WO2008081331A2 - Enhancing the level of antibody expression by framework re-engineering - Google Patents

Abstract

Methods for enhancing antibody expression by re-engineering immunoglobulins by selecting a set of framework sequences that provides high expression and high specific binding of a target antigen. The methods include providing a set of amino acid sequences corresponding to the heavy or light chain variable region framework segments derived from two or more immunoglobulin heavy or light chains, replacing the heavy or light chain framework segments of a parent immunoglobulin with sequences from the set of amino acid sequences to obtain framework re-engineered immunoglobulins. In some embodiments, the framework sequences selected for use in the re-engineered immunoglobulin are not all derived from the same immunoglobulin. The methods provide a new approach to humanizing antibodies.

Description

ENHANCING THE LEVEL OF ANTIBODY EXPRESSION BY FRAMEWORK RE-ENGINEERING

PRIORITY This application claims the benefit of U.S. Provisional Application Serial No.

60/878,030, filed December 29, 2006, the contents of which are incorporated by reference herein in their entirety.

BACKGROUND The use of cell fusion for the production of monoclonal antibodies from immunized mice described by Kohler and Milstein in 1975 was an important step in the development of antibody technology. Monoclonal antibodies are highly specific and will bind and affect disease-specific targets, thereby sparing normal cells, and presumably causing fewer toxic side-effects than less specific chemical drugs. OKT3, an anti-CD3 murine monoclonal antibody, was the first therapeutic antibody approved by the FDA for uses in the prevention of organ graft rejection in 1986. However, the development of appropriate therapeutic products has been severely hampered due to a number of drawbacks inherent in monoclonal antibodies of murine origin, such as short serum half-life, inability to trigger human effector functions and the production of human anti-mouse antibodies (the HAMA response). Morrison et al. (1984) employed the techniques of molecular engineering to reduce the immunogenicity of murine antibodies by splicing the heavy and light chain variable regions of the murine parent antibody onto human constant regions, making the resultant "chimeric" antibody 67% more human, and thereby less immunogenic, than the murine parent antibody (Morrison et al., Proc. Natl. Acad. Sci. 21:6851-6855, 1984, incorporated herein by reference). Moreover, since chimeric antibodies retain their specificity and affinity against target antigens, they are able to elicit other human effector functions such as antibody-dependent cellular cytotoxicity (ADCC) and complement-dependent cytotoxicity (CMC) through the human constant Fc regions. Since then, several chimeric antibodies have been marketed and clinically used. Although chimeric antibodies are less immunogenic than murine monoclonal antibodies, human anti -chimeric antibody (HACA) responses have been observed in some patients (Saito et al., Rheumatology 44: 1462-1464, 2005; Miele et al., Journal of Pediatric Gastroenterology and Nutrition 38:502-508, 2004, each of which is incorporated herein by reference), limiting the number of repeated uses of these antibodies for patients. By "grafting" the murine complementarity-determining regions (CDRs) onto the corresponding framework regions of a human antibody, Jones et al. (1986) were able to "humanize" antibodies, reducing the murine portion of the resultant antibodies to less than 10% while maintaining the affinity to within three-fold that of their parent antibodies (Jones et al., Nature 321:522-525; 1986, incorporated herein by reference). However, direct grafting of murine CDRs onto human immunoglobulin frameworks does not always result in successful humanization due to substantial loss of binding affinity. Queen et al. (U.S. Pat. No. 5,585,089; U.S. Pat. No. 5,693,762; U.S. Pat. No. 5,693,761, each of which is incorporated herein by reference) proposed the re -introduction of important murine framework residues at critical positions which are identified by a set of criteria such as computer modeling, X-ray crystallography, etc. To date, all humanized antibodies used clinically have contained "back- mutated" murine framework residues believed to be important for maintaining the proper conformation and contacts of the antigen-binding sites.

The use of CDR-grafting and back-mutation in the conventional humanization of antibodies as proposed by Winter et al. (U.S. Pat. No. 5,225,539, incorporated herein by reference) and Queen et al. (U.S. Pat. No. 5,585,089; U.S. Pat. No. 5,693,762; U.S. Pat. No. 5,693,761) aims to make the re-engineered antibody look human in appearance. Yet it is not uncommon that as many as seven back-mutated murine residues are needed for restoring the original immunoreactivity of the humanized immunoglobulin (Figure 1). Although immunoglobulins humanized using this approach would appear human based on their overall amino acid sequence and structure, the immune system may see it differently (Figure 2). A humanized immunoglobulin containing back-mutated murine residues interspersed within the context of a human framework, when internalized into an antigen presenting cell (APC) such as Dendritic cells, will be chopped up into peptide pieces. Peptides of the appropriate sizes and sequences will be presented by the MHCII to helper T cells, and when the back-mutated murine residues just happen to be of the proper conformation of the peptides, the immune system will consider this to be a new T- or B-cell epitope and will eventually develop immune responses including human anti-humanized antibody (HAHA) responses against the humanized antibody. So, a visually humanized antibody may not necessarily be functionally humanized, due to the presence of back-mutated murine residues.

Therapeutic monoclonal antibodies represent one of the fastest growing areas of the pharmaceutical industry, with 48.1% growth between 2003 and 2004. There are currently 18 therapeutic antibodies approved by the FDA for clinical uses and over 150 in various stages of clinical trials. The market is expected to have a value of $30.3 billion in 2010. Oncology products will continue to dominate the market, but the sales of arthritis, immune and inflammatory disorder products are forecasted to grow strongly and account for 40% of the market by 2010. Among the marketed and developing therapeutic antibodies, a number of them are expected to be used in the form of "naked antibodies," meaning that they do not have a radioisotope or toxin attached to them. The elimination of the target of these antibodies depends entirely on the recruitment of the body's own effector mechanisms. For example, Rituximab, which had over $3 billion in global sales in 2005, is used in its naked form for the treatment of non-Hodgkin's lymphoma. Compared to immunotoxin or radiolabeled antibodies (in mg quantities per patient), naked antibodies require substantially higher doses (in gram quantities per patient) to achieve clinical efficacies. Therefore, other than generating antibodies that are target specific and low in immunogenicity, the bottleneck for the development of a commercially viable antibody lies in production. The ability to produce large quantities of antibodies of interest has become the key success factor for the monoclonal antibody pharmaceutical industry.

SUMMARY OF THE INVENTION

The invention provides methods for enhancing the level of antibody expression by humanization of antibodies. The level of antibody production is affected by many factors, including the amino acid sequence of the antibody, the type of expression vector used, the host cell, and the culture conditions. The present invention presents a new approach to enhancing the level of expression for re-engineered antibodies and simultaneously minimizing their potential immunogenicity through a process named "framework re- engineering."

In one aspect, the invention provides a method of improving the level of expression of an immunoglobulin. The method involves systematically replacing the compartmentalized variable region framework segments, known as FRl, FR2, FR3, and FR4, of the parent immunoglobulin with corresponding compartmentalized variable region framework segments from one or more immunoglobulins of different origin, i.e., different species, different immunoglobulin molecules, different immunoglobulin isotypes, or different immunoglobulin allotypes. In some embodiments of this method, the framework segments of the parent immunoglobulin are replaced by the corresponding framework segments from a combination of different immunoglobulins of the same species as the parent immunoglobulin. In other embodiments, the framework segments of the parent immunoglobulin are replaced by the corresponding framework segments from a combination of different immunoglobulins of a species different from that of the parent immunoglobulin. In yet other embodiments, the framework segments of the parent immunoglobulin are replaced by the corresponding framework segments from a combination of different immunoglobulins derived from more than one species. The method also includes determining the expression level of the framework re-engineered immunoglobulin and comparing it to that of the parent immunoglobulin. The method further includes selecting a framework re-engineered immunoglobulin having a high level of expression from among a set of framework re- engineered immunoglobulins. Still other embodiments of the method include determining the binding affinity of the framework re-engineered immunoglobulin and, optionally, selecting a framework re-engineered immunoglobulin having a high binding affinity from among a set of framework re-engineered immunoglobulins.

In another embodiment, the invention provides methods for enhancing the level of antibody expression by functional humanization of antibodies. From the perspective of the immune system, an immunoglobulin containing framework segments, namely, FRl, FR2, FR3 and FR4, from different immunoglobulins of human origin is considered to be functionally human even though the immunoglobulin sequence is not that of a naturally occurring human antibody (Figure 3). This "functional humanization" approach represents a change of concept from conventional humanization of antibodies.

