Computational Methods for Mapping Mimotopes to Protein Antigens
Mimotopes are molecules - often peptides - that are recognized by antibodies or T cells that were primed to recognize
a different epitope. For antibody epitope mapping, peptide mimotopes can be selected from a combinatorial library (usually
peptides from a specific protein antigen) on binding to a specific monoclonal antibody with high affinity; they are commonly
obtained using page display. Experimental data on mimotopes can be found in the IEDB, MimoDB
(http://immunet.cn/mimodb/index.html)
and journal publications. Mimotopes can mimic essential features of genuine epitopes and therefore their mapping to the antigen
is a logical and essential step of the mimotope analysis. This page provides information on available methods for mimotope mapping,
how to search the IEDB for mimotopes, and an example of a mimotope dataset and the results of its mapping, using the available
web servers hosted outside the IEDB.
Methods available as web servers and operational as of July 2014 (they all require the user submitting an antigen protein 3D structure, providing either PDB ID of the structure or the pdb file)
Other published methods (unavailable as web servers)
If an antigen sequence is not available
Publications on mimotope mapping tools evaluation
How to search for mimotopes in the IEDB (note that IEDB does not contain HIV-related data)
Example of the mimotope mapping to the protein with known structure
For PMID: 23006741, the IEDB provides 16 mimotopes recognized by a monoclonal antibody directed against the vaccinia virus A33 protein, MAb-1G10. The mimotope mapping and subsequent experimental analysis presented in this paper showed that D115 and L118 were essential epitope residues. Inspection of the GenBank record for VACV A33, or the EEV membrane phosphoglycoprotein, http://www.ncbi.nlm.nih.gov/protein/38349023, reveals that the 3D structure of this protein is available and has PDB ID 3K7B, in which chains A and B represent a protein dimer.
Here is the list of 16 mimotopes as provided in IEDB:
Below are the results of the mapping of the above provided mimotopes using the MIMOX, Pepitope (three methods) and EpiSearch servers. The task of mapping these particular mimotopes is challenging as peptides were obtained from a combinatorial library and had no matches in the VACV A33 protein. Five epitope residues predicted by all methods are in bold; they however do not include residues D115 and L118, which are functionally essential epitope residues.
MIMOX
We used http://immunet.cn/mimox/cgi-bin/mimox2.pl, submitting to its form the peptide sequences (above) and specifying PDB ID 3K7B and chain A.
Results:
Best cluster (37 out of 88 protein residues), Center Residue P134, Score 0.98:
L103, Y105, Q106, S108, C109, Y110, I111, L112, H113, A123, S131, T132, L133, P134, N135, K136, S137, D138, V139, T142, I145, Y147, E149, D150, T151, S154, D155, G156, N157, P158, R176, K177, Y178, F179, C180, V181, K182
The inspection of the 3D image of the structure with the mapped predicted epitope showed that the epitopes is too large and lies at both the protein surface and interior and on opposite sides of the protein.
Pepitope server at http://pepitope.tau.ac.il/
As above, we used PDB ID 3K7B and chain A. The server requires submitting peptides in the fasta-format and doesn't accept sequences longer than 14 aa, therefore the following 14 peptides were submitted:
Results
PepSurf:
Best cluster (21 out of 88 protein residues), Score 28.016: CYS100, ASN101, GLY102, LEU103, SER108, CYS109, ILE111, LEU112, HIS113, SER114, ASP115, TYR116, THR127, ALA128, GLU129, SER130, SER131, THR132, LYS177, VAL181, LYS182
MAPITOPE:
Best cluster (10 out of 88 protein residues): GLU98, SER99, CYS100, ASN101, GLY102, LEU103, TYR104, TYR105, GLN106, GLY107, SER108, CYS109, ILE111, VAL181
Combined:
(8 out of 88 protein residues, intersects cluster number 1 of PepSurf with cluster number 1 of Mapitope): CYS100, ASN101, GLY102, LEU103, SER108, CYS109, ILE111, VAL181
The predicted epitope is compact and located on the protein surface.
EpiSearch (http://curie.utmb.edu/episearch.html)
The server doesn't allow to provide PDB ID, the user therefore needs to download pdb text file from http://www.pdb.org/pdb/explore/explore.do?structureId=3k7b and modify it in any text editor by removing all lines starting with ATOM and containing chain B, leaving only chain A.
