Title:  Compositions and methods for generating chimeric heteromultimers

United States Patent:  6,833,441

Issued:  December 21, 2004

Inventors:  Wang; Caili (San Francisco, CA); Zhong; Pingyu (Mountain View, CA); Liu; Shengjiang (Mountain View, CA); Luo; Peizhi (Sunnyvale, CA); Li; Shengfeng (Belmont, CA); Wang; Xinwei (San Jose, CA)

Assignee:  Abmaxis, Inc. (Santa Clara, CA)

Appl. No.:  921144

Filed:  August 1, 2001

Abstract

The present invention provides a technique for specific assembly of monomeric polypeptides to form a heteromultimer. This technique is particularly useful for generating a genetically diverse repertoire of heteromultimers such as antigen-binding units. The invention also provides both non-single-chain and single-chain antigen-binding units that are assembled by the technique described herein. The present invention also provides recombinant polynucleotides, vectors, host cells, and kits for producing the subject antigen-binding units. Further provided by the invention are methods of using the subject antigen-binding units.

Description of the Invention

TECHNICAL FIELD

This invention is in the field of immunology. Specifically, the invention relates to the generation of chimeric heteromultimers such as non-single-chain antigen-binding units using unique heterodimerization sequences. This invention also relates to the generation of single-chain antigen-binding units stabilized by the subject heterodimerization sequences. The compositions and methods embodied in the present invention are particularly useful for identifying antigen-binding units that are of major diagnostic and/or therapeutic potential.

BACKGROUND OF THE INVENTION

Antibodies or immunoglobulins are molecules that recognize and bind to specific cognate antigens. Because of their exclusive specificities, antibodies, particularly monoclonal antibodies, have been widely used in the diagnosis and treatment of a variety of human diseases.

The basic immunoglobulin (Ig) in vertebrate systems is composed of two identical light ("L") chain polypeptides (approximately 23 kDa), and two identical heavy ("H") chain polypeptides (approximately 53 to 70 kDa). The four chains are joined by disulfide bonds in a "Y" configuration. At the base of the Y, the two H chains are bound by covalent disulfide linkages. The L and H chains are organized in a series of domains. The L chain has two domains, corresponding to the C region ("CL") and the other to the V region ("VL"). The H chain has four domains, one corresponding to the V region ("VH") and three domains (CH1, CH2 and CH3) in the C region. The antibody contains two arms (each arm being a Fab fragment), each of which has a VL and a VH region associated with each other. It is this pair of V regions (VL and VH) that differ, from one antibody to another (due to amino acid sequence variations), and which together are responsible for recognizing the antigen and providing an antigen-binding site. More specifically, each V region is made up from three complementarity determining regions (CDR) separated by four framework regions (FR). The CDR's are the most variable part of the variable regions, and they perform the critical antigen binding function. The CDR regions are derived from many potential germ line sequences via a complex process involving recombination, mutation and selection.

Research in recent years has demonstrated that the function of a binding antigen can be performed by fragments of a whole antibody. Exemplary antigen binding fragments are (i) the Fab fragment consisting of the VL, VH, CL and CH1 domains; (ii) the Fd fragment consisting of the VH and CH1 domains; (iii) the dAb fragment (Ward, E. S. et al, Nature 341, 544-546 (1989) which consists of a VH domain; (iv) isolated CDR regions; and (v) F(ab')₂ fragments, a bivalent fragment comprising two Fab fragments linked by a disulfide bridge at the hinge region; and (vi) the Fv fragment consisting of the VL and VH domains of a single arm of an antibody. The Fv fragment is the smallest functional unit required for high affinity binding of antigen.

One major challenge in the antibody field has been to reconstitute a vast diverse repertoire of immunoglobulins that mimics the immunoglobulin pool in the human immune system. Such a repertoire generally has a complexity ranging from 10⁸ to 10¹³ distinct immunoglobulins. The generation of such a repertoire would greatly facilitate the identification and production of immunoglobulins capable of interacting specifically with therapeutic targets. However, the design and production of such a repertoire has traditionally been hampered by the lack of a stabilizing means for assembly of the minimal functional unit, namely the Fv fragment. It is a well-known problem in the art that the VH and VL regions, when expressed alone, have very low interaction energy (Glockshuber et al (1990) Biochemistry 29(6):1362-1367). The two components dissociate at low protein concentrations and are too unstable for many applications at physiological body temperature. It is also a long-recognized technical obstacle that large proteins, such as whole antibodies (albeit extremely stable), do not express at an appreciable level in the host cell, thus rendering the construction of a highly diverse antibody repertoire very difficult.

