Role of antibody heavy and light chain interface residues in affinity maturation of binding antigens
Antibodies are diverse, modular proteins that play a vital role in the immune response. While many studies are aimed at elucidating the binding profiles of individual antibodies against viral proteins, several integral questions still remain. Antibody-antigen interactions are not a two-way interaction (antibody/antigen), but a three-way interaction, as antibodies have both a heavy and a light chain. Recent studies have shown that mature antibodies inhabit a wide range of angles, relative to how the heavy and light chains are oriented with respect to one another. I postulate that a well-defined relative orientation between the variable heavy chain (VH) and variable light chain (VL) is needed for an antibody to engage its target with maximum affinity, as the relative orientation defines the geometry of the paratope. I hypothesize that during Ab maturation mutations in the interface are introduced in an “allosteric” manner in that these interactions, though removed from the antibody-antigen interface, rigidify the VH-VL orientation as an additional mechanism of affinity maturation. I further hypothesize that germline antibodies have a more flexible VH-VL interface. This concept is consistent with our understanding that germline antibodies are capable of binding to a wide variety of epitopes. One thermodynamic consequence of affinity maturation in the VH-VL interface would then be to decrease the entropic cost inherent to antigen binding, increasing affinity for the epitope by locking in the VH-VL orientation.
The long-term goals for this project are to identify naturally occurring pairing patterns of antibody heavy and light chains following infection with influenza or vaccination, to execute and experimentally verify a protocol for the computational design of novel antibodies, and ultimately inform rational antibody design and vaccine development by interrogating the relationship between thermodynamic stability of an antibody and affinity to its target. This study will further elucidate the sequence, structure, and function relationship of proteins, which is key for the design of stable and functional antibodies. While an increasing number of co-crystal structures are becoming available, it is still unknown whether the antibodies in complex are optimal in sequence and structure in terms of affinity for the target. HAs of many influenza subtypes have not yet been co-crystallized with an antibody, and these studies have the ability to discern whether co-crystal structures illuminate epitopes on the surface of certain HA subtypes that can be exploited to create novel antibodies for other HA subtypes. This study will also provide important new structural data for formulating hypotheses about how antibodies interact with their paratope; complementing the known antibody space with novel, designed antibody sequences will identify the elements required for HA binding and neutralization of influenza, which in turn has the potential to guide vaccination strategies directed towards the production of broadly neutralizing antibodies.
The aims described in this proposal seek to accomplish three goals: 1) execute and experimentally verify a protocol for the computational design of antibodies that exhibit broad cross-reactivity in binding, 2) explore the relationship between the thermodynamic stability of an antibody and the affinity for its target, and 3) determine if restrictions or patterns for the pairing of heavy and light chains limit antibody-antigen interactions.
Progress Report April 2018
Previous studies have shown that mutations in the interface between heavy and light chains can disrupt the binding affinity of the antibody to its target protein; these residues are integral in determining the geometry of the paratope, and modifications of residues in the interface as far away as 12 Å from the CDRs can cause a loss of binding. Dunbar et al. (PEDS 2013) developed a computational tool (ABangle) that characterizes VH–VL orientation using five angles (HL, HC1, LC1, HC2 and LC2) and a distance metric (dc). This study found that antibodies are capable of a high degree of flexibility between bound and unbound states. Aim 2 uses highly mutated anti-HIV antibody co-crystals to explore how interactions in the VH-VL interface alter binding affinity; instead of approaching this problem through the introduction of new residues in the interface, the main focus of this aim is to interrogate interfaces that have already included mutations in this region. After characterization of naturally occurring mutations in the interface, this aim proposes three modes of design: 1) The VH side will be optimized and mutants will be generated. 2) The VL region will be optimized and these mutants will be characterized. 3) Rosetta will be employed to optimize both sides of the interface simultaneously and the resulting candidates will be characterized. Using ABangle, I was able to identify a subset of anti-HIV antibodies that maintain a unique orientation in their bound conformation. VRC01-like antibodies are those that bind the gp120 CD4-binding site, mimicking the CD4 immunoglobulin fold by inserting the heavy chain variable domain into the binding site. These antibodies show a markedly different HC1 angle, and the VH-VL interface is on average tighter than that of a non-HIV antibody. Using a panel of 10 highly mutated anti-HIV antibody co-crystal structures, Rosetta was employed to revert mutated residues in the VH-VL interface and determine the effect of these mutations on the stability of the interface (ΔΔG). This was accomplished through rigid-body modeling, which prevents disruption of the backbone and preserves the VH-VL orientation. This allows me to look at the contribution of each residue in the interface while the antibody is in its bound conformation. Fifty models are generated for each complex, and an average is taken of the top 2%. Using the germline revertant interface ΔΔG as a baseline, it was determined that the wild-type interface has a higher overall stability, which suggests that these mutations may be incorporated to compensate for the entropic loss inherent to binding. Using Biolayer Interferometry, the binding kinetics of pairs of interface revertant and mature antibodies were determined. Antibodies with reversions in the VH-VL interface show a decreased Kon (association), and this may in turn be caused by an increase in conformational flexibility; the antibodies Identifying the effects of VH-VL mutations on orientationIn order to explore the effects of mutations in the VH-VL interface, I expanded the dataset to 15 antibodies and used Rosetta’s relax function to model the effects of these changes. Rosetta’s relax function energetically minimizes the structure by introducing small perturbations in the backbone and allowing the sidechains to repack; this protocol allows for antibodies to alter their conformation in a manner that accommodates changes in sequence. However, it was determined that the relax function created models with unrealistic orientations. In order to rectify this problem, rigid-body small-perturbation docking was performed on both the apo structures and antigen-bound conformations. In order to quantify the change in orientation that occurs upon reversion of mutations in the interface 1000 models were made for each antibody and structure (native or germline revertant). The top scoring (most thermodynamically stable) 5% of each condition was used to calculate the magnitude of shift in mean angle and the tightening of each distribution. The average of these measures were used as a means of estimating the overall shift and tightening in orientation that an antibody undergoes during maturation in the interface. The change in overall thermodynamic stability (ΔE), a normalized value for the shift in orientation, and a normalized value for the differences in standard deviation between distributions for each angle were also calculated. Using these calculations as a way to estimate the change in orientation upon acquisition of the mutations in the interface, we now have a way to rank the effects we expect to see in vitro and the models that were generated may serve as a valuable tool for explaining any differences in their thermodynamic binding profiles. I expect antibodies like VRC06 (3se9), which show a large shift in the orientation, significant tightening in the distribution of angles, and an increase in thermodynamic stability to bind their targets with greater affinity than their germline-reverted counterparts due to a reduction of conformational entropy.
Primary: James E. Crowe, Jr.Secondary: Jens MeilerType of Trainee