DK175581B1 - Polynucleotide(s) encoding immunoglobulin molecules - used for efficient prodn. of chimeric human or non-human- or class switched-antibodies - Google Patents

Abstract

Following are claimed (A) a polynucleotide molecule comprising a promoter region in operable linkage to a dicistronic transcription unit, the unit encoding a chimeric heavy chain immunoglobulin or fragment and a chimeric light chain immunoglobulin, (B) a chimeric heavy chain immunoglobulin molecule or fragment linked to a polypeptide secretion signal; (C) a chimeric light chain immunoglobulin molecule linked to a polypeptide secretion signal; (D) a chimeric Fab immunoglobulin molecule, (E) a chimeric Fd immunoglobulin molecule, (F) a polynucleotide molecule encoding the immunoglobulin molecule of (D) or (E), (G) a polynucleotide molecule encoding the Fd fragment of a chimeric heavy chain immunoglobulin comprising a restriction site at or downstream from a stop codon.

Description

DK 175581 B1

The present invention relates to DNA recombination methods for the production of immune and obulins, genetic sequences encoding them, as well as methods for obtaining such sequences.

5 The use of cell-to-cell fusion to produce monoclonal antibodies by Kohier and Milstein (Nature (London) 256: 495,1975) has given rise to a revolution in biology that is substantially similar to the invention of recombinant DNA cloning. Monoclonal antibodies prepared from hybridomas have already been widely used in clinical diagnoses and basic scientific studies. The use of 10 human B-cell hybridoma-produced monoclonal antibodies is very promising for the clinical treatment of cancer, viral and microbial infections, B-cell immunodeficiencies with diminished antibody production, and other disorders and disorders of the immune system.

Unfortunately, yields of monoclonal antibodies from human hybridoma cell lines are relatively low (1 µg / ml in human x human compared to 100 µg / ml in mouse hybrids) and production costs are high for antibodies produced in large scale human tissue cultures. The mouse x mouse hybridomas, on the other hand, are suitable as they produce copious amounts of protein and these cell lines are more stable than the 20 human lines. However, repeated injections of "foreign" antibodies, such as a mouse antibody, into humans can lead to harmful hypersensitivity reactions.

Therefore, there has recently been exploration of the possibility of producing antibodies which have the advantages of monoclonal antibodies from mouse-mouse hybridomas and yet the species-specific properties of human monoclonal antibodies.

Another problem encountered by immunologists is that most human monoclonal antibodies (i.e., antibodies that have human recognition properties) obtained in cell culture are of the IgM type. However, when it is desired to obtain human monoclonal antibodies of the IgG type, it has been necessary to use such techniques as cell sorting to separate the few cells which have switched to produce antibodies of DK 175581 B1 2.

The IgG type or other type from the majority producing IgM type antibodies.

Therefore, there is a need for a more accessible method of switching antibody classes for any given antibody with a predetermined or desired antigen specificity.

5

The present invention bridges hybridoma technology with monoclonal antibody technology, and discloses a rapid and effective method as well as products thereof for improved production of chimeric human / non-human antibodies or for "class-shifted" antibodies.

10

Approaches to the problem of chimeric antibody production have been described by various authors.

S. L. Morrison et al., Proc. Natl. Acad. Sci., USA, 81: 6851-6855 (November 1984), 15 discloses the preparation of a mouse / human antibody molecule which has defined antigen binding specificity and which is prepared by combining the variable region genes from a mouse antibody producing myeloma cell line with known antigen binding specificity to human constant immunoglobulin region genes using recombinant DNA technique. Chimeric genes were constructed in which the heavy chain variable region neon from the myeloma cell line S107 connected well to the human IgG1 or IgG2 heavy chain constant region exons, and the light chain variable region exon from the same myeloma cell to the human kappa kædeexon. These genes were transfected into mouse myeloma cell lines, and transformed cells producing chimeric mouse human antiphosphocholine antibodies were thus developed.

25 S. L. Morrison et al., EP Publication No. 173494 (published March 5, 1986), discloses chimeric "receptors" (e.g., antibodies) with variable regions of one species and constant regions of another. Mention is made of the use of cDNA cloning to construct the genes, although details are not provided regarding cDNA cloning or priming (see pages 5, 7 and 8).

Boulianne et al., Nature, 312: 643 (December 13, 1984) also produced antibodies consisting of variable mouse regions linked to constant human regions. They constructed immunoglobulin genes in which the DNA segments encoding mouse variable regions specific for the hapten trinitrophenyl (TNP) were linked to 5 segments encoding human constant mouse and kappa regions. These chimeric genes were expressed as functional TNP-binding chimeric IgM.

See Munro, Nature, 312: 597 (December 13, 1984), Dickson, Genetic Engineering News, 5, No. 3 (March 1985), or Marx, Science, 229: 455 (August 1985) for a comment on the work done by Boulianne et al. and Morrison et al.

M.S. Neuberger et al., Nature, 314: 268 (March 25, 1986) also constructed a chimeric heavy chain immunoglobulin in which a DNA segment encoding a mouse variable region specific for the hapten 4-hydroxy-3-nitrophenacetyl (NP) was linked to a segment encoding the human epsilon region. When this chimeric gene was transfected into the J558L cell line, an antibody was bound to the NP hapten and had human IgE properties.

M.S. Neuberger et al. has also published work showing the production of cell lines secreting hapten-specific antibodies in which the Fc portion has been replaced either with an active enzyme moiety (G. Williams and MS Neuberger, Gene 43: 319, 1986) and with a polypeptide having c-myc antigenic determinants (Nature, 312: 604, 1984).

M. Neuberger et al., PCT publication WO 86/01533 (published March 13, 1986) also discloses the production of chimeric antibodies (see page 5) and suggests among the many uses of the technique the term "class change" (see page 6).

M. Taniguchi in European Patent Publication No. 171,496 (published February 19, 1985) discloses the preparation of variable antibodies with variable regions with tumor specificity derived from laboratory animals and constant regions derived from humans. The corresponding heavy and light chain genes corresponding to the genomic form are expressed in mammalian cells.

S. Takeda et al., Nature, 314: 452 (April 4, 1985) have described a potential method for constructing chimeric immunoglobulin genes which have inconsistencies removed by the use of a retroviral vector. However, an unexpected spl est sedon site caused the deletion of the V region leader sequence. Therefore, this experiment did not yield complete chimeric antibody molecules.