More recently, three approaches have been developed to generate stable VL and VH complexes. However, each of these techniques bears a number of intrinsic limitations; and none of them circumvents the aforementioned technical hurdles completely. The first approach uses a peptide linker to connect the VL and VH as a single-chain ("scFv") (Huston et al (1988) Proc. Natl. Acad. Sci. U.S.A 85:5879-5883). While the resulting scFv exhibits substantial antigen-binding activity, not all antibodies can be made as single chains and still retain high binding affinity (Huston et al (1988) Proc. Natl. Acad. Sci. U.S.A. 85:5879-5883; Stemmer et al (1993) Biotechniques 14(2): 256-265). In part, this is due to the interference of linker sequences with the antigen binding sites. The second approach involves inserting a pair of cysteine residues in the VL and VH regions to generate a disulfide-bond stabilized Fv ("dsFv") (Brinkmann et al (1993) Proc. Natl. Acad. Sci. U.S.A. 90(16): 7538-7542). The incorporated disulfide linkage, however, is unstable under reducing conditions in many host cells. For instance, in cytosol of E. Coli, the inter-molecular disulfide bond is often insufficient to stabilize the VL and VH complex. Moreover, this method typically requires 3-dimensional structural information of the V regions to ensure that the cysteine pair is inserted in a proper place without disruption the binding activity. Because the 3-dimensional information of a vast majority of the existing antibodies is unknown, this approach has little practical utility, and is particularly unsuited for antibody library construction, especially for constructing antibody repertoires derived from B cells. The third approach for stabilizing the VL and VH regions utilizes the disulfide bonds native to the CH1 and CL domains. This method proceeds with grafting a disulfide-bond linked CH1 and CL domains to the C-termini of the VL and VH regions in order to reconstitute a Fab fragment. While the resulting Fab fragment is generally more stable and often exhibits higher binding affinity than scFv, Fab is not optimal for high level expression and antibody repertoire construction due to its large size.

Certain dimerization sequences that form coiled-coil structures have also been employed to assemble multivalent antibodies. Specifically, U.S. Pat. No. 5,932,448 describes a bispecific F(ab')₂ heterodimer linked by the Fos and Jun leucine zippers. The Fos and Jun leucine zippers are well-characterized sequences known to preferentially form heterodimers. However, they still exhibit significant propensity to form homodimers under physiological buffer conditions and/or at physiological body temperature (O'Shea et al. (1992) Cell 68: 699-708; Vidal et al. (1996) Proc. Natl. Acad. Sci. U.S.A.). In fact, the Jun/Jun homodimer is so stable that formation of Fos/Jun heterodimer in vitro requires dissociation of the Jun/Jun homodimer by first heating or reduction with 2-mercaptoethanylamine (see U.S. Pat. No. 5,910,573 column 7 lines 35-37; U.S. Pat. No. 5,932,448, column 16 lines 15-30). When tested in vivo, both Fos and Jun yield detectable amounts of homodimers (see, e.g. column 15, lines 41-43 of U.S. Pat. No. 5,932,448; and Vidal et al. (1996) Proc. Natl. Acad. Sci. U.S.A). While the existence of some homodimerization propensity may not be of substantial concern for the production of a single antibody species, such propensity presents a serious problem for antibody repertoire construction, where high efficiency of heterodimerization between VL and VH regions is required.

Aside from Fos and Jun leucine zippers, U.S. Pat. No. 5,824,483 by Houston et al. describes the construction of a combinatorial library of coiled-coil dimerization peptides. Houston et al. proposes that the library is useful for identifying a polypeptide that is capable of interacting specifically with a selected macromolecule ligand such as antibodies (see last paragraph bridging pages 8 and 9). Apparently, Houston et al. concerns the selection of "antigen peptides" that bind to targeted antibodies, rather than the construction and selection of target antibodies. Focusing on an entirely different purpose, Houston et al. does not describe or even suggest the use of coiled-coil sequences to generate stable antigen-binding units.

