Biology Reference
In-Depth Information
of proteins, should be modified to specification to achieve favorable
structural features for a predetermined function. The mathematical
abstraction of favorable property features for the design of artificial
amino acids, along with the testing of these engineered peptides/proteins
centered on a particular function(s), could make possible smart and
fast engineering of biomolecules with pre-specified function. To
search and test for potential artificial antigens/immunogens, the
peptide/protein library synthesized with modified amino acids
would be mixed with the viral particle of interest. Those that bind
(
win
) would stick to the viral particle, and those that do not (
lose
)
would be washed away. The winning ligands can be selectively used
as seed candidates for another round to build more
in silico
lead
libraries. With enough candidate samples, their activity-based infor-
mation can be used to train the ANN. ANN-selected (predicted)
candidates can then be validated experimentally. Thus, the ANN
“knowledge” and the peptide/protein library, evolve through the
iterative process: predicting binding properties and building
in silico
libraries from ANN-selected “seed” candidates. Such a strategy may
allow researchers to develop agents that cover the entire sequence
space of the virus. It may also help generate more stable and long-
lasting agents because, in contrast to peptides/proteins made from
the naturally occurring amino acids, modified amino acids may not
be as quickly recycled by the body.
Schultz and colleagues have been able to expand the genetic code
by building new building blocks (primarily in
E. coli
) for site-specific
incorporation of artificial amino acids into peptides/proteins.
233-239
Presently, more than 30 artificial amino acids have been genetically
encoded through unique triplet,
234,235
quadruplet,
235,240-243
and quin-
tuplet
244,245
codons. Artificial peptides have already been successfully
tested for their binding affinities to proteins of the MHC class I in
T cells. Krebs, Folkers and Rognan (1998) synthesized and modified
peptides to optimally bind to the MHC I molecule HLA-B*2705 by
filling a hydrophobic binding pocket (pocket D) with artificial aro-
matic amino acids (alpha-naphthylalanine, betanaphthylalanine, and
homophenylalanine), whose side chains in position 3 had been
altered.
246
