HUMANIZATION OF NON-HUMAN, MAMMALIAN ANTIBODIES Cross Reference to Related Applications This application is a continuation-in-part of United States Patent Application Serial No. 09/543,004, filed April 4,2000, which is in turn a continuation-in-part of United States Patent Application Serial No. 09/425,638, filed October 22,1999. The disclosures of both these applications are incorporated herein by reference.

Technical Field of the Invention The field of this invention is immunology. More particularly, the present invention pertains to a process of humanizing a rabbit antibody.

Background of the Invention Use of antibodies as therapeutic agents is gaining acceptance as an important and valuable approach in the treatment of various conditions. The specificity of antibodies makes them particularly useful in treating conditions where a"marker"or "markers"characterize abnormal cells. Antibodies effectively target such cells by binding to these markers, which are molecules present in, or preferably on, the cell type of interest.

Initial forays into the production of antibodies used mice as subject animals. In vivo use of murine antibodies in humans has been curtailed, however, for a number of reasons. Murine antibodies, recognized as foreign by a human host, elicit the so-called "human anti-mouse antibody"or"HAMA"response (See, e. g., Schiff, et al., Canc.

Res. 45: 879-885 (1985)). In addition, the Fc portion of murine antibodies is not as efficacious in stimulating human complement or cell mediated toxicity.

There have been extensive and intensive efforts to circumvent such problems.

One such approach is the development of chimeric antibodies (See e. g., European Patent Applications 120694 and 125023). Chimeric antibodies contain portions of antibodies from two or more different species, such as the variable regions of a mouse antibody, and the constant regions of a human antibody. The advantage of such chimeras is that they retain the specificity of murine antibodies, but also stimulate human Fc complement fixation. Such chimeras can still elicit a HAMA response,

however (See. e. g., Bruggemann, et al., J. Exp. Med 170: 2153-2157 (1989)).

Additional approaches have been sought which would alleviate these problems.

British Patent Application GB 2188638A and US Patent No. 5,585,089 disclose processes wherein recombinant antibodies are produced where the only portion of the antibody that is substituted is the complementarity determining region, or"CDR."The CDR grafting technique has been used to generate antibodies which consist of murine CDRs, and human variable region framework and constant regions (See. e. g., Riechmann, et al., Nature 332: 323-327 (1988)). These antibodies retain the human constant regions that are necessary for Fc dependent effector function, but are much less likely to evoke a HAMA response.

Substitution of murine CDRs for human CDRs is not generally sufficient to generate an efficacious humanized antibody. The humanized antibodies must include a small number of critical murine antibody residues in the human variable region. The particular residues of importance depend upon the structure of both the murine antibody and human antibody (See. e. g., WO 04381 to Harris et al.). Notwithstanding these issues, humanized antibodies have become much more available (U. S. Patent No.

5,952,484 to Wallace et al. and U. S. Patent No. 5,958,412 to Welt et al., both of which are incorporated herein by reference).

The rabbit Ig gene repertoire has been well characterized (See e. g., Knight, et al., Adv. Immunol 56 : 179-218 (1994)). This characterization has permitted selection of monoclonal antibodies, by screening combinatorial antibody libraries displayed on phage (Ridder, et al., Biotechnology 95 (15): 8910-15 (1998); Vaughan, et al., Nat.

Biotechnol. 16 (6): 535-539 (1998)). This information, together with information discussed infra, has been used to develop the invention described herein.

The structure of an immunoglobulin is discussed in standard textbooks such as Paul, W. E, Fundamental Immunology, Raven Press, New York, New York, 1993.

Incorporated herein by reference.

Herein we exploit a new route to human antibodies. We use display technology to select and humanize antibodies from rabbits that were immunized with a human antigen. Monoclonal antibodies from multiple species are accessible by screening combinatorial antibody libraries. The accessibility of antibody repertoires of multiple

species is substantial in the search for new therapeutic antibodies. In particular the rabbit antibody repertoire, which in form of polyclonal antibodies has been used in diagnostic applications for decades, is an attractive source for therapeutic antibodies.

The humanization of rabbit antibodies, however, has not previously been reported.

Brief Summary of the Invention In one aspect, the present invention provides a process of preparing a humanized rabbit antibody that specifically immunoreacts with a particular antigen.

The process includes the steps of : expressing a library of rabbit light and heavy chain variable domain sequences obtained from a rabbit immunized with the antigen; screening the library to identify those variable domain sequences that specifically immunoreact with the antigen; expressing a library of antibodies comprising CDR regions from the variable domain sequences selected in the previous step with framework regions derived from a human; and screening the expressed library from the previous step to identify antibodies that specifically immunoreact with the antigen.

Preferably, one or both of the expressing steps in the process involve expression using display technology. The first expression step in the process can be of a library containing only rabbit sequences or of a library of rabbit variable domain sequences grafted to human constant domain sequences. In a preferred embodiment, expressing is accomplished in vitro in a microorganism such as E. coli, yeast or cells derived from a higher organism such as mammalian cell lines.

Preferably, the human framework regions are selected from human light chain and human heavy chain framework regions having homology with the rabbit variable domain sequences. The human framework regions can be a consensus human sequence. The human framework region sequences are preferably diversified so that at least one amino acid residue in the sequence is altered. The position of the diversified residue (s) corresponds to a position known to be involved in either antigen binding or CDR conformation. More preferably, at least 5 residues are altered. The diversified/altered residues can be found in either the light or heavy chain framework region sequence. In another embodiment, select positions in the CDR regions can also be diversified.

The humanized antibody or a portion of the humanized antibody can consist of rabbit CDR and human constant regions. The humanized antibody can be an antibody or fragment thereof (e. g., Fab fragment). Further, the antigen can be a molecule or portion of a molecule presented on a cell surface. For example, the antigen can be a molecule or portion of a molecule presented on a cell surface of a neoplastic (e. g., cancer) cell. Another embodiment of the invention is directed to a humanized antibody made by any of the methods described above.

Antibody engineering technology has facilitated the generation of antibodies to virtually any antigen of interest as well as their improvement in terms of affinity, specificity, and immunogenicity. The generation of human antibodies to human antigens is of particular interest for the treatment of a variety of diseases. Here we report a new route to therapeutically relevant human antibodies. Using phage display, we selected and humanized antibodies from rabbits immunized with a human antigen.

This has allowed the generation of combinatorial rabbit antibody libraries displayed on phage whose screening resulted in the selection of rabbit monoclonal antibodies. While the generation of rabbit monoclonal antibodies by hybridoma technology has also been reported, the phage display approach with its inherent linkage of phenotype and genotype provides ready access to antibody sequences and facilitates further in vitro optimizations such as humanization or affinity maturation.

Compared to the other existing sources of human antibodies, immune rabbits are an attractive alternative for several reasons. Firstly, human antibodies from immune rabbits extend the accessible epitope repertoire of a given antigen. Epitopes that are not immunogenic in mice, a species from which the vast majority of monoclonal antibodies to human antigens has been generated, might be immunogenic in rabbits. This is of particular interest for the development of therapeutic human antibodies that are evaluated in mouse models and are required to recognize both the human antigen and its mouse homologue. In contrast, human antibodies that are derived from immune mice, either indirectly through humanization or directly through transgenic mice containing human Ig loci, are negatively selected against epitopes displayed by the mouse homologue. Secondly, in contrast to human antibodies derived from large naive combinatorial antibody libraries that are selected in vitro, human antibodies derived

from immune animals have been subjected to in vivo selection and, thus, are more likely to recognize a given antigen selectively. Lastly, as we demonstrate here for the first time, rabbit antibodies can be converted to human antibodies that retain both high specificity and affinity to the antigen.

Using Fab display we selected rabbit antibodies with dissociation constants as low as 390 pM. While different antigens as well as different immunization schedules are likely to contribute to these differences, Fab display in general can result in the selection of antibodies with higher affinity. This is based on the fact that Fab are displayed monovalently at the phage surface and are selected based on affinity and expression, whereas the selection of scFv is also influenced by avidity due to their tendency to dimerize or form higher order aggregates. As a result, antibodies selected as scFv do not generally bind their antigen well when they are converted in the Fab format. In contrast, we found that the opposite conversion from Fab to scFv format yielded scFv with high apparent affinity. Fab are also readily convertible to whole antibody formats by simple fusion with Fc coding sequences which does not affect the antigen binding site. In summary, selections based on the Fab format generate antibodies that are well suited for therapeutical evaluation in multiple formats.

We chose human A33 antigen as target for the development of human antibodies from immune rabbits (See Examples that follow). The human antibodies described here compare favorably to the previously described human antibody A33.

They were found to recognize a surface accessible epitope of the A33 antigen with a specificity as high as mouse and human antibody A33. The monovalent affinity of human clone B is 5-to 10-fold higher than the monovalent affinity reported for mouse and human antibody A33. Using surface plasmon resonance, we further found that the human antibodies described here (i) bind to an epitope on the A33 antigen distinct from the epitope recognized by mouse and human antibody A33, and (ii) are not recognized by sera from patients who had developed an antiidiotypic immune response after repeated treatment with human antibody A33. Thus, regarding therapeutic application, our human antibodies derived from the rabbit antibody repertoire complement human antibody A33 derived from the mouse antibody repertoire. Based on these findings, immune rabbits are shown to be an important source for the generation of therapeutic

human antibodies.

Brief Description of the Drawings In the drawings, which form a portion of the specification Figure 1 depicts the amino acid sequence of the V regions of rabbit anti A33 antigen antibodies. In particular, three rabbit antibodies, rabbit 1, rabbit 2 and rabbit 3 are shown. Further, the V sequence of the humanized antibodies is shown. Finally, the amino acid sequence of six human antibodies, labelled human A to F are listed. The framework regions, corresponding to about amino acids 1-22 (FR1), 35-49 (FR2), 57- 88 (FR3) and 98-107 (FR4) of the VL chain and amino acids 1-30 (FR1), 36-49 (FR2), 66-94 (FR3), 103-113 (FR4) of the VH chain. The CDR regions correspond to about amino acids 24-34 (CDR1), 50-56 (CDR2), 89-97 (CDR3) of the VL chain and about amino acids 31-35 (CDR1), 50-65 (CDR2), and 95-109 (CDR3) of the VH chain.

Figure 2 depicts Western blot reactivity of human Fab B with Triton X-100 extracts of human A33 antigen expressing (LIM 1215, SW1222) and nonexpressing (SW620) human colon cancer cell lines. Specific binding was detected by alkaline- phosphatase-conjugated goat anti-human F (ab') 2polyclonal antibodies and visualized using chemiluminescence. Numbers on the left indicate molecular masses of standard proteins in kilodaltons ("kDs").

Figure 3 depicts flow cytometry histograms demonstrating that the selected rabbit clones 1 and 2 as well as the selected human clones A-F bind specifically to native human A33 antigen expressed on the cell surface. For indirect immunofluorescence staining, cells were incubated with Fab (except for the control) followed by FITC-conjugated secondary antibodies. Human colon cancer cell lines LIM1216 (bold line) and SW1222 (fine line) are known to express human A33 antigen, whereas HT29 (dotted line) is known not to. The y axis gives the number of events in linear scale, the x axis the fluorescence intensity in logarithmic scale.

Figure 4 depicts the results of affinity measurements on human Fab. Human A refers to human VLA and VHA, Human B refers to human VLB and VHB, Human C refers to human VLC and VHC, Human D refers to human VLD and VHD, Human E refers to human VLE and VHE, Human F refers to human VLF and VHF.

Figure 5 depicts analysis of purified rabbit and humanized Fab by SDS-PAGE and Coomassie Blue staining. Fabs were purified from E. coli cultures by Protein G affinity chromatography. Numbers on the right indicate molecular masses of standard proteins in kD.

Figure 6 depicts representative Biacore sensorgrams obtained for the binding of rabbit Fab 1 to immobilized human A33 antigen. For association, Fab were injected at 5 different concentrations (200 nM, 150 nM, 125 nM, 100 nM, 75 nM, top to bottom) between t = 125 and t = 370 seconds, using a flow rate of 5 pl/minute. For dissociation, the flow rate was increased to 50 pl/minute resonance units (RU).