In another aspect of the method of the invention, the particular variable region framework segment chosen for replacing the parent framework segment either: (i) exhibits at least 60% sequence identity to the parent framework segment; (ii) exhibits the highest degree of amino acid sequence homology available within a given set of donor sequences; (iii) exhibits the highest degree of amino acid sequence homology to the parent framework segment available within a given set of donor sequences at the three (or in some embodiments four) amino acids immediately adjacent to the flanking CDRs; or (iv) contains identical amino acids to the corresponding parent framework segment at positions known to be close to, or have interactions with, a CDR or an antigen binding site, as evaluated by computer modeling or crystal structure.

In another aspect of the method of the invention, the particular variable region framework segment chosen for replacing the parent framework segment contains one or more conservatively similar amino acids compared to the parent framework segment at the three (or in some embodiments four) amino acids immediately adjacent to the flanking CDRs, such as gly and ala; val, ile, and leu; asp and glu; asn and gin; ser and thr; lys and arg; and phe and tyr. In another aspect of the method of the invention, the particular variable region framework segment chosen for replacing the corresponding parent framework segment contains one or more back-substituted amino acids from the corresponding positions of the replaced parent framework segment. Each of these back-substituted amino acids is chosen such that either: (i) the substituted amino acid is adjacent to a CDR in the parent immunoglobulin sequence; (ii) the substituted amino acid contains an atom within a distance of 4 A (or in some embodiments 5 or 6 A) of a CDR in the framework re-engineered immunoglobulin; (iii) the donor framework segment exhibits at least 60% amino acid sequence homology to the corresponding framework segment of the parent immunoglobulin; (iv) the amino acid sequence homology at the three amino acids immediately adjacent to the flanking CDRs to the corresponding parent framework segment is increased; (v) the donor framework segment contains identical amino acids to the corresponding parent framework segment at positions known to be close to, or have interactions with, a CDR or an antigen binding site, as evaluated by computer modeling or crystal structure; or (vi) the substituted amino acid is typical for the parent species at its position in the parent framework segment and rare for the donor species at its position in the donor framework segment.

In certain embodiments, the framework re-engineered immunoglobulin of the invention specifically binds to an antigen with an affinity of between 10⁷ M^"1 and 10¹¹ M^"1. In other embodiments, the framework re-engineered immunoglobulin of the invention specifically binds to an antigen with an affinity of between 10⁸ M^"1 and 10¹⁰ M^"1. In some embodiments, the framework re-engineered immunoglobulin binds specifically to the intended antigen with affinity comparable to, or within 3 -fold of, that of the parent immunoglobulin.

BRIEF DESCRIPTION OF THE DRAWINGS

Other features and advantages of the present invention will be apparent from the following detailed description of the invention, taken in conjunction with the accompanying drawings of which:

Figure 1 shows schematically the use of conventional CDR grafting technologies to achieve humanization;

Figure 2 shows schematically how the immune system sees a conventionally humanized antibody from a functional perspective;

Figure 3 shows schematically how functional humanization of an antibody by framework re-engineering contains no back-mutated murine residues; Figure 4 displays the DNA sequences of the heavy chain (Figure 4(a)) and light chain (Figure 4(b)) variable region of 1F5 antibody;

Figure 5 displays the amino acid sequences of the heavy chain (Figure 5 (a)) and light chain (Figure 5(b)) variable region of 1F5 antibody; the CDR regions are boxed; Figure 6 displays the amino acid sequences of the heavy chain (Figure 6(a)) and light chain (Figure 6(b)) variable regions of framework re-engineered 1F5 antibody (flF5); the CDR regions are boxed;

Figure 7 displays the DNA sequences of the heavy chain (Figure 7 (a)) and light chain (Figure 7(b)) variable regions of framework re-engineered 1F5 antibody (flF5); Figure 8 shows SDS-PAGE of flF5 antibody under reduced (Figure 8(a)) and non- reduced (Figure 8(b)) conditions; Lane 1, molecular weight marker; Lane 2, human IgG standard; Lane 3, flF5 antibody;

Figure 9 shows the binding of flF5 antibody to Raji cells as determined by flow cytometry; Figure 10 shows the binding of flF5 antibody to Raji cells as determined by competition flow cytometry;

Figure 11 shows the alignment of VH domain FR3 sequences of the different versions of reengineered anti-liver cancer antibody;

Figure 12 shows the alignment of selected VH domain sequences for re-engineering of antibody 1F5; and

Figure 13 shows the alignment of selected VK domain sequences for re-engineering of antibody 1F5.

DETAILED DESCRIPTION OF THE INVENTION A "naturally occurring immunoglobulin" as used herein is an antibody, usually in the form of a tetramer consisting of two identical pairs of immunoglobulin chains, each pair having one light and one heavy chain. In each pair, the light and heavy chain variable regions are together responsible for binding to an antigen, and the constant regions are responsible for the effector functions typical of an antibody. Each of the variable regions of both the heavy and light chain is divided into segments comprising four framework sub-regions (FRl, FR2, FR3 and FR4), interrupted by three stretches of hypervariable sequences, also known as the complementarity determining regions (CDRs), as defined in Kabat's database (Kabat et al., Sequences of proteins of immunological interest. US Department of Health and Human Services, NIH publication, 91-3242, 1991, incorporated herein by reference). Without specifying the particular sub-regions as FRl, FR2, FR3, or FR4, a framework region represents the combined frameworks within the variable region of a single, naturally occurring immunoglobulin chain. The sequences of the framework regions of different light and heavy chains are relatively conserved within a species. The "framework region of an antibody" is the combined framework regions of the constituent light and heavy chains and serves to position and align the CDRs. The CDRs are primarily responsible for forming the binding site of an antibody, conferring binding specificity and affinity to an epitope of an antigen.

Immunoglobulins may be in different forms, including naturally occurring, chemically modified, or genetically-engineered immunoglobulins, such as single chain Fv, single domain antibodies (dAbs), diabodies, mini-antibodies, Fab, Fab', F(ab')₂ and bifunctional hybrid antibodies.

A "chimeric antibody" is an antibody in which variable regions are linked, without significant sequence modification from the parent V-region sequences, to the corresponding heavy and light chain constant regions of a different species. The most common types of chimeric antibodies are those containing murine variable regions and human constant regions. In this case, the expressed hybrid molecule will have the binding specificity and affinity of the parent murine anybody, and the effector functions of a human antibody. In addition, since two-thirds of the amino acids of a chimeric antibody are of human origin, a reduced immunogenicity is expected when used therapeutically in humans.

A "humanized" antibody as used herein is an immunoglobulin comprising one or more human framework regions with one or more CDRs from a non-human immunoglobulin. The selected human immunoglobulin, into which the CDRs are grafted, is conventionally derived from a single human immunoglobulin species. This direct grafting method of inserting donor CDRs onto human acceptor frameworks without further sequence modifications usually results in substantial loss of antigen affinity as well as expression level.

A "parent immunoglobulin" as used herein refers to an immunoglobulin that is the starting point of the process of immunoglobulin re-engineering according to the invention.

Usually, a parent immunoglobulin is an antibody that has been identified or developed due to certain useful properties, such as a specificity or a high binding affinity for a target antigen, e.g., an antigen that plays a role in a disease or condition which can be treated with an antibody against the target antigen. A parent immunoglobulin is one whose structure can be altered, e.g., by a process of functional humanization. A parent immunoglobulin is often a murine antibody, or some other non-human immunoglobulin, that would raise an antigenic response if introduced into a human or another species that is different from the parent immunoglobulin species. Typically, it is the CDR segments of the parent antibody which have desirable properties, i.e., binding affinity for a target antigen, that can be translated into a humanized antibody for therapeutic uses. The present invention describes a framework re-engineering approach to modify an immunoglobulin for expression improvement and immunogenicity reduction. The variable region of an immunoglobulin heavy chain or light chain is divided into different structural or sequence segments such as FRl, CDRl, FR2, CDR2, FR3, CDR3, and FR4, according to the classification of Kabat et al. (1991). By swapping corresponding framework segments (i.e., FRl of one heavy chain with FRl of a different heavy chain, FR2 of one heavy chain with FR2 of a different heavy chain, FRl of one light chain with FRl of a different light chain, etc.), one can alter the level of expression of the re-engineered immunoglobulin, and at the same time convert the sequence of the re-engineered immunoglobulin to a less immunogenic protein, e.g., for use in treatment of a given organism, such as a human, which may be a different species than the species of the parent immunoglobulin. For example, donor framework sequences for immunoglobulin re-engineering can be derived from immunoglobulins of the same species as the parent immunoglobulin, but different variable region subgroups, different immunoglobulin isotypes, or different immunoglobulin allotypes. Immunoglobulin isotypes are classes of immunoglobulins having different types of heavy or light chains. For example, heavy chain isotypes include heavy chains derived from IgA, IgD, IgE, IgG, and IgM; light chain isotypes include kappa and lambda chains. In different embodiments of the invention, donor framework sequences can be derived from an immunoglobulin of either the same or different isotype as the parent immunoglobulin. Immunoglobulin allotypes are immunoglobulins containing one or more allelic polymorphisms. Alternatively, donor framework sequences can be derived from immunoglobulins of different species (same or different isotypes or allotypes) as that of the parent immunoglobulin.

