WO2003054021A2

WO2003054021A2 - Semi-synthetic antibody fragment combinatorial library based on avian immunoglobulin genes

Info

Publication number: WO2003054021A2
Application number: PCT/ZA2002/000218
Authority: WO
Inventors: Wouter Van Wyngaardt; Terry Ramodike; Cordelia Mashau; Dubravka Miltiadou; Frances Jordaan; Jeanni Fehrsen; Dion H. Du Plessis
Original assignee: Agricultural Research Council
Current assignee: Agricultural Research Council
Priority date: 2001-12-20
Filing date: 2002-12-20
Publication date: 2003-07-03
Anticipated expiration: 2004-06-20
Also published as: AU2002359913A8; WO2003054021A3; AU2002359913A1

Abstract

According to the invention there is provided a semi-synthetic antibody fragment combinatorial library based on avian immunoglobulin genes. The avian immunoglobulin genes are poultry or chicken immunoglobulin genes. The antibodies are single chain antibody fragments and the library is displayed on large phages. At least some of the third complementarity determining region (CDR 3) of the variable (V) domain of the heavy (H) chains of the antibody fragments are replaced by synthetically randomised peptides. Synthetically randomised variable heavy (VH) regions are paired with the natural repertoire of avian VH regions and not with defined light chains, the arrangement being such that combinatorial diversity is contributed by both the variable light (VL) and VH regions and further such that the natural VH repertoire of the avian VH regions adds further diversity.

Description

SEMI-SYNTHETIC ANTIBODY FRAGMENT COMBINATORIAL LIBRARY

INTRODUCTION

This invention relates to a semi-synthetic antibody fragment combinatorial library based on avian immunoglobulin genes. More particularly this invention relates to a large phage-displayed semi-synthetic single chain antibody fragment combinatorial library based on avian immunoglobulin genes.

OBJECT OF THE INVENTION It is an object of the present invention to provide a semi-synthetic antibody fragment combinatorial library based on avian immunoglobulin genes.

SUMMARY OF THE INVENTION

According to the invention there is provided a semi-synthetic antibody fragment combinatorial library based on avian immunoglobulin genes.

Further according to the invention, at least some of the third complementarity determining region (CDR 3) of the variable (V) domain of the heavy (H) chains of the antibody fragments are replaced by synthetically randomised peptides.

Yet further according to the invention, synthetically randomised variable heavy (VH) regions are paired with the natural repertoire of avian VH regions and not with defined light chains, the arrangement being such that combinatorial diversity is contributed by both the variable light (VL) and VH regions and further such that the natural VH repertoire of the avian VH regions adds further diversity.

Both modified and unmodified heavy chains may be paired with VL chains from non-immune birds.

Preferably the specific antibodies are selected by panning on immobilised avian antigens.

The antibody fragments may be displayed on large phages.

Preferably the large phages are filamentous bacteriophages and the antibody fragments may be displayed on the surface of a population of the filamentous bacteriophages.

Further according to the invention the antibody fragments are single chain antibody fragments.

Further according to the invention the avian immunoglobulin genes are poultry immunoglobulin genes. Further according to the invention the avian immunoglobulin genes are chicken immunoglobulin genes.

A recombinant antibody library according to the invention can be used to bypass the immune system and provide a virtually unlimited repertoire of antibody specificities within a volume of liquid as small as a few microlitres^M This invention therefore relates to a "universal" single chain fragment variable (scFv) antibody library from which antibodies of different specificities may be obtained without the need for immunisation. The library is novel in that it is based on avian immunoglobulin genes with the third complementarity determining region (CDR 3) of the variable (V) domain of its heavy (H) chain replaced by a synthetically randomised peptide in some of its members. This adds to its diversity, supplementing that provided by the natural non-immune avian immunoglobulin repertoire.

Both modified and unmodified heavy chains are paired with VL light (L) chains from non-immune birds. The resulting repertoire of antibody fragments is displayed on the surface of a population of filamentous bacteriophages. Specific antibodies are selected by panning on immobilised antigens. The library described here is based on chicken immunoglobulin genes. Protection is also sought for similar libraries based on ostrich and other avian immunoglobulin genes. Uniqueness

Most existing combinatorial antibody libraries are based on either human or mouse immunoglobulin sequences ^1_4. Antibody libraries based on chicken immunoglobulin genes have been constructed, but without synthetically randomised sequences^5,6 No antibody library based on the ostrich, guinea fowl etc. genes could be located by the Applicant. CDR 3 normally plays a major role in determining antibody specificity. While synthetically randomised CDR 3s have been incorporated into a library based on human immunoglobulin genes, its modified VH regions are paired with a single light chain V region specific for bovine serum albumin³. In contrast, the library in accordance with the present invention pairs its synthetically randomised VH regions with the natural repertoire of chicken VH regions and not with a defined light chain. Combinatorial diversity is thus contributed by both the VL and VH regions. In addition, it draws on the natural VH repertoire of the chicken to further add diversity.