Figure 7 shows immunohistochemical reactivity of humanized Fab B in human colon cancer tissue sections. A and B, xenograft of human colon cancer cell line SW 1222 in nude mice, C-F, serial sections of moderately differentiated human colon adenocarcinoma. Scale bar = 300 film. Specific binding was detected by biotinylated goat-anti human F (ab') 2 polyclonal antibodies and visualized using an avidin-biotin- complex system and diaminobenzidine tetrahydrochloride as a chromogen. (A) Humanized Fab B showed intense staining in SW1222 xenograph. (B) Buffer only without application of humanized Fab (negative control) showing no staining in SW 1222 xenograft. (C) Mouse monoclonal antibody A33 showing intense staining of dysplastic glandular structures in human colon adenocarcinoma. (D) Humanized Fab B revealing similar staining in corresponding carcinoma areas after blocking of endogenous human immunoglobulins. No staining of additional tissue components due to endogenous human immunoglobulins is detectable. (E) Buffer only without application of humanized Fab (negative control) but with blocking of endogenous human immunoglobulins showing no staining. (F) Buffer only without application of humanized Fab (negative control) and omitting the blocking of endogenous human immunoglobulins showing intense staining of endogenous human immunoglobulins.

Detailed Description of the Invention A. Definitions "Antibody"as used herein means single monoclonal antibodies (including agonist and antagonist antibodies), antibody compositions with polyepitopic specificity

and fragments thereof."Antibody fragments"comprise a portion of an intact antibody, generally the antigen binding site or variable region of the intact antibody. Examples of antibody fragments include Fab, Fab', Fab'-SH, F (ab'). sub. 2, and Fv fragments; diabodies; any antibody fragment that is a polypeptide having a primary structure consisting of one uninterrupted sequence of contiguous amino acid residues (referred to herein as a"single-chain antibody fragment"or"single chain polypeptide"), including without limitation (1) single-chain Fv (scFv) molecules (2) single chain polypeptides containing only one light chain variable domain, or a fragment thereof that contains the three CDRs of the light chain variable domain, without an associated heavy chain moiety and (3) single chain polypeptides containing only one heavy chain variable region, or a fragment thereof containing the three CDRs of the heavy chain variable region, without an associated light chain moiety; and multispecific antibodies formed from antibody fragments.

"Antibodies" (Abs) and"immunoglobulins" (Igs) are glycoproteins having the same structural characteristics. While antibodies exhibit binding specificity to a specific antigen, immunoglobulins include both antibodies and other antibody-like molecules which lack antigen specificity. Polypeptides of the latter kind are, for example, produced at low levels by the lymph system and at increased levels by myelomas.

Bispecific antibody (or heteroantibodies): A multivalent antibody containing binding sites specific for two different antigenic determinants. A bispecific antibody can be chemically synthesized as antibody heteroconjugates (AHCs) by covalently attaching two whole monoclonal antibodies ("whole AHCs") (B. Karpovsky, et al.

(1984) J. Exp. Med. 160 (6): 1686-1701) or by attaching two monoclonal antibody Fab or Fab'fragments ("monovalent AHCs") (M. Brennan, et al., Science (1985) 229: (1708): 81-83), where each antibody or antibody fragment has a different antigenic specificity. Alternatively, bispecific antibodies can be produced from a"hybrid hybridoma,"a cell fusion of two monoclonal antibody-producing cells (C. L. Reading, in HYBRIDOMAS AND CELLULAR IMMORTALITY, B. H. Tom et al., eds., 1984, (New York: Plenum Press), p. 235; U. D. Staerz et al., Proc. Natl. Acad. Sci. (1986) 83: 1453-1457; A. Lanzavecchia et al., Eur. J. Immunol. (1987) 17: 105-111; D. B. Ring et al., in BREAST EPITHELIAL ANTIGENS: MOLECULAR BIOLOGY TO CLINICAL

APPLICATIONS, R. Cedani, ed., 1991, (New York: Plenum Press), pp. 91-104).

Epitope: A portion of a molecule that is specifically recognized by an immunoglobulin product. It is also referred to as the determinant or antigenic determinant.

Fab fragment: A protein consisting of the portion of an immunoglobulin molecule containing the immunologically active portions of an immunoglobulin heavy chain and an immunoglobulin light chain covalently coupled together and capable of specifically combining with antigen. Fab fragments are typically prepared by proteolytic digestion of substantially intact immunoglobulin molecules with papain using methods that are well known in the art; however, a Fab fragment can also be prepared by expressing the desired portions of immunoglobulin heavy chain and immunoglobulin light chain in a host cell, using methods well known in the art.

F,, fragment: A protein consisting of the immunologically active portions of an immunoglobulin heavy chain variable region and an immunoglobulin light chain variable region covalently coupled together and capable of specifically combining with antigen. F fragments are typically prepared by expressing the desired portions of immunoglobulin heavy chain variable region and immunoglobulin light chain variable region in a host cell using methods well known in the art.

"Humanized"forms of non-human (e. g., murine) antibodies are specific chimeric immunoglobulins, immunoglobulin chains or fragments thereof (such as Fv, Fab, Fab', F (ab') sub 2, or other antigen-binding subsequences of antibodies) which contain minimal sequence derived from non-human immunoglobulin. For the most part, humanized antibodies are human immunoglobulins (recipient antibody) in which residues from a complementary-determining region (CDR) of the recipient are replaced by residues from a CDR of a non-human species (donor antibody) such as mouse, rat, or rabbit having the desired specificity, affinity, and capacity. In some instances, Fv framework residues of the human immunoglobulin are replaced by corresponding non-human residues. Furthermore, humanized antibodies can comprise residues which are found neither in the recipient antibody nor in the imported CDR or framework sequences. These modifications are made to further refine and maximize antibody performance. In general, the humanized antibody will comprise substantially all of at

least one, and typically two, variable domains, in which all or substantially all of the CDR regions correspond to those of a non-human immunoglobulin and all or substantially all of the FR regions are those of a human immunoglobulin consensus sequence. The humanized antibody optimally also will comprise at least a portion of an immunoglobulin constant region (Fc), typically that of a human immunoglobulin. For further details see Jones et al., Nature 321: 522 (1986); Reichmann et al., Nature 332: 323 (1988); and Presta, Curr. Op. Struct. Biol. 2: 593 (1992).

Immunoglobulin superfamily molecule: A molecule that has a domain size and amino acid residue sequence that is significantly similar to immunoglobulin or immunoglobulin related domains. The significance of similarity is determined statistically using a computer program such as the Align program described by Dayhoff et al., Meth Enzymol., 91: 524-545 (1983) incorporated by reference. A typical Align score of less than 3 indicates that the molecule being tested is a member of the immunoglobulin gene superfamily. Exemplary of immunoglobulin superfamily molecules include the following members: immunoglobulin heavy chains (i. e., the heavy chain of IgM, IgD, IgG, IgA or IgE and light chains K and k), T cell receptors (a, P, y, X, CD3), major histocompatibility antigens (Class I H-chain, (32-microglobulin, Class II (cc and ß)), ß2-microglobulin associated antigens (TL H chain, Qa-2 H chain, CDIa H chain), T lymphocyte antigens (CD2, CD4, CD7, CD8 chain I, CD8 Chain IId, CD28 and CTLA4), haemopoietic/endothelium antigens (LFA-3, MRC OX-45), brain/lymphoid antigens (Thy-1, MRC OX-2), immunoglobulin receptors (Poly Ig R, Fc gamma 2b/gamma 1R, FcERI (a)), neural molecules (Neural adhesion molecule, Myelin associated gp, Po myelin protein, Tumor antigen (carcinoembryonic antigen (CEA)), growth factor receptors (platelet-derived growth factor (PDGF) receptor, colony stimulating factor-1 (CSF1) receptor), non-cell surface molecules (°61 B-glycoprotein, basement membrane link protein) and A33 antigen (Heaths et al., Proc Natl Acad Sci 94: 469-474 (1997)) (See e. g., Williams and Barclay, in Immunglobulin Genes, p 361, Academic Press, NY (1989); and Sequences of Proteins of Immunological Interest, 4th ed., U. S. Dept. of Health and Human Serving (1987)).

Inducible promoter: A promoter where the rate of RNA polymerase binding and initiation is modulated by external stimuli. Such stimuli include light, heat, anaerobic

stress, alteration in nutrient conditions, presence or absence of a metabolite, presence of a ligand, microbial attack, wounding and the like.

Insert: A DNA sequence foreign to the host, consisting of a structural gene and optionally additional DNA sequences.

"Mammal"for purposes of treatment refers to any animal classified as a mammal, including humans, domestic and farm animals, and zoo, sports, or pet animals, such as dogs, horses, cats, cows, etc.

"Monoclonal antibody" (mAb) as used herein refers to an antibody obtained from a population of substantially homogeneous antibodies, i. e., the individual antibodies comprising the population are identical except for possible naturally occurring mutations that may be present in minor amounts. Monoclonal antibodies are highly specific, being directed against a single antigenic site. Furthermore, in contrast to conventional (polyclonal) antibody preparations which typically include different antibodies directed against different determinants (epitopes), each mAb is directed against a single determinant on the antigen. In addition to their specificity, the monoclonal antibodies are advantageous in that they can be synthesized by hybridoma culture, uncontaminated by other immunoglobulins. The modifier"monoclonal" indicates the character of the antibody as being obtained from a substantially homogeneous population of antibodies, and is not to be construed as requiring production of the antibody by any particular method. For example, the monoclonal antibodies to be used in accordance with the present invention may be made by the hybridoma method first described by Kohler et al., Nature, 256: 495 (1975), or may be made by recombinant DNA methods (see, e. g., U. S. Pat. No. 4,816,567 to Cabilly et al.). The"monoclonal antibodies"also include clones of antigen-recognition and binding-site containing antibody fragments (Fv clones) isolated from phage antibody libraries using the techniques described in Clackson et al., Nature, 352: 624-628 (1991) and Marks et al., J. Mol. Biol., 222: 581-597 (1991), for example. The monoclonal antibodies herein include hybrid and recombinant antibodies produced by splicing a variable (including hypervariable) domain of an anti-IL-8 antibody with a constant domain (e. g."humanized"antibodies), or a light chain with a heavy chain, or a chain from one species with a chain from another species, or fusions with heterologous

proteins, regardless of species of origin or immunoglobulin class or subclass designation, as well as antibody fragments (e. g., Fab, F (ab') sub 2, and Fv), so long as they exhibit the desired biological activity. (See, e. g., U. S. Pat. No. 4,816,567 to Cabilly et al.; Mage and Lamoyi, in Monoclonal Antibody Production Techniques and Applications, pp. 79-97 (Marcel Dekker, Inc., New York, 1987).) The monoclonal antibodies herein specifically include"chimeric"antibodies (immunoglobulins) in which a portion of the heavy and/or light chain is identical with or homologous to corresponding sequences in antibodies derived from a particular species or belonging to a particular antibody class or subclass, while the remainder of the chain (s) is identical with or homologous to corresponding sequences in antibodies derived from another species or belonging to another antibody class or subclass, as well as fragments of such antibodies, so long as they exhibit the desired biological activity (Cabilly et al., supra; Morrison et al., Proc. Natl. Acad. Sci. U. S. A. 81: 6851 (1984)).

Multimeric protein: A globular protein containing more than one separate polypeptide or protein chain associated with each other to form a single protein. Both heterodimeric and homodimeric proteins are multimeric proteins.