Thus, there remains a considerable need for improved compositions and methods to generate stable antigen-binding units and repertoires thereof to effect identification of therapeutic antigen-binding units. An ideal antigen-binding unit would be more stable than a Fv fragment, but would preferably be smaller than a Fab fragment to allow large-scale production and efficient display. Such antigen-binding unit would also serve as a building block for constructing multivalent and/or multispecific antibodies. The present invention satisfies these needs and provides related advantages as well.

SUMMARY OF THE INVENTION

A principal aspect of the present invention is the design of a technique for specific assembly of monomeric polypeptides to form a stable heteromultimer. This technique of heteromultimer production facilitates high throughput production of functional heteromultimers and avoids the assembly of undesired homodimers. The method is particularly useful for generating a genetically diverse repertoire of heteromultimers such as antigen-binding units. The technique can readily be adapted to a variety of "genetic package display" technologies that facilitate the selection of antigen-binding units possessing the desired binding specificities. Such genetic package display technologies are detailed in U.S. Pat. Nos. 6,248,516, 5,969,108, 5,885,793, 5,837,500, 5,571,698, 5,223,409, 5,514,548, WO9005144, EP0368684, WO09201047, WO09311236, and WO09708320.

The subject antigen-binding unit is assembled and stabilized by the pairwise affinity of a distinct pair of heterodimerization sequences. The sequences are distinct in that at least one member of the heterodimerization pair is essentially incapable of forming homodimers under physiological buffer conditions and/or at physiological body temperatures. In certain embodiments, the stabilized antigen-binding unit not only has a molecular size smaller than a Fab fragment, but also exhibits the required binding specificity and affinity. Moreover, certain non-single-chain antigen-binding units of the present invention bear higher binding affinities than the corresponding conventional single-chain antibodies (scFv). The antigen-binding unit is particularly suited for antibody library construction and display. Certain configurations of the subject antigen-binding unit serve as convenient building units for multivalent and multispecific immunoglobulins.

Specifically, the present invention provides a non-single-chain antigen-binding unit comprising: (a) a light (L) chain polypeptide comprising a light (L) chain variable region fused to a first heterodimerization sequence; (b) a heavy (H) chain polypeptide comprising a heavy (H) chain variable region fused to a second heterodimerization sequence; wherein the L chain and the H chain polypeptides dimerize via pairwise affinity of the first and second heterodimerization sequences; and wherein at least one of the heterodimerization sequences is essentially incapable of forming a homodimer under physiological buffer conditions and/or at physiological body temperatures. Preferably, both of the first and second heterodimerization sequences are essentially incapable of forming homodimers under physiological buffer conditions and at physiological body temperatures.

In another aspect, the present invention provides a non-single-chain antigen-binding unit comprising: (a) a light (L) chain polypeptide comprising a light (L) chain variable region fused to a first heterodimerization sequence; (b) a heavy (H) chain polypeptide comprising a heavy (H) chain variable region fused to a second heterodimerization sequence; wherein the L chain and the H chain polypeptides dimerize via pairwise affinity of the first and second heterodimerization sequences which are derived from heterodimeric receptors. In one aspect, the first and second heterodimerization sequences comprising heterodimerization receptor sequences that mediate heterodimerization of the receptors. In yet another aspect, the subject heterodimerization sequences form a coiled-coil dimer. In still another aspect, the L and the H chain polypeptides dimerize via non-covalent pairwise affinity of the two heterodimerization sequences. Preferably, the L or the H chain polypeptide further comprises a flexon that is flanked by the variable region and the heterodimerization sequence. Both the L and H polypeptide sequences may be derived from human L and H chains. To further stabilize the heterodimeric Abus, cysteine residues can be introduced to provide disulfide bonds between the first and the second heterodimerization sequences. The non-single-chain antigen-binding units may be monovalent or multivalent. They may be monospecific or multispecific. Preferred multispecific Abus are bispecific, trispecific and tetraspecific molecules.

