epitope prediction algorithms

Epitope prediction algorithms

Urmila Kulkarni-KaleBioinformatics Centre

University of Pune

October 2K5 © Bioinformatics Centre, UoP 2

Vaccine developmentIn Post-genomic era: Reverse Vaccinology Approach.

• Rappuoli R. (2000). Reverse vaccinology. Curr Opin Microbiol. 3:445-450.

October 2K5 © Bioinformatics Centre, UoP 3

Genome Sequence

Proteomics Technologies

In silico analysis

DNAmicroarrays

High throughputCloning and expression

In vitro and in vivo assays forVaccine candidate identification

Global genomic approach to identify new vaccine candidates

October 2K5 © Bioinformatics Centre, UoP 4

In Silico Analysis

Gene/Protein Sequence Database

Disease related protein DB

Candidate Epitope DB

VACCINOME

PeptideMultiepitope

vaccines

Epitope prediction

October 2K5 © Bioinformatics Centre, UoP 5

What Are Epitopes?

Antigenic determinants or Epitopes are the portions of the antigen molecules which are responsible for specificity of the antigens in antigen-antibody (Ag-Ab) reactions and that combine with the antigen binding site of Ab, to which they are complementary.

October 2K5 © Bioinformatics Centre, UoP 6

Types of Epitopes• Sequential / Continuous epitopes:

• recognized by Th cells

• linear peptide fragments

• amphipathic helical 9-12 mer

• Conformational / Discontinuous epitopes:

• recognized by both Th & B cells

• non-linear discrete amino acid sequences, come together due to folding

• exposed 15-22 mer

October 2K5 © Bioinformatics Centre, UoP 7

Properties of Epitopes

• They occur on the surface of the protein and are more flexible than the rest of the protein.

• They have high degree of exposure to the solvent.

• The amino acids making the epitope are usually charged and hydrophilic.

October 2K5 © Bioinformatics Centre, UoP 8

Methods to identify epitopes

1. Immunochemical methods• ELISA : Enzyme linked immunosorbent assay• Immunoflurorescence• Radioimmunoassay

2. X-ray crystallography: Ag-Ab complex is crystallized and the structure is scanned for contact residues between Ag and Ab. The contact residues on the Ag are considered as the epitope.

3. Prediction methods: Based on the X-ray crystal data available for Ag-Ab complexes, the propensity of an amino acid to lie in an epitope is calculated.

October 2K5 © Bioinformatics Centre, UoP 9

Antigen-Antibody (Ag-Ab) complexes• Non-obligatory heterocomplexes that are made

and broken according to the environment • Involve proteins (Ag & Ab) that must also exist

independently • Remarkable feature:

– high affinity and strict specificity of antibodies for their antigens.

• Ab recognize the unique conformations and spatial locations on the surface of Ag

• Epitopes & paratopes are relational entities

October 2K5 © Bioinformatics Centre, UoP 10

Antigen-Antibody complex

October 2K5 © Bioinformatics Centre, UoP 11

Ab-binding sites:Sequential & Conformational Epitopes!

Sequential Conformational

Ab-binding sites

Paratope

October 2K5 © Bioinformatics Centre, UoP 12

B cell epitope prediction algorithms :

• Hopp and Woods –1981• Welling et al –1985• Parker & Hodges - 1986 • Kolaskar & Tongaonkar – 1990• Kolaskar & Urmila Kulkarni - 1999

T cell epitope prediction algorithms :• Margalit, Spouge et al - 1987 • Rothbard & Taylor – 1988• Stille et al –1987• Tepitope -1999

October 2K5 © Bioinformatics Centre, UoP 13

Hopp & Woods method• Pioneering work• Based on the fact that only the hydrophilic

nature of amino acids is essential for an sequence to be an antigenic determinant

• Local hydrophilicity values are assigned to each amino acid by the method of repetitive averaging using a window of six

• Not very accurate

October 2K5 © Bioinformatics Centre, UoP 14

Welling’s method• Based on the % of each aa present in

known epitopes compared with the % of aa in the avg. composition of a protein.

• assigns an antigenicity value for each amino acid from the relative occurrence of the amino acid in an antigenic determinant site.

• regions of 7 aa with relatively high antigenicity are extended to 11-13 aa depending on the antigenicity values of neighboring residues.