"Native antibodies and immunoglobulins"are usually heterotetrameric glycoproteins of about 150,000 daltons, composed of two identical light (L) chains and two identical heavy (H) chains. Each light chain is linked to a heavy chain by one covalent disulfide bond, while the number of disulfide linkages varies between the heavy chains of different immunoglobulin isotypes. Each heavy and light chain also has regularly spaced intrachain disulfide bridges. Each heavy chain has at one end a variable domain (V. sub. H) followed by a number of constant domains. Each light chain has a variable domain at one end (V. sub. L) and a constant domain at its other end; the constant domain of the light chain is aligned with the first constant domain of the heavy chain, and the light chain variable domain is aligned with the variable domain of the heavy chain. Particular amino acid residues are believed to form an interface between the light-and heavy-chain variable domains (Clothia et al., J. Mol. Biol. 186: 651 (1985); Novotny and Haber, Proc. Natl. Acad. Sci. U. S. A. 82: 4592 (1985)). The term "variable"refers to the fact that certain portions of the variable domains differ

extensively in sequence among antibodies and are used in the binding and specificity of each particular antibody for its particular antigen. However, the variability is not evenly distributed throughout the variable domains of antibodies. It is concentrated in three segments called complementarity-determining regions (CDRs) or hypervariable regions both in the light-chain and the heavy-chain variable domains. The more highly conserved portions of variable domains are called the framework (FR). The variable domains of native heavy and light chains each comprise four FR regions, largely adopting a beta-sheet configuration, connected by three CDRs, which form loops connecting, and in some cases forming part of the beta.-sheet structure. The CDRs in each chain are held together in close proximity by the FR regions and, with the CDRs from the other chain, contribute to the formation of the antigen-binding site of antibodies (see Kabat et al., Sequences of Proteins of Immunological Interest, Fifth Edition, National Institute of Health, Bethesda, Md. (1991)). The constant domains are not involved directly in binding an antibody to an antigen, but exhibit various effector functions, such as participation of the antibody in antibody-dependent cellular toxicity. Papain digestion of antibodies produces two identical antigen-binding fragments, called"Fab"fragments, each with a single antigen-binding site, and a residual"Fc"fragment, whose name reflects its ability to crystallize readily. Pepsin treatment yields an F (ab'). sub. 2 fragment that has two antigen-combining sites and is still capable of cross-linking antigen. The Fab fragment also contains the constant domain of the light chain and the first constant domain (CH1) of the heavy chain. Fab' fragments differ from Fab fragments by the addition of a few residues at the carboxy terminus of the heavy chain CH1 domain including one or more cysteines from the antibody hinge region. Fab'-SH is the designation herein for Fab'in which the cysteine residue (s) of the constant domains bear a free thiol group. F (ab'). sub. 2 antibody fragments originally were produced as pairs of Fab'fragments which have hinge cysteines between them. Other chemical couplings of antibody fragments are also known. The"light chains"of antibodies (immunoglobulins) from any vertebrate species can be assigned to one of two clearly distinct types, called kappa (k) and lambda (1), based on the amino acid sequences of their constant domains. Depending on the amino acid sequence of the constant domain of their heavy chains, immunoglobulins

can be assigned to different classes. There are five major classes of immunoglobulins : IgA, IgD, IgE, IgG, and IgM, and several of these can be further divided into subclasses (isotypes), e. g., IgG subl, IgG sub 2, IgG sub 3, IgG sub 4, IgA sub 1, and IgA sub 2.

The heavy-chain constant domains that correspond to the different classes of immunoglobulins are called alpha., delta., epsilon., gamma., and mu, respectively. The subunit structures and three-dimensional configurations of different classes of immunoglobulins are well known.

"Polymerase chain reaction"or"PCR"refers to a procedure or technique in which minute amounts of a specific piece of nucleic acid, RNA and/or DNA, are amplified as described in U. S. Pat. No. 4,683,195 issued Jul. 28,1987. Generally, sequence information from the ends of the region of interest or beyond needs to be available, such that oligonucleotide primers can be designed; these primers will be identical or similar in sequence to opposite strands of the template to be amplified. The 5'terminal nucleotides of the two primers can coincide with the ends of the amplified material. PCR can be used to amplify specific RNA sequences, specific DNA sequences from total genomic DNA, and cDNA transcribed from total cellular RNA, bacteriophage or plasmid sequences, etc. See generally Mullis et al., Cold Spring Harbor Symp. Quant. Biol. 51: 263 (1987); Erlich, ed., PCR Technology (Stockton Press, New York, 1989). As used herein, PCR is considered to be one, but not the only, example of a nucleic acid polymerase reaction method for amplifying a nucleic acid test sample comprising the use of a known nucleic acid as a primer and a nucleic acid polymerase to amplify or generate a specific piece of nucleic acid.

Polypeptide and peptide: A linear series of amino acid residues connected one to the other by peptide bonds between the alpha-amino and carboxy groups of adjacent residues.

Promoter: A recognition site on a DNA sequence or group of DNA sequences that provide an expression control element for a gene and to which RNA polymerase specifically binds and initiates RNA synthesis (transcription) of that sequence.

Protein: A linear series of greater than about 50 amino acid residues connected one to the other as in a polypeptide.

As used herein, protein, peptide and polypeptide are used interchangeably to

denote an amino acid polymer or a set of two or more interacting or bound amino acid polymers.

Structural gene: A nucleic acid molecule coding for a polypeptide and being in operable linkage with a suitable promoter, termination sequence and optionally other regulatory DNA sequences.

VHCDR1, VHCDR2, and VHCDR3 denote immunoglobulin heavy chain complementarity determining region 1,2 and 3 respectively.

VHFR1, VHFR2, and VHFR3 VHFR4 denote immunoglobulin heavy chain framework region 1,2,3 and 4 respectively.

VLCDR1, VLCDR2, and VLCDR3 denote immunoglobulin light chain complementarity determining region 1,2 and 3 respectively.

VLFR1, VLFR2, and VLFR3 VLFR4 denote immunoglobulin light chain framework region 1,2,3 and 4 respectively.

B. Process of Humanizing Rabbit Immunoglobulins The present invention provides a process of humanizing a non-human antibody and humanized antibodies produced in accordance with that process. Preferred non- human, mammalian antibodies include mouse, rabbit and chicken antibodies. The present process is preferably used to humanize rabbit antibodies. The process of the present invention differs from processes of the prior art by using a unique combination of library expression using display technologies and high throughput screening.

In accordance with the present process, a library of antibody variable domain sequences obtained from a non-human, mammal immunized with a particular antigen (preferably a human antigen) are expressed as a library using display technology well known in the art (See, e. g., Fitzgerald, Drug Discovery Today, 5: 253-258,2000).

Expression using the well known technique of phage display is preferred. The antibody variable domain sequences can be expressed as antibodies or fragments thereof together with other non-human or human constant domains. In a preferred embodiment, the library is expressed as a chimeric molecule containing non-human variable domain sequences grafted to human constant domain sequences. Grafting can be accomplished using any technique well known in the art. By way of example, a rabbit antibody

library displayed on phage is generated as follows. RNA is isolated from bone marrow and spleen of the immune rabbits, retro-transcribed, and VL and VH coding sequences amplified using a variety of primer combinations designed to amplify most of the known rabbit antibody sequences. Importantly, the rabbit antibody library is based on a chimeric Fab format. Variable domains from rabbit light and heavy chains are fused to the corresponding human constant domains. The use of human constant domains suggests several advantages. Firstly, while antigen binding is confined to the variable domains and, thus, is not expected to be influenced by constant domain swapping, the human constant domains confer established and standardized detection and purification means on Fab derived from multiple species. Secondly, the use of human constant regions improves the E. coli expression level of Fab. Lastly, a Fab with human constant domains is already partially humanized and can be readily channeled into recently reported strategies for complete humanization.

Selection of preferred variable domain sequences for subsequent steps in the process is accomplished by screening the library for immunoreactivity (binding specificity and affinity) to the particular antigen. By way of example, the phage library displaying chimeric rabbit/human Fab is selected by panning against immobilized antigen. Clones that demonstrate strong immunoreactivity with the antigen are selected. Clones are produced as soluble Fab in E. coli and purified by protein G affinity chromatography.

As a further step in the generation of therapeutic antibodies directed to the antigen, the selected variable domains are humanized using a humanization strategy that combines CDR grafting with framework and CDR fine tuning by phage display.

This type of strategy has been used for the humanization of mouse antibodies. The selected non-human variable domain sequences (six rabbit CDR sequences, 3 from each variable domain as defined by Kabat et al. (Kabat, et al. (1991) Sequences of Proteins oflmmunological Interest, 5th Ed., Public Health Service, National Institutes of Health, Bethesda, MD)) are grafted into the appropriate human framework sequences. Any human framework sequences can be used. One of skill in the art will recognize that grafting occurs by maintaining the proper alignment of light and heavy chain CDR and framework sequences. Either human kappa or lambda framework

sequences can be used. In a preferred embodiment, human framework sequences are obtained from human antibodies that show homology to the selected non-human variable domain sequences. Preferably, the human sequence is greater than about 50% identical to the selected non-human sequences. More preferably, the human gene is greater than about 70% identical and, even more preferably, greater than about 75% identical. By way of example, selected rabbit variable domain sequences are aligned with human V and J genes to identify the greatest homology to the human V and J genes.

Optimization of obtaining humanized antibodies having the greatest immunoreactivity with the antigen is accomplished by framework and CDR fine tuning or diversification. As used herein,"fine tuning"means identifying and selecting specific sequences that result in humanized antibodies having the greatest specificity and affinity for the antigen (See, e. g., PCT Publication WO 98/45332). As disclosed therein, a particularly preferred method for producing a humanized murine antibody involves the following: preparing an antibody phagemid vector for monovalent display of Fab fragments having CDR sequences transplanted by site-directed mutagenesis onto a vector which codes for a human VLKI-CK, light chain and human VH E-CH 15/heavy chain Fd; constructing the antibody Fab phagemid library by random mutagenesis of a small set of selected critical framework residues; expressing and purifying the humanized Fab fragments; selecting humanized Fab variants; and, determining binding affinities. These steps do not have to be performed in any particular order. These steps are generally performed as described below.

Based on the cumulative results from humanizing a number of non-human antibodies onto a human VLKI-VHE framework, framework changes required to optimize antigen binding are likely limited to some subset of the residues. See, Carter et al., PNAS USA 89,4285 (1992); Presta et al., J Immunol. 151,2623 (1993); Eigenbrot et al., Proteins 18,49-62 (1994); Shalaby et al., J. Exp. Med. 175,217 (1992). Accordingly, a novel group of residues is selected for randomization.

Randomizing these identified key framework residues provides the desired library of Fab variants to be displayed on the surface of filamentous phage. Specifically, VL residues 4 and 71 and V residues 24,37,67,69,71,71,75,76,78,93 and 94 have been

selected as key framework residues important for antigen binding and targeted for randomization.

By way of specific example relating to the instant disclosure, a tryptophan in a rabbit sequence is converted to serine. By way of further example, for framework fine tuning, residues at 6 positions in the human VH framework sequences and residues at 4 positions in the human Vk framework sequences are diversified. These residues are chosen out of a set of key framework residues that are known to be involved in antigen binding, either indirectly, by supporting the confirmation of the CDR loops, or directly, by contacting the antigen. Key framework residues of rodent monoclonal antibodies have been identified structurally by crystallization and molecular modeling as well as empirically by antibody humanization. No such data are available for rabbit monoclonal antibodies. However, the similarity of the primary structure of rabbit antibodies with rodent antibodies makes an analogous set of key framework residues likely. From this set, positions that differed between the rabbit and the human frameworks are identified.

These positions are then diversified to allow the selection of basically either the non- human sequence, the human sequence or a consensus sequence.

As an alternative to optimizing the framework residues to accept novel CDR sequences, the CDR regions may be directly modified and selected to work within defined human frameworks directly. Rabbit anti-human A33 antigen, clone 2 (Rader et al. (2000) J. Biol. Chem. 275,13668-13676) was used as the starting point for humanization by CDR fine tuning. The genes for the vh and vl regions of a humanized antibody were synthesized from overlapping oligonucleotides. The oligonucleotides that encoded the rabbit CDR regions were synthesized wherein the codons encoding the N-terminal amino acids in each of the six CDRs, i. e., leucine 24 in LCDR1, glycine 50 in LCDR2, leucine 89 in LCDR3, histidine 31 in HCDR1, tyrosine 50 in HCDR2, and aspartic acid 95 in HCDR3 were randomized by replacing them with NNK codons = A, C, G, or T; K = G or T). The size of the corresponding synthetic antibody library allowed the selection of any of the 20 natural amino acids in all six randomized positions. A more limited randomization of the amino-terminal residues in each CDR was also provided by using a VNS randomization scheme where V= C, A, or G and S=G or C in a separate library. Two different human frameworks were chosen to provide the scaffolding for the CDRs and these regions were synthesized to provide

oligonucleotides that would overlap and allow for gene assembly with oligonucleotides synthesized to encode the CDR regions. These were the frameworks derived from the anti-HIV antibody bl2 (Barbas in, C. F., Collet, T. A., Roben, P., Binley, J., Amberg, W., Hoekstra, D., Cababa, D., Jones, T. M., Williamson, R. A., Pilkington, G. R., Haigwoods, N. L., Satterthwait, A. C., Sanz, I., and Burton, D. R. (1993) Molecular profile of an antibody response to HIV-1 as probed by combinatorial libraries. J. Mol.