In a separate embodiment, the present invention provides a single-chain antigen-binding unit comprising a light (L) chain variable region and a heavy (H) chain variable region connected by a first and a second heterodimerization sequence spanning the distance between the C-terminus of one of the region to the N-terminus of the other region, wherein the two regions form an intra-molecular dimer via pairwise affinity of the first and second heterodimerization sequences; and wherein at least one of the heterodimerization sequences is essentially incapable of forming a homodimer under physiological buffer conditions and/or at physiological body temperatures. Preferably, both of the first and second heterodimerization sequences are essentially incapable of forming homodimers under physiological buffer conditions and at physiological body temperatures.

In another aspect, the present invention provides a single-chain antigen-binding unit, wherein the VL and VH regions form an intra-molecular dimer via pairwise affinity of the first and second heterodimerization sequences which are derived from heterodimeric receptors. In one aspect, the first and second heterodimerization sequences comprising heterodimerization receptor sequences that mediate heterodimerization of the receptors.

In yet another aspect, first and second heterodimerization sequences form a coiled-coil dimer. In another aspect, the first and second heterodimerization sequences dimerize via non-covalent pairwise affinity. Both the VL and VH regions can be derived from the corresponding sequences in a human L and H chains, respectively.

Both the non-single-chain and single-chain antigen-binding units can be conjugated to a chemically functional moiety. Exemplary functional moieties include but are not limited to signal peptides, agents that enhance immunologic reactivity, agents that facilitate coupling to a solid support, vaccine carriers, bioresponse modifiers, toxins, detectable labels, paramagnetic labels, and drugs.

Preferred heterodimerization sequences contained in the subject antigen-binding units are derived from C-terminal sequences of GABA_B receptor 1 and GABA_B receptor 2, respectively. More preferably, the first heterodimerization sequence is linked to a cysteine residue, said first heterodimerization comprising a GABA_B receptor 1 polypeptide of at least 30 amino acid residues that is essentially identical to a linear peptide sequence of comparable length depicted in SEQ ID NO. 2; and the second heterodimerization sequence is linked to a cysteine residue, said second heterodimerization comprising a GABA_B receptor 2 polypeptide of at least 30 amino acid residues that is essentially identical to a linear peptide sequence of comparable length depicted in SEQ ID NO. 4. Alternatively, the first heterodimerization sequence is linked to a cysteine residue, said first heterodimerization comprising a GABA_B receptor 2 polypeptide of at least 30 amino acid residues that is essentially identical to a linear peptide sequence of comparable length depicted in SEQ ID NO. 4; and the second heterodimerization sequence is linked to a cysteine residue, said second heterodimerization comprising a GABA_B receptor 1 polypeptide of at least 30 amino acid residues that is essentially identical to a linear peptide sequence of comparable length depicted in SEQ ID NO. 2.

The present invention provides a recombinant polynucleotide comprising a coding sequence that encodes the L and/or H polypeptide of a non-single-chain antigen-binding unit. The invention also provides a recombinant polynucleotide comprising a coding sequence that encodes the VL or VH regions of a single-chain antigen-binding unit. Also provided is a vector comprising any one of the recombinant polynucleotides described herein. The vector can be an expression vector, e.g. a phage display vector. Further provided in this invention is a selectable library of expression vectors encoding a repertoire of antigen binding units, comprising more than one subject vector. Preferably, the selectable library comprises a plurality of phage display vectors.

The present invention also provides a host cell comprising the subject recombinant polynucleotides. The recombinant polynucleotide encoding the L chain polypeptide and the polynucleotide encoding the H chain polypeptide may be present in a single vector or in separate vectors. The host cell may be eukaryotic or prokaryotic.

In yet another embodiment, the present invention provides a method of producing a non-single-chain antigen-binding unit. The method involves the following steps: (a) expressing in a host cell a first recombinant polynucleotide encoding a light (L) chain polypeptide comprising a light (L) chain variable region fused to a first heterodimerization sequence, and a second recombinant polynucleotide encoding a heavy (H) chain polypeptide comprising a heavy (H) chain variable region fused to a second heterodimerization sequence; wherein the L chain and the H chain polypeptides dimerize via pairwise affinity of the first and second heterodimerization sequences; and wherein at least one of the heterodimerization sequences is essentially incapable of forming a homodimer under physiological buffer conditions and/or at physiological body temperatures; and optionally (b) isolating the antigen-binding unit expressed in the host cell.