Results:
Best cluster (24 out of 88 protein residues), Center Residue V181, Score 14.22: C100, G102, L103, Y104, Y105, Q106, G107, S108, C109, I111, L112, H113, A123, T127, E129, S130, S131, T132, N135, D138, S154, V181, K182, T183
Methods available as web servers and operational as of July 2014 (they all require the user submitting an antigen protein 3D structure, providing either PDB ID of the structure or the pdb file)
- MIMOX (http://immunet.cn/mimox/)
- Pepitope server (http://pepitope.tau.ac.il/) provides three algorithms: Mapitope, PepSurf, or combined
- EpiSearch (http://curie.utmb.edu/episearch.html)
Other published methods (unavailable as web servers)
- PepMapper (https://pubmed.ncbi.nlm.nih.gov/22701536) (2012) - provided URL didn't respond
- MimoPro (https://pubmed.ncbi.nlm.nih.gov/21609501) (2011) - provided URL didn't respond
- Pep-3D-Search (https://pubmed.ncbi.nlm.nih.gov/19087303) (2008) - available for download at http://kyc.nenu.edu.cn/Pep3DSearch/
- MEPS (https://pubmed.ncbi.nlm.nih.gov/17430573) (2007)
- MIMOP (https://pubmed.ncbi.nlm.nih.gov/16434442) (2006)
- 3D-Epitope-Explorer (3DEX) (https://pubmed.ncbi.nlm.nih.gov/15834923) (2005) - available for download at http://www.schreiber-abc.com/3dex/
- Enshell-Seijffers et al. (https://pubmed.ncbi.nlm.nih.gov/14596802) (2003)
- FINDMAP (https://pubmed.ncbi.nlm.nih.gov/12935344) (2003)
- SiteLight (https://pubmed.ncbi.nlm.nih.gov/12824481) (2003)
If an antigen sequence is not available
- Use The Protein Modeling Portal at http://www.proteinmodelportal.org/ to attempt to obtain the 3D model of the antigen.
- Use tools described in the protocol published at https://pubmed.ncbi.nlm.nih.gov/25048128, consider the protocol for initial data cleaning (depending on the experimental system).
Publications on mimotope mapping tools evaluation
- http://pepitope.tau.ac.il/overview.html#Validation
- https://pubmed.ncbi.nlm.nih.gov/21681149 (2011) describes a benchmark dataset of 18 peptide sets with 13 known 3D structures of antibody-protein complexes and 29 peptide sets with 19 known 3D structures of protein-protein complexes (other than antibody); five methods have been evaluated: Mapitope, PepSurf, EpiSearch, Pep-3D-Search (authored by the paper's authors), Pepitope (combined Mapitope and PepSurf). For antibody-protein complexes, the best MCC (Mathew Correlation Coefficient) of 0.143 demonstrated Pepitope (sensitivity of 0.212, specificity 0.990), and the best sensitivity of 0.455 showed Pep-3D-Search (specificity 0.804 and MCC of 0.085).
How to search for mimotopes in the IEDB (note that IEDB does not contain HIV-related data)
- To retrieve all curated mimotopes, use "B Cell Search" in "Advanced Search" drop-down menu, applying the following parameters:
- Epitope Object Subtype equals 'Linear peptide'
- Related Object Related Object equals 'The epitope is a mimotope of:'
- To retrieve mimotopes for a specific protein use the following search parameters:
- Epitope Object Subtype equals 'Linear peptide'
- Related Object Related Object equals 'The epitope is a mimotope of:'
- Related Object Subtype equals 'Protein'
- To retrieve mimotopes from proteins published in journals after 2009 and obtained using monoclonal antibodies use the following search parameters:
- Year > 2009
- Epitope Object Subtype equals 'Linear peptide'
- Related Object Related Object equals 'The epitope is a mimotope of:'
- Related Object Subtype equals 'Protein'
- Antibody Purification Status equals 'Monoclonal'
Example of the mimotope mapping to the protein with known structure
For PMID: 23006741, the IEDB provides 16 mimotopes recognized by a monoclonal antibody directed against the vaccinia virus A33 protein, MAb-1G10. The mimotope mapping and subsequent experimental analysis presented in this paper showed that D115 and L118 were essential epitope residues. Inspection of the GenBank record for VACV A33, or the EEV membrane phosphoglycoprotein, http://www.ncbi.nlm.nih.gov/protein/38349023, reveals that the 3D structure of this protein is available and has PDB ID 3K7B, in which chains A and B represent a protein dimer.
