Organic compounds -- part of the class 532-570 series – Organic compounds – Carbohydrates or derivatives
Reexamination Certificate
2000-01-19
2002-06-04
Guzo, David (Department: 1636)
Organic compounds -- part of the class 532-570 series
Organic compounds
Carbohydrates or derivatives
C536S023100, C435S320100, C435S091400, C530S387100, C530S387300
Reexamination Certificate
active
06399763
ABSTRACT:
FIELD OF THE INVENTION
The present invention relates to an expression library comprising a repertoire of nucleic acid sequences derived from the natural sequence repertoire but modified to enhance the extent of sequence variability, each said nucleic acid sequence encoding at least part of a variable domain of a heavy chain derived from an immunoglobulin naturally devoid of light chains and its use in producing antibodies, or more particularly fragments thereof.
In particular, the invention relates to a method for the preparation of antibodies or fragments thereof, having binding specificity for a target antigen which avoids the need for the donor previously to have been immunised with the target antigen.
BACKGROUND TO THE INVENTION
Monoclonal antibodies, or binding fragments thereof, have traditionally been prepared using hybridoma technology (Kohler and Milstein, 1975, Nature 256, 495). More recently, the application of recombinant DNA methods to generating and expressing antibodies has found favour. In particular, interest has concentrated on combinatorial library techniques with the aim of utilizing more efficiently the antibody repertoire.
The natural immune response in vivo generates antigen-specific antibodies via an antigen-driven recombination and selection process wherein the initial gene recombination mechanism generates low specificity, low-affinity antibodies. These clones can be mutated further by antigen-driven hypermutation of the variable region genes to provide high specificity, high affinity antibodies.
Approaches to mimicking the first stage randomisation process which have been described in the literature include those based on the construction of ‘naive’ combinatorial antibody libraries prepared by isolating panels of immunoglobulin heavy chain variable (VH) domains and recombining these with panels of light variable chains (VL) domains (see, for example, Gram et al, Proc. Natl. Acad. Sa, USA, 89, 3576-3580, 1992). Naive libraries of antibody fragments have been constructed, for example, by cloning the rearranged V-genes from the IgM RNA of B cells of un-immunised donors isolated from peripheral blood lymphocytes, bone marrow or spleen cells (see, for example, Griffiths et al, EMBO Journal, 12(2), 725-734, 1993, Marks et al, J. Mol. Biol., 222, 581-597, 1991). Such libraries can be screened for antibodies against a range of different antigens.
In combinatorial libraries derived from a large number of VH genes and VL genes, the number of possible combinations is such that the likelihood that some of these newly formed combinations will exhibit antigen-specific binding activity is reasonably high provided that the final library size is sufficiently large.
Given that the original B-cell pairings between antibody heavy and light chain, selected by the immune system according to their affinity of binding, are likely to be lost in the randomly, recombined repertoires, low affinity pairings would generally be expected. In line with expectations, low affinity antibody fragments (Fabs) with K
a
s of 10
4
-10
5
M
−1
for a progesterone-bovine serum albumin (BSA) conjugate have been isolated from a small (5×10
6
) library constructed from the bone marrow of non-immunised adult mice (Gram et al, see above).
Antibody fragments of higher affinity (K
a
s of 10
6
-10
7
M
−1
range) were selected from a repertoire of 3×10
7
clones, made from the peripheral blood lymphocytes of two healthy human volunteers (Marks et al, see above) comprising heavy chain repertoires of the IgM (naive) class. These were combined with both Lamda and Kappa light chain sequences, isolated from the same source. Antibodies to more than 25 antigens were isolated from this library, including self-antigens (Griffiths et al, see above) and cell-surface molecules (Marks et al, Bio/Technology, 11, 1145-1149, 1993).
The second stage of the natural immune response, involving affinity maturation of the selected specificities by mutation and selection has been mimicked in-vitro using the technique of random point mutation in the V-genes and selecting mutants for improved affinity.
Recently, the construction of a repertoire of 1.4×10
10
scfv clones, achieved by ‘brute force’ cloning of rearranged V genes of all classes from 43 non-immunised human donors has been reported (Vaughan et al 1996) and Griffiths et al, see above. Antibodies to seven different targets (including toxic and immunosuppressant molecules) were isolated, with measured affinities all below 10 nM.
The main limitation in the construction of combinatorial libraries is their size, which consequently limits their complexity. Evidence from the literature suggests that there is a direct link between library size and diversity and antibody specificity and affinity (see Vaughan., et al, Nature Biotechnology, 14, 309-314, 1996), such that the larger (and more diverse) the library, the higher the affinity of the selected antibodies.
The optimisation of binding affinity through random recombination of a heavy and light chain in combinatorial libraries is complicated by sequence variations in the two framework regions, (i.e. the parts of the variable domains that serve as a scaffold in supporting the regions of hypervariability which are in turn termed the complementary determining regions or CDRs.
Only some combinations of framework sequences are compatible with the folding and interaction required for the correct orientation of the 6 CDRs that is necessary for good binding affinity. Consequently, conventional combinatorial libraries are likely to contain a high percentage of molecules that are non-functional.
The affinity of antibodies may also be improved by the process of “chain shuffling”, whereby a single heavy or light chain is recombined with a library of partner chains (Marks et al, Bio/Technology, 10, 779-782, 1992).
EP-A-0368684 (Medical Research Council) discloses the construction of expression libraries comprising a repertoire of nucleic acid sequences each encoding at least part of an immunoglobulin variable domain and the screening of the encoded domains for binding activities. It is stated that repertoires of genes encoding immunoglobulin variable domains are preferably prepared from lymphocytes of animals immunised with an antigen.
The isolation of single VH domains having antigen binding activities, facilitated by immunisation, is exemplified (see Example 6).
These results illustrate that although the VH part alone of a classical antibody binding domain can exhibit binding activity, the specificity and affinity are generally very low. This may be explained by the absence of the functional involvement of the missing light chain such that only half of the intended binding pocket is present, leading to binding with related or homologous targets.
EP-A-0368684 further describes the cloning of heavy chain variable domains with binding activities generated by mutagenesis of one or each of the CDRs. The preparation of a repertoire of CDR3s is described by using “universal” primers based in the flanking sequences, and likewise repertoires of other CDRs singly or in combination. These synthetic mutant VH clones can then be recombined with VL chains to produce a synthetic combinatorial library.
Construction of libraries by such synthetic recombinatorial techniques produces a repertoire of molecules that collectively exhibit a good degree of binding diversity, wherein the variability is focused into the sequences that encode the CDRs of each chain. However this technique does not overcome the problems previously discussed with respect to random recombination of heavy and light chains and production of non-functional molecules. Furthermore there are still six separate regions (3 CDRs in VH and another 3 in VL) determining the binding capacity of the molecule hence the repertoire of possible binding variants is encoded within a rather diffuse stretch of coding sequence, thus making a focused approach to altering the binding affinity of these binding domains very difficult.
There remains a continuing need for the
Frenken Leon
van der Logt Cornelis P.
Guzo David
Leffers, Jr. Gerald G.
Morgan & Lewis & Bockius, LLP
Unilever Patent Holdings B.V.
LandOfFree
Method for producing antibody fragments does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method for producing antibody fragments, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method for producing antibody fragments will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2950349