Biol. 230 : 812-823.) (both light and heavy chain framework regions), as an example of the use of a nonhomologous framework derived from an affinity matured human antibody and the germline derived frameworks derived in the heavy chain variable region from human VH gene DP-77/3-21 and JH1 and in the light chain from human V kappa gene DPK-4/A20 and Jk4.

Assembly of the encoding genes by PCR and addition of the light and heavy chain constant regions derived from a human Fab provided the genes encoding the Fab antibody libraries which were then displayed on phage and selected by panning against human A33 antigen. DNA sequencing of selected antibody sequences that bound A33 antigen revealed the selection of amino acids different from the original rabbit amino acids at the N-terminal position of the CDRs. Both template frameworks provided novel anti-A33 binding antibodies.

The phage library displaying Fab with a diversified human framework sequence and rabbit CDR sequences is selected by panning against immobilized antigen using highly stringent conditions. Clones demonstrating strong reactivity in ELISA on immobilized antigen are selected.

The Examples that follow illustrate preferred embodiments of the present invention and are not limiting of the specification and claims in any way.

Example 1: Method of Humanizing Non-human Antibodies The following steps summarize one embodiment for humanizing a non-human antibody.

-Immunize animal (e. g., mouse, rabbit, chicken) with (human) antigen.

-Generate a nonhuman antibody library from peripheral blood cells and/or spleen and/or bone marrow of the immune animal.

-Select or screen the nonhuman antibody library against (human) antigen.

-Evaluate the identified nonhuman antibodies in terms of specificity and affinity toward the (human) antigen.

-Proceed with nonhuman antibodies that have the desired properties.

-Align the nucleotide sequence of the variable domains of the nonhuman antibodies with the corresponding variable Immunoglobulin domains of human antibodies in a database.

-Identify human sequences with homology to the nonhuman sequences. These human sequences can be but do not have to be most homologous to the nonhuman sequences.

-Graft the nonhuman complementarity-determining regions (CDRs) from the nonhuman antibodies into the human framework regions of a homologue human antibody.

-Identify a set of key residues in the human framework and non-human CDR residues that are likely to be involved in antigen binding, either indirectly, by supporting the confirmation of the CDRs, or directly, by contacting the antigen (Foote and Winter (1992) J. Mol. Biol. 224,487-499.). Use the current knowledge about key framework residues obtained from structural studies including crystallization, site-directed mutagenesis, humanziation etc. The diversification can be but does not have to be limited to one framework region in either variable Immunoglobulin domain of light and heavy chain. At least one residue has to be diversified. Typically, the diversification at a particular position in the human framework is designed to allow the selection of either the residue from the original nonhuman framework or the residue from the human framework. However, the diversification at a particular position does not have to be limited to two residues. The upper limit is a total randomization allowing any of the 20 natural amino acids to be selected at a particular position.

-Select or screen the resulting humanized antibody library with the diversified human framework against (human) antigen.

-Evaluate the identified humanized antibodies in terms of specificity and affinity toward the (human) antigen.

Example 2: Generation of antibodies to human A33 Antigen.

In order to generate monoclonal antibodies to human A33 antigen, New Zealand white rabbits were immunized, over a 4-5 month period, with human colon carcinoma cell line LIM 1215, which is known to express large amounts of A33 antigen. It should be noted that LIM 1215 was chosen because it expresses A33 antigen. Any other cell line which expresses A33 antigen may be substituted for LIM 1215. Subject animals received three subcutaneous injections of 106 LIM 1215 cells followed by three subcutaneous injections of 1 ig of extracellular domain of human A33 that had been purified from LIM 1215 cells. The A33 was administered in the form of a 1 ml emulsion of RIBI adjuvant in phosphate buffered saline.

This approach was taken in order to target hormonal immune responses to native epitopes of protein accessible on cell surfaces, which is key to developing therapeutically useful antibodies.

Antisera from the subject animals were tested following the three injections of LIM 1215 cells, and then the three injections of antigen. Testing was carried out by combining the antisera with recombinant human A33 and alkaline phosphatase conjugated, goat anti-rabbit Fc polyclonal antibodies.

The result indicated that there was a weak immune response following the injections with cells, and a strong immune response was observed following the three injections with antigen.

Example 3: Amplification of Variable Region Sequences and Generation of Chimeric Antibodies.

Five days after the last of the six immunizations referred to supra, spleen and bone marrow cells from one leg were harvested from each animal. Total RNA was extracted from the cells, using standard methodologies. First strand cDNA was then synthesized from the RNA, using standard techniques. The cDNA was then amplified via PCR (35 cycles). Various primers were used, i. e.: VK 5'sense primers : 1. 5'-gggcccaggcggccgagctcgtgmtgacccagactcca-3' (SEQ ID NO : 1) 2.5'-gggcccaggcggccgagctcgatmtgacccagactcca-3' (SEQ ID NO : 2)

3.5'-gggcccaggcggccgagctcgtgatgacccagactgaa-3' (SEQ ID NO : 3) VK 3'antisense primers : 1.5'-acagatggtgcagccacagttaggatctccagctcggtccc-3' (SEQ ID NO : 4) 2.5'-gacagatggtgcagccacagttttgatttccacattggtgcc-3' (SEQ ID NO : 5) 3.5'-gacagatggtgcagccacagttttgacsaccacctcggtccc-3' (SEQ ID NO : 6) V ; 5'sense primer : 5'-gggcccaggcggccgagctcgtgctgactcagtcgccctc-3' (SEQ ID NO : 7) VA 3'antisense primer : 5'-cgagggggcagccttgggctggcctgtgacggtcagctgggtccc-3' (SEQ ID NO : 8) To carry out the PCR, all nine possible combinations for amplification of VK were used, as well as the single combination provided for V. In addition, the four possible combinations provided by SEQ ID NOS: 9-13, i. e., VH 5'sense primers : 1.5'-gctgcccaaccagccatggcccagtcggtggaggagtccrgg-3' (SEQ ID NO: 9) 2.5'-gctgcccaaccagccatggcccagtcggtgaaggagtccgag-3' (SEQ ID NO : 10) 3.5'-gctgcccaaccagccatggcccagtcgytggaggagtccggg-3' (SEQ ID NO : 11) 4.5'-gctgcccaaccagccatggcccagsagcagctgrtggagtccgg-3' (SEQ ID NO : 12) VH 3'antisense primer : 5'-cgatgggcccttggtggaggctgargagayggtgaccagggtgcc-3' (SEQ ID NO : 13) were used to amplify VH. It should be noted that the antisense primers (SEQ ID NOS: 4-6,8 and 13) represent hybrids of rabbit and human sequences, and were designed to permit fusion of rabbit, variable domains to human constant domains (i. e., fusion of rabbit VA or VH to human CK and CH1). These human constant regions had been amplified from an expression vector containing a human Fab directed to tetanus toxoid.

See, e. g., Rader, et al., Curr. Opin. Biotechnol 8 (4): 503-508 (1997).

The procedure permitted assembly and fusion of chimeric rabbit/human light chain and Fd fragment coding sequences and two sequential overlap extension PCR steps. In the first step, the rabbi VA and human CK fragments were fused using: gaggaggagg aggaggaggc ggggcccagg cggccgagct c (SEQ ID NO: 14), and gccatggctg gttgggcagc (SEQ ID NO: 15), and

rabbit VH and human CH1 were fused using: gctgcccaac cagccatggc c (SEQ ID NO : 16) and gaggaggagg aggaggagag aagcgtagtc cggaacgtc (SEQ ID NO : 17). Then, assembled chimeric light chain and Fd fragment coding sequences were fused using SEQ ID NO: 14 and SEQ ID NO: 17. Only light chain and Fd fragment coding sequences from the same animal were combined. Final constructs were cloned into a phagemid vector, in accordance with Rader, et al., supra, to yield 2x 107 independent transformants. This methodology has several advantages over approaches using a uniform Fab format with original, constant domains from a given species. First, notwithstanding the fact that antigen binding is confined to variable domains, and should not be expected to be influenced by constant domain swapping, the human constant domains provide established and standardized modes for detecting and purification, as compared to Fabs derived from multiple species. In addition, Ulrich et al., Proc. Natl. Acad Sci USA 92 (25): 11907-11 (1995), have shown that this approach improves E. Coli expression levels of Fab. Also, Fab molecules with human constant domains are partially humanized, and can be readily channeled into strategies for complete humanization, as reported by, e. g., Rader, et al., Proc. Natl. Acad Sci USA 95 (15): 8910-8915 (1998), incorporated by reference.

Example 4: Screening the Chimera Antibody.

The phage library prepared in example 2, supra, was then panned against recombinant human A33 antigen using 200ng of protein in 25p1 of TBS for coating on 1 well of a 96 well plate, 0.05% (v/v) Tween 20 in TBS for washing, and 10mg/ml of trypsin in TBS for elution. Trypsinization was carried out for 30 minutes at 37°C. The number of washing steps increased from 5 (first round) to 10 (second round), to 15 in the third and fourth rounds.

Output phage pool of each round was monitored, via phage ELISA, using horseradish peroxidase labelled sheep anti-M13 phage polyclonal antibodies. Increased signal above background from round to round was observed, and output numbers increased strongly after the third and fourth rounds, indicating successful selection.

Forty clones from final output were grown and induced with lmM IPTG.

Supernatants from the clones were tested for binding to immobilized, recombinant human A33 via ELISA, using alkaline phosphatase-conjugated goat, anti-human F (ab') 2 polyclonal antibodies. All clones gave a strong signal, above background, and were subjected to DNA fingerprinting using standard methodologies. In brief, flanking primers : AAGACAGCTA TCGCGAATTG CAC (SEQ ID NO: 18) and GCCCCCTTAT TAGCCTTTGC CATC (SEQ ID NO: 19) were used, and digested with 4 base pair cutter BstXI. Three different but highly similar fingerprints were obtained. One was found in 13 clones, the second in 26 clones, and the third, in one clone. Figure 1 presents these. Also see SEQ ID NOS: 20- 22.

Analysis indicated that the sequences corresponding to variable domains were rabbit, and that the three clones were highly related. Clones 1 and 2 (SEQ ID NOS: 20 and 21) had identical VK coding sequences, and 90% identity in the VH sequence. SEQ ID NO: 22 had a VK coding sequence 90% identical to SEQ ID NOS: 20 and 21, and its VH sequence was identical to that of SEQ ID NO : 22. The hypervariable VDJ and VJ joint regions HCDR3 and LCDR3 were highly similar, suggesting that all the selected sequences originated from a single B cell clone that had undergone diversification by somatic mutation.

Example 5: Characterization of the Expressed Fabs.

Soluble Fabs from rabbit VH1, VL1 and rabbit VH2 and VL2 were produced from E. coli, in accordance with Rader, et al., supra. Fab molecules were purified from concentrated supernatants and from sonicated lysates of overnight cultures that had been induced with 1mM IPTG, followed by affinity chromatography, using PBS as equilibration and washing buffer, and U. 5M acetic acid for elution. The eluted fractions were neutralized immediately using 0.5 volumes 1M Tris-HCl, pH 9.0, followed by pooling. The materials were concentrated, and combined with PBS.

Quality was analyzed via SDS-PAGE and Coomassie Blue staining, using standard

methods (Figure 5). They were then subjected to flow cytometry, using FACS scan.

For each determination, 1x104 cells were analyzed. Indirect immunofluorescence staining was carried out using 2mg/ml of Fab, in 1% w/v BSA, 25mM Hepes, 0.05% (w/v) sodium azide in PBS. Dilutions (1: 100) of FITC conjugated donkey anti-human F (ab') z polyclonal antibodies were used for detection. Incubation was carried out for 1 hour with primary antibodies and 30 minutes with the secondary antibodies, at room temperature. The results are plotted in Figure 3.

The flow cytometry revealed that both Fabs specifically bound to cells that express natural A33 antigen. The binding strength was determined by surface plasmon resonance in accordance with Rader, et al., Proc. Natl. Acad. Sci USA 95 (15): 8910- 8915 (1998), incorporated by reference. Briefly, the determination of association (kon) and dissociation (koff) rate constants for binding of rabbit and humanized Fab to recombinant human A33 antigen was performed on a Biacore instrument (Biacore AB, Uppsala, Sweden). A CM5 sensor chip (Biacore AB) was activated for immobilization with Nhydroxysuccinimide and N-ethyl-N'- (3-dimethylaminopropyl) carbodiimide according to standard methods. Recombinant human A33 antigen was coupled at a low density to the surface by injection of 30 111 to 40 p1 of a 1 ng/lll sample in 10 mM sodium acetate (pH 3.5). Approximately 500 resonance units were immobilized.