October 2K5 © Bioinformatics Centre, UoP 15

Parker & Hodges method• Utilizes 3 parameters :

– Hydrophilicity : HPLC– Accessibility : Janin’s scale– Flexibility : Karplus & Schultz

• Hydrophilicity parameter was calculated using HPLC from retention co-efficients of model synthetic peptides.

• Surface profile was determined by summing the parameters for each residue of a seven-residue segment and assigning the sum to the fourth residue.

• One of the most useful prediction algorithms

October 2K5 © Bioinformatics Centre, UoP 16

Kolaskar & Tongaonkar’s method• Semi-empirical method which uses

physiological properties of amino acid residues

• frequencies of occurrence of amino acids in experimentally known epitopes.

• Data of 169 epitopes from 34 different proteins was collected of which 156 which have less than 20 aa per determinant were used.

• Antigen: EMBOSS

October 2K5 © Bioinformatics Centre, UoP 17

CEP Server

• Predicts the conformational epitopes from X-ray crystals of Ag-Ab complexes.

• uses percent accessible surface area and distance as criteria

October 2K5 © Bioinformatics Centre, UoP 18

An algorithm to map sequential and conformational epitopes of protein antigens of known structure

October 2K5 © Bioinformatics Centre, UoP 19

October 2K5 © Bioinformatics Centre, UoP 20

CE: Beyond validation• High accuracy:

– Limited data set to evaluate the algorithm– Non-availability of true negative data sets

• Prediction of false positives? – Are they really false positives?

• Limitation:– Limited by the availability of 3D structure data of antigens

Different Abs (HyHEL10 & D1.3) have over-lapping binding sites

October 2K5 © Bioinformatics Centre, UoP 21

CE: Features• The first algorithm  for the prediction of

conformational epitopes or antibody binding sites of protein antigens

• Maps both: sequential & conformational epitopes

• Prerequisite: 3D structure of an antigen

October 2K5 © Bioinformatics Centre, UoP 22

CEP: Conformational Epitope Prediction Serverhttp://bioinfo.ernet.in/cep.htm

October 2K5 © Bioinformatics Centre, UoP 23

T-cell epitope prediction algorithms

• Considers amphipathic helix segments, tetramer and pentamer motifs (charged residues or glycine) followed by 2-3 hydrophobic residues and then a polar residue.

• Sequence motifs of immunodominant secondary structure capable of binding to MHC with high affinity.

• Virtual matrices which are used for predicting MHC polymorphism and anchor residues.

October 2K5 © Bioinformatics Centre, UoP 24

• Case study: Design & development of peptide vaccine against Japanese encephalitis virus

October 2K5 © Bioinformatics Centre, UoP 25

We Have Chosen JE Virus, Because

JE virus is endemic in South-east Asia including India.

JE virus causes encephalitis in children between 5-15 years of age with fatality rates between 21-44%.

Man is a "DEAD END" host.

October 2K5 © Bioinformatics Centre, UoP 26

We Have Chosen JE Virus, Because

• Killed virus vaccine purified from mouse brain is used presently which requires storage at specific temperatures and hence not cost effective in tropical countries.

• Protective prophylactic immunity is induced only after administration of 2-3 doses.

• Cost of vaccination, storage and transportation is high.

October 2K5 © Bioinformatics Centre, UoP 27

Predicted structure of JEVSMutations: JEVN/JEVS

October 2K5 © Bioinformatics Centre, UoP 28

October 2K5 © Bioinformatics Centre, UoP 29

CE of JEVN Egp

October 2K5 © Bioinformatics Centre, UoP 30

• Loop1 in TBEV: LA EEH QGGT• Loop1 in JEVN: HN EKR ADSS • Loop1 in JEVS: HN KKR ADSS

Species and Strain specific properties:TBEV/ JEVN/JEVS

Antibodies recognising TBEV and JEVN would require exactly opposite pattern of charges in their CDR regions.

Further, modification in CDR is required to recognise strain-specific region of JEVS.

