Information Technology Reference
In-Depth Information
Fig. 4.
EpiMatrix analysis of human beta-interferon predicted epitope clusters. For this analy-
sis, the complete sequence of human beta-interferon was downloaded from Genbank (acces-
sion 1AU1_A). The sequence was parsed into 158 overlapping 9-mer frames. Each frame was
then evaluated for its potential to bind to 8 common Class II HLAs (DRB1*0101, DRB1*0301,
DRB1*0401, DRB1*0701, DRB1*0801, DRB1*1101, DRB1*1301, and DRB1*1501). The
entire dataset was then scanned for frames that contain more predicted ligands than would be
expected by chance alone. In this case any 9-mer frame that is predicted to react to 3 or more
alleles defines a cluster. The binomial probability of observing 3 positive signals (a positive
signal is defined as an EpiMatrix Z-score in excess of 1.64) in 8 trials is 0.0058. Contiguous
and nearly clusters were then combined to yield the 7 clusters displayed in the graph.
6.4.2 Modifying Epitopes
Regions of epitope clustering are often highly immunogenic (Pevear, Luo, and
Lipton 1988), thus the first step in deimmunizing a protein is to address the clustered
regions defined by EpiMatrix. For example, Tables 4a and 4b show the last 20 over-
lapping frames of an analysis for human CLIP (an HLA binding portion of Invariant
chain) and a modified version of CLIP. Putative epitopes (HLA Class II binding
regions) located in the last cluster of epitopes are shown. Each overlapping frame in
the entire amino acid sequence is scored against each of the eight common HLA
alleles. The total number of epitopes in the natural and modified versions and the
highest score per allele are shown at the bottom of each table. For example, by sub-
stituting alanine for critical residues the total number of DRB1*0101 epitopes is
reduced from three to none and no increase in the number of putative HLA binding
regions develops. Confirmation of
in silico
immunogenicity prediction is subse-
quently performed
in vitro
.