At the gene level, the swapping of framework segments will affect the mRNA sequence, which in turn will affect the transcription rate and/or stability of the mRNA encoding the recombinant protein. At the protein level, the swapping of framework segments will affect the folding, domain contacts, and stability of the resultant protein. The aggregated effect, when the right combination of framework segments is chosen, will be an enhancement of protein expression.

While production enhancement represents one of the aspects of the invention, selection of the appropriate framework segments should be performed so as to mitigate possible immunogenicity problems.

One embodiment of the invention is a method of enhancing antibody expression through re-engineering a parent immunoglobulin by replacing the framework segments of the parent heavy chain variable region. A set of amino acid sequences corresponding to the heavy chain variable region framework segments FRl, FR2, FR3, and FR4, derived from two or more immunoglobulin heavy chains is provided. Such sequences are known in the art. The amino acid sequences of the heavy chain variable region framework segments FRl, FR2, FR3, and/or FR4 of the parent immunoglobulin are replaced with sequences of corresponding heavy chain variable region framework segments selected from the set of amino acid sequences to obtain a set of two or more framework re-engineered immunoglobulins. In one aspect of the invention, not all of the framework segment sequences selected from said set are derived from the same immunoglobulin; at least one framework segment sequence from FRl, FR2, and FR3 is derived from a different immunoglobulin chain than the other FRl, FR2, and FR3 framework segment sequences. The expression level of each of the set of two or more framework re-engineered immunoglobulins is determined using methods generally known in the art. For example, a nucleotide sequence encoding a re-engineered immunoglobulin heavy chain and a nucleotide sequence encoding a light chain are co-transfected into a suitable host cell, e.g., by incorporating them into one or two expression vectors, and the transfected host cell is cultured under conditions known to promote expression of the introduced immunoglobulin sequences from the expression vector(s). The expression level of each re- engineered immunoglobulin can be determined by methods well known in the art, e.g., by an ELISA assay using an antibody specific for the secreted re-engineered immunoglobulin. Using the expression data, a framework re-engineered immunoglobulin from the set of two or more framework re-engineered immunoglobulins is selected according to its expression level. Criteria for selection can be determined by the user. For example, the framework re- engineered immunoglobulin having the highest level of expression can be selected. Alternatively, all re-engineered immunoglobulins having a level of expression above a certain threshold (e.g., 10%, 20%, 30%, 50%, 70%, 100%, or 150% or more of the expression level of the parent antibody in the same system) can be selected. Re-engineered antibodies that are selected are available for subsequent use, e.g., for further investigation of their suitability as therapeutic antibodies.

In some embodiments, the heavy chain variable region of the parent immunoglobulin is re-engineered according to a process as outlined above. In other embodiments, the light chain variable region of the parent immunoglobulin is re-engineered according to a process as outlined above. In still other embodiments, both the heavy and light chain regions of the parent immunoglobulin are re-engineered.

Because not only the expression level, but also the affinity of a re-engineered immunoglobulin is an important aspect of its usefulness, the above described method for improving the level of expression can be combined with, or performed subsequent to, a selection of re-engineered antibodies based on their binding affinity for a target antigen. In such a case, the selection based on expression level can be made either separately or in combination with selection based on binding affinity. For example, only re-engineered immunoglobulins having a sufficiently high level of expression (e.g., 10%, 20%, 30%, 50%, 70%, 100%, or 150% or more of the expression level of the parent antibody in the same system) can then be selected based on their binding affinity for the target antigen. Alternatively, only re-engineered immunoglobulins having a sufficiently high affinity for the target (e.g., 10%, 20%, 30%, 50%, 70%, 100%, or 150% or more of the Ka of the parent antibody) are screened for expression level.

In certain embodiments, a framework sequence with the highest degree of homology to the corresponding parent framework, within the set of available donor framework sequences, is selected. In other embodiments, a framework sequence with at least 60% amino acid sequence homology to the corresponding parent framework is selected. In some embodiments, only human immunoglobulins are used as a source of donor framework sequences.

In other embodiments, the framework sequence with the highest homology at the three or four amino acids immediately adjacent to the neighboring CDRs, within the set of available donor framework sequences, is selected. For example, the FR3 having the highest homology to the parent FR3, preferably 100%, over three or more amino acids at both ends immediately adjacent to the flanking CDR2 and CDR3, can be chosen.

In case no framework segment fulfilling the previous criterion is identified within the set of available donor framework sequences, a framework segment with the closest homology at these positions and containing conservatively similar amino acids can be selected. Conservatively similar amino acids are well known in the art. For example, amino acids can be substituted within the following groups: gly, ala; val, ile, leu; asp, glu; asn, gin; ser, thr; lys, arg; and phe, tyr.

Alternatively, libraries of framework segments from different species can be constructed. For example, one can construct a library of human FRl, a library of human FR2, a library of human FR3, and a library of human FR4 for the heavy chain variable region, and do the same for the light chain variable region, and use recombinant methods to assemble the different frameworks in conjunction with the CDRs of a murine antibody of interest according to their natural order (i.e., FRl, CDRl, FR2, CDR2, FR3, CDR3, FR4 for the heavy chain V region; same configuration for the light chain V region).

To maintain the affinity of a humanized immunoglobulin, one or more donor amino acid residues may have to be incorporated into the framework regions of the acceptor immunoglobulin. A set of criteria have been developed for selecting a limited number of amino acids within the acceptor immunoglobulin for conversion into amino acid residues, as reported in a series of publications (U.S. Pat. No. 5,85,089; U.S. Pat. No. 5,693,762; U.S. Pat. No. 5,693,761, incorporated herein by reference ).

The level of expression for a re-engineered antibody can be modified by manipulating certain amino acid residues in the framework regions. A number of investigators found that certain amino residues in the framework region have significant effects on the folding, expression and stability of the antibodies (Liu et al., Sheng Wu Gong Cheng Xue Bao, 19:272-276, 2003; Jung et al., J. MoI. Biol., 309:701-716, 2001; Saldanha et al., MoI. Immunol., 36:709-719, 1999; Forsberg et al., J. Biol. Chem., 272: 12430-12436, 1997, each of which is incorporated herein by reference). Forsberg et al. (1997) suggested that at least some of the framework residues have to be replaced in order to obtain a high level of production as they found that by substituting five residues in the VL chain increased the level of secretion of a monoclonal antibody Fab protein to almost 15 times. In addition, Liu et al. (2003) showed that the level of expression was increased 3 fold after the amino acid of the light chain gene of the parent anti-CD20 antibody was mutated. In some cases, back- mutations involved in the process of humanization elevate the level of antibody expression while restoring its immunoreactivity. For example, Saldanha et al. (1999) demonstrated that by back mutating Asp at position 9 in the FRl of the kappa light chain, not only the binding affinity of the humanized L-25 antibody was restored, but the levels of secretion of the antibody in Cos cells were also increased. So, in theory, one should be able to identify individual amino acid residues in the framework of a humanized antibody that can be altered so as to change the level of expression of that antibody.

For those skilled in the art of antibody engineering, it is common knowledge that the primary amino acid sequences of the variable region of an antibody will have effects on the stability and level of production for cell lines transfected with the sequences. For example, in the construction of chimeric and conventionally humanized (CDR-grafted) antibodies, it is common practice that variable region sequences of different antibodies are spliced onto the same human constant region sequences, such as human IgGl for the heavy chain, and human kappa for the light chain. Yet when the same expression vector system and cell line expression system are used, variations in the cell line stability, and level of antibody expression are observed with different chimeric and humanized antibodies. Everything else being equal, such variations will only be attributable to differences in the primary amino acid sequences of the respective variable regions. It is conceivable that the presence of some amino acids at particular positions along the framework sequence of an antibody will be favorable for cell line stability and antibody production, and the presence of other amino acids at different positions can have the opposite effects. While it is difficult to develop predictive guidelines in identifying the positions and amino acid residues to be altered in order to achieve improved productivity and cell line stability, there are ways to systematically narrow down the regions and identity of the amino acids to be modified with the intended effects. One way of achieving this goal is to compare the primary amino acid sequences for the variable regions of different chimeric and humanized antibodies in relationship to their levels of antibody expression and cell line stabilities. Preferably, it would be more predictive if variable region sequences of chimeric and different humanized versions of the same parent antibody observed to have different levels of expression and cell line stabilities are compared. If a sufficiently large database is generated or made available, one should be able to narrow down the positions within the framework regions of the heavy and light chain variable regions and the preferred amino acid residues at these positions that would have a positive influence on the level of antibody production and cell line stability; similarly, one should also be able to identify framework positions and amino acid residues at these position that would have deleterious effects on the level of antibody production and cell line stability.

