Biased Antibody Repertoires:From Concept to Implementation

Department of Biological SciencesFlorida International University

Miami, Florida

May - July 2005

Juan C. Almagro, Ph.D

1. Statement of the problem

3. Design and Validation of a VH Repertoire with Tailored Diversity for Protein and Peptide Antigens

Plan of the talk

4. Design and Validation of Topography-Biased Antibody Libraries

2. Structure-Function relationships in antibodies

However, we cannot predict the specificity of a given antibody sequence or structure.

Hence, Our understanding of the evolution of the antibody repertoire is limited and

antibodies cannot be designed de novo.

0

100

200

300

400

1976

1978

1980

1982

1984

1986

1988

1990

1992

1994

1996

1998

2000

year

Num

ber

of s

truc

ture

s

cumulative

releases per year

• Today: ~ 600 structures are available at PDB (ABG; Antibody: Structure-function web site and IMGT)

0

5000

10000

15000

20000

25000

1965

1967

1969

1971

1973

1975

1977

1979

1981

1983

1985

1987

1989

1991

1993

1995

1997

1999

year

Num

ber

of s

eque

nces

Cumulative

Sequences per year

• Today: ~ 40 K sequences are available in databases, including the complete repertoire of germline genes from several species (IMGT)

Statement of the problem

1. Statement of the problem

3. Design and Validation of a VH Repertoire with Tailored Diversity for Protein and Peptide Antigens

Plan of the talk

4. Design and Validation of Topography-Biased Antibody Libraries

2. Structure-Function relationships in antibodies

Taken from Andreas Plückthun’s home page, with permission

Canonical structures(Chothia and Lesk, J. Mol. Biol. 196: 901, ‘87)

Type 1

Type 3

L1

~ 4,000 antibody sequences

Complete sequences

VL:VH dimmers

Canonical structure in L1, L2, L3, H1 and H2

381 VL:VH sequences

Predicting the Specificity of Antibody Sequences Based on the Structure of the Antigen-Binding Site

Found 10

0

5

10

15

20

25

1-2-

2-1-

1

1-2-

4-1-

1

1-3-

4-1-

1

1-3-

2-1-

1

1-2-

1-1-

1

1-4-

3-1-

1

1-4-

4-1-

1

1-1-

4-1-

1

1-1-

2-1-

1

1-2-

3-1-

1

othe

rs

ExpectedL1: 5

L2: 1

L3: 5 x

H1: 3

H2: 4 x25 12= 300x

Canonical structure classes in the known sequences

Vargas-Madrazo et al., J. Mol. Biol. 254: 497, ‘95

Only L1 and H2 contribute

to the structural diversity

Canonical Structure Classes

make ~ 90% of the sample

Canonical Structure class

H1-H2-L1-L2-L3 Frequency

(%)

Protein(169)

Surface Antigen(22)

Polysaccharide(17)

Nucleic acid(42)

Peptide(19)

Hapten(112)

1-1-2-1-1 3.2 56d 0 0 25 0 19

1-1-4-1-1 3.7 10 0 33 13 0 44

1-2-1-1-1 7.1 5 5 80 5 0 4

1-2-2-1-1 24.5 16 44 0 18 0 23

1-2-3-1-1 2.9 57 43 0 0 0 0

1-2-4-1-1 14.2 11 4 5 24 52 4

1-3-2-1-1 7.9 15 26 0 31 20 8

1-3-4-1-1 10 43 0 14 11 25 6

1-4-3-1-1 6.8 11 0 0 0 0 89

1-4-4-1-1

others

6.6

13.1

4

26

0

17

0

17

41

6

0

21

55

13

Canonical structure classes

Antigen size

classified in gross specificities

Vargas-Madrazo et al., J. Mol. Biol. 254: 497, ‘95

Anti-protein Anti-peptide Anti-hapten

Topography-specificity relationship

Model to correlate loop lengths (in particular L1 and H2) with the specificity

Predicting the Specificity of Antibody Sequences Based on the Structure of the Antigen-Binding Site

59 unique antibody structures

19 anti-protein

18 anti-peptide

22 anti-hapten

Determine residues in contact

~ 300 structures

0.0

0.5

1.0

Protein

Peptide

Hapten

Residues in contact with proteins, peptides and haptens

L1 L2 L3

SD

R u

sage

0.0

0.5

1.0

Protein

Peptide

Hapten

SD

R u

sage

H1 H2 H3Almagro. J. Mol. Recognit. 17:132, ‘04

Some positions in the antigen-binding siteinteract with the antigen very often (> 70%of the antibodies).