Advantages

Although not commonly used to provide a framework for recombinant antibody library construction, avian immunoglobulin genes have one important advantage; unlike mice and humans, birds use gene conversion of rearranged V(D)J gene sequences to generate antibody diversity^5,7 Nucleotide sequences of the heavy and light chains of chicken immunoglobulins show that virtually all V regions in mature B cells have identical ends. Consequently, these conserved termini make it possible to use PCR to amplify the V region repertoires using a single set of primers for the light chains and another set for the heavy chains⁵. This is in contrast to the mouse or human immunoglobulin genes where a complex and expensive cocktail of primers for each variable region gene family is essential. Incorporating a synthetically randomised CDR3 adds to the natural repertoire since it has not been subjected to biological process (e.g. clonal deletion or affinity maturation) during B cell development. In addition, it allows better control over the construction of the library by incorporating predetermined amounts of a particular CDR 3 size.

Possible applications

The library is expected to yield highly specific binding reagents for use in immunoassays for the diagnosis of human, animal and plant disease and for environmental monitoring for e.g. pesticide residues. Suitable antibody fragments can be incorporated any one of a number of immunoassay formats. These include various ELISAs, agglutination assays, immunoblot assays, immunofluorescence and other immunochemical tests. They may also be useful in avian therapeutics. Antibody fragments can also be used in immunoaffinity purification of high value industrial products. The CDR sequences of appropriate antibodies may be able to be transferred to human antibody or other suitable frameworks for medical applications, eg cancer targeting. METHODS

Construction of the avian antibody library

The bursa of Fabricius is in birds a reservoir for the avian immune system's diverse repertoire of rearranged and converted V region genes. To access this repertoire, total RNA was isolated from the bursal lymphocytes of five 3-8 week old non-immunised white leghorn chickens using TRI - Reagent (Molecular Research Center, USA). To construct the non-synthetic (i.e. unmodified non- immune) component of the final library, the natural repertoire of chicken VH and VL gene sequences was amplified separately and without modification. To achieve this, sets of oligonucleotide primers based on published V region sequences^5,6,8 were designed and used to amplify these regions by means of the polymerase chain reaction (PCR). Primers annealing at the 3' end of the VH and at 5' the end of the VL genes and that included codons for a (Gly Ser)₃ flexible linker¹⁰ were designed with the aim of covalently joining the amplified VH and VL repertoires. These primers overlapped sufficiently with each other in the linker region to permit the repertoires to be joined using splicing by overlap extension (SOE). Once joined, primers annealing at the ends and containing S//1 and Not 1 restriction enzyme sites were next used to amplify by PCR the resulting construct which now encoded the scFv repertoire. This construct was cloned into the appropriate sites in the phagemid vector pHEN and electroporated into E. coli TG 1 cells. Transformants were selected on antibiotic-containing nutrient agar plates. The size of the library was determined by counting appropriate dilutions of antibiotic-resistant colonies. This portion of the library consisted of a total of 3.8 X 10⁸ independent clones.

For the second (semi-synthetic) component, synthetic CDR 3s of defined size, but with random sequences were introduced into the VH domain. A set of primers was designed to anneal to the 3' end of the sequence encoding the VH region. These primers incorporated synthetically randomised sequences at the positions that code for the natural chicken VH CDR 3. After amplification by PCR, the VH genes with their synthetic CDR 3 repertoires were linked via a separate oligonucleotide coding for the flexible linker to the VL gene repertoire using an SOE protocol similar to that described above. After amplification (see above), the constructs were ligated into the phagemid display vector pHEN. Transformation and antibiotic selection were as described above. A sublibrary for each size of CDR was constructed. The synthetic CDRs ranged in size from six to 14 amino acid residues. The sublibraries were pooled in appropriate proportions so that each was fully represented. The semi-synthetic portion of the library consisted of approximately 10⁹ clones. Sequence analysis of these scFv DNA constructs was performed in order to confirm that the random synthetic CDRs of the predetermined sizes were in fact present.

The final phagemid library was obtained by pooling the two major subcomponents (natural and semi-synthetic) described above so that both were fully represented. The pool consists of more thanlO⁹ different clones. This figure is not adjusted for out of frame or for non-insert bearing clones.

Screening for antigen-specific binders

To prove its usefulness as a source of antibody fragments that recognise disparate antigens, the library was screened by panning ⁹ on three model antigens (proteins and a hapten). First, phages displaying the antibody fragment repertoire were rescued by infecting the bacterial phagemid library with M13KO7 helper phages. Each antigen was individually adsorbed passively to the plastic surface of an immunotube (Nunc Maxisorp). Non-specific binding was abrogated by the use of milk powder as a blocking agent.