Such information will be particularly useful in the design of humanized antibody sequences. In the conventional humanization approach, one should either select framework sequences that carry the preferred amino acid residues at the preferred positions for CDR-grafting, or alternatively, introduce mutations with the preferred amino acid residues at the preferred positions in the framework sequences selected for CDR-grafting.

This constitutes one embodiment of the current invention by developing a large database with chimeric and humanized antibody sequences in relation to their levels of antibody production and cell line stabilities, and aligning these sequences to identify unusual amino acids at the framework positions that appear to have impacts on the levels of antibody production and cell line stabilities. Whether the identified amino acid residues at the identified positions will have impacts on the levels of expression and cell line stabilities of antibodies can be evaluated by systematically introducing these residues at the proper positions to different antibodies. However, by introducing unnatural amino acids in the framework sequence of an antibody, two potentially immunogenic epitopes might result, namely, "T cell epitopes" and "B cell epitopes." T cell epitopes are short peptide sequences released during the degradation of proteins within cells and subsequently presented by molecules of the major histocompatibility complexes (MHC) in order to trigger the activation of T cells. For peptides presented by MHC class II, such activation of T cells can then give rise to an antibody response by direct stimulation of B cells to produce such antibodies. Therefore, the introduction of donor framework residues to the acceptor framework region during the process for humanization, or the manipulation of expression levels by changing framework residues, will have the possibility of generating new, immunogenic T cell epitopes, resulting in the elicitation of immune responses against the "humanized" antibody. B cell epitopes refer to exposed structural differences that have become recognizable by a B cell receptor in the host system when the framework sequence contains unnatural (for example, back-mutated) amino acid residues.

Also, it is possible that some "favorable" amino acids can only exert their effects when placed within the context of the proper neighboring framework sequences, or overall sequences; and therefore, when a "favorable" amino acid is placed within a wrong framework or overall sequence environment by back mutation, its positive effect on the level of antibody expression and cell line stability may be diminished or offset. As an alternative approach, it would be more effective if stretches of framework sequences, such as FRl, FR2, FR3 and FR4 as defined in the Kabat database (Kabat et al. Sequences of proteins of immunological interest. US Department of Health and Human Services, NIH publication, 91-3242, 1991, incorporated herein by reference) are used for constructing a reengineered antibody with the intended effects of enhancing its level of antibody expression and improving cell line stability. Such FR-reengineering approach serves the dual purposes of reducing the immunogenicity of the resultant antibody and enhancing its level of expression that will facilitate production of the reengineered antibody for clinical uses.

In the construction of a FR-reengineered immunoglobulin of the invention, sequence design for the variable regions of the immunoglobulin can be performed using the criteria and principles illustrated above. The designed FR-reengineered variable region sequence can be assembled using a variety of standard recombinant techniques well known to those skilled in the art. Preferably, the designed sequence, usually of a size of about 350 base pairs, will be synthesized (Leung et al., Molecular Immunol. 32: 1413-1427, 1995; Daugherty et al., Nucl. Acid Res. 19:2471-2476; DeMartino et al., Antibody Immunoconj. Radiopharmaceut. 4:829, 1991; Jones et al., op. cit, all of which are incorporated herein by reference), or the individual frameworks can be introduced to replace the parent frameworks by methods of site- or oligonucleotide-directed mutagenesis (Gillman and Smith, Gene 8:81-97, 1979; and Roberts et al., Nature 328:731-734, both of which are incorporated herein by reference).

The DNA segment encoding the FR-reengineered immunoglobulin can be joined to DNA sequences encoding the human heavy and light chain regions in DNA expression vectors suitable for bacterial propagation and expression in different host cells. There are a variety of DNA vectors suitable for expression in a variety of host cell systems. Appropriate DNA vectors can be chosen for the expression of the FR-reengineered immunoglobulins. Typically, the vector includes a suitable expression control DNA sequence that is linked operably to DNA segments encoding the immunoglobulin chains. Preferably, the expression control sequences will be eukaryotic promoter systems and will be used in vectors capable of transforming or transfecting eukaryotic host cells, but control sequences for prokaryotic hosts may also be used. The sequence encoding the FR-reengineered heavy and light immunoglobulin chains can be incorporated into one single DNA expression vector, or into separate heavy and light chain expression vectors. In the latter case, host cells will have to be simultaneously transfected with both vectors in order to produce a FR-reengineered antibody with the properly paired heavy and light chain polypeptides. In general, a leader sequence allowing the transportation of the immunoglobulin polypeptide into the Golgi apparatus for later secretion is included at the N-terminal end of each immunoglobulin chain for expression in eukaryotic host cells. Once the vector has been incorporated into the appropriate host, the host is maintained under conditions suitable for high level expression of the nucleotide sequences, and, as desired, the collection and purification of the re-engineered light chains, heavy chains, light/heavy chain dimers or intact antibodies, binding fragments, single chain (sFv) antibodies, diabodies, domain antibodies, or derivatives thereof, or other immunoglobulin forms may follow (see Beychok, Cells of Immunoglobulin Synthesis, Academic Press, NY, 1979, which is incorporated herein by reference).

It is well-known that there are different human constant regions for the heavy and light chains. A particular isotype will have specific effector characteristics that can be chosen for use for different purposes. Human constant region DNA sequences can be isolated in accordance with well known procedures from a variety of human cells, but preferably immortalized B-cells (see, Kabat op. cit. and WO87/02671, incorporated by reference herein). The CDRs for producing the immunoglobulins of the present invention will be similarly derived from monoclonal antibodies capable of binding to the predetermined antigens, CD20, for example, and produced by well known methods in any convenient mammalian source including, mice, rats, rabbits, and other vertebrates capable of producing antibodies. Suitable source cells for the constant region as well as framework DNA and secretion can be obtained from a number of sources such as the American Type Culture Collection ("Catalogue of Cell Lines and Hybridomas," Sixth Edition, 1988, Rockville, Md., USA, which is incorporated herein by reference).

DNA expression vectors containing the coding sequences for the FR-reengineered immunoglobulin chains operably linked to an expression control sequence (including promoter and enhancers) are typically replicable in the host organisms either as episomes or as an integral part of the host chromosomal DNA. Selectable markers such as tetracycline- resistance, neomycin-resistance, beta-lactamase, etc., are included in the vector to allow detection of cells transformed with the DNA vectors (see, for example, U.S. Pat. No. 4,704,362, which is incorporated herein by reference).

Bacterial hosts are suitable for propagating the DNA vectors as well as expressing the incorporated immunoglobulin DNA. For example, E. coli is the most preferably used prokaryotic host used for cloning the DNA sequence for the present invention. Other microbial hosts that are useful for the same purposes include, as examples, bacilli (for example Bacillus subtilis), and other enterobacteriaceae (for example Salmonella, Serratia), and various Pseudomonas species. Expression of cloned sequences in these hosts require the presence of expression control sequences compatible with the host cell (for example an origin of replication), and functional promoters to be included in the DNA vector. Example of well- known promoter systems include, but are not limited to, tryptophan (trp) promoter systems, beta-lactamase promoter systems, phage lambda promoter systems, etc. These promoters are responsible for controlling expression, or transcription, of the functional gene sequence downstream of the promoter system, which contains, in addition to all necessary motifs, and optionally an operator sequence, ribosome binding site sequences and the like, necessary for transcription initiation and translation.

Similarly, other microbes, such as yeast, may also be used for expression. For example, a preferred host is Saccharomyces , which is a suitable host for expression vectors containing the appropriate expression control elements, such as promoters, including 3- phosphoglycerate kinase or other glycolytic enzymes, and origin of replication or termination sequences and the like as desired.

Eukaryotic host cells of invertebrate origin, including insect cells, such as hi-5, SF9, and SF21, also can be used for expression. Appropriate expression vectors containing convenient cloning sites, promoters, termination sequences, etc., that are important for high- level expression in the host cells are available commercially (for example from Invitrogen, San Diego, CA).