The produced antigen-binding unit may also contain heterodimerization sequences that are derived from heterodimeric receptors. Additionally, the non-single-chain antigen-binding expressed in step (a) can be displayed on surface of the host cell. Preferably, the non-single-chain antigen-binding expressed in step (a) is displayed on a phage particle.

In still another embodiment, the present invention provides a method of producing a non-single-chain antigen-binding unit, the method comprises the steps of (a) preparing a first recombinant polynucleotide encoding a light (L) chain polypeptide comprising a light (L) chain variable region fused to a first heterodimerization sequence, and a second recombinant polynucleotide encoding a heavy (H) chain polypeptide comprising a heavy (H) chain variable region fused to a second heterodimerization sequence; wherein the L chain and the H chain polypeptides dimerize via pairwise affinity of the first and second heterodimerization sequences; and wherein at least one of the heterodimerization sequences is essentially incapable of forming a homodimer under physiological buffer conditions and/or at physiological body temperatures; and (b) allowing the first and second polypeptides to dimerize via pairwise affinity of the first and second heterodimerization sequences. The step of dimerization may take place in vitro or in vivo.

This invention also includes a method of producing a single-chain antigen-binding unit. The methods involves the steps of (a) expressing in a host cell a polynucleotide comprising a coding sequence that encodes the subject single-chain antigen-binding unit; and optionally (b) isolating the single-chain antigen-binding unit expressed in the host cell.

This invention further includes a method of displaying a chimeric heteromultimer comprising at least two polypeptides on a surface of a host cell. This method comprises expressing in the host cell (i) a first recombinant polynucleotide encoding a first polypeptide fused to a first heterodimerization sequence and a surface presenting sequence; (ii) a second recombinant polynucleotide encoding a second polypeptide fused to a second heterodimerization sequence; wherein the first and second polypeptides dimerize via pairwise affinity of the first and second heterodimerization sequences; wherein at least one of the heterodimerization sequences is incapable of forming a homodimer under physiological buffer conditions and/or at physiological body temperatures. In one aspect, the first and second polynucleotides are expressed by a single phage display vector. In another aspect, the first and second polynucleotides are expressed by separate phage display vectors. The chimeric heteromultimer is preferably a non-single-chain antigen-binding unit of the present invention.

The invention also encompasses a method of identifying a non-single-chain antigen-binding unit that is immunoreactive with a desired antigen. The method comprises the steps of: (a) preparing a genetically diverse repertoire of antigen-binding units, wherein the repertoire comprises more than one subject antigen-binding unit; (b) contacting the repertoire of antigen binding units with the desired antigen; and (c) detecting a specific binding between antigen binding units and the antigen, thereby identifying the antigen-binding unit that is immunoreactive with the desired antigen. In one aspect of this embodiment, the repertoire of antigen-binding units is prepared by expressing a library of vectors encoding a plurality of the antigen-binding units. Preferably, the library of vectors comprises a plurality of phage vectors.

Finally, the present invention provides a kit comprising a vector of this invention in suitable packaging.

Claim 1 of 38 Claims

What is claimed is:

1. A non-single-chain antigen-binding unit comprising:

(a) a light (L) chain polypeptide comprising a light (L) chain variable region fused in-frame to a first heterodimerization polypeptide;

(b) a heavy (H) chain polypeptide comprising a heavy (H) chain variable region fused in-frame to a second heterodimerization polypeptide;

wherein the L chain and the H chain polypeptides dimerize via pairwise affinity of the first and second heterodimerization polypeptides; and wherein at least one of the heterodimerization polypeptides is essentially incapable of fanning a homodimer under physiological buffer conditions and/or at physiological body temperatures.

____________________________________________
If you want to learn more about this patent, please go directly to the U.S. Patent and Trademark Office Web site to access the full patent.