Others do so with a frequency between 30% and 70%.

A third group interacts with the antigen infrequently (<30%).

The frequency of contacts differs depending upon the type of antigen with which the antibody interacts.

Anti-protein Anti-peptide Anti-hapten

Contact usage - specificity relationship

Model to create diversity in the antigen-binding site as a function of the specificity

Conclusions I

1. Model to correlate the structure of the antigen binding site

with its specificity.

2. Guide for tailoring the antigen-binding site diversity depending upon the type of antigen

the antibody interacts with.

1. Statement of the problem

3. Design and Validation of a VH Repertoire with Tailored Diversity for Protein and Peptide Antigens

Plan of the talk

4. Design and Validation of Topography-Biased Antibody Libraries

2. Structure-Function relationships in antibodies

Full diversity: 20 aa + 1 amber codon in positions often found in contact with

proteins and peptides

Limited Diversity: YDAS (Felluose et al., PNAS 34, 12467, ‘04)

R/K

To simplify the diversity in positions of medium

usage while avoiding stop codons

To explore all amino acid variants in

positions with high contact usage

All the germline genes have R at

this position, except dp47 that has K

Dp47 scaffold

VH Repertoire with Tailored Diversity

Theoretical diversity: 720 x 94 x 2 = 6.7 x 1014 variants

Construction of the VH repertoire

1 10 20 30 40 50 60 70 80 90 100 110 |...|....|....|....|....|....|....|....|....|....|..a..|....|....|....|....|....|..abc..|....|....|a....|....|....|.. 1 3 5 7 9 Leader |||||||<<<<||||||| |||||||<<<<<<||||||||||||||<<<<<<<<<<||||||| |||||||<<<||||||| |||||||<<<<<||||||| |||||||<<<LLAAQPAMAEVQLLESGGGLVQPGGSLRLSCAASGFTFOOOOMOWVRQAPGKGLEWVSOIOOOOGOTOYADSVKGRFTISRDNSKNTLYLQMNSLRAEDTAVYYCAOOOOOYFDYWGQGTLVTVSSGGG>>>>>>>>>||||||| |||||||>>>>||||||| |||||||||||||| |||||||>>>>||||||| |||||||>>>||||||| |||||||>|||||||linker

2 4 6 8 10

Size:

3 x 108

members

The repertoire was synthesized by overlapping PCR (Stemmer et al., Gene. 164:49, ‘95)

in a single-step PCR reaction by using 10 internal oligonucleotides and two amplification primers

pIT

D1.3-VLVH repertoire

*(His)5linker

Selection

0.00

1.00

2.00

3.00

4.00

1.00E+10 1.00E+11 1.00E+12 1.00E+13 1.00E+14

Titer (pfu/mL)

OD

(450

nm

)

KM13

D1.3

Round 1

Round 2

HEL-coated immunotubes

Validation of the VH repertoire

Polyclonal ELISA

Chimeric library

Washed away

non-bound Ф

Trypsin to elute

bound Ф

Characterization

In Round 2, the ELISA signal was similar to D1.3 displayed on the phage

Frequency of scFv’s (hit rate)

0.000

0.500

1.000

1.500

2.000

2.500

1 11 21 31 41 51 61 71 81 91 101 111 121 131 141 151 161

Clone number

OD

(45

0 nm

)

Clone Frequency OD

H130 35| . . . . |

H250 55| . . a . | . . . .