Preferably, mammalian tissue cell cultures may be used to express and produce the polypeptides of the present invention (see, Winnacker, "From Genes to Clones," VCH Publisher, NY, N.Y., 1987, which is incorporated herein by reference). The most preferably used mammalian host cells for antibodies are Chinese Hamster Ovary (CHO) cell lines, various COS cell lines, HeLa cells, and myeloma cell lines such as SP2/0 cell lines, NSO cell lines, YB2/0 cell lines, etc, and transformed B-cells or hybridomas. These cell lines are capable of conferring the right glycosylation at appropriate sites such as amino acid 297 in the heavy chain CH2 domain. These cells are also capable of secreting full-length immunoglobulins, and are the preferred host cell systems for this invention. Similar to expression vectors for other host cells, a eukaryotic cell expression vector will contain the appropriate expression control sequences including promoter (for example, those derived from immunoglobulin genes, metallothionein genes, SV40, Adenovirus, cytomegalovirus, Bovine Papilloma Virus, and the like), enhancers, usually with a leader sequence for directing the expressed polypeptide to the Golgi apparatus for glycosylation and export, the DNA segments of interest (for example, the heavy and light chain coding sequences and expression control sequences), a termination codon, other necessary processing information sites (such as ribosome binding sites, RNA splice sites, a polyadenylation sequence, and transcriptional terminator sequences), and a selection marker (such as mutant Dhfr, glutamine synthetase (GS), hygromycin, neomycin) (see Kellems, "Gene Amplification in mammalian cells", Marcel Dekker Inc., NY, N.Y, 1993, which is incorporated herein by reference).

There exist a plethora of established and well-known methods for introducing vectors containing the DNA segments of interest into the host cell, either transiently or stably integrated into the host cell genome. They include, but are not limited to, calcium chloride transfection, calcium phosphate treatment, electroporation, lipofection, etc. (See, Maniatis et al., Molecular Cloning: A Laboratory Manual, Cold Spring Harbor Press, 1982, which is incorporated herein by reference). Identification of host cells incorporated with the appropriate expression vector is achievable typically by first growing cells under selection pressure in accordance with the selectable marker used in the vector, and detection of secreted proteins. For example, whole antibodies containing two pairs of heavy and light chains or other immunoglobulin forms of the present invention can be identified by standard procedures such as ELISA and Western analysis. Purification of the expressed immunoglobulin can be purified according to standard procedures known in the art, including ammonium sulfate precipitation, affinity columns, column chromatography, gel electrophoresis and the like (see, generally, R. Scopes, "Protein Purification", Springer- Verlag, NY, 1982). Substantially pure immunoglobulins of at least about 90 to 95% homogeneity are preferred, and 98 to 99% or more homogeneity are most preferred, for pharmaceutical uses. Once purified, partially or to a desired homogeneity, the polypeptides may then be used therapeutically (including extracorporeally) or in developing and performing assay procedures, immunofluorescent stainings, and the like (see, generally, Immunological Methods VoIs. I and II, Lefkovits and Pernis, eds., Academic Press, New York, N.Y, 1979 and 1981, incorporated by reference herein). The antibodies of the present invention will typically find use individually, or in combination with other treatment modalities, in treating diseases susceptible to antibody- based therapy. For example, the immunoglobulins can be used for passive immunization, or the removal of unwanted cells or antigens, such as by complement mediated lysis, all without substantial adverse immune reactions (for example, anaphylactic shock) associated with many prior antibodies.

A preferable usage of the antibodies of the present invention will be in the treatment of diseases using their naked forms (naked antibodies) at dosages ranging from 50 mg to 400 mg/m², administered either locally at the lesion site, subcutaneously, or parenterally (intravenously, intramuscularly, etc). Such dosages represent exemplary therapeutically effective amounts of the antibodies of the invention.

The antibodies of the invention can also be administered in the form of compositions, but may also be formulated into well known drug delivery systems. For example, antibodies of the invention can be administered orally, rectally, parenterally (intravenously, intramuscularly, or subcutaneously), intracisternally, intravaginally, intraperitoneally, locally (powders, ointments or drops) or as a buccal or nasal spray. Administration of an antibody of the invention may also be local or systemic and accomplished intravenously, intraarterially, intrathecally (via the spinal fluid) or the like. A composition for administration can comprise a pharmaceutically acceptable carrier for the antibody of the invention. A pharmaceutically acceptable carrier includes such carriers as, for example, aqueous solutions, non-toxic excipients including salts, preservatives, buffers and the like (Remington's Pharmaceutical Sciences, 15th Ed. Easton, Mack Publishing Co., pp. 1405-1412 and 1461-1487 (1975); The National Formulary XIV., 14th Ed., Washington: American Pharmaceutical Association (1975), each of which incorporated by reference herein. Exemplary pharmaceutically acceptable carriers for an antibody of the invention can also include non-aqueous solvents such as propylene glycol, polyethylene glycol and vegetable oil or injectable organic esters such as ethyl oleate. An aqueous carrier can also include, without limitation, water, alcoholic/aqueous solutions, saline solutions and parenteral vehicles such as sodium chloride or Ringer's dextrose. Intravenous carriers for administration of an antibody of the invention include, for example, fluid and nutrient replenishers. The pH and exact concentration of the various components for a pharmaceutical composition can also be adjusted according to routine skills in the art (Goodman and Gilman's The Pharmacological Basis for Therapeutics (7th Edition), incorporated by reference herein). An antibody of the invention may be administered to a patient in an amount or dosage suitable for treatment of a disease (e.g., a therapeutically effective amount). Generally, a unit dosage comprising an antibody of the invention will vary depending on patient considerations. Such considerations include, for example, age, condition, sex, extent of disease, contraindications, concomitant therapies and the like. An exemplary unit dosage based on these considerations can also be adjusted or modified by a physician skilled in the art.

The invention also comprises administering, either sequentially or in combination with one or more antibodies of the invention, a conventional therapeutic agent in an amount that can potentially be effective for the treatment or prophylaxis of a disease. Multiple dosing at different intervals can also be performed to achieve optimal therapeutic or diagnostic responses, for example, at weekly intervals, once a week, for four weeks, etc. Usage of the antibodies derived from the present invention can also be combined with different treatment modalities, such as therapeutic agents including, but not limited to, chemotherapeutic or immunosuppressive drugs (for example, methotrexate, CHOP, Dox, 5-Fu, etc), radiotherapy, radioimmunotherapy, vaccines, enzymes, toxins/immunotoxins, or other antibodies derived from the present invention or others (for example, Rituxan) as well as combinations thereof. The antibodies of the present invention, if specific for the idiotype of an anti-tumor antibody (Ab2β), can be used as tumor vaccines for the elicitation of antibodies against the tumor antigen. Numerous additional agents, or combinations of agents, well-known to those skilled in the art may also be utilized.

In certain embodiments, the antibodies can be used in their naked forms, or as conjugated proteins with drugs, radionuclides, toxins, cytokines, soluble factors, hormones, enzymes (for example, carboxylesterase, ribonuclease), peptides, antigens (as a tumor vaccine), DNA, RNA, or any other effector molecules having a specific therapeutic function with the antibody moiety serving as the targeting agents or delivery vehicles. Exemplary conjugates can include polyethylene amine (PEI), polyglycine, hybrids of PEI and polyglycine, polyethylene glycol (PEG) or methoxypolyethylene glycol (mPEG). A conjugate for use with an antibody of the invention improve the in vivo half-life or stability of the antibody. Moreover, the antibodies or antibody derivatives, such as antibody fragments, single-chain Fv, diabodies, single domain antibodies, etc. of the present invention can be used in fusion proteins with other functional moieties, such as, antibodies or antibody derivatives (for example, as bispecific antibodies), toxins, cytokines, soluble factors, hormones, enzymes, peptides, etc. Different combinations of known pharmaceutical compositions may also be utilized.

Framework re-engineered antibodies of the present invention can also be used for in vitro purposes, for example, as diagnostic tools for the detection of specific antigens, or the like. The following examples are offered by way of illustration, and are not intended to limit the invention in any way. The examples herein may also illustrate advantages of the present invention that have not been previously described and to further assist a person of ordinary skill in the art with practicing the invention. The examples can include or incorporate any of the variations or embodiments of the invention described above. The embodiments described above may also further each include or incorporate the variations of any or all other embodiments of the invention.

