By Dean A. Haycock
Special To BioWorld Today
It has been years since computers joined pipettes and test tubes as standard lab equipment in biology and biotechnology labs.
If, however, you want a place at the forefront of certain areas of biological research today, it is advisable - and could soon become mandatory - that you get to know your computer better. Vaccine researchers already know this.
The technological developments in this field are made clear in a paper by Tiziana Sturniolo and others in the June issue of Nature Biotechnology. The article, "Generation of tissue-specific and promiscuous HLA ligand databases using DNA microarrays and virtual HLA class II matrices," describes a computer program that predicts whether a peptide will be able to induce an important type of immune response. This new tool could have significant effects on vaccine development.
"It's a different way of identifying epitopes [the simplest forms of chemical groups, present in complex foreign molecules, that can induce an immune response by combining with an antibody or T cell receptor]," said senior author J|rgen Hammer, section head of genomics at Hoffmann-La Roche Inc. in Nutley, N.J.
Ann De Groot, assistant professor of medicine at Brown University and CEO and president of the vaccine design company, Epivax, both in Providence, R.I., described the development of the successful algorithm as "a beautiful piece of work." She noted that there are a number of different research groups doing this type of work, each with its own approach that must be validated in vitro.
"I would say that vaccines are really taking off right now because of the new technologies, not just informatics, but also the availability of sequences to screen with the Genome Project," De Groot told BioWorld Today.
Specifically, the newly described software, named Tepitope, uses a matrix-based prediction algorithm to foretell whether a particular peptide derived from a pathogenic microbe or a cancer cell, for example, will be able to bind to major histocompatibility complex (MHC) class II molecules. Sticking up from the surfaces of immune cells called antigen-presenting cells, they "hold up" or display a peptide obtained from a protein produced by an invading, abnormal or otherwise unwanted entity in the body. Other immune cells, T cells, learn to recognize the displayed foreign peptide. This recognition results in an immune response that, if effective, kills anything that expresses such peptides. In humans, MHC molecules are designated HLAs, or human lymphocyte antigens.
Prediction Capability Speeds Prediction
The ability to predict which peptides will bind to an MHC molecule is important because not any peptide will bind well to any MHC molecule. It could take a long time at the lab bench trying to figure out if a particular foreign peptide binds well and would make a good immunogen or antigen for stimulating an immune response. The new approach will decrease that time considerably.
"A project that used to take a post-graduate student one year to complete - one protein painstakingly analyzed over the course of one year using mice and then human samples - can now be reduced to one month or even less; probably two weeks is more reasonable," De Groot said. "This is because you are able to reduce not only the time frame that you have to perform the work in but also the number of compounds you screen.
"Your initial screen would be the computer and that reduces by between 10- to 20-fold, if not more, the number of compounds that have to be screened in the test tube."
The program uses a small set of binding profiles to identify other peptides that are likely to bind to specific MHC molecules. Most importantly, the computer-based predictions described by co-author Hammer and his colleagues correlate well with the results of traditional binding experiments.
Sequences On Structures Key To Approach
"What's different about J|rgen Hammer's paper, is that he actually executed something we have been talking about for a while," De Groot said. "He recognized that the HLA molecules that present epitopes to T cells are structures. These structures, being constructed from genetic codes, borrow from one another," De Groot explained. "So, from within each of the structures, you might see [different] pockets. One structure might have borrowed a pocket from another one.
"He lined them all up - this wouldn't have been possible five years ago when we didn't know the sequences of those structures - figured out which pockets were the same for all structures and then basically derived a pattern matching each of the pockets. Then all he had to do was look at the sequences of the structures and figure out which pocket was present. Then he could mix and match the information he derived from his in vitro work and construct a new pattern."
The software might be used for identifying promising peptide structures now stored in databases containing genomic data of pathogens or cancer cells. To illustrate the potential power of their approach, the authors have used their software to directly analyze the information provided by DNA chip sequencing technology applied to colon cancer cells. The peptides selected as a result of this application might prove to be the basis of highly immunogenic cancer vaccines, according to the authors.
"Instead of starting from even a potential antigen, you could start on the genomic level to systematically profile them. You now can do this based on both the DNA chip technology and this new software," Hammer said.
The technology might profitably be applied to infectious disease research as well as to cancer biology, according to Hammer. "You can cover a large population and start looking for potential promiscuous T cell epitopes," he said. "Promiscuous epitopes are epitopes that could be presented by a large percentage of the human population, where vaccinations should go."