Today, he answers the question:
What are the techniques that your group has used to better understand polyspecificity?
Dr. Crowe: In order to sample the interacting energies of a large number of antibodies and antigens, it’s currently not possible to express thousands or millions of antibodies, at least not in an economic manner to measure energies. So, we’ve been focusing on using computational modeling experiments to determine the predicted interacting energies. In particular, we’ve been using a computational suite called Rosetta that was developed in Seattle. They have a Baker Lab and he’ll be talking at the IBC Conferences and going over the impact of Rosetta and structural modeling.
We’ve used Rosetta to ask: “What are the ideal or the optimal framework residues that can make antibodies have the ability to bind more than one antigen?” So, within Rosetta you cannot only model the binding to one antigen, but you can do a process called “multi-state design” in which you are asking the antibody to have more than one state – more than one bound state – which is binding to more than one antigen. By doing this, we asked Rosetta at many residues in the framework that we saw mutated in naturally occurring antibodies to just --- of the 20 amino acids that are available to predict which amino acids would allow binding to more than one antigen. Very remarkably, Rosetta predicted amino acids that are the germline encoded residues at those positions, which it is highly unexpected that a computer could pick at multiple positions the correct amino acid of 20 – an ensemble that would allow binding to more than one antigen.
So, the computational modeling is done on supercomputing. It is very computationally intensive. It predicted what the optimal sequences are and lo and behold, the optimal sequences for binding more than one antigen turned out to be the very germline sequences that we’re using in our genome. So, it looks like germline sequences are already in a state. It’s as if they are designed perfectly to be polyspecific and it may be that many – if not most – antibodies are polyspecific when they start.
The process of somatic mutation is, in essence, a two-step process. One in which the framework residues are mutated to reduce flexibility and rigidify the framework to orient the loops in a proper configuration. Then somatic mutations occur in the tips of the loops to adjust the interface so that their shape is complementarity and other interactions. So, really reducing flexibility achieves specificity and then you get some optimization of surface by somatic mutation.
So, we’ve focused on computational modeling, but the validation of the results of those computational experiments was the observation that the residues that are predicted are, in fact, the ones that are encoded by the genome. Even though the computer program didn’t know what those sequences were, it predicted them and they are the naturally occurring ones. So, we’ve been using Rosetta, trying to develop new modeling techniques within Rosetta and furthermore developing statistics to ask how expected or unexpected are the findings. It’s really quite remarkable that the computer can pick out the naturally occurring residues.
Dr. Crowe will be presenting Deep Sequencing the Human Antibody Response to Viral Infections and Human Germline Antibody Gene Segments Encode Polyspecific Antibodies. For more information on these sessions and the rest of the program, download the agenda. The Antibody Engineering & Therapeutics Event will take place December 8-12, 2013 in Huntington Beach, California. If you'd like to join us, as a reader of this blog,when you register to join us and mention priority code XD13172BLOGJP to save 20% off the standard rate.
|
Share this article with your social network, just click below to share now!
|
|
|
No comments :
Post a Comment