10 S. Cabilly et al., Proc. Natl. Acad. Sci., USA, 81: 3273-3277 (June 1984) discloses plasmids that direct the synthesis in E. coli of heavy chains and / or light chains of antibody to carcino embryo antigen (CEA). Another plasmid was constructed to express a truncated form of heavy chain (Fd ') phage in E. coli. Functional CEA binding activity was obtained by in vitro reconstitution in E. coli extracts of a 15-part light chain heavy chain.

S. Cabilly et al. in European Patent Application 125023 (published November 14, 1984) discloses chimeric immunoglobulin genes and their putative products as well as other modified forms. On pages 21, 28 and 33, cDNA cloning and priming are discussed.

M. A. Boss, European Patent Application 120694 (published October 3, 1984) shows expression in E. coli of non-chimeric immunoglobulin chains with 4-nitrophenyl specificity.

There is a broad description of chimeric antibodies but no details (see page 9).

25 C.R. Wood et al., Nature, 314: 446 (April 1985) describe plasmids which direct the synthesis of mouse antibody proteins against NP in yeast. Heavy mu chain antibody proteins were found to be glycosylated in the yeast cells. When both heavy and light chains were synthesized in the same cell, some of the protein was assembled into functional antibody molecules, as detected by anti-NP binding activity in soluble protein prepared from yeast cells.

A. Alexander et al., Proc. Night. Acad. Sci. USA, 79: 3260-3264 (1982) discloses the preparation of a cDNA sequence encoding an abnormally short human Ig gamma heavy chain (OMM gamma3 HCD serum protein) containing a 19-amino acid leader followed by the first 15 residues of V region. Extensive internal deletion removes the rest of the 5 V region and the entire CH1 region. This is cDNA which encodes an internal deleted molecule.

T. W. Dolby et al., Proc. Natl. Acad. Sci., USA, 77: 6027-6031 (1980) describe the preparation of a cDNA sequence and recombined plasmids containing the same coding for human mu and kappa immunoglobulin polypeptides. One of the recombined DNA molecules contained codons for part of the CH3 constant region region and the entire 3 'non-coding sequence.

M. Seno et al., Nucleic Acids Research, 11: 719-726 (1983), describe the preparation of a cDNA sequence and recombined plasmids containing the same coding for part of the variable region and the entire constant region of the human heavy IgE chain (epsilon chain).

T. Kurokawa et al., Ibid., 11: 3077-3085 (1983), shows the construct using 20 cDNA of three expression plasmids encoding the constant portion of the human heavy IgE chain.

F. T. Lieu et al., Proc. Night. Acad. Sci., USA, 81: 5369-5373 (September 1984) discloses the preparation of a cDNA sequence and recombined plasmids containing the same which encode approximately 25 2/3 of the CH2 and entire CH3 and CH4 regions of human heavy IgE chain.

Y. Tsujimoto et al., Nucleic Acids Res., 12: 8407-8414 (November 1984) disclose the preparation of a human V lambda cDNA sequence from an Ig lambda producing 30 human Burkitt lymphoma cell line using a cloned constant region gene as a primer for cDNA synthesis.

DK 175581 B1 6 J. Murphy, PCT publication WO 83/03971 (published November 24, 1983) discloses hybrid proteins made from fragments comprising a toxin and a cell-specific ligand (which may be suggested to be an antibody).

5

Tan et al., J. Immunol. 135: 8564 (November, 1985) obtained expression of a human mouse chimeric immunoglobulin gene gene after transfection into mouse myeloma cells.

10 P. T. Jones et al., Nature 321: 552 (May 1986) constructed and expressed a genome construct in which variable region CDR regions from a monoclonal mouse antibody were used to replace the corresponding regions of a human antibody.

L. K. Sun et al., Hybridoma 5 suppl. 1 SI7 (1986) discloses a chimeric human / -15 mouse antibody with potential tumor specificity. The chimeric heavy and light chain genes are genome constructs and expressed in mammalian cells.

Sahagan et al., J. Immun. 137: 1066-1074 (August 1986) discloses a chimeric antibody with specificity to a human tumor-associated antigen for which the genes are assembled from genome sequences.

For a more recent overview of the area, see also S.L. Morrison, Science 229: 1202-1207 (September 20, 1985) and V. T. Oi et al., BioTechniques 4: 214 (1986).

The Oi et al description is relevant as it states that the preparation of chimeric antibodies from cDNA constructs in yeast and / or bacteria is not necessarily advantageous.

See also the commentary on page 835 of Biotechnology 4 (1986).

The present invention provides a novel method for producing genetically engineered antibodies with desired variable region specificity and constant response properties by gene cloning and expression of light and heavy chains. The cloned immunoglobulin products can be prepared by expression in genetically engineered organisms.

5 The use of chemical gene synthesis, recombined DNA cloning and production of specific immunoglobulin chains in various organisms provides an effective solution for the efficient production of large-scale human monoclonal antibodies.

According to the present invention, there is provided a method for producing an immunoglobulin molecule with heavy and light chains or fragments thereof, which are linked such that the parent molecule exhibits desired binding and recognition properties, the method comprising (a) cultivating a transformed or transfected prokaryotic host under appropriate conditions such that heavy chain immunoglobulin and light chain immunoglobulin or fragments thereof are expressed as a result of transcription of polynucleotide sequences in the host, wherein the polynucleotide sequences encode at least one functionally active region of a variable region of an antibody, and prokaryotic secretion signal peptide, wherein the variable region under expression is operably linked 20 at its N-end to a prokaryotic secretion signal peptide to enable secretion from the host; and (b) obtaining the immunoglobulin chain or fragment of an immunoglobulin chain which, prior to secretion from the host, was thus initially bound to the signal peptide in the form of a heavy and light chain immunoglobulin molecule or the parent molecule exhibiting the desired binding and recognition properties.

The invention can be used to provide individual chimeric immunoglobulin chains as well as completely assembled molecules and immunoglobulin phagments (such as Fab) having human constant regions and non-human variable regions, both variable regions having the same binding specificity.