A representative aliquot of the phage displayed antibody fragment library was added and time was allowed for binding. Irrelevant non-binders were removed by washing 20 X with phosphate buffered saline supplemented with 0.1% Tween 20 and another 20 X with PBS. Specifically binding phages were eluted at high pH (pH 12) and amplified by reinfecting TG1 bacterial host cells. Infective phage particles were rescued by co-infection with M13K07 helper phages. After several such rounds of panning, amplification and rescue, the resulting phage were tested and characterised as polyclonal preparations. ELISAs were performed using purified antigens adsorbed to NUNC Polysorp microtitre plates. Milk powder was used as blocking agent. The preparations were adjusted to contain approximately 10¹¹ phage particles /ml and added to blocked plates. After washing, binding phages were detected by the addition of an anti-M13KO7 mouse monoclonal antibody conjugated to horseradish peroxidase and a chromogenic substrate. Absorbances were determined at 492 nm.

RESULTS Construction

Natural repertoire component

Fig 1A illustrates agarose gel electropherograms of the PCR products obtained after amplifying the individual chicken VH and VL sequences which were used to construct the natural non-immune antibody repertoire. Fig 1B shows the product of splicing by overlap extension (SOE) of the VH and VL chain constructs (approx 800 bp). A 10Obp ladder is included in both figures.

Semi-synthetic repertoire component

Fig 2 shows (from left to right) agarose gel electropherograms of the PCR products obtained after amplifying the individual chicken VH (with randomised CDR3), VL and linker sequences. The gel labelled "SOE" shows the product obtained after the aforementioned had been spliced by overlap extension (approx 800bp) and an amplification step that added appropriate restriction sites for phage display cloning. The right-hand track in each gel shows a 100bp marker ladder.

Sequence analysis

Fig 3 illustrates incorporation of synthetically randomised amino acid sequences of three defined sizes into the CDR3 region of the VH frameworks together with part of the framework region and the flexible linker. Only the region into which the defined-length synthetic CDRs were introduced is shown. The unmodified CDRs 1 and 2 and the L chains are not illustrated.

Screening to prove that binders are present

ELISA results obtained with polyclonal preparations with a number of model antigens are shown in Fig 4. Test antigens were porcine thyroglobulin, the hapten 4-hydroxy-5-iodo-3-nitrophenylacetic acid (NIP) conjugated to bovine serum albumin and cytochrome c. Phage preparations were tested after each round of selection by panning. The antigen against which the library was panned is given in brackets above each diagram. The Y axes depict the A ₄g₂ ELISA values. The X axes show the different panning rounds while the Z axes show the different antigens upon which the selected phage populations were tested. In all cases a population of specifically binding antibodies was selected by round three of the panning procedure. The signal shown with MBP in the bottom graph represents a positive control using a maltose binding protein (MBP) antibody fragment³ (provided by MRC, Cambridge, UK). DISCUSSION

A large universal phage displayed library consisting of more than 10⁹ members is described. It has yielded antibodies to a number of different antigens, of which three examples are presented here. Because of its size and diversity, it is expected to yield many more. While it is based on chicken genes, a similar approach can be used in principle for any avian species, provided the relevant VH ant VL region sequences can be amplified. Preliminary results show that his may be possible with the ostrich.

It will be appreciated that variations in detail are possible with a semi-synthetic antibody fragment combinatorial library based on avian immunoglobulin genes according to the invention without departing from the scope of the appended claims.

Claims

1. A semi-synthetic antibody fragment combinatorial library based on avian immunoglobulin genes.

2. A semi-synthetic antibody fragment combinatorial library according to claim 1 wherein at least some of the third complementarity determining region (CDR 3) of the variable (V) domain of the heavy (H) chains of the antibody fragments are replaced by synthetically randomised peptides.

3. A semi-synthetic antibody fragment combinatorial library according to claim 2 wherein synthetically randomised variable heavy (VH) regions are paired with the natural repertoire of avian VH regions and not with defined light chains, the arrangement being such that combinatorial diversity is contributed by both the variable light (VL) and VH regions and further such that the natural VH repertoire of the avian VH regions adds further diversity.

4. A semi-synthetic antibody fragment combinatorial library according to claim 3 wherein both modified and unmodified heavy chains are paired with VL chains from non-immune birds.

5. A semi-synthetic antibody fragment combinatorial library according to any one of the preceding claims wherein the specific antibodies are selected by panning on immobilised avian antigens.

6. A semi-synthetic antibody fragment combinatorial library according to any one of the preceding claims wherein the antibody fragments are displayed on large phages.

7. A semi-synthetic antibody fragment combinatorial library according to claim 6 wherein the large phages are filamentous bacteriophages and wherein the antibody fragments are displayed on the surface of a population of the filamentous bacteriophages.

8. A semi-synthetic antibody fragment combinatorial library according to any one of the preceding claims wherein the antibody fragments are single chain antibody fragments.

9. A semi-synthetic antibody fragment combinatorial library according to any one of the preceding claims wherein the avian immunoglobulin genes are poultry immunoglobulin genes.

10. A semi-synthetic antibody fragment combinatorial library according to claim 9 wherein the avian immunoglobulin genes are chicken immunoglobulin genes.

1. A semi-synthetic antibody fragment combinatorial library substantially as herein described and exemplified.