H3 100. | a b c

D1.3 1.230 T G Y G V N M I W - G D G N T D R E R D Y RJCAV-II-HEL-C6-1 1 0.434 A A S Y M S S I A S Y D G A T D R E D V Y YJCAV-II-HEL-E3-1 1 0.597 Y N D S . A . . Y . S Y . D . D K . G A M .JCAV-II-HEL-B3-1 1 0.597 . H D A . A T . H A . S . S . D . . E . F .JCAV-II-HEL-C7-1 2 1.254 . G . S . . Y . G A S S . D . D . . K P Q .JCAV-II-HEL-F8-1 1 1.657 . S A A . A . . Y . S Y . D . S . . G A M .JCAV-II-HEL-B8-1 2 1.779 D D . D . . Y . G Y . Y . D . A . . R A T .JCAV-II-HEL-B7-1 5 1.747 S K A A . A . . S A S S . D . D . . M P T .JCAV-II-HEL-F8-2 3 1.925 Y N . A . D G . L . D S . D . Y . . E A T .JCAV-II-HEL-E8-2 1 1.984 . F D S . . . . G A A A . D . S . . M P L .JCAV-II-HEL-H6-1 3 2.276 S T D A . A D . T A D S . D . A . . R P L .

select clones

Grow in 2xTY

IPTG

Test in ELISA

Expression and relative affinity

0

0.25

0.5

0.75

1

1 10 100 1000

Log (Dilution)

Rel

ativ

e Sc

ale* D1.3

HELII-B7

HELII-H6

*Relative Scale:

The chimeric scFvs are 7-9 times better expressed than D1.3 as suggested by the ED50

1- (Max-O.D. / Max - Min)

The affinity for HEL may be similar to D1.3 as suggested by the slope of the curves

HEL-D1.3 affinity 5 nM ( Foote and Winter, J. Mol. Biol. 224: 487, ‘92)

Conclusions II

4. The scFvs dominating the population are well expressed in E. coli and may have affinities in the nM range.

1. A VH repertoire with tailored diversity to recognize proteins and peptides was designed and constructed

3. After the second round of selection on HEL-coated Immunotubes, diverse scFvs against HEL were obtained, thus validating the library

as source of VH domains.

2. It was cloned with the VL chain chain of D1.3 to yield a chimeric library of 3 x108 members.

1. Statement of the problem

3. Design and Validation of a VH Repertoire with Tailored Diversity for Protein and Peptide Antigens

Plan of the talk

4. Design and Validation of Topography-Biased Antibody Libraries

2. Structure-Function relationships in antibodies

Dp47 scaffold with tailored diversity for proteins and peptides (Almagro et al., J. Mol. Biol. Submitted)

Topography biased antibody libraries

Invariant VL chain with a long L1 Invariant VL chain with a short L1

Theoretical diversity: 2.1 x 1010

Invariant VL chains

Short L1

0

10

20

30

0 5 10 15 20 25 30 35 40 45

A27

Use

Fre

quen

cy (

%)

Human Germline Genes (IGMT)

The difference between repertoires is reduced to one insertion of 5 amino acids at the tip of L1 and 5 mutations in L1,

positions: 28, 29, 30, 30a and 31

Long L1

Numbering 30 ..|....|abcdef....|A27 CRASQSVSS-----SYLAW | |||| ||||1-4 (B3)CKSSQSVLYSSNNKNYLAW |||||||||||||||||||A27md CKSSQSVLYSSNNKNYLAW

Graft L1 of B3 in A27

Combination with different invariant VL chains

Size:

3.6 x 108

memberspIT

A27/Jk1VH repertoire

*(His)5linker

Anti-protein repertoire

Size:

6 x 107

memberspITA27/Jk1modVH repertoire

*(His)5linker

Anti-peptide repertoire

Insertion of 5 aminoacids at L1

Panel of Selectors

V3 loop gp120(V3)

14 aa; 1.6 KDa

Hen Egg White Lysozyme(HEL)