Among other immunoglobulin chains and / or molecules that can be provided according to the invention are: (a) a fully functional immunoglobulin molecule comprising: (i) two identical chimeric heavy chains comprising a non-human variable region and a human constant region, and 1 ° (ii) two identical fully (i.e., non-chimeric) human light chains, and (b) a fully functional immunoglobulin molecule comprising: (i) two identical chimeric heavy chains comprising a non-human variable region and a human constant region, and (ii) two identical completely (i.e., non-chimeric) non-human light chains; and (c) a monovalent antibody, i. a fully functional immunoglobulin molecule comprising: (i) two identical chimeric heavy chains comprising a non-human variable region and a human constant region, and (ii) two different light chains, only one having the same specificity as the variable region of the heavy chains. The resulting antibody molecule binds only to one end thereof and is therefore incapable of divalent binding, and (d) an antibody having two different specificities, viz. a fully functional immunoglobulin molecule comprising: (i) two different chimeric heavy chains, the first comprising a non-human variable region and a human constant region, and the second comprising a different non-human variable region and a human constant. and (ii) two different chimeric light chains, the first comprising a non-human variable region with the same specificity as the variable region of the first heavy chain and a human constant region, and the second comprising a non-human variable region. with the same specificity as the second heavy chain variable region and a human constant region.

The antibody molecule obtained binds to two different antigens.

The invention also provides the preparation of functionally active chimeric immunoglobulin fragments secreted by prokaryotic hosts or fully folded and reassembled chimeric immunoglobulin chains.

Genetic sequences, especially cDNA sequences encoding the aforementioned combinations of chimeric or non-chimeric chains, are also provided herein.

20

The use of cDNA sequences is particularly advantageous compared to genome sequences (which contain introns) in that cDNA sequences can be expressed in bacteria or other hosts lacking RNA splicing systems.

Among preferred specific antibodies are those which have specificities for cancer-related antigens.

FIG. Figure 1 shows DNA rearrangements and expression of heavy immunoglobulin my and gamma gastric genes. This is a schematic representation of the human heavy chain gene complex, not shown on a scale. Formation of heavy chain variable V region occurs at the junction of VH, D and JH gene segments. This produces an active mosquito.

10 DK 175581 B1

Another type of DNA rearrangement called "class shift" relocates the unified VH, D, and JH region from the constant my-C region to another heavy chain C region (shift to gamma is shown here). The diagram shows that the J region is a common feature of all expressed heavy chain genes. The J region is also a common feature of expressed light chain 5 genes.

FIG. Figure 2 shows the known nucleotide sequence of human J regions and mouse J regions. Consensus sequences for the J regions are shown below, the actual sequences.

The oligonucleotide sequence under the mouse kappa J region consensus sequence is a universal immunoglobulin gene (UIG) oligonucleotide used in the present invention.

FIG. Figure 3 shows a scheme showing the use of the UIG oligonucleotide primer for the synthesis of cDNA which is complementary to the variable region of immunoglobulin messenger RNA or the use of oligo-dT as a primer for cDNA synthesis followed by an in vitro mutagenesis.

FIG. Figure 4 shows the synthesis and analysis of human IgG1 genes comprising three isolated clones (A.b), one of which (pGMH-6) is used as a cloning vector (B). A 1.5 kb deletion of pBR322 sequence between BamHI and PvuII is shown. Not to scale.

FIG. 5 shows the cloning vector pQ23, a modified pBR322 suitable for cDNA cloning at the KpnI site. This vector also contains the appropriate restriction enzyme sites BglII and SalI. Not to scale.

FIG. 6 shows in A. the synthesis and analysis of human light kappa chain genes. The figure also shows in B. (not to scale) construction of a human Ci <region cloning vector pING2001.

30

II

DK 175581 B1

FIG. Figure 7 shows primers designed for immunoglobulin V region synthesis. (A) shows the heavy chain J-C regions and primers. A DNA version of each mouse J heavy region is shown directly over primers designed from that sequence. Muse-J regions are 5 'to 3', left to right, while primers are 3 'to 5', left to right. Prime names are shown in parentheses and the number of nucleotides (N) and the number of non-matching with each JH region are indicated on the right. Primers introducing a BstEII site are underlined. (B) shows the light chain J regions and primers. Same as for (A) except for light chains. Primers designed to introduce a BgLII site are underlined as the BclI site present in pING2016Es. (C) shows variable mouse region consensus UIG primers. The actual primer sequence is shown below that consensus sequence. The human CK-HindIII vector pGML60 is shown below. (D) shows a mouse gamma-2α-J / C compound primer.

FIG. Figure 8 shows the synthesis of heavy chain V region modulator genes using oligonucleotide primed cDNA synthesis. Not to scale.

15

FIG. Figure 9 shows the construction of hybrid mouse / human immunoglobulin genes. Panel A shows the construction of a heavy chain gene. Dashed regions show C region modules, while shaded or black regions show V region modules. Not to scale.

FIG. 10 shows the construction of cDNA cloning expression pendulvectors for mammalian cells. The vectors pING2003 and pING2003E are derived from pL1, pUC12, pSV2-neo and M8-otRX12. Dashed regions show mouse heavy chain promoter DNA, shaded regions show SVL-40 DNA from pL1 and double-shaded regions show SV40 DNA from pSV2 neo. In the vectors pING2003 and pING2003E, thick lines show pBR322 DNA from 25 pSV2 neo, while thin lines show pUC12 DNA. Arrows show locations and directions of SV-40 early region promoters and show a complete SV-40 intron sequence. Not to scale.

FIG. 11 shows the construction of the heavy chain expression plasmid pING2006E. Arrow 30 shows SV-40 promoter locations and directions for transcription. Shaded and black areas show mouse V region modules, while dashed areas show human C region modules. Not to scale.

FIG. Figure 12 shows the structure of chimeric anti-hepatitis heavy chain genes in the expression plasmids pENG2006E and pING2012E. Panel A shows the structure of chimeric mouse / human anti-hepatitis heavy chain genes. The structure of human IgG1 mRNA and cDNA is shown in A.a. The human heavy constant chain region cDNA clone pGMH-6 and the mouse heavy variable chain region cDNA clones pBS13-1 and pJ3-11 were used to prepare the hybrid gene used in pING2006E. Shaded gene blocks show mouse variable region sequences, while open gene blocks show human constant IgG1 region sequences. Panel B shows the nucleotide sequence of the anti-hepatitis B heavy chain variable region in pING2006E and pING2012E. pING2012E was constructed by first inserting a BglII site at the SalI site into pING1202 (see Fig. 16) to form pING1202BglII. The chimeric heavy chain gene from this plasmid was inserted into the expression vector 15 pING2003E, resulting in pING2012E. pING2012E is different from pING2006E in the region immediately upstream of the initiator ATG. Underlined nucleotides show human J region sequences from the cDNA clone pGMH-6. Starred amino acid 117 shows a single change in this site between mouse and human sequence (Ala to Ser) introduced into the chimeric gene J region. Sequencing was performed according to Sanger's method on plasmid template (open circle) and M13 template (closed circle).