EXAMPLES

Construction of chimeric and "functionally humanized" anti-human CD20 antibodies is illustrated and the expression levels of the parent murine antibody and the chimeric and functionally humanized antibody were compared. Manipulation of individual amino acids within the VH FR3 region of an anti-liver cancer antibody, which can affect the expression level and antigen binding affinity of the resultant antibody, is also discussed. EXAMPLE 1

Construction of chimeric anti-human CD20 antibody (clF5)

1F5 is a murine antibody specific for the human CD20 B cell antigen, which is of an IgG2a/kappa isotype. The hybridoma for the antibody was obtained from American Type Culture Collection (ATCC # HB-9645; lot # 221900). RNA was extracted from 3 x 10⁷ hybridoma cells using a Track mRNA Isolation Kit (Invitrogen). cDNA was then prepared by a cDNA Cycle Kit (Invitrogen) using the primer CHlB (5'ACA GTC ACT GAG CTG G 3') that is specifically prime to the CHl constant region of mouse IgG and the primer Ck3BHl (5'GCC GGA TCC TCA CTG GAT GGT GGG AAG ATG GAT ACA 3') that is prime to the kappa constant region of mouse IgG The first-stranded cDNA were then amplified by RACE and PCR methods. The DNA fragments after PCR were cloned into a TA Cloning Vector (Invitrogen) and then sequenced by the dideoxy chain termination method. The DNA and amino acid sequences of 1F5 heavy and light chain variable genes are shown in Figure 4 and Figure 5, respectively. To construct the chimeric antibody, the murine 1F5 heavy and light chain variable regions were inserted into appropriate expression vectors that contained the DNA sequences of human heavy and light chain constant regions. The 1F5 heavy chain variable region DNA sequence was used to replace the VH sequence of a heavy chain expression vector containing an Ig promoter, an Ig enhancer, a human IgGl constant region genomic sequence, and a selectable marker, gpt. The final heavy chain expression vector was designated as clF5pEgammal. Similarly, the 1F5 light chain variable region DNA sequence was used to replace the VL sequence of a light chain expression vector containing an Ig promoter, an Ig enhancer, a human kappa constant region genomic sequence, and a selectable marker, hyg. The final light chain expression vector was designated as clF5pEkappa.

Expression of clF5 antibody

The expression plasmids clF5pEgammal and clF5pEkappa were linearized and co- transfected into SP2/0 cells by electroporation. Cells were recovered in a culture medium for two days and then selected by standard methods in the presence of mycophenolic acid and/or hygromycin B conferred by the gpt and hyg genes on the plasmids. The surviving cell clones (from 96-well plates) were selected and scaled up to 24-well plates. The levels of antibody expression were checked by the ELISA method. ELISA strips were coated with goat anti- human F(ab)'₂ (Jackson ImmunoResearch), 100 μL of culture supernatants were added to each well of the ELISA strip and the antibodies were detected by an HRP conjugated goat anti-human Fc antibody (Jackson ImmunoResearch). Human IgG was diluted to various concentrations and used to generate a standard curve. To determine the concentration of the murine antibody, goat anti-mouse F(ab)'₂, HRP conjugated goat anti-mouse Fc antibody and mouse IgG were used. The chimeric 1F5 antibody (clF5) expression level was measured and compared to its murine parent (Table 1). The 1F5 hybridoma cells had a productivity of 0.5 mg/L, but after chimerization, the ability of SP2/0 cells to express the clF5 antibody had dropped to positive, but barely detectable, levels.

Table 1

EXAMPLE 2

Construction of Framework Re-engineered Anti-human CD20 Antibody (flF5) Design of genes for framework re-engineering anti-human CD 20 antibody (flF5)

The heavy and light chain variable regions of 1F5 were compartmentalized into FRl, CDRl, FR2, CDR2, FR2, CDR3 and FR4. All four frameworks were compared to the corresponding frameworks of human immunoglobulins listed in the Kabat database. According to the criteria and methods of the invention, various frameworks with high homology to 1F5 were aligned and are shown in Figure 12. The CDR regions are boxed and amino acids different from the parent murine 1F5 antibody are in bold. The FRl sequence of the murine VH was compared with the FRl sequences of human

VH from the Kabat database. Human FRl sequences of the highest homology are preferred, particularly at the sequences closest to the CDRl. For example, the human FRl sequence from LS2'CL has close to 80% sequence homology to that of the murine 1F5 antibody, and the 10 residues adjacent to the CDRl are identical to the murine parent sequence. Accordingly, the human FRl sequence from LS2'CL was chosen for the FRl of the anti- CD20 antibody. The FR2 sequence of the human NEWM was chosen for replacing the FR2 sequence of the anti-CD20 antibody. It should be noted that although the third residue of the NEWM FR2 closest to the CDRl is not identical to that of the murine parent sequence, it is a conserved K to R conversion. The human heavy chain FR3 sequence of 783CCL was used to replace the FR3 sequence of murine 1F5. The human FR3 from 783C'CL has 78% sequence homology and seven identical residues adjacent to the CDR3 compared to murine. Although the human residues adjacent to the CDR2 were not the same as that of the murine sequence, the differences were conserved. For example, the K, A, and L at positions 57, 58 and 60 (Kabat numbering), which are the first, second, and fourth residues closest to the CDR2, were replaced by the conserved human residues R, V and I, respectively. Accordingly, 783CCL was chosen for the replacement of murine FR3.

There are many human FR4 sequences with the amino acids closest to the CDR3 being identical and of a high degree of homology to the murine parent. In this example, the human 4Gl 2' CL sequence was selected for replacing the FR4 of the anti-CD20 antibody.

The final design of the framework re-engineered VH sequence for the anti-CD20 antibody contains the human LS2'CL FRl, NEWM FR2, 783CCL FR3 and 4G12'CL FR4 sequences replacing the original VH frameworks of the murine 1F5 antibody (Figure 6a).

An alignment of human framework sequences homologous with the VK domain of the murine FRl antibody is presented in Figure 13. The CDR regions are boxed and amino acids different from the parent murine 1F5 antibody are in bold. Human BJl 9 was chosen to replace the FRl of the murine VL as it is the human FRl sequence with the highest homology

(61%) to the murine parent. Moreover, some of the human residues that are different from that of the murine parent are conserved. For example, the E to D and K to R conversions at positions 18 and 19, respectively, are conserved changes.

Human MOT was chosen for replacement of the FR2 of the murine VL. MOT was found to be of high sequence homology (73%) to the murine FR2, with four human residues adjacent to the CDRl identical to those of the murine parent. Although only two human residues adjacent to the CDR2 are identical to those of the murine parent, and the third residue W was replaced by V, both W and V are neutral amino acids. It was therefore determined to replace the murine FR2 of the VL domain with the human MOT.

Human WES was chosen for replacing the FR3 of the murine VL. FR3 has the longest sequence and its homology between WES and the murine FR3 is 71%, with the three human residues flanking CDR2 and CDR3 being identical to the corresponding murine- residues.

Human lambda FR4 sequence from NIG-58 was chosen to replace FR4 of murine VL, for similar reasons. These sequences are 72% homologous to the stretch of seven residues adjacent to the CDR3, which are identical between the human and murine. The final design of the VL sequence for the anti-CD20 antibody includes the human

BJl 9 FRl, MOT FR2, WES FR3, and NIG-58 FR4, replacing the original VL frameworks of the anti-CD20 antibody. (Figure 6b).

Construction of the framework re-engineered heavy and light chain genes The designed heavy and light chain variable region sequences of the flF5 antibody were assembled by a combination of oligonucleotide synthesis and PCR using a variety of conventional methods.

To construct the heavy chain variable region sequence (Figure 7a), the full DNA sequence was divided into two halves. The N- and C-terminal halves were constructed separately by PCR and the complete variable region sequence was formed by joining the N- and C-terminal halves at the Spel site.

The N-terminal half was constructed as follows: The N-template (5'AAT AAG CCT GGG GCC TCA GTG AAG GTC TCC TGC AAG GCT TCT GGC TAC ACA TTT ACC AGT TAC AAT ATG CAC TGG GTA CGG CAG CCT CCT GGA AGG GGC CTG GAA TGG ATT GGA 3 ') (Seq. Id. No.: 30) has a synthetic sense-strand oligonucleotide (114-mer) encoding amino acids 12-49 of the VH region (Figure 6a). The template was PCR-amplified by two primers: the 5' primer (5'GTG CAA CTG CAG GCT TCC GGG GCT GAG GTA AAT AAG CCT GGG GCC TCA GTG AAG 3'), (Seq. Id. No.: 31) and the 3' primer (5'TGT AAC TAG TAT CAC CAT TTC CTG GAT AAA TAG CTC CAA TCC ATT CCA GGC CCC T 3') (Seq. Id. No.: 32). The N-template was PCR-amplified using the 5' and 3' primer set via standard techniques.