Subsequently, the sensor chip was deactivated with 1 M ethanolamine hydrochloride (pH 8.5). Binding of Fab to immobilized A33 antigen was studied by injection of Fab at 5 different concentrations ranging from 75 nM to 200 nM. PBS was used as the running buffer. The sensor chip was regenerated with 20 mM HCl and remained active for at least 50 measurements. The kon and kff values were calculated using Biacore AB evaluation software. The equilibrium dissociation constant Kd was calculated from kofJkon. Data obtained from different sensor chips revealed a high consistency and were further validated according to procedure as described in Rader et al., (Rader, C., Cheresh, D. A., and Barbas, C. F., III (1998) Proc. Natl. Acad. Sci. U. S. A. 95,8910- 8915, incorporated herein by reference).

The binding of the Fab was very strong, i. e., with affinity in the 1nM range (Figure 4). Kd values for SEQ ID NOS: 20 and 21 were 390 pM and 1. 6nM, respectively (Figure 6, Table I). While SEQ ID NO: 20 showed higher association and slower dissociation rates, SEQ ID NO: 21 gave consistently higher yields. This, taken with the fact that the majority of clones contained SEQ ID NO: 21, suggests that the higher expression level competes well with the stronger affinity of SEQ ID NO: 20.

Table I

Clone kon/104 (M-1 s-1) koff/10-4 Kd rabbit 130. 7+/-0. 91. 2+/-0. 10. 39 rabbit 2 17. 4+/-1. 0 2. 8+/-0. 2 1. 6 humanized A 10.5+/-0. 7 5. 9+/-O. l 5. 6 humanized B 3s. 2 +/-1. S 6. 1 +/-o. l 1. 7 humanized C 10. 9 +/-0. 3 19. 7 +/-0. 3 18. 1 humanized D 6. 5+/-0. 3 19. 0+/-0. 6 29. 2 humanized E 6. 5 +/-0. 2 5.0+/-0. 17.7 humanized F 19. 2 +/-1. 2 6. 8 +/-0. 2 3. 5 Association (ko") and dissociation (kob) rate constant were determined using surface plasmon resonance.

Human antigen A33 was immobilized on the sensor chip. The dissociation constant (Kd) was calculated from korY'bm Example 6: Humanization of Selected Rabbit Variable Domains.

These experiments describe the humanization of the selected, rabbit variable domains described supra. First the VBASE Directory of Human V Gene sequences (http ://www. mrc-cpe. cam. ac. uk/imt-doc 1, incorporated by reference) was screened, using amino acid sequence alignment, to identify human germ-line VA and VK sequences having the highest degree of homology with the rabbit sequences described herein. To elaborate, the rabbit sequences were first aligned with human V and J genes.

Human V gene DP-77 (3-21), from the VH3 family, and human J gene JH1 showed highest homology. The VK sequence (rabbit) used gave a best match with human V gene DPK-4 (A20) from VK1 family, and human J gene JK4. These human sequences not only gave the best alignment with the rabbit sequences, but are found, frequently, in the human antibody repertoire. See deWildt, et al., J. Mol. Biol 285 (3): 895-901 (1999).

Further, they are highly related to human V genes DP-47 (3-23), and DPK-9 (02), the

frameworks of which have both been used for mouse antibody hybridizations, and both of which give high yields when expressed in E. coli. See, e. g., Presta, et al., Canc. Res.

57 (20): 4593-9 (1997). Indeed, pairs of VH3 family heavy chains and VK1 family light chains are the most frequent combination found in native human antibodies. This suggests that the combination is immuno silent.

The CDR sequence of SEQ ID NO: 2 was used because of high expression. The six variable domains described by Kabat, et al., supra, were grafted into human framework sequences. There was a potentially immunogenic tryptophan at position 62, in rabbit"HCDR2" (Kabat et al., supra), was converted to serine.

"Fine tuning"of frameworks was accomplished by diversifying 6 positions in human VH framework, and 4 in human VK framework (Table II). The residues chosen were selected from key framework residues known to be involved in antigen binding.

Analysis of these human sequences indicated that they are diversified at positions that are potentially involved in antigen binding. These sequences were used as framework for grafting of the six rabbit CDRs described by Kabat, et al., Sequences of Proteins of Immunological Interest) (5"edition, US Dept. of Health and Human Services, Public Health Services, National Institutes of Health, 1991), incorporated by reference.

Table II Key framework residues targeted for diversification.

Position Human Rabbit Diversification VL 43 A P A V 46 L F L F 71 F Y F Y 80 P A P A VH 27 F I F I 28 T D T D 71 R L R L 75 K Q K Q 78 L V L V 91 Y F Y F Linked positions (VH 27-28 and VH 71-75, respectively) indicate a coupled diversification that limits the selection to either all-human or all rabbit sequence.

Overlapping oligonucleotides were designed, synthesized, and then assembled to create synthetic VA and VH coding sequences, using PCR. The procedure described, supra, for the generation of rabbit antibody library was followed, and when the final constructs were completed, they were Sfi I cloned into a vector carrying a chloramphenicol resistence gene, to avoid contamination with phage from the rabbit

antibody. The resulting library consisted of 1X107 independent transformants with a theoretical complexity of 2x10'.

The following oligonucleotides were used for humanization, L denotes primers for the VL assembly, H denotes primers for the VH assembly : L1, 5'-gagctccaga tgacccagtc tccatcctcc ctgtctgcat ctgtaggaga cagagtcacc atcacttgcc tggccagtga gttccttttt aatggtgtat cc-3 ; (SEQ ID NO : 23) L2, 5'-agatgggacc ccagattcta aattggatgc accatagatc aggarcttag garctttccc tggtttctgc tgataccagg atacaccatt aaaaaggaac tc-3 ; (SEQ ID N0 : 24) L3, 5'-aatttagaat ctggggtccc atctcggttc agtggcagtg gatctgggac agattwcact ctcaccatca gcagcctgca gsctgaagat gttgcaact -3' ; (SEQ ID NO : 25) L4, 5'-tttgatctcc accttggtcc ctccgccgaa agtcaaacca ctactaccac tataaccgcc tagacagtaa taagttgcaa catcttcags ctgcag- 3' (SEQ ID N0 : 26) L flank sense, 5'-gaggaggagg aggagggccc aggcggccga gctccagatg acccagtctc

ca - 3'; (SEQ ID NO : 27) L antisense flank, 5'-gacagatggt gcagccacag ttcgtttgat ctccaccttg gtccctcc -3'; (SEQ ED NO : 28) H1, 5'-gaggtgcagc tggtggagtc tgggggaggc ctggtcaagc ctggggggtc cctgagactc tcctgtgcag cctctgga - 3'; (SEQ ID NO: 29) H2A, 5'-ccagctaatg ccatagtgac tgaaggtgaa tccagaggct gcacaggaga gtct - 3'; (SEQ ID NO : 30) H2B, 5'-ccagctaatg ccatagtgac tgaagtcgat tccagaggct gcacaggaga gtct - 3'; (SEQ ID NO:31) H3, 5'-ttcagtcact atggcattag ctgggtccgc caggctccag ggaaggggct ggagtgggtc gcctacattt atcctaatta tgggagtgta gactacgcga gc-3' ; (SEQ ID N0 : 32) H4A, 5'-gttcatttgc agatacastg agttcttggc gttgtctctg gagatggtga atcggccatt cacgctgctc gcgtagtcta cactcccata-3' ; (SEQ ID NO : 33)

H4B, 5'-gttcatttgc agatacastg agttctgggc gttgtcgagg gagatggtga atcggccatt cacgctgctc gcgtagtcta cactcccata-3' ; (SEQ m N0 : 34) H5, 5'-aactcastgt atctgcaaat gaacagcctg agagccgagg acacggccgt atattwctgt gcgagagatc ggggttatta ttctggtagt-3' ; (SEQ ID NO : 35) H6, 5'-tgaggagacg gtgaccaggg tgccctggcc ccagagatcc aaccgagtcc ccctactacc agaataataa ccccgatc-3' (SEQI1DNO : 36) H flank sense, 5'-gctgcccaac cagccatggc cgaggtgcag ctggtggagt ctggggga-3' ; (SEQ ID N0 : 37) H flank antisense, 5'-gaccgatggg cccttggtgg aggctgagga gacggtgacc agggtgcc-3'. (SEQ ID N0 : 38) The transformants were panned as described supra, but the amount of antigen employed was decreased over the course of panning. In the first two rounds, 100ng were used, followed by two rounds at 50ng, and two rounds at 25ng. Ten washing steps were carried out for each round, using 0.5% (v/v) Tween 20 in TBS. Rounds 3 and 4, and rounds 5 and 6, were linked without phage amplification. To do this, phages from rounds 3 and 5 were eluted, using 50gl of 100mM HCl-glycine (pH 2.2), incubated for

10 minutes at room temperature, collected, neutralized with 3p1 of 2M Tris base, and 50 of 1% (w/v) BSA in TBS. The phages were than directly subjected to another round of panning. Phages from rounds 1,2,4 and 6 were eluted by trypsinization, as described supra.

Seventy clones resulted from final output. All were found to be positive via ELISA. Twenty-four of the seventy clones were further analyzed via DNA sequencing.

Sequences for the heavy and light chain of 6 of these clones (total of 12 sequences) are presented as human VLA, VLB, VLC, VLD, VLE, VLF, VHA, VHB, VHC, VHD, VHE, VHF in Figure 1. A consensus sequence was found for the diversified framework of VH, with positions 27 and 28 in framework 1, and positions 71 and 75 in framework 3 being found to contain original rabbit residues isoleucine, aspartic acid, leucine, and glutamine, respectively, in 16 of 24 clones. Three clones contained human residues phenylalanine and threonine at positions 27 and 28, and none contained human residues at positions 71 and 75. Two of the diversified positions contained mutations. Both appeared to be due to a single point mutation, probably generated via misincorporation during oligonucleotide synthesis, or assembly. Three clones had glycine at position 28, and phenylalanine was found in two clones at position 71. These 5 clones, notably, demonstrated the strongest reactivity in ELISA.

The two remaining diversified positions in the framework, i. e., positions 78 and 91, did not give significant consensus sequence, but random selection of human/rabbit residues. This was also the case for 3 of 4 diversified positions in the VA framework (positions 43 and 46 in framework 2, positron 71 in framework 3). Proline, a human residue, was found at position 80 in framework 3, in 18 of 24 clones, including the 5 mutated clones showing strongest reactivity via ELISA.

The six clones (human A to F, wherein each comprise a VH and VL as shown in Figure 1) referred to supra were then produced as soluble Fab molecules via E. coli, and purified as described supra. Yields ranged from 0.5 to 2mg per 1 liter shake flash culture. When subjected to flow cytometry, all Fabs bound to cells expressing native A33 antigen. Those cells which did not express human A33 were not recognized.

There were slight differences in fluorescence intensity, which correlated to differences in affinity to immobilized recombinant human A33, measured by surface

plasmon resonance carried out as described supra. This suggests strongly that the antibodies, which were selected on immobilized, recombinant antigen, bind to a native epitope fully accessible on the cell surface, thereby constituting a relevant therapeutic target.

Example 7: Characterization of Novel A33 Antibodies.

Preparation of recombinant A33 antigen : A 1.6 kb XhoI/PstI cDNA fragment, containing the full length coding sequence of A33, was subcloned into pBlueBac4 transfer vector. To generate the transfer vector harboring only extracellular domain of A33 (ECD-A33) the 340bp BglII/PstI fragment was removed from the pBlueBac4/A33 vector and the resulting plasmid was religated with the use of two overlapping oligonucleotides (gatctccctccatgaaccat catcatcatcatcattgactgca and gtcaatgatgatgatgatgatggttcatggaggga (SEQ ID NO : 127)). When annealed, these oligonucleotides would create BglII and PstI sites at the 5'and 3'end respectively and sequences encoding SPSMHHHHHH (SEQ ID NO : 128) and stop codon between both restriction sites. Transfection of Sf9 cells with pBlueBac4/A33 and pBlueBac4/A33- ECD transfer vectors and isolation of recombinant viruses was performed according to the manufacturer's recommendations (Invitrogen). For large-scale expression, Sf9 cells were infected with the recombinant viruses at a multiplicity of infection (MOI) of 10.