FIG. Figure 13 shows in panel A the J-C linking region nucleotide sequence in light chain clones derived from pING2001 (pMACK-3, pING2013E, pING2007E, pING2010E-gpt and 25 pING2014E-gpt). The J region sequence, which originates from pK2-3, is labeled human JK4. The G nucleotide, which is not predicted by genome sequencing, is marked with an asterisk. The oligonucleotide primer (K2-4BCLI) used to modify this sequence is shown under the human JK4 sequence. Panel B shows the site-directed mutagenesis method used to prepare pING2016E-gpt. Not to scale.

30 13 DK 175581 B1

FIG. 14. Gene copy number of the transfected sequences in two transformants. nDNA from 2AE9, 2BH10 was digested with the enzymes shown. The concentration of DNA is titrated down the lanes by the amount indicated above them. The probe contains human C-gamma-1 sequences (pmvHc24 Apal-BamHI). The reference is germ line or 5 GM2146 nDNA digested with Apal. The 3 'Apal site is 2 bp on the other side of poly (A) addition (3).

FIG. 15 shows the nucleotide sequence of the V region of the L6-VH cDNA clone pH3-6a. The sequence was determined by the dideoxy termination method using M13 sub-10 clones of gene fragments (shown below). Open circles show amino acid residues confirmed by peptide sequence. A sequence homolog with DSP.2 in the CDR3 region is underlined.

FIG. 16 shows the nucleotide sequence of the V region of the L6-VK cDNA clone pL3-12a. The oligonucleotide primer used for site-directed mutagenesis is shown under the JK5 segment.

Open circles show amino acid residues confirmed by peptide sequence.

FIG. 17 shows the construction of chimeric L6-VH as well as human C-gamma-1 expression plasmids. Panel (a) shows the sequences of the BAL-31 deletion clones M13mpl9-Cl-del-ta 4 (Cl-delta 4) and M13mpl9-Cl-delta 21 (Cl-delta 21). The S 'end of the cDNA clone pH3-6a is indicated by an arrow. M13 sequences are underlined. The oligonucleotide primer used for this experiment is H3-6a (5'- GACTGCACCAACTGG-3 '), which primes in FRI near the fully developed N end. Panel (b) shows the site-directed mutagenesis strategy of 1 μg of Cl-delta 4 and Cl-delta 21, each joined together with 20 ng of the 31-mer oligonu-clototide MJH2-ApaI. Complementary strand synthesis with the Klenow fragment of 25 DNA polymerase was carried out at room temperature for 30 minutes, then 15 ° C for 72 hours. Transfected phage plaques were adsorbed on nitrocellulose fixed with NaOH and hybridized to 32 P-labeled MJH2-ApaI oligonucleotide at 65 ° C, 18 h, in 4xTBS (0.6 M NaCl, 0.04 M Tris-HCl (pH -value 7.4), 0.004 M EDTA) plus 10% dextran sulfate. Final washing of the filters was at 65 ° C 4xSSPE, 0.1% SDS for 15 minutes (Maniatis, 30 T., et al., Molecular Cloning: A Laboratory Manual, 1982). Positive plaques were shown by overnight exposure to Kodak XAR films and were directly sampled for growth and restriction enzyme analysis of RF DNA. Incorrectly matched between the MJH2 ApaI oligonucleotide and mouse CH1 is shown, resulting in the coding changes shown during the oligonucleotide. Panel (c) shows the strategy of replacing the resident VH in the chimeric expression plasmid pING2012 with each of the mutagenized 5 L6-VH modules to form pING2111 and pING2112.

FIG. Figure 18 shows the construction of the chimeric L6 expression plasmid pING2119. The SalI to -BamHI phage fragment from pING2100 is identical to the SalI to -BamHI-A fragment of pING2012E.

10

FIG. 19 shows the modification of the VK gene and its use in construction of chain and heavy chain and light chain expression plasmids.

(a) Deletion of the oligo-d [GC] segment 5 'for VK in L6. The oligonucleotide is a 22-mer 15 and contains a SalI site. The 3 incorrectly matched are shown. The VK gene, after mutagenesis, is linked as a SalI-HindIII fragment to the human C-K module. The expression plasmid thus formed is pING2119.

(b) pING2114, a heavy as well as light chain expression piasmid. Expression plasmid 20 pfNG2114 contains the heavy chimeric L6 chain gene from pING2111 and the chimeric light chain from pING2114 (highlighted line).

FIG. 20 shows a summary of the sequence changes made in the construct of the chimeric L6 antibody expression plasmids. Underlined residues in the 25 5 'untranslated region are derived from the cloned mouse kappa and heavy chain genes.

Circular residues around the V / C boundary result from mutagenesis operation to form restriction enzyme sites in this region. Residues shown with small circles above them in the heavy chain L6 chimera are also a result of mutagenesis. They are quiet changes.

15 DK 175581 B1

FIG. 21 shows the 2H7 VH sequence. The VH gene contains JH1 sequences and DSP.2 sequence elements. Small circles of amino acid residues are those that match peptide sequences.

FIG. 22 shows the 2H7 VL sequence. The VK gene contains JK5 sequences. A 22-mer oligonucleotide was used to place a SalI site 5 'of the ATG initiator codon.

Small circles above the amino acid residues are those that match peptide sequences.

FIG. 23 shows the chimeric immunoglobulin gene expression plasmids with the 10 2 H7 specificity. A gene plasmids are pING2011 (VH, neo), pING2106 (VK, neo) and pING2107 (VK, gpt). The others are two-gene plasmids. Their construction involved the ligation of the larger Ndel phage fragments of pING2101 and pING2107 to linearized pING2106 partially digested with Ndel. pHL2-1 and pHL2-26 were obtained from pING2101 and pING2106, pLL2-25 was obtained from pING2107 and pING2106.

15

FIG. 24 shows a summary of nucleotide changes introduced into VH and VK in the construction of the chimeric plasmids. The associated VH and VK nucleotide residues at the 5 'end are underlined. Circular residues in the J-C junctions derive from the human C-modules.