129 aa; 14.3 KDa

Bovine Serum Albumin(BSA)

583 aa; 66.4 KDa

V3-BSA conjugate ND

Selections conducted as described for VH-D1.3

Polyclonal ELISA after Round 3V3 selections

V3-BSA selections

BSA selections

HEL selections

0.000

1.000

2.000

3.000

4.000

1.0E+07 1.0E+09 1.0E+11 1.0E+13Titer (cfu/mL)

O.D

. 450

nm

0.000

1.000

2.000

3.000

4.000

1.0E+07 1.0E+09 1.0E+11 1.0E+13

Titer (cfu/mL)

O.D

. 45

0 nm

0.000

1.000

2.000

3.000

4.000

1.0E+07 1.0E+09 1.0E+11 1.0E+13

Titer (cfu/mL)

O.D

. 450

nm

0.000

1.000

2.000

3.000

4.000

1.0E+07 1.0E+09 1.0E+11 1.0E+13

Titer (cfu/mL)

O.D

. 450 n

m

Red: Anti-peptide library Blue: Anti-protein library

Frequency of positive clones

32/96

42/192

0/960/96

91/96 92/96

64/192

3/96

Anti-peptide library yields more scFvs for V3 and V3-BSA than for proteins

Anti-protein library yields more scFvs for proteins than forV3 or V3-BSA

0

20

40

60

80

100

V3 V3-BSA BSA HEL

Fre

quen

cy (

%)

LL1 library

SL1 library

select clones

Grow in 2xTY

IPTG

Test in ELISA

Unique scFvsDetermined by DNA sequencing

Library Selector Unique / Totalpep V3 1/6pep V3-BSA 1/6pep BSA -pep HEL 2/5pro V3 -pro V3-BSA 1/3pro BSA 2/5pro HEL 2/6

Specificity of unique clones O

.D. 4

5 0 n

m

O.D

. 45 0

nm

V3 and V3-BSA selections

BSA selections

HEL selections

V3 V3-BSA BSA HEL

0.000

1.000

2.000

LL1-B2

LL1-B1

O.D

. 45 0

nm

V3 V3-BSA BSA HEL

0.000

1.000

2.000 SL1-H1

SL1-H2

LL1-H1

LL1-H2

The scFv selected from the anti-peptide lib. on V3 is specific for V3 and V3-BSA

The scFv selected from the anti-protein lib. on V3-BSA is specific for the carrier

ScFvs selected on proteins are specifics

V3 V3-BSA BSA HEL

0.000

1.000

2.000

SL1-VB1

LL1-V1

Expression and Relative Affinity

HEL selectionsO

.D. 4

5 0 n

m

0.00

0.50

1.00

1.50

2.00

2.50

3.00

1 10 100 1000

SL1-H1

SL1-H2

LL1-H1

LL1-H2

D1.3

Different dynamic ranges, better slopes and higher ED50 indicating differences in binding (different epitopes?)

ScFvs from SL1 (anti-protein library) look better than the those isolated from LL1 (anti-peptide library)

General Conclusions

1. Anti-protein library produced diverse specific scFvs against two protein models: BSA and HEL.

2. Anti-protein library did not produce scFvs against the peptide model, free or conjugated. Only against the carrier.

3. Anti-peptide library produced specific scFvs against the peptide.

5. Anti-peptide library did not produce scFvs against BSA and against HEL produced less binders than the anti-protein library.

6. Together, these results suggest that antibody libraries can be biased toward the recognition of different kinds of antigens based on structural principles

4. Anti-peptide library produced less binders against HEL than the anti-protein library.

AcknowledgmentsFlorida International UniversityAlvaro Velandia

Matt Osentoski

Sylvia L Smith

National University of MexicoLuisa Fernadez

Alejandra Blancas

Enesto Ortiz

Baltazar Beceril

Lourival Possani

Alejandro Alagon

This work was supported by:

•Grant 1R03AI057752-01 from NIH/NIAID •Sub-Contract DAAD13-03-C-0065 from CBD/USF.