The C-terminal half was constructed as follows: The C-template (5'ATC ACT GCA GAC AAA TCC ACT AGC ACA GCC TAC ATG GAG CTC AGC AGT CTG AGG TCT GAG GAC ACT GCG GTC TAT TAC TGT GCA AGA TCG CAC TAC GGT AGT AAC TAC GTA GAC TAC TTT GAC TAC 3 ') (Seq. Id. No.: 33) has a synthetic sense-strand oligonucleotide (126-mer) encoding amino acids 70-111 of the VH region (Figure 6a). The template was PCR-amplified by two primers: the 5' primer (5'TGA TAC TAG TTA CAA TCA GAA ATT CAA GGG CAG GGT CAC AAT CAC TGC AGA CAA ATC CAC T 3') (Seq. Id. No.: 34) and the 3' primer (5'GGA GAC GGT GAC CGT GGT GCC TTG GCC CCA GTA GTC AAA GTA GTC TAC GTA 3 ') (Seq. Id. No.: 35). The C-template was PCR-amplified using the 5' and 3' primer set via standard techniques.

Double-stranded PCR-amplifϊed products for the N- and C-templates were gel- purified and restriction-digested with an Spel restriction enzyme. The N- and C-double stranded DNA were ligated at the Spel site, and the ligated products were subjected to another round of PCR amplification using the 5' primer for the N-template and the 3' primer for the C-template. The PCR product with a size of 350 bases was directly cloned into a TA Cloning Vector (Invitrogen). The sequence of the cloned fragment was confirmed by Sanger's method (Sanger et al., op. cit.) to be identical to the designed VH sequence. The confirmed sequence was used to replace the VH sequence of a heavy chain expression vector containing an Ig promoter, an Ig enhancer, a human IgGl constant region genomic sequence, and a selectable marker, gpt. The final heavy chain expression vector was designated as flF5pEgammal. To construct the light chain variable region sequence (Figure 7b), the full length VL variable region sequence is divided into two halves. The N-terminal and C-terminal halves were assembled separately by PCR and joined together via the BspEI site.

The N-terminal half was constructed as follows: an N-template (5'TCA AGT CTT TCT GCA TCT GTG GGG GAC AGA GTC ACA ATT ACT TGC AGG GCC AGC TCA AGT TTA AGT TTC ATG CAC TGG TAC CAG CAG AAG CCA GGA CAG GCT CCC GTC CCC GTAATT TAT GCC ACA TCC 3') (Seq. Id. No.: 36) has a synthetic sense-strand oligonucleotide (129-mer) encoding amino acids 9-51 of the VL region (Figure 6b). The template was PCR-amplified by two primers: the 5' primer (5'GAT ATT CAG CTG ACA CAG TCT CCA TCA AGT CTT TCT GCA TCT GTG 3') (Seq. Id. No.: 37) and the 3' primer (5'GGA CTC CGG AAG CCA GGT TGG ATG TGG CAT AAA TTA CGG G 3') (Seq. Id. No.: 38). The N-template was PCR-amplified using the 5' and 3' primer set via standard techniques.

The C-terminal half was constructed as follows: The C-template (5'TTC AGT GGC AGT GGG TCT GGG ACC GAG TTC ACT CTC ACA ATC AGC AGT TTG CAG CCT GAA GAT TTC GCC ACT TAT TTC TGC CAT CAG TGG AGT AGT AAC CCG CTC ACG TTC GGT GCT GGG 3 ') (Seq. Id. No.: 39) has a synthetic sense-strand oligonucleotide (120-mer) encoding amino acids 61-100 of the VH region (Figure 6b). The template was PCR-amplified by two primers: the 5' primer (5'GGC TTC CGG AGT CCC TAG TCG CTT CAG TGG CAG TGG GTC TGG G 3') (Seq. Id. No.: 40) and the 3' primer (5'CCG TTT GAT CAC CAG CTT GGT CCC AGC ACC GAA CGT GAG CGG 3') (Seq. Id. No.: 41). The C-template was PCR-amplified using the 5' and 3' primer set via standard techniques and procedures. Double-stranded PCR-amplifϊed products for the N- and C-templates were gel- purified and restriction-digested with a BspEI restriction enzyme. The N- and C-double stranded DNA were ligated at the BspEI site and amplified using the 5' primer for the N- template and the 3 ' primer for the C-template. The PCR product with a size of around 350 bp was directly cloned into a TA Cloning Vector (Invitrogen). The sequence of the cloned fragment was confirmed by Sanger's method to be identical to the designed VL sequence. The confirmed sequence was used to replace the VL sequence of a light chain expression vector containing an Ig promoter, an Ig enhancer, a human kappa constant region genomic sequence, and a selectable marker, hyg. The final light chain expression vector was designated as flF5pEkappa.

Expression offlF5 antibody

The expression plasmids flF5pEgammal and flF5pEkappa were linearized and co- transfected into SP2/0 cells by electroporation. Cells recovered in a culture medium for two days and then were selected in the presence of hygromycin. The surviving SP2/0 clones (from 96-well plates) were selected and scaled up to 24-well plates. The levels of antibody expression were checked by an ELISA method. ELISA strips were coated with goat anti- human F(ab)'₂ (Jackson ImmunoResearch), 100 μL of culture supernatants were added to each well of the ELISA strip and the antibodies were detected by HRP conjugated goat anti- human Fc antibody (Jackson ImmunoResearch). Human IgG was diluted to various concentrations and used to generate a standard curve. The framework re-engineered 1F5 antibody (flF5) expression level was measured and compared to its murine parent and chimeric counterpart (Table T). Table 2

The results indicated that the productivity of the flF5 antibody after framework re- engineering was restored to a high level when compared to its chimeric counterparts. The expression level, after framework re-engineering, was even higher than its parent hybridoma cell line.

Clones that were identified to have a higher expression level by an ELISA were expanded for production in 500 ml roller bottles. Antibodies were purified using standard protein A columns. The purified antibodies were analyzed in an SDS-PAGE gel under both reducing and non-reducing conditions (Figure 8). The affinity of the framework re- engineered antibody (flF5) was first evaluated by flow cytometry. Raji cells (2.5 x 10⁶) were incubated with different concentrations of purified flF5 antibody in a final volume of 400 μL of PBS supplemented with 1% FCS and 0.01% (w/v) sodium azide (PBS-FA). The mixtures were incubated for 30 minutes at 4°C and washed three times with PBS to remove unbound antibodies. The bound levels of the antibodies to Raji cells were assessed by the addition of FITC-labeled goat anti-human IgGl, Fc fragment-specific antibodies (Jackson ImmunoResearch) in a final volume of 400 μL in PBS-FA, and incubated for 30 minutes at 4°C. The mixture was washed three times with PBS and fluorescence intensities were measured by a FACSCAN fluorescence-activated cell sorter (Becton Dickinson) (Figure 9). The results indicated that the flF5 antibodies bound well to Raji cells and showed a dose- dependant S-shaped curve.

To compare the affinity of the antibody after framework re-engineering, a competitive binding assay was performed. A fixed amount (200-fold dilution from stock) of FITC- conjugated flF5 was mixed with varying concentrations of either 1F5 or flF5. The mixtures were added to Raji cells in a final volume of 400 μL in PBS-FA, and incubated for 30 minutes at 4°C. After washing three times with PBS, the fluorescence intensities of Raji cells bound with the FITC-flF5 were measured by FASCAN (Becton Dickinson). The results indicated that flF5 antibody after framework re-engineering did not have significant effects on the affinity of the framework re-engineered antibody (Figure 10).

EXAMPLE 3

In the humanization of an anti-liver cancer antibody, manipulation of individual amino acids within the VH FR3 region can affect the expression level and antigen binding affinity of the resultant antibody. For the purpose of illustration, only the mutational studies within the FR3 of the heavy chain variable region of the anti-liver cancer antibody are discussed. Figure 11 lists out the original murine sequence and the different humanization sequence designs at the FR3 segment of the heavy chain variable region.

In the original humanization design, the human Lay FR3 sequence was found to exhibit the highest degree of homology to that of the parent antibody, and was therefore chosen for humanizing the heavy chain FR3 of the antibody. When the reengineered antibody carrying the heavy chain Lay FR3 was expressed, the antigen binding affinity of the resultant antibody dropped approximately 10 fold compared to that of the parent antibody. Moreover, the level of antibody expression was substantially lower than that of the chimeric antibody. Sequence comparison had identified two groups of amino acids that might be responsible for the drop in affinity and expression levels. The DNT group encompasses N->D mutation at position 72, G->N mutation at position 82B, and A->T mutation at position 93 of the Lay FR3 sequence (Kabat's numbering). The DTG group encompasses V->D mutation at position 86, S->T mutation at position 87, and A->G mutation at position 88 of the Lay FR3 sequence (Figure 11). Table 3 summarizes the comparative levels of antibody expression and immunoreactivities of the different versions of humanized antibodies.