20

FIG. Figure 25 shows the strategy used to combine the fully developed L6 chimeric light chain sequence to the yeast invertase signal sequence and truncated PGK promoter. The open double line represents yeast invertase signal sequence DNA.

The filled double line represents yeast PGK DNA; -> represents the PGK promoter; - | represents the PGK terminator; RF = replicative form.

pING1225 was derived by associating human CK DNA with the PGK promoter. pING1149 was derived by associating the yeast invertase signal sequence with the yeast PGK promoter. (A) shows the strategy of introduction by in vitro mutagenesis of an AatII site at the signal sequence process site. (B) shows the DNA sequence of the 30 single-stranded mutagenesis primer and the corresponding non-mutagenized DNA sequence. (C) shows the strategy used to construct a plasmid 16 containing the fully developed light chain sequence joined to the invertase signal sequence and truncated PGK promoter.

FIG. Figure 26 shows the strategy used to combine the fully developed L6 chimeric heavy chain sequence with the yeast invertase signal sequence and truncated PGK promoter. pING1288 contains the chimeric heavy chain gene together with the variable region of the mouse monoclonal 2H7 antibody (see Example IV). All symbols are as defined in the description of FIG. 25. (A) shows the strategy of introduction by in vitro mutagenesis of an Sstl site into the signal sequence process site. (B) shows the DNA sequence 10 of the single-stranded mutagenesis primer and the corresponding non-mutagenized DNA sequence. (C) shows the strategy used to construct a plasmid containing the fully developed heavy chain sequence joined to the invertase signal sequence and truncated PGK promoter.

FIG. 27 shows the strategy used to remove non-yeast 3 'untranslated DNA sequences from the L6 chimeric light chain gene and to construct a plasmid containing the light chain gene linked to the invertase signal sequence and truncated PGK promoter, wherein all sequences are either are known by DNA sequence analysis or have been found to be functional. pBR322NA is derived from pBR322 by deletion of 20 DNA from Ndel to Aual. Symbols are as defined in the description of FIG. 25th

FIG. 28 depicts the strategy used to remove non-yeast 3 'untranslated DNA sequence from the L6 chimeric heavy chain gene and construct a plasmid containing the heavy chain gene associated with the invertase signal sequence and truncated PGK promoter in which all sequences are either are known by DNA sequence analysis or have been found to be functional. Symbols are as defined in the description of FIG.

25th

FIG. 29 shows the strategy used for cloning the L6 chimeric light chain gene 30 associated with the invertase signal sequence and truncated PGK promoter in yeast E, transformants. Symbols are as defined in the description of FIG. 25th

FIG. Figure 30 shows the strategy used for cloning the L6 chimeric heavy chain gene 5 associated with the invertase signal sequence and truncated PGK promoter in yeast E. coli shuttle vectors containing PGK transcription termination polyadenylation signal, yeast replication sequences and genes for selection of transformants. Symbols are as defined in the description of FIG. 25th

FIG. 31 shows a schematic diagram of the structure of human IgG1.

FIG. 32 (A) shows the strategy used to insert a stop codon and BclI site into the "hinge region" of human gamma 1. (B) shows the DNA sequence of the single-stranded primer used for in vitro mutagenesis of 15 gamma. 1 hinge region and the corresponding non-mutagenized sequence. Vertical arrows represent inter-chain disulfide bonds. Symbols are as defined in the description of FIG. 25th

FIG. 33 shows the strategy used to associate the L6 chimeric heavy chain gene 20 containing a hinge region (Fd chain) stop codon with the yeast invertase signal sequence and truncated PGK promoter. Symbols are as defined in the description of FIG. 25th

FIG. 34 depicts the strategy used to remove non-yeast, 3 'untranslated 25 sequences from the L6 chimeric Fd chain and construct a plasmid containing the Fd chain linked to the invertase signal sequence and truncated PGK promoter wherein all sequences are either is known by DNA sequence analysis or has been shown to be functional. Symbols are as defined in the description of FIG. 25th

FIG. 35 shows the strategy used for cloning the L6 chimeric Fd chain gene associated with the invertase signal sequence and truncated PGK promoter in yeast E.

18 DK 175581 B1 coli shuttle vectors containing PGK transcription termination polyadenylation signal, yeast replication sequence and genes for selection of transformants. Symbols are as defined in the description of FIG. 25th

FIG. 36 (A) shows the nucleotide sequence surrounding the N-terminus of the Erwinia caratovora pelB gene (Lei, S.P. et al., J. Bacteriol. (1987, reprinted)). The Ndel and Haelll sites used for cloning are shown. The arrows show the leader pep at the dot cleavage site for pectal lysis. (B) shows the cloning strategy for constructing the pelB leader cassette. pSS1004 contains a 1.9 kb Dral phage fragment cloned into the 10 SmaI site of pUC8. Symbols are as defined in the description of FIG. 39th

FIG. Fig. 37 shows the construction of delayed chain expression plasmids pRR177-8, pRR180, pRR190 and pRR191. In addition to the plasmids described in the text, M13mp18 and pIT 181 were used. pIT181 contains the fully developed light chain gene joined 15 directly after the ATG initiation codon of the araB gene in pIT2 (see Fig. 40).

FIG. 38 shows the construction of Fd expression plasmids pRR178-5, pRR.186 and pRR196.

FIG. Figure 39 shows the restriction maps for the light chain and the Fd gene cassette in pFK10, pFK10, pFK102, pFK103 and pFK104. These plasmids were constructed as described in the text using plasmids shown in FIG. 37 and 38. The arrows show the direction of transcription from the lac promoter. E. caratovora and eukaryotic non-coding sequences are shown as open bars. the pelB leader sequence is double-stranded and the 25-filled bar represents the antibody genes Fd and light chain (K).

FIG. 40 (A) shows the construction of a vector for arabinose inducible Fab expression.

Plasmid plT2 (Mason and Ray, Nucl. Acids. Res. 14: 5696 (1986)) is a 6431 bp plasmid encoding the araC gene, araB promoter and a portion of the araB gene from 30 pING1 (Johnston, S.A. et al., Gene 34 34: 134 (1985)) in a derivative of pBR322. Et 19 DK 175581 B1

NcoI site has been generated by the ATG initiation codon of the araB gene. (B) shows the introduction of the lacigen in pFK102.