After three days of infection cells were harvested by centrifugation and used immediately for the purification of recombinant proteins. Expressed protein was purified by immunoaffinity chromatography using mouse mAb A33 immobilized to protein A conjugated Sepharose 4B beads with dimethylpimelimidate as previously described (Moritz, R. L. et al., J. Chromatogr. A, 798: 91-101).

Western Blots : Triton X-100 (0. 3% in PBS pH 7.5) lysates of colon cancer cells were resolved by SDS-PAGE on 10-20% polyacrylamide Tris-glycine pre-cast gels under reducing (5% P-ME) and non-reducing conditions. Proteins were blotted to PVDF and incubated with 0.5 ug/ml murine A33 mAb or humanized Fab B overnight at 4°C. Specific binding was detected by alkaline phosphatase conjugated species specific

secondary Abs and visualized using chemiluminescent detection. Blocking and washing steps were carried out as per manufacturer's instructions.

Hemadsorption assay : The protein A, rabbit anti-human F (ab') 2 mixed hemadsorption assay which detects surface bound Fab by adherence of protein A coated human RBC (blood group O) to target cells was performed as previously described (Pfreundschuh, M. et al., Proc. Natl. Acad. Sci. (Wash.), 75,5122-5126 (1978)).

Results: Fabs A, B, C, E, and F were analyzed for reactivity with A33 antigen extracted from colon cancer cell lines by Western blot assays (Figure 2). All new Fabs reacted with a band of about 43 kD protein under non reducing conditions. No Western blot reactivity was observed using reducing conditions (Figure 2). These Western blot reactivities of the Fabs prepared from a rabbit phage display library are identical with those obtained with mouse mAb A33 suggesting recognition of a conformational epitope on the A33 antigen as previously described for mAb A33 (Catimel, B. et al., J.

Biol. Chem. 271: 25664-25670).

Mixed hemadsorption assays : Fabs A, B, C, E, and F were analyzed for reactivity with A33 antigen expressed on the cell surface of human cancer cell lines using a mixed hemadsorption assay. All five Fabs bound to A33+ but not to A33- cancer cells (listed below). Fabs A and B showed the strongest reactivity with cell surface expressed A33 antigen.

Mixed hemadsorption titer (ng Ig/ml) Table m Ig LIM1215 SW1222 NCI-H508 HT29 SW620 Fab A 10# 1 5 FabB10510 Fab C 80 nd nd-- Fab E 40 20 20-- Fab F 10 20 20-- HmAbA33555- "Lowest concentration of Fab or human mAb A33 giving 50% rosetting. nd = not determined.

The difference between the humanized clones were found to correlate with their differences in affinity to immobilized recombinant human A33 antigen. The humanized Fab were further analyzed for reactivity with human A33 antigen extracted

from colon cancer cell lines by Western blotting. As shown for humanized clone B (Figure 2), the humanized Fab strongly reacted with a band of about 43 kD under nonreducing conditions. No reactivity was observed using reducing condition, suggesting the recognition of a conformation epitope on human A33 antigen (Catimel, B. et al., (1996) J. Biol. Chem. 271,25664-25670). Taken together, these results demonstrate that the selected humanized antibodies bind to a native epitope on human A33 antigen that is fully accessible on the cell surface.

Example 8: Immunohistochemistrv In order to evaluate the selectivity of the humanized Fab in an independent system of higher complexity, their reactivity with tumor tissue sections was analyzed by immunohistochemistry. All immunochemical stainings were done on snap-frozen tissue samples, embedded in O. C. T. compound (Tissue Tek, Torrance, CA) 0.5 llm cuts (HM503 cryostat, Zeiss, Walldorf, Germany) were mounted on slides for immunohistochemistry (Superfrost Plus, Fisher Scientific, Pittsburgh, PA). Serial sections were used, so as to compare staining results of the different antibody preparations. After cutting, the slides were fixed in cold acetone for 10 min and then air dried. Reactivity of the humanized Fab was analyzed using the colon cancer cell line SW1222 xenografted into nude mice. A working concentration of Fab (1 pg/ml) was established by titering. The humanized Fab was detected by biotinylated goat-anti human F (ab) 2 polyclonal antibodies (1: 200; Vector, Burlingame, CA) and an avidin- biotin-complex system (ABC/Elite kit, Vector). Diaminobenzidine tetrahydrochloride (DAB, Biogenex, San Ramon, CA) was used as a chromogen. Reactivity of the humanized Fab was also evaluated in human colonic adenocarcinoma samples. In order to prevent immunoreactivity of endogenous human immunoglobulin, a special technique for the detection of humanized Fab was utilized. Prior to addition to tissue, the humanized Fab (1 pg/ml) was incubated with biotinylated goat-anti human F (ab) z polyclonal antibodies in a test tube. The optimal ratio of humanized Fab to secondary antibody was determined in separate titration assays. Incubation of humanized Fab and secondary antibody was done at room temperature for 1 hour and followed by an addition of human serum in order to block the activity of unbound secondary antibody.

Again, the optimal ratio of human serum to secondary antibody was determined in separate titration assays.

As can be seen in Figure 8, the humanized Fab reacted strongly with xenografts of the human colon cancer cell line SW1222 grown in nude mice (Figure 8A and 8B).

The humanized Fab revealed immunoreactivity similar to the mouse monoclonal antibody A33 with an intense staining of dysplastic glandular structures in tissue sections of human colon adenocarcinoma after blocking endogenous human immunoglobulins (Figure 8C and 8D). A comparison of corresponding tissue sections stained with the full blocking step and stained without blocking of the endogenous human immunoglobulins illustrates the amount of internal reactivity (Figure 8F) and its complete blocking (Figure 8E).

Example 9: Immunoglobulin Products Against A33 Antigen One embodiment of the invention is directed to an immunoglobulin product that binds with specificity to an A33 antigen. An immunoglobulin product is a polypeptide, protein or multimeric protein containing at least the immunologically active portion of an immunoglobulin heavy chain or an immunologically active portion of an immunoglobulin light chain and is thus capable of specifically combining with an antigen. Exemplary immunoglobulin products are an immunoglobulin heavy chain, immunoglobulin light chain, immunoglobulin molecules, bispecific antibodies, substantially intact immunoglobulin molecules, any portion of an immunoglobulin that contains the paratope, including those portions known in the art as Fab fragments, Fab' fragment, F (ab') 2 fragment and Fv fragment. The structures of immunoglobulin products are well known to those skilled in the art and described in Basic and Clinical Immunology, by Stites, et al., 4th ed., Lange Medical Publications, Los Altos, Calif.

Another embodiment of the invention is directed to an immunoglobulin product such as an immunoglobulin molecule that binds with specificity to an A33 antigen. An immunoglobulin molecule is a multimeric protein containing the immunologically active portions of an immunoglobulin heavy chain and immunoglobulin light chain covalently coupled together and capable of specifically combining with antigen. It

should be noted that the immunoglobulin molecule may be a bispecific antibody with affinity for A33 and a second non-A33 epitope.

Another embodiment of the invention is directed to a single-chain antigen-binding protein that binds with specificity to an A33 antigen. A single chain antigen binding protein is a polypeptide composed of an immunoglobulin light-chain variable region amino acid sequence (V) tethered to an immunoglobulin heavy-chain variable region amino acid sequence (VH) by a peptide that links either (1) the carboxyl terminus of the VL sequence to the amino terminus of the VH sequence or (2) the carboxyl terminus of the VH sequence to the amino terminus of the VL sequence. A single-chain antigen-binding protein-coding gene, a recombinant gene coding for a single-chain antigen-binding protein, which encodes a single-chain antigen-binding protein that bind with specificity to an A33 antigen is also contemplated by this invention. The structure of single chain antigen binding proteins has been described by, e. g., Bird et al., Science, 242: 423-426 (1988) and U. S. Pat. No. 4,704,692 by Ladner.

The immunoglobulins, or antibody molecules, are a large family of molecules that include several types of molecules, such as IgD, IgG, IgA, IgM and IgE. The antibody molecule typically includes two heavy (H) and two light (L) chains, each of which has a variable (V) and constant (C) region. Several different regions of an immunoglobulin molecule contain conserved sequences useful for isolating the immunoglobulin genes using the polymerase chain reaction. Extensive amino acid and nucleic acid sequence data displaying exemplary conserved sequences is compiled for immunoglobulin molecules by Kabat et al., in Sequences of Proteins of Immunological Interest, National Institute of Health, Bethesda, Md. (1987), incorporated by reference.

The V region of the H or L chain typically comprises four framework (FR) regions (FIG. 1) each containing relatively lower degrees of variability that includes lengths of conserved sequences.

One particularly useful immunoglobulin product is an immunoglobulin heavy chain. An immunoglobulin heavy chain consists of an immunoglobulin heavy chain variable region and an immunoglobulin heavy chain constant region. The immunoglobulin heavy chain variable region is a polypeptide containing an antigen binding site (and antibody combining site). The immunoglobulin heavy chain variable

region is capable of specifically binding a particular epitope. Preferably, the VH will be from about 110 to about 125 amino acid residues long. The amino acid residue sequence will vary widely, depending on the particular epitope the VH is capable of binding.

One embodiment of the invention is directed to a method of reducing the effects of colon cancer in a subject. In the method, a pharmaceutically effective amount of an anti-cancer agent is conjugated to an immunoglobulin product that binds with specificity to A33 antigens. This anti-cancer agent-immunoglobulin product is conjugate is administered to a subject which has colon cancer to reduce the effects of the cancer. In particular, the immunoglobulin product comprises one or more CDRs having a sequence selected from the group consisting of LASEFLFNGVS (SEQ ID NO : 68), LASDFLFNGVS (SEQ ID NO : 69), GASNLES (SEQ ID NO : 70), GASDLET (SEQ ID NO : 71), LGGYSGSSGLT (SEQ ID NO : 72), LGGYSGSAGLT (SEQ ID NO : 73), HYGIS (SEQ ID NO : 74), NNGIS (SEQ ID NO : 75), YIYPNYGSVDYASSVNG (SEQ ID NO : 76), YIYPNYGSVDYASWVNG (SEQ ID NO : 77), YIYPDYGSTDYASWVNG (SEQ ID NO : 78), DRGYYSGSRGTRLDL (SEQ ID NO : 79), and DRGAYAGSRGTRLDL (SEQ ID NO : 80).

The anti-cancer agent may be a drug selected from the group consisting of calicheamicin, BCNU, streptozoicin, vincristine and 5-fluorouracil. In addition, the anti-cancer agent may be a peptide that specifically inhibits DNA activity of said colon cancer. Other anti-cancer agents that may be used include a radioactive isotope such as 1251, 1311, 99Tc, 90Y or 11'In.

The immunoglobulin product of this invention may also comprise an immunologically active portion of an immunoglobulin light chain which has, for

example, a VLCDR1 region with a sequence of LASEFLFNGVS (SEQ ID NO : 68) or LASDFLFNGVS (SEQ ID NO : 69); a VLCDR2 region with sequence GASNLES (SEQ ID NO : 70) or GASDLET (SEQ ID NO : 71); and a VLCDR3 region with a sequence consisting of LGGYSGSSGLT (SEQ ID NO : 72) or LGGYSGSAGLT (SEQ ID NO : 73). In a preferred embodiment, VLCDR1 has sequence LASEFLFNGVS (SEQ ID NO : 68), VLCDR2 has sequence GASNLES (SEQ ID NO : 70) and VLCDR3 has sequence LGGYSGSSGLT (SEQ ID NO : 72).

