & R. BEHRINGER, DUKE UNIV.
simplicity, the mini-domain V1/V2 epitope designed by McLellan et al. may well be a more straightforward target for re-eliciting potent neutralizing antibodies by active vaccination. However, uncertainties remain. First, does the scaffolded V1/V2 mimic reported here accurately represent the respective molecular architecture of the actual gp120 trimer? The answer would come with the structure of the trimeric envelope glycoprotein, although the reduced affinity of PG9 for the scaffolded mimic3 compared with that estimated for the trimeric glycoproteins suggests differences. Second, it is not clear how common this type of interaction between antibodies and glycopeptide antigens is. But there is reason to be optimistic. McLellan et al. propose3 that the epitopes of two other related, but distinct, neutralizing antibodies (CH04 and PGT145) have similar structures. Moreover, another potent neutralizing antibody, PGT128, binds a remarkably similar type of glycopeptide
QUENTIN J. SATTENTAU VACCINOLOGY ‘Hammerhead’ CDR H3 324 | NATURE | VOL 480 | 15 DECEMBER 2011 Glycan contact surfaces CDR H3 contact surfac...