In general, antibodies are composed of two light and two heavy chain molecules. Light and heavy 5 chains are divided into areas of structural and functional homology. The variable regions in both light (VL) and heavy (VH) chains determine recognition and specificity. The constant region regions of light (CL) and heavy (CH) chains confer important biological properties such as antibody chain association, secretion, transplacental mobility, complement binding and the like.

10

A complex series of events leads to immunoglobulin gene expression in B cells. The V-region gene sequences conferring antigen specificity and binding are located in separate germ line gene segments, called VH, D and JH; or VL and JL. These gene segments are linked by 15 DNA rearrangements to form the complete V regions expressed in heavy and light chains, respectively (Fig. 1), the rearranged associated (VL-JL and VH-D-JH) V segments encode then for the completely variable regions or antigen binding regions of light and heavy chains, respectively.

20 DEFINITIONS

Certain terms and terms are used throughout the description and claims. The following definitions are stated for the purpose of clarity and consistency.

1. Expression vector - a plasmid DNA containing necessary regulatory signals for the synthesis of mRNA derived from gene sequences which can be inserted into the vector.

2. Module vector - a plasmid DNA containing a constant or variable region gene module.

3. Expression plasmid - an expression vector containing an inserted gene, such as a chimeric immunoglobulin gene.

4. Gene cloning - synthesis of a gene, insertion into DNA vectors and identification by hybridization or the like.

5. Transfection - transfer of DNA into mammalian cells.

6. Promoter region - a nucleotide sequence that gives a cell the necessary regulatory sequences to express an operably linked cistron or operon.

7. Secretion signal - a polypeptide present at the N-terminus of a chimeric immunoglobulin chain suitable for contributing to the secretion of the chain to that surrounding the host. Also called "leader peptide" or "leader".

15

GENETIC PROCESSES AND PRODUCTS

The invention provides a novel method for cloning and producing human antibodies of the desired specificity. In general, the method combines five elements: (1) Isolation of messenger RNA (mRNA) ffa B cell hybridoma lines producing monoclonal antibodies to specific antigens, cloning and cDNA production therefrom.

(2) Preparation of universal immunoglobulin gene (UIG) oligonucleotides suitable as primers and / or probes for cloning the variable region gene segments in the light and heavy chain mRNA from ffa specific human or non-human hybridoma cell lines and cDNAs. production thereof.

(3) Preparation of constant region gene segment modules by cDNA preparation and cloning or genome gene preparation and cloning.

21 DK 175581 B1 (4) Construction of complete heavy or light chain coding sequences by binding of the cloned specific variable region gene immunoglobulin segments from part (2) above to cloned human constant region gene segment modules.

5 (5) Expression and production of light and heavy chains in selected hosts comprising prokaryotic and eukaryotic hosts, either by separate fermentations followed by collection of antibody molecules in vitro or by production of both chains in the same cell.

The invention utilizes cloned hybridoma B cell lines that produce monoclonal antibodies of defined specificity for isolating mRNA for cDNA cloning. Because many lymphoid cell lines contain highly active nucleases that degrade mRNA during isolation, the invention utilizes mRNA preparation methods specifically developed for the isolation of intact mRNA from cells and tissues containing 15 active nucleases. One such method which provides complete RNA preparations by cell or tissue destruction is an ethanol perchlorate dry ice mixture which reduces nuclease action (Lizardi, P. M. et al., Anal. Biochem., 98: 116 (1979)). This method yields intact transducible mRNA.

Other methods which have been used according to the invention include extraction of cells with lithium chloride and urea (Aufffay, C. and Rougeon, F., Eur. J. Biochem., 107: 303 (1980)) or guanidine thiocyanate (Chirgwin, J.M. et al., Biochemistry, 18: 5294 (1979)) for the preparation of complete RNA.

A common feature of all expressed light and heavy immunoglobulin chain genes and messenger RNAs is the so-called J region (i.e., the junction region, see Fig. 1). Heavy and light chain J regions have different sequences, but a high degree of sequence homology (more than 80%) occurs within the heavy JH or kappa light chain regions. The invention provides consensus sequences of light and heavy chain J regions useful in the design of oligonucleotides (referred to herein as UIGs) for use as primers or probes for cloning light or heavy immunoglobulin chain mRNAs. or genes (Figs. 2 or 7). Depending on the nature of the design of a particular UIG, it may be able to hybridize to all immunoglobulin mRNAs or genes containing a single specific J sequence, such as UIG-MJH3, which detect only mouse JH3 sequences (FIG. 7).

5

Another use of a particular UIG probe may be hybridization to light chain or heavy chain mRNAs with a specific constant region, such as UIG-MJK, which detects all mouse JK-containing sequences (Fig. 7). The UIG design may also comprise a sequence for introducing a restriction enzyme site into the cDNA copy of an immunoglobulin (see Figure 7). The invention may e.g. utilize chemical gene synthesis to produce the UIG probes for the cloning of V regions in immunoglobulin mRNA from hybridoma cells producing monoclonal antibodies with desired antigen specificities.

A multi-step method is used to generate complete 15 V + C region cDNA clones from hybridoma cell light and heavy chain mRNAs. In the first step, the invention utilizes UIG probes as "primers" for reverse transcriptase copying of the complete V region and leader coding sequences of heavy and light chain mRNAs (Fig. 3). The complementary strand of the primer-extended cDNA is then synthesized and this double-stranded cDNA is cloned into appropriate cDNA cloning vectors such as pBR322 (Gubier and Hoffman, Gene 25: 263 (1983)) or pQ23 (Fig. 5, Maniatis, T., et al., Molecular Cloning: A Laboratory Manual, Cold Spring Harbor Laboratory Publications, New York, page 224 (1982)).

Clones are screened for specific hybridization with UIG oligonucleotide probes. Positive heavy and light chain clones identified by this screening procedure are mapped and sequenced to select those containing V region and leader coding sequences.

An alternative approach is to use cDNA clones using oligo-dT as a primer followed by selection of light and heavy chain clones by standard hybridization methods.

23 DK 175581 B1

Another step utilizes cloning of C region gene segments to form light and heavy chain module vectors. In one method, human heavy and light chain immunoglobulin rriRNA cDNA clones are prepared. These cDNA clones are then converted into C region modulator vectors by site-directed mutagenesis to place a restriction site 5 at a desired location near a boundary to the constant region. An alternative method utilizes genome C region clones as a source for C region module vectors.