The immunoglobulin product may contain an immunoactive portion of an immunoglobulin light chain. This light chain may contain a sequence in the VLFR1 region which corresponds to one of the following sequences: ELQMTQSPSSLSASVGDRVTITC (SEQ ID NO : 81), EFDMTQTPPSLSASVGETVRIRC (SEQ ID NO : 82), ELVMTQTPPSLSASVGETVRIRC (SEQ ID NO : 83), or ELVLTQTPPSLSPSVGETVRIRC (SEQ ID NO : 84); or a VLFR2 region which corresponds to one of the following sequences: WYQQKPGKAPKLLIY (SEQ ID NO : 85), WYQQKPGKAPKLLIY (SEQ ID NO : 86) WYQQKPGKVPKFLIY (SEQ ID NO : 87), WYQQKPGKAPKFLIY (SEQ ID NO : 88), WYQQKPGKVPKLLIY (SEQ ID NO : 89), WYQQKPGKPPKFLIS (SEQ ID NO : 90), or WYQQKPEKPPTLLIS. (SEQ ID NO : 91); or a VLFR3 region which corresponds to one of the following sequences: GVPSRFSGSGSGTDFTLTISSLQPEDVATYYC (SEQ ID NO : 92), GVPSRFSGSGSGTDYTLTISSLQPEDVATYYC (SEQ ID NO : 93), GVPSRFSGSGSGTDFTLTISSLQPEDVATYYC (SEQ ID NO : 94), GVPPRFSGSGSGTDYTLTIGGVQAEDVATYYC (SEQ ID NO : 95), or

GVPPRFSGSGSGTDYTLTIGGVQAEDAATYYC (SEQ ID NO : 96); or a VLFR4 region which corresponds to one of the following sequences: FGGGTKVEIK (SEQ ID NO : 97) or FGAGTNVEIK. (SEQ ID NO : 98).

In another embodiment, the immunoglobulin product may also comprise an immunologically active portion of an immunoglobulin heavy chain which has, for example, a VHCDR1 having a sequence of HYGIS (SEQ ID NO : 74) or NNGIS (SEQ ID NO : 75); a VHCDR2 sequence of YIYPNYGSVDYASSVNG (SEQ ID NO : 76), YIYPNYGSVDYASWVNG (SEQ ID NO : 77), or YIYPDYGSTDYASWVNG (SEQ ID NO : 78); and a VHCDR3 sequence of DRGYYSGSRGTRLDL (SEQ ID NO : 79) or DRGAYAGSRGTRLDL (SEQ ID NO : 80). In a preferred embodiment, VHCDR1 has a sequence of HYGIS (SEQ ID NO : 74), VHCDR2 has a sequence of YIYPNYGSVDYASSVNG (SEQ ID NO : 76); and VHCDR3 has a sequence of DRGYYSGSRGTRLDL (SEQ ID NO : 79).

The immunoglobulin product may contain an immunoactive portion of an immunoglobulin heavy chain. This heavy chain may contain a sequence in the VHFR1 region which corresponds to one of the following sequences: EVQVMESGGGLVKPGGSLRLSCAASGFTFS (SEQ ID NO : 99), EVQVMESGGGLVKPGGSLRLSCAASGIDFS (SEQ ID NO : 100), EVQVMESGGGLVKPGGSLRLSCAASGIGFS (SEQ ID NO : 101), QQQVMESGGGLVTLGGSLTLTCKASGIDFS (SEQ ID NO : 102), QEQLMESGGGLVTLGGSLKLSCKASGIDFS (SEQ ID NO : 103), or QEQVMESGGGLVTLGGSLKLSCKASGIDFS (SEQ ID NO : 104); or a VHFR2 region which corresponds to one of the following sequences:

WVRQAPGKGLEWIL (SEQ ID NO : 105), WVRQAPGKGLEWIA (SEQ ID NO : 106), or WVRQAPGKGLEWVS. (SEQ ID NO : 107); or a VHFR3 region which corresponds to one of the following sequences: RFTISRDNAKNSLYLQMNSLRAEDTAVYYCAR (SEQ ID NO : 108), RFTISFDNAQNSLYLQMNSLRAEDTAVYYCAR (SEQ ID NO : 109), RFTISLDNAQNSLYLQMNSLRAEDTAVYFCAR (SEQ ID NO : 110), RFTISLDNAQNSLYLQMNSLRAEDTAVYYCAR (SEQ ID NO : 111), RFTISFDNAQNSVYLQMNSLRAEDTAVYYCAR (SEQ ID NO : 112), RFTISRDNAKNSLYLQMNSLRAEDTAVYYCAR (SEQ ID NO : 113), RFTISRDNAKNSLYLQMNSLRAEDTAVYFCAR (SEQ ID NO : 114), RFTISRDNAKNSVYLQMNSLRAEDTAVYYCAR (SEQ ID NO : 115), RFTISRDNAKNSVYLQMNSLRAEDTAVYFCAR (SEQ ID NO : 116), RFTISLDNAQNSLYLQMNSLRAEDTAVYYCAR (SEQ ID NO : 117), RFTISLDNAQNSLYLQMNSLRAEDTAVYFCAR (SEQ ID NO : 118), RFTISLDNAQNSVYLQMNSLRAEDTAVYYCAR (SEQ ID NO : 119), RFTISLDNAQNSVYLQMNSLRAEDTAVYFCAR (SEQ ID NO : 120),

RFTISSDNAQNSLYLQMNSLRAEDTAVYYCAR (SEQ ID NO : 121), RFTISSDNAQNSLYLQMNSLRAEDTAVYFCAR (SEQ ID NO : 122), RFTISSDNAQNSVYLQMNSLRAEDTAVYYCAR (SEQ ID NO : 123), or RFTISSDNAQNSVYLQMNSLRAEDTAVYFCAR (SEQ ED NO : 124); or a VHFR4 region which corresponds to one of the following sequences: WGQGTLVTISS (SEQ ID NO : 125), or WGQGTLVTVSS. (SEQ ID NO : 126) In another embodiment, the immunoglobulin product may comprise an immunologically active portion of an immunoglobulin light chain which has (A) a VLCDR1 region with a sequence of LASEFLFNGVS (SEQ ID NO : 68) or LASDFLFNGVS (SEQ ID NO : 69); a VLCDR2 region with a sequence of GASNLES (SEQ ID NO : 70) or GASDLET (SEQ ID NO : 71); and a VLCDR3 region with a sequence of LGGYSGSSGLT (SEQ ID NO : 72) or LGGYSGSAGLT (SEQ ID NO : 73); and (B) a VHCDR1 having a sequence of HYGIS (SEQ ID NO : 74) or NNGIS (SEQ ID NO : 75); a VHCDR2 sequence of YIYPNYGSVDYASSVNG (SEQ ID NO : 76), YIYPNYGSVDYASWVNG (SEQ ID NO : 77), or YIYPDYGSTDYASWVNG (SEQ ID NO : 78); and a VHCDR3 sequence of DRGYYSGSRGTRLDL (SEQ ID NO : 79) or DRGAYAGSRGTRLDL (SEQ BD NO : 80).

In a preferred embodiment, the immunoglobulin product of the invention binds to A33 antigen with an affinity which is stronger than 500 pM. More preferably, the immunoglobulin product of the invention binds to A33 antigen with an affinity which is stronger than 100 pM.

Another embodiment of the invention is directed to a substantially pure immunoglobulin product that binds with specificity to A33 antigen. The immunoglobulin product may comprise one or more sequences of amino acids having the sequence of

LASEFLFNGVS (SEQ ID NO : 68), LASDFLFNGVS (SEQ ID NO : 69), GASNLES (SEQ ID NO : 70), GASDLET (SEQ ID NO : 71), LGGYSGSSGLT (SEQ ID NO : 72), LGGYSGSAGLT (SEQ ID NO : 73), HYGIS (SEQ ID NO : 74), NNGIS (SEQ ID NO : 75), YIYPNYGSVDYASSVNG (SEQ ID NO : 76), YIYPNYGSVDYASWVNG (SEQ ID NO : 77), YIYPDYGSTDYASWVNG (SEQ ID NO : 78), DRGYYSGSRGTRLDL (SEQ ID NO : 79), or DRGAYAGSRGTRLDL (SEQ ID NO : 80).

In one embodiment, the substantially pure immunoglobulin product which binds the A33 antigen contains an immunologically active portion of an immunoglobulin light chain that in turn contains one or more light chain CDRs. For example, the immunoglobulin light chain, VLCDR1 may have sequence LASEFLFNGVS (SEQ ID NO : 68) or LASDFLFNGVS (SEQ ID NO : 69); VLCDR2 may have sequence GASNLES (SEQ ID NO : 70) or GASDLET (SEQ ID NO : 71); and VLCDR3 may have a sequence LGGYSGSSGLT (SEQ ID NO : 72) or LGGYSGSAGLT (SEQ ID NO: 73).

In a preferred embodiment, VLCDR1 is LASEFLFNGVS (SEQ ID NO : 68), VLCDR2 is GASNLES (SEQ ID NO : 70) and VLCDR3 is LGGYSGSSGLT (SEQ ID NO : 72).

In another embodiment, the substantially pure immunoglobulin product which binds the A33 antigen contains an immunologically active portion of an immunoglobulin heavy chain that in turn contains one or more CDRs of a heavy chain.

For example, in the immunoglobulin heavy chain, VHCDRl may have sequence HYGIS (SEQ ID NO : 74) or NNGIS (SEQ ID NO : 75); VHCDR2 may have sequence YIYPNYGSVDYASSVNG (SEQ ID NO : 76), YIYPNYGSVDYASWVNG (SEQ ID NO : 77), or YIYPDYGSTDYASWVNG (SEQ ID NO : 78); and VHCDR3 may have sequence DRGYYSGSRGTRLDL (SEQ ID NO: 79) or DRGAYAGSRGTRLDL

(SEQ ID NO : 80). In a preferred embodiment V, _, CDR1 is HYGIS (SEQ ID NO : 74), VHCDR2 is YIYPNYGSVDYASSVNG (SEQ ID NO : 76), VHCDR3 is DRGYYSGSRGTRLDL (SEQ ID NO : 79).

In another preferred embodiment, the immunoglobulin product comprises at least two polypeptide sequences selected from the following: rabbit VL1 and rabbit VH1; rabbit VL2 and rabbit VH2; rabbit VL3 and rabbit VH3; human VLA and human VHA, human VLB and human VHB, human VLC and human VHC, human VLD and human VHD, human VLE and human VHE, or human VLF and human VHF.

In another embodiment, the substantially pure immunoglobulin product may comprise an immunologically active portion of an immunoglobulin heavy chain and an immunologically active portion of an immunoglobulin light chain. For example, in the active portion immunoglobulin light chain, VLCDR1 may have sequence LASEFLFNGVS (SEQ ID NO : 68) or LASDFLFNGVS (SEQ ID NO : 69); VLCDR2 may have sequence GASNLES (SEQ ID NO : 70) or GASDLET (SEQ ID NO : 71); and VLCDR3 may have sequence LGGYSGSSGLT (SEQ ID NO : 72) or LGGYSGSAGLT (SEQ ID NO : 73). Further, in the active portion immunoglobulin heavy chain, VHCDRl may have sequence HYGIS (SEQ ID NO : 74) or NNGIS (SEQ ID NO : 75); VHCDR2 may have sequence YIYPNYGSVDYASSVNG (SEQ ID NO : 76), YIYPNYGSVDYASWVNG (SEQ ID NO : 77), or YIYPDYGSTDYASWVNG (SEQ ID NO : 78); and VHCDR3 may have sequence DRGYYSGSRGTRLDL (SEQ ID NO : 79) or DRGAYAGSRGTRLDL (SEQ ID NO : 80).

An immunoglobulin product of the invention may be an antibody, a Fv fragment, a Fab fragment, a Fab2 fragment, or a single chain antibody or a combination or multimer thereof. A multimer may be any linked combination of immunoglobulin products. For example, a multimer may contain more than 2, preferably more than 4, or even more than 6 antibodies, antibody fragments, or single chain antibodies linked together. Linkage may be by covalent bonds. Methods of linking antibodies and polypeptides, and proteins are known. Further, the linkage may be ionic. For example, one antibody linked to avidin may be linked by ionic bond to another antibody linked to biotin. The linked immunoglobulin products need not have the same affinity. For example, one linked immunoglobulin product may have a high affinity for A33 antigen,

another linked immunoglobulin product may have a low affinity for A33 antigen, and a third linked immunoglobulin product may have an affinity to a toxic or therapeutic chemical such as ricin.

The immunoglobulin product may be an antibody molecule such as a IgM, IgD, IgG, IgA or IgE or a fragment of these molecules. The immunoglobulin product may bind A33 antigen with an affinity that is stronger than 1 pM, preferably stronger than 10 pM, more preferably stronger than 100 pM, even more preferably stronger than 300 pM such as, for example, stronger than 500 pM.

