Information Technology Reference
In-Depth Information
6.3.1
Available Epitope Mapping Tools
Alteration of T-cell epitopes is known to result in reduced binding to the MHC
and/or altered binding to the T-cell receptor (TCR). This effect has been observed
both for Class I and Class II MHC ligands in the context of both tumor cell (Scanlan
and Jager 2001) and pathogen (Mullbacher 1992; Hill, Jepson, Plebanski, and Gilbert
1997) escape from immune response. Mapping, confirming, and modifying T-cell
epitopes may reduce the immunogenicity of therapeutics.
Based on the hypothesis that the immunogenicity of therapeutic proteins is
probably linked to both (1) the presence of T helper epitopes and (2) an event that
triggers immune response (the “danger signal”), it follows that removal of T-cell
epitopes from a protein that is intended to be used as a therapeutic may reduce the
protein's overall potential to stimulate T cells. In order to modify T-cell epitopes, it
is important to first identify those epitopes that are responsible for stimulating an
immune response and then determine their amino acid sequences. Currently, a num-
ber of epitope-mapping tools are available for discovering T-cell epitopes contained
within protein sequences. Reviews of these tools have been published (De Groot and
Martin 2003a). The following paragraphs provide background on mapping tools
developed by the authors of this article.
Prior to the development of tools for T-cell epitope selection, the cost and
effort required to identify T-cell epitopes from protein sequences was a significant
barrier to the deimmunization of therapeutic proteins. Computational immunology
(immunoinformatics) methods dramatically reduce the time and effort involved in
screening proteins for potential epitopes, ranging from a reduction of 10- to 20-
fold (Kast, Brandt, Sidney, Drijfhout, Kubo, Grey, Melief, and Sette 1994;
Schafer, Jesdale, George, Kouttab, and De Groot 1998) to a 95% reduction
(De Groot, Bosma, Chinai, Frost, Jesdale, Gonzalez, Martin, and Saint-Aubin
2001a; De Groot, Saint Aubin, Rayner, and Martin 2001b).
EpiMatrix, an algorithm developed by the Brown University TB/HIV Research
Lab and licensed to EpiVax, ranks 9- to 10- amino-acid-long segments overlapping
by 8 to 9 amino acids derived from any protein sequence by estimated probability of
binding to a selected MHC molecule. The EpiMatrix method for ranking prospective
epitopes has been published (Schafer et al. 1998; De Groot, Jesdale, Szu, and Schafer
1997). Matrix motifs for 24 HLA Class I alleles are available for use with EpiMatrix.
EpiVax used the pocket profile approach first described by Sturniolo and Hammer
(Sturniolo, Bono, Ding, Raddrizzani, Tuereci, Sahin, Braxenthaler, Gallazzi, Protti,
Sinigaglia, and Hammer 1999) to generate predictive matrices for 74 Class II alleles
(De Groot et al. 1997) These new Class II matrices are now included in the EpiMa-
trix repertoire at EpiVax. Previous studies have demonstrated that EpiMatrix accu-
rately predicts published MHC ligands and T-cell epitopes (De Groot et al. 2001a;
De Groot et al.1997; De Groot, Jesdale, Martin, Saint-Aubin, Sbai, Bosma, Lieber-man,
Skowron, Mansourati, and Mayer 2003b).
ClustiMer, which is an optional feature of EpiMatrix, can measure and store
the MHC binding potential for a 9- or 10- amino-acid sequence to a number of
human HLAs. ClustiMer can therefore be used to identify clustered or “promiscuous”