A third cDNA cloning step involves the formation of complete light and heavy chain coding sequences with linked V and C regions. The cloned V region segments formed as described above are cut and ligated to light or heavy chain C region modulator vectors. Eg. For example, one can clone the complete human kappa light chain C region and the complete human gamma 1-C region. Furthermore, one can modify a human gamma-1 region and introduce a termination codon, thereby obtaining a gene sequence encoding the heavy chain portion of a Fab molecule.

The coding sequences having operationally bound V and C regions are then transferred to appropriate expression systems for expression in appropriate hosts, prokaryotes or eukaryotes. Operationally bound means in-frame association of coding sequences to obtain a continuous translatable gene sequence without alterations or perturbations of the triplet read frame.

A particular advantage of using genetic cDNA sequences according to the invention is the fact that they continuously encode immunoglobulin chains, either heavy or light. By "continuous" is meant that the sequences do not contain introns (i.e., are not genome sequences, but rather, as they derive from mKNA by reverse transcription, are sequences of adjacent exons). This property of the cDNA sequences of the invention allows them to be expressed in prokaryotic hosts such as bacteria or in lower eukaryotic hosts such as yeast.

30 24 DK 175581 B1

Another advantage of cDNA cloning methods is the ease and simplicity of obtaining V region gene modules.

The term "non-human" as used in the invention is intended to include any animal other than a human in which an immune response can be produced which then leads to useful B cells resulting in corresponding hybridomas, or B-cells. cell clones obtained by viral transformation and the like. Such animals commonly include rodents such as the mouse or rat. Because of how easy it is to obtain them and their high availability, the mouse is currently the preferred 10 non-human animal. Thus, mouse-mouse hybridomas are utilized as the preferred sources for heavy and light chain variable regions.

Preferably, the invention provides whole V and / or C region cDNA sequences. This means that the sequences encode substantially operable V and / or 15 C regions without missing any significant structural parts thereof.

The terms "constant" and "variable" are used to functionally designate such regions of the immunoglobulin chain either heavy or light chain, which encode properties and traits of the variable and constant regions of naturally occurring non-chimeric antibodies. As noted, it is not necessary that the complete coding region for variable or constant regions be present, as long as a functional operable region is present and available.

There are a variety of sources of hybridomas available for mRNA production. See e.g. catalog ATCC CELL LINES AND HYBRIDOMAS, December 1984,

American Type Culture Collection, 12301 Parklawn Drive, Rockville, Maryland 20852, U.S.A., pages 5-9 and ECACC Catalog, 2nd Edition, PHLS CAMR Porton Down, Salisbury, Wills; SP40JG, U.K. pages 30-35 and 40-46. Hybridomas secreting monoclonal antibodies responsive to a wide variety of antigens are listed therein, are available from the collection and useful in the invention. Of particular interest are hybridomas that secrete antibodies that are reactive with viral viruses including Dengue complex spread (ATCC HB 114), Dengue type 1 virus (ATCC HB 47), Dengue type 2 virus (ATCC HB 46), Dengue type 3 virus (ATCC HB 49), Dengue type 4 virus (ATCC HB 48), Epstein-Barr receptor (ATCC HB 135), Flavivirus group (ATCC HB 112) , hepatitis B surface antigen (ATCC CRL 8017 5 and 8018), herpes simplex type I (ATCC HB 8068), herpes simplex type II (ATCC HB 8067), influenza virus (ATCC CL 189), influenza A virus, matrix protein (ATCC HB 64), influenza A virus, nucleoprotein (ATCC HB 65), influenza A Bangkok / 1 / 79HA (ATCC HB 66), influenza AWSN NP (ATCC HB 67), SV40 major T antigen (ATCC TIB 115), SV40 large T antigen, C-terminal end (ATCC TIB 117), and SV40 non-viral T antigen (ATCC TIB 116). Examples of other hybridomas include those that secrete antibodies to tumor-associated antigens or to human lymphocyte antigens, such as those that are reactive to human tumor-associated CEA, high mw (ATCC CRL 8019); human tumor-associated α-footprotein, IgG1-K (ATCC HB 134); human B lymphocyte HLA-DR, monomorph, IgG2b (ATCC HB 104); Human T lymphocyte T cell precursors, IgG1 (ATCC CRL 8022); human T lymphocyte T cell subset, helper, IgG2b (ATCC CRL 8002); T-subset, suppressor / cytotoxic, human, IgG1 (ATCC CRL 8013); T cell subset, suppressor / cytotoxic, human, IgG2a (ATCC CRL 8014); T cells, peripheral, human, IgG1 (ATCC CRL 8000); T cells, peripheral, human, IgG2a (ATCC CRL 8001); thymocytes, "common", human, IgG120 (ATCC CRL 8020).

These lines and others of similar nature can be utilized to copy the variable region mRNA coding using the UIG probes. Of particular interest are antibodies with specificity to human tumor antigens.

25

Expression vehicles include plasmids or other vectors. Preferred among these are vehicles which carry a functionally complete human constant heavy or light chain sequence with appropriate restriction sites formed such that any variable heavy or light chain sequence with the appropriate cohesive ends can be readily inserted therein. Thus, human constant heavy or light chain sequence-containing vehicles are an important embodiment of the invention. These vehicles can be used as intermediates in the expression of any desired completely heavy or light chain in any suitable host.

A preferred host is yeast. Yeast provides significant benefits in the preparation of lightweight and 5 heavy immunoglobulin chains. Yeast performs post-translational peptide modifications including glycosylation. A number of DNA recombination strategies now exist that utilize strong promoter sequences and high copy number plasmids that can be used for open production of the desired proteins in yeast. Yeast recognizes leader sequences on cloned mammalian gene products and secretes peptides that have leader sequences (i.e.

10 prepeptides) (Hitzman et al., 11th International Conference on Yeast, Genetics and Molecular Biology, Montpelier, France, September 13-17, 1982).

Yeast gene expression systems can be routinely assessed for heavy and light chain production levels, protein stability and secretion. Any of a variety of 15 yeast gene expression systems comprising promoter and termination elements of the actively expressed genes encoding glycolytic enzymes produced in large quantities when yeast is grown in glucose rich media can be used. Known glycolytic genes can also provide highly efficient transcriptional control signals. Eg. For example, promoter and termination signals from the iso-1 cytochrome C (CYC-1) gene can be used.