Here is the list of 16 mimotopes as provided in IEDB:
ACCDDWPHEC
ACGDDWPHECGGGSAK
ACKPTWPFEC
ACNLYNEPFC
ACNLYNEPLC
ACQLKWPFEC
ACQTSYPYEC
ACSNSWPHEC
ACTSYNEPLC
ACTSYNEPLCGGGSAK
ACTVFNEPFC
ACWYDSLIFC
MKPALCEPLCG
NPYCEPVCQDWA
QLTAQHRDSLSS
SLMRELCEPRCE
Below are the results of the mapping of the above provided mimotopes using the MIMOX, Pepitope (three methods) and EpiSearch servers. The task of mapping these particular mimotopes is challenging as peptides were obtained from a combinatorial library and had no matches in the VACV A33 protein. Five epitope residues predicted by all methods are in bold; they however do not include residues D115 and L118, which are functionally essential epitope residues.
MIMOX
We used http://immunet.cn/mimox/cgi-bin/mimox2.pl, submitting to its form the peptide sequences (above) and specifying PDB ID 3K7B and chain A.
Results:
Best cluster (37 out of 88 protein residues), Center Residue P134, Score 0.98:
L103, Y105, Q106, S108, C109, Y110, I111, L112, H113, A123, S131, T132, L133, P134, N135, K136, S137, D138, V139, T142, I145, Y147, E149, D150, T151, S154, D155, G156, N157, P158, R176, K177, Y178, F179, C180, V181, K182
The inspection of the 3D image of the structure with the mapped predicted epitope showed that the epitopes is too large and lies at both the protein surface and interior and on opposite sides of the protein.
Pepitope server at http://pepitope.tau.ac.il/
As above, we used PDB ID 3K7B and chain A. The server requires submitting peptides in the fasta-format and doesn't accept sequences longer than 14 aa, therefore the following 14 peptides were submitted:
>
ACCDDWPHEC
>
ACKPTWPFEC
>
ACNLYNEPFC
>
ACNLYNEPLC
>
ACQLKWPFEC
>
ACQTSYPYEC
>
ACSNSWPHEC
>
ACTSYNEPLC
>
ACTVFNEPFC
>
ACWYDSLIFC
>
MKPALCEPLCG
>
NPYCEPVCQDWA
>
QLTAQHRDSLSS
>
SLMRELCEPRCE
Results
PepSurf:
Best cluster (21 out of 88 protein residues), Score 28.016: CYS100, ASN101, GLY102, LEU103, SER108, CYS109, ILE111, LEU112, HIS113, SER114, ASP115, TYR116, THR127, ALA128, GLU129, SER130, SER131, THR132, LYS177, VAL181, LYS182
MAPITOPE:
Best cluster (10 out of 88 protein residues): GLU98, SER99, CYS100, ASN101, GLY102, LEU103, TYR104, TYR105, GLN106, GLY107, SER108, CYS109, ILE111, VAL181
Combined:
(8 out of 88 protein residues, intersects cluster number 1 of PepSurf with cluster number 1 of Mapitope): CYS100, ASN101, GLY102, LEU103, SER108, CYS109, ILE111, VAL181
The predicted epitope is compact and located on the protein surface.
EpiSearch (http://curie.utmb.edu/episearch.html)
The server doesn't allow to provide PDB ID, the user therefore needs to download pdb text file from http://www.pdb.org/pdb/explore/explore.do?structureId=3k7b and modify it in any text editor by removing all lines starting with ATOM and containing chain B, leaving only chain A.
Results:
Best cluster (24 out of 88 protein residues), Center Residue V181, Score 14.22: C100, G102, L103, Y104, Y105, Q106, G107, S108, C109, I111, L112, H113, A123, T127, E129, S130, S131, T132, N135, D138, S154, V181, K182, T183