The immunoglobulin product may be an anti A33 antigen immunoglobulin product that is derived from a rabbit. A rabbit derived anti A33 antigen immunoglobulin product may be made, for example, by injecting a rabbit with A33 antigen. Another method for producing rabbit anti A33 antigen immunoglobulin product is shown in the Example section.

Another embodiment of the invention is directed to a CDR peptide and proteins that contain one or more CDR peptides with a sequence of LASEFLFNGVS (SEQ ID N0 : 68), LASDFLFNGVS (SEQ ID NO : 69), GASNLES (SEQ ID NO : 70), GASDLET (SEQ ID NO : 71), LGGYSGSSGLT (SEQ ID NO : 72), LGGYSGSAGLT (SEQ ID NO : 73), HYGIS (SEQ ID NO : 74), NNGIS (SEQ ID NO : 75), YIYPNYGSVDYASSVNG (SEQ ID NO : 76), YIYPNYGSVDYASWVNG (SEQ ID NO : 77), YIYPDYGSTDYASWVNG (SEQ ID NO : 78), DRGYYSGSRGTRLDL (SEQ ID NO : 79), or DRGAYAGSRGTRLDL (SEQ ID NO : 80).

The immunoglobulin product of the invention may be a member of an immunoglobulin gene superfamily such as a immunoglobulin heavy chain, a T cell receptor, a major histocompatibility antigen, a D2-microglobulin associated antigen, a T lymphocyte antigens, a haemopoietic/endothelium antigens, a brain/lymphoid antigen, an immunoglobulin receptor, a neural molecule, a tumor antigen and the like.

In addition, the immunoglobulin product of the invention may contain immunologically active portion of an immunoglobulin light chain. The active portion may be VLFR1 with a sequence of ELQMTQSPSSLSASVGDRVTITC (SEQ ID NO : 81), EFDMTQTPPSLSASVGETVRIRC (SEQ ID NO : 82), ELVMTQTPPSLSASVGETVRIRC (SEQ ID NO : 83), or

ELVLTQTPPSLSPSVGETVRIRC (SEQ ID NO : 84). Also, the active portion may be VLFR2 having sequence WYQQKPGKAPKLLIY (SEQ ID NO : 85), WYQQKPGKAPKLLIY (SEQ ID NO : 86), WYQQKPGKVPKFLIY (SEQ ID NO : 87), WYQQKPGKAPKFLIY (SEQ ID NO : 88), WYQQKPGKVPKLLIY (SEQ ID NO : 89), WYQQKPGKPPKFLIS (SEQ ID NO : 90), or WYQQKPEKPPTLLIS (SEQ ID NO : 91).

The active portion may be VLFR3 with a sequence GVPSRFSGSGSGTDFTLTISSLQPEDVATYYC (SEQ ID NO : 92), GVPSRFSGSGSGTDYTLTISSLQPEDVATYYC (SEQ ID NO : 93), GVPSRFSGSGSGTDFTLTISSLQPEDVATYYC (SEQ ID NO : 94), GVPPRFSGSGSGTDYTLTIGGVQAEDVATYYC (SEQ ID NO : 95), or GVPPRFSGSGSGTDYTLTIGGVQAEDAATYYC (SEQ ID NO : 96). The active portion may also be VLFR4 with sequence FGGGTKVEIK (SEQ ID NO : 97) or FGAGTNVEIK (SEQ ID NO: 98).

The immunoglobulin product of the invention may contain immunologically active portion of an immunoglobulin heavy chain. The active portion may be VHFR1 with a sequence of EVQVMESGGGLVKPGGSLRLSCAASGFTFS (SEQ ID NO : 99), EVQVMESGGGLVKPGGSLRLSCAASGIDFS (SEQ ID NO : 100), EVQVMESGGGLVKPGGSLRLSCAASGIGFS (SEQ ID NO : 101), QQQVMESGGGLVTLGGSLTLTCKASGIDFS (SEQ ID NO : 102), QEQLMESGGGLVTLGGSLKLSCKASGIDFS (SEQ ID NO : 103), or QEQVMESGGGLVTLGGSLKLSCKASGIDFS (SEQ ID NO: 104). The active portion may also be VHFR2 with sequence WVRQAPGKGLEWIL (SEQ ID NO: 105), WVRQAPGKGLEWIA (SEQ ID NO : 106) or WVRQAPGKGLEWVS (SEQ ID NO : 107). The active portion may also be VHFR3 with sequence RFTISRDNAKNSLYLQMNSLRAEDTAVYYCAR (SEQ ID NO : 108), RFTISFDNAQNSLYLQMNSLRAEDTAVYYCAR (SEQ ID NO : 109), RFTISLDNAQNSLYLQMNSLRAEDTAVYFCAR (SEQ ID NO : 110), RFTISLDNAQNSLYLQMNSLRAEDTAVYYCAR (SEQ ID NO: 111), RFTISFDNAQNSVYLQMNSLRAEDTAVYYCAR (SEQ ID NO : 112), RFTISRDNAKNSLYLQMNSLRAEDTAVYYCAR (SEQ ID NO : 113), RFTISRDNAKNSLYLQMNSLRAEDTAVYFCAR (SEQ ID NO : 114),

RFTISRDNAKNSVYLQMNSLRAEDTAVYYCAR (SEQ ID NO : 115), RFTISRDNAKNSVYLQMNSLRAEDTAVYFCAR (SEQ ID NO : 116), RFTISLDNAQNSLYLQMNSLRAEDTAVYYCAR (SEQ ID NO : 117), RFTISLDNAQNSLYLQMNSLRAEDTAVYFCAR (SEQ ID NO : 118), RFTISLDNAQNSVYLQMNSLRAEDTAVYYCAR (SEQ ID NO : 119), RFTISLDNAQNSVYLQMNSLRAEDTAVYFCAR (SEQ ID NO : 120), RFTISSDNAQNSLYLQMNSLRAEDTAVYYCAR (SEQ ID NO : 121), RFTISSDNAQNSLYLQMNSLRAEDTAVYFCAR (SEQ ID NO : 122), RFTISSDNAQNSVYLQMNSLRAEDTAVYYCAR (SEQ ID NO : 123), or RFTISSDNAQNSVYLQMNSLRAEDTAVYFCAR (SEQ ID NO : 124). The active portion may also be VHFR4 with sequence WGQGTLVTISS (SEQ ID NO : 125) or WGQGTLVTVSS (SEQ ID NO : 126).

In an embodiment of the invention, the substantially pure immunoglobulin product may be a humanized immunoglobulin.

Another embodiment is directed to a purified nucleic acid molecule encoding the substantially pure immunoglobulin product of the invention. A nucleic acid molecule encoding an immunoglobulin product of the invention may be made using conventional techniques. For example, oligonucleotides may be synthesized and ligated together to form a functional open reading frame that encodes an immunoglobulin product of the invention. The nucleic acid molecule, once synthesized, may be cloned into a nucleic acid vector. A nucleic acid vector such as a plasmid, cosmid, phagemid, yeast plasmid, phage vectors, TI plasmid and the like are known in the art. The vector may be an expression vector. Expression vectors and expression systems are available commercially.

Another embodiment of the invention is directed to a cell comprising a nucleic acid of the invention. A cell may be made by transfection. Methods of transfection are known and kits for transfection of prokaryotic and eukaryotic cells may be purchased from commercial sources.

Another embodiment of the invention is directed to a method for detecting or diagnosing a disorder comprising the steps of contacting a tissue sample from a subject to the substantially pure immunoglobulin product of the invention under condition that

permits the formation of a complex between said immunoglobulin product and an A33 antigen, and determining the formation of said complex.

Another embodiment of the invention is directed to a method of treating a patient with a neoplastic disorder comprising administering an immunoglobulin product of invention or a nucleic acid of the invention to said patient. Methods for immunotherapy for cancer are known. See for example Old, L. J. Immunotherapy for Cancer, Scientific American, September 1996, US Patent 5,851,526 and 5,712,369; all incorporated herein by reference.

Another embodiment is directed to a therapeutic composition comprising an immunoglobulin product of the invention. The immunoglobulin products of the invention may be provided in the form of a composition comprising the immunoglobulin and a pharmaceutically acceptable carrier or diluent. The therapeutic composition may be used for the treatment of disorders in a mammal such as a human.

The invention also provides a method for treating a mammal comprising administering a therapeutically effective amount of the immunoglobulin products of the invention to the mammal, wherein the mammal has a disorder, such as cancer, requiring treatment with the antibody.

In its use as a therapeutic agent, the immunoglobulin product of the invention may be linked to an agent. Linkage may be by covalent bonds or by antibody-epitope bond. For example, an immunoglobulin product may be crosslinked to a second antibody wherein the second antibody may have an affinity for the agent. The agent may be a cytotoxic agent. The term"cytotoxic agent"as used herein refers to a substance that inhibits or prevents the function of cells and/or causes destruction of cells. The term is intended to include radioactive isotopes (e. g.,'1,''1, "Tc, Y,'"In), chemotherapeutic agents, and toxins such as enzymatically active toxins of bacterial, fungal, plant or animal origin, or fragments thereof. The agent may be a chemotherapeutic agent. A"chemotherapeutic agent"is a chemical compound useful in the treatment of cancer. Examples of chemotherapeutic agents include Adriamycin, Doxorubicin, 5-Fluorouracil, Cytosine arabinoside ("Ara-C"), Cyclophosphamide, Thiotepa, Busulfan, Cytoxin, Taxol, Methotrexate, Cisplatin, Melphalan, Vinblastine, Bleomycin, Etoposide, Ifosfamide, Mitomycin C, Mitoxantrone, Vincreistine,

Vinorelbine, Carboplatin, Teniposide, Daunomycin, Carminomycin, Aminopterin, Dactinomycin, Mitomycins, Esperamicins (see U. S. Pat. No. 4,675,187), Melphalan and other related nitrogen mustards. The agent may be a cytokine. The term "cytokine"is a generic term for proteins released by one cell population which act on another cell as intercellular mediators. Examples of such cytokines are lymphokines, monokines, and traditional polypeptide hormones. Included among the cytokines are growth hormones such as human growth hormone, N-methionyl human growth hormone, and bovine growth hormone; parathyroid hormone; thyroxine; insulin; proinsulin; relaxin; prorelaxin; glycoprotein hormones such as follicle stimulating hormone (FSH), thyroid stimulating hormone (TSH), and luteinizing hormone (LH); hepatic growth factor; fibroblast growth factor; prolactin; placental lactogen; tumor necrosis factor; mullerian-inhibiting substance; mouse gonadotropin-associated peptide; inhibin; activin; vascular endothelial growth factor; integrin; thrombopoietin (TPO); nerve growth factors such as NGF; platelet-growth factor; transforming growth factors (TGFs); insulin-like growth factor-I and-11 ; erythropoietin (EPO); osteoinductive factors; interferons such as interferon-cx,-P, and-y ; colony stimulating factors (CSFs); granulocyte-macrophage-CSF (GM-CSF); and granulocyte-CSF (G-CSF); interleukins (ILs) such as IL-1, IL-loc, IL-2, IL-3, IL-4, IL-5, IL-6, IL-7, IL-8, IL9, IL-11, IL-12 ; a tumor necrosis factor; and other polyp eptide factors including LIF and kit ligand (KL).

As used herein, the term cytokine includes proteins from natural sources or from recombinant cell culture and biologically active equivalents of the native sequence cytokines.

For diagnosis, the immunoglobulin product of the invention may be attached to a label. The word"label"when used herein refers to a detectable compound or composition which is conjugated directly or indirectly to the antibody. The label may be detectable by itself (e. g. radioisotope labels or fluorescent labels) or, in the case of an enzymatic label, may catalyze chemical alteration of a substrate compound or composition which is detectable.

The invention also contemplated the generation of mutants of the disclosed CDRs by mutating one or more amino acids in the sequence of one or more of the CDRs. It is known that a single amino acid substitution appropriately positioned in a

CDR can be sufficient to raise the affinity. Researchers have used site directed mutagenesis to increase affinity of some immunoglobulin products by about 10 folds.

This method of increasing or decreasing affinity of antibodies by mutating CDRs is common knowledge (see, e. g., Chapter 23, Paul, W. E., Fundamental Immunology, Raven Press, NY, NY 1993). Thus, the substitution, deletion, or addition of amino acids to the CDRs of the invention to increase or decrease binding affinity or specificity is also within the contemplation of this invention.