October 2K5 © Bioinformatics Centre, UoP 31

Multiple alignment of Predicted TH-cell epitope in the JE_Egp with corresponding epitopes in Egps of other Flaviviruses

426 457JE DFGSIGGVFNSIGKAVHQVFGGAFRTLFGGMSMVE DFGSVGGVFNSIGKAVHQVFGGAFRTLFGGMSWNE DFGSVGGVFTSVGKAIHQVFGGAFRSLFGGMSKUN DFGSVGGVFTSVGKAVHQVFGGAFRSLFGGMSSLE DFGSIGGVFNSIGKAVHQVFGGAFRTLFGGMSDEN2 DFGSLGGVFTSIGKALHQVFGAIYGAAFSGVSYF DFSSAGGFFTSVGKGIHTVFGSAFQGLFGGLNTBE DFGSAGGFLSSIGKAVHTVLGGAFNSIFGGVGCOMM DF S GG S GK H V G F G

Multiple alignment of JE_Egp with Egps of other Flaviviruses in the YSAQVGASQ region.

151 183JE SENHGNYSAQVGASQAAKFTITPNAPSITLKLGMVE STSHGNYSTQIGANQAVRFTISPNAPAITAKMGWNE VESHG‑‑‑‑KIGATQAGRFSITPSAPSYTLKLGKUN VESHGNYFTQTGAAQAGRFSITPAAPSYTLKLGSLE STSHGNYSEQIGKNQAARFTISPQAPSFTANMGDEN2 HAVGNDTG‑‑‑‑‑KHGKEIKITPQSSTTEAELTYF QENWN‑‑‑‑‑‑‑‑TDIKTLKFDALSGSQEVEFITBE VAANETHS‑‑‑‑GRKTASFTIS‑‑SEKTILTMG

Peptide ModelingInitial random conformationForce field: AmberDistance dependent dielectric constant 4rij

Geometry optimization: Steepest descents & Conjugate gradientsMolecular dynamics at 400 K for 1nsPeptides are:

SENHGNYSAQVGASQ NHGNYSAQVGASQ YSAQVGASQ

YSAQVGASQAAKFT NHGNYSAQVGASQAAKFTSENHGNYSAQVGASQAAKFT149 168

October 2K5 © Bioinformatics Centre, UoP 33

October 2K5 © Bioinformatics Centre, UoP 34

October 2K5 © Bioinformatics Centre, UoP 35

Relevant Publications & Patent• Urmila Kulkarni-Kale, Shriram Bhosale, G. Sunitha Manjari, Ashok

Kolaskar, (2004). VirGen: A comprehensive viral genome resource. Nucleic Acids Research 32:289-292.

• Urmila Kulkarni-Kale & A. S. Kolaskar (2003). Prediction of 3D structure of envelope glycoprotein of Sri Lanka strain of Japanese encephalitis virus. In Yi-Ping Phoebe Chen (ed.), Conferences in research and practice in information technology. 19:87-96.

• A. S. Kolaskar & Urmila Kulkarni-Kale (1999) Prediction of three-dimensional structure and mapping of conformational antigenic determinants of envelope glycoprotein of Japanese encephalitis virus. Virology. 261:31-42.

Patent: Chimeric T helper-B cell peptide as a vaccine for Flaviviruses. Dr. M. M. Gore, Dr. S.S. Dewasthaly, Prof. A.S. Kolaskar, Urmila Kulkarni-Kale Sangeeta Sawant WO 02/053182 A1

October 2K5 © Bioinformatics Centre, UoP 36

Important references• Hopp, Woods, 1981, Prediction of protein antigenic determinants from amino

acid sequences, PNAS U.S.A 78, 3824-3828 • Parker, Hodges et al, 1986, New hydrophilicity scale derived from high

performance liquid chromatography peptide retention data: Correlation of predicted surface residues with antigenicity and X-ray derived accessible sites, Biochemistry:25, 5425-32

• Kolaskar, Tongaonkar, 1990, A semi empirical method for prediction of antigenic determinants on protein antigens, FEBS 276, 172-174

• Men‚ndez-Arias, L. & Rodriguez, R. (1990), A BASIC microcomputer program forprediction of B and T cell epitopes in proteins, CABIOS, 6, 101-105

• Peter S. Stern (1991), Predicting antigenic sites on proteins, TIBTECH, 9, 163-169• A.S. Kolaskar and Urmila Kulkarni-Kale, 1999 - Prediction of three-dimensional

structure and mapping of conformational epitopes of envelope glycoprotein of Japanese encephalitis virus,Virology, 261, 31-42