Table 3

Expressions of different versions of the reengineered antibody carrying the DNT mutations (Figure 11) at the Lay VH FR3 segment were low with substantial reduction in their antigen binding affinities. However, the expression levels for the reengineered antibody carrying the DTG mutations at the Lay VH FR3 segment were somewhat restored; and more importantly, the antigen binding affinity of the DTG mutant was comparable to that of the chimeric antibody.

In order to reduce the number of mutations needed for affinity and expression level restoration, further single mutational studies had revealed that a single V->D mutation at position 86 was necessary and sufficient to restore the yield and binding affinity of the reengineered antibody carrying a Lay VH FR3 segment (Lay(D), Figure 11, and Table 3). Since introduction of an unnatural mutation in the context of human sequence can potentially create new T cell epitope, thereby enhancing the immunogenicity of the resultant reengineered antibody, attempt was made to search for natural human VH FR segments that fulfill most of our preset criteria for FR selection, and at the same time carry a natural D amino acid at position 86 in the Kabat database; the Tur VH FR3 sequence was found to be a likely candidate (Figure 11). When the VH FR3 of the antibody was replaced with that of the Tur sequence (no other mutations were introduced), the level of antibody expression and the antigen binding affinity of the resultant reengineered antibody were found to be comparable to that of the chimeric antibody (Table 3). The VH FR3 of the Tur sequence was therefore used for the construction of the final reengineered antibody.

While the present invention has been described herein in conjunction with a preferred embodiment, a person with ordinary skill in the art, after reading the foregoing specification, can effect changes, substitutions of equivalents and other types of alterations to that set forth herein. Each embodiment described above can also have included or incorporated therewith such variations as disclosed in regard to any or all of the other embodiments. Thus, it is intended that protection granted by Letters Patent hereon be limited in breadth and scope only by definitions contained in the appended claims and any equivalents thereof.

Claims

CLAIMS What is claimed is:

1. A method of enhancing antibody expression comprising: providing a set of amino acid sequences corresponding to the heavy chain variable region framework segments FRl, FR2, FR3, and FR4 derived from two or more immunoglobulin heavy chains; replacing the amino acid sequences of the heavy chain variable region framework segments FRl, FR2, FR3, and/or FR4 of a parent immunoglobulin with sequences of corresponding heavy chain variable region framework segments selected from said set of amino acid sequences to obtain a set of two or more framework re-engineered immunoglobulins; determining the expression level of each of said set of two or more framework re- engineered immunoglobulins; and selecting a framework re-engineered immunoglobulin from said set of two or more framework re-engineered immunoglobulins according to its expression level.

2. The method of claim 1 further comprising: determining the binding affinity of each of said set of two or more framework re- engineered immunoglobulins to a target antigen; and selecting a framework re-engineered immunoglobulin from said set of two or more framework re-engineered immunoglobulins according to its expression level and binding affinity for said target antigen.

3. The method of claim 1 or 2, wherein said set of amino acid sequences is derived entirely from one or more species different from the species of the parent immunoglobulin.

4. The method of claim 3, wherein said set of amino acid sequences is derived from two or more different species.

5. The method of claim 1 or 2, wherein said set of amino acid sequences is derived entirely from the same species as the parent immunoglobulin.

6. The method of claim 1 or 2, wherein said set of amino acid sequences is derived entirely from a single immunoglobulin isotype.

7. The method of claim 1 or 2, wherein said set of amino acid sequences is derived from two or more different immunoglobulin isotypes.

8. The method of claim 1 or 2, wherein said set of amino acid sequences comprises sequences derived from two or more different immunoglobulin allotypes.

9. The method of claim 1 or 2, wherein said set of amino acid sequences is derived entirely from human immunoglobulins.

10. The method of claim 1 or 2, wherein not all of the FRl, FR2, and FR3 framework segment sequences selected from said set are derived from the same immunoglobulin.

11. A method of enhancing antibody expression comprising: providing a set of amino acid sequences corresponding to the light chain variable region framework segments FRl, FR2, FR3, and FR4 derived from two or more immunoglobulin light chains; replacing the amino acid sequences of the light chain variable region framework segments FRl, FR2, FR3, and/or FR4 of a parent immunoglobulin with sequences of corresponding light chain variable region framework segments selected from said set of amino acid sequences to obtain a set of two or more framework re-engineered immunoglobulins; determining the expression level of each of said set of two or more framework re- engineered immunoglobulins; and selecting a framework re-engineered immunoglobulin from said set of two or more framework re-engineered immunoglobulins according to its expression level.

12. The method of claim 11 further comprising: determining the binding affinity of each of said set of two or more framework re- engineered immunoglobulins to a target antigen; and selecting a framework re-engineered immunoglobulin from said set of two or more framework re-engineered immunoglobulins according to its expression level and binding affinity for said target antigen.

13. The method of claim 11 or 12, wherein said set of amino acid sequences is derived entirely from one or more species different from the species of the parent immunoglobulin.

14. The method of claim 13, wherein said set of amino acid sequences is derived from two or more different species.

15. The method of claim 11 or 12, wherein said set of amino acid sequences is derived entirely from the same species as the parent immunoglobulin.

16. The method of claim 11 or 12, wherein said set of amino acid sequences is derived entirely from a single immunoglobulin isotype.

17. The method of claim 11 or 12, wherein said set of amino acid sequences is derived from two or more different immunoglobulin isotypes.

18. The method of claim 11 or 12, wherein said set of amino acid sequences comprises sequences derived from two or more different immunoglobulin allotypes.

19. The method of claim 11 or 12, wherein said set of amino acid sequences is derived entirely from human immunoglobulins.

20. The method of claim 11 or 12, wherein not all of the FRl, FR2, and FR3 framework segment sequences selected from said set are derived from the same immunoglobulin.

21. A method of enhancing antibody expression comprising providing a first set of amino acid sequences corresponding to the heavy chain variable region framework segments FRl, FR2, FR3, and FR4 derived from two or more immunoglobulin heavy chains; providing a second set of amino acid sequences corresponding to the light chain variable region framework segments FRl, FR2, FR3, and FR4 derived from two or more immunoglobulin light chains; replacing the amino acid sequences of the heavy chain variable region framework segments FRl, FR2, FR3, and/or FR4 of a parent immunoglobulin with sequences of corresponding heavy chain variable region framework segments selected from said first set of amino acid sequences to obtain a set of two or more framework re-engineered immunoglobulin heavy chains; replacing the amino acid sequences of the light chain variable region framework segments FRl, FR2, FR3, and/or FR4 of a parent immunoglobulin with sequences of corresponding light chain variable region framework segments selected from said second set of amino acid sequences to obtain a set of two or more framework re-engineered immunoglobulin light chains; combining said framework re-engineered immunoglobulin heavy chains with said framework re-engineered immunoglobulin light chains to yield a set of two or more re- engineered immunoglobulins; determining the expression level of each of said set of two or more framework re- engineered immunoglobulins; and selecting a framework re-engineered immunoglobulin from said set of two or more framework re-engineered immunoglobulins according to its expression level.

22. The method of claim 21 further comprising: determining the binding affinity of each of said set of two or more framework re- engineered immunoglobulins to a target antigen; and selecting a framework re-engineered immunoglobulin from said set of two or more framework re-engineered immunoglobulins according to its expression level and binding affinity for said target antigen.

23. A framework re-engineered immunoglobulin selected by the method of any one of claims 1 to 22,.

24. A composition, comprising the framework re-engineered immunoglobulin or antigen binding fragment thereof of claim 23 and a carrier.

25. The composition of claim 24, wherein the carrier is a pharmaceutically acceptable carrier.

26. The composition of claim 24, further comprising a second therapeutic agent.

27. A framework re-engineered immunoglobulin or an antigen binding fragment selected by the method of any one of claims 1 to 22, wherein the framework re-engineered immunoglobulin binds to human CD2, CD3, CD4, CD5, CD7, CD8, CDlO, CD14, CD15, CD19, CD20, CD21, CD22, CD23, CD25, CD30, CD33, CD37, CD38, CD40, CD40L, CD46, CD52, CD54, CD72, CD74, CD80, CD126, CD138, CD147, B7, EGF receptor, HER2, HER3, HER4, LFA-I, VEGF, PlGF, ED-B fibronectin, TRAIL-Rl (DR4), TRAIL- R2 (DR-5), HLA-DR, HM 1.24, MUCl, Ia, IL-2, TAC, IL-6, tenascin, CTLA-4, CDIIa, CDIIb, VLA-4, ICAM-I or avb3 integrin.