20

The following ways can be used to evaluate optimal expression plasmids for the expression of cloned immunoglobulin cDNAs in yeast.

(1) The cloned immunoglobulin DNA-binding V and C regions are linked to 25 different transcription promoters and termination DNA fragments.

(2) The chimeric genes are placed on yeast plasmids used for protein overproduction (see, e.g., Beggs, J. D., Molecular Genetics and Yeast, Alfred Benzon Symposium, 16, Copenhagen (1981)).

(3) Additional genetic units, such as a yeast leader peptide, may be included on immunoglobulin DNA constructs to achieve antibody secretion.

(4) A portion of the sequence, often the first 6 to 20 codons in the gene sequence, may be modified 5 to represent preferred yeast codon use.

(5) The chimeric genes are placed on plasmids used for integration into yeast chromosomes.

The following ways can be used to simultaneously express both light and heavy chain 10 genes in yeast.

(1) The light and heavy chain genes are each linked to a yeast promoter and a termination sequence and placed on the same plasmid. This plasmid can be designed for either autonomous replication in yeast or integration at specific sites in the yeast chromosome.

(2) The light and heavy chain genes are each linked to a yeast promoter and termination sequence on separate plasmids containing various selectable markers. Eg. For example, the light chain gene may be placed on a plasmid containing the trpl gene as a selectable marker, while the heavy chain gene may be placed on a plasmid containing ura3 as a selectable marker. The plasmids can be designed for either autonomic replication in yeast or integration at specific sites in yeast chromosomes. A yeast strain deficient in both selectable markers is transformed either simultaneously or sequentially with the light chain gene plasmid 25 and the heavy chain gene plasmid. 1

The light and heavy chain genes are each linked to a yeast promoter and termination sequence on separate plasmids, each containing different selectable markers as described in (2) above. A yeast mating type "a" strain lacking the selectable markers found on the light and heavy chain expression plasmids (trp1 and ura3 in the above example) is transformed with the light chain gene plasmid by selecting one of the two selectable markers (trpl in the above example). A yeast pairing type "" "strain which is deficient in the same selectable markers as the" a "strain (i.e., trpl and ura3 as examples) 5 is transformed with a plasmid containing the heavy chain gene by selecting the alternative selectable marker (i.e., ura3 in the above example). The "a" strain containing the light chain plasmid (phenotype: Trp + Ura- in the above example) and the strain containing the heavy chain plasmid (phenotype: Trp- Ura + in the above example) are paired and diploids selected, which are prototropic for both of the above selectable markers (Trp + Ur a + in the above example).

Among bacterial hosts which can be used as transformation hosts, E. coli K12 strain 294 (ATCC 31446) is particularly suitable. Other microbial strains which may be used include E. coli XI776 (ATCC 31537). The aforementioned strains as well as E. coli W3110 (ATCC 27325) and other enterobacteria such as Salmonella typhimurium or Serratia marcescens and various Pseudomona species may be used.

In general, plasmid vectors containing replicon and control sequences derived from species compatible with a host cell are used in association with these hosts. The vector usually carries a site of replication as well as specific genes capable of providing phenotypic selection in transformed cells. Eg. E. coli is readily transformed using pBR322, a plasmid derived from an E. coli species (Bolivar et al., Gene, 2: 95 (1977)). pBR322 contains genes for ampicillin and tetracycline resistance and thus provides an easy way to identify transformed cells. The pBR322 plasmid or other microbial plasmids must also contain or be modified to contain promoters which can be used by the microbial organism to express its own proteins. The promoters most commonly used in DNA recombination construction include 6-1 actamase promoter systems (penicilinase) and lactose promoter systems (6-galactosidase) (Chang et al., Nature, 275: 615 (1978); Itakura et al., Science 198: 1056 (1977)); and tryptophan promoter systems 29 (Goeddel et al., Nucleic Acids Research, 8: 4057 (1980); EPO Publication No. 0036776). While these are the most commonly used, other microbial promoters have been discovered and utilized.

For example. For example, a genetic construct for any heavy or light chimeric immunoglobulin chain can be brought under control by the left-sided promoter of bacteriophage λ (PL). This promoter is one of the strongest known promoters that can be regulated. Control is exercised by the λ repressor and nearby restriction sites are known.

10

The expression of the immunoglobulin chain sequence may also be brought under control by other regulatory sequences which may be "homologous" to the organism in its untransformed state. Eg. For example, lactose-dependent E. coli chromosomal DNA comprises a lactose or lac operon which mediates lactose digestion by forming the enzyme β-galactosidase. The lac controls can be obtained from bacteriophage λ pLAC5, which is infecting E. coli. The lac promoter / operator system can be induced by IPTG.

Other promoter / operator systems or portions thereof may also be used. Eg. For example, arabinose, colicin E1, galactose, alkaline phosphatase, trypthophane, xylose, tac and the like may be used. Other bacterial gene expression controls can be used to achieve the expression of immunoglobulin proteins. Eg. For example, a gene with a bacterial secretion signal peptide coding region can be expressed in bacteria, resulting in secretion of the immunoglobulin peptide originally linked to the signal peptide.

Other preferred hosts are mammalian cells cultured in vitro in tissue culture or in vivo in animals. Mammalian cells provide post-translational modifications of immunoglobulin protein molecules including leader peptide removal, proper folding and assembly of heavy and light chains, glycosylation at correct sites, and secretion of functional antibody protein from the cell as H2L2 molecules.

30 30 DK 175581 B1

Mammalian cells which may be useful as hosts in the production of antibody proteins include cells of fibroblast origin, such as Vero (ATCC CRL 81) or CHO-K1 (ATCC CRL 61), or cells of lymphoid origin, such as the hybridoma Sp2 / 0-Agl4 ( ATCC CRL 1581) or the myeloma P3X63Ag8 (ATCC TIB 9) and 5 derivatives thereof, /

There are several possible vector systems available for the expression of cloned heavy chain and light chain genes in mammalian cells. A class of vectors utilizes DNA elements that produce an autonomously replicating extrachromosomal plasmid derived from animal viruses such as bovine papillomavirus (Sarver, N. et al., Proc. Natl.

Acad. Sci., USA, 79: 7147 (1982)), polyomavirus (Deans, R. J. et al., Proc. Natl. Acad.

Sci., USA, 81: 1292 (1984)), or SV40 virus (Lusky, M. and Botchan, M., Nature, 293: 79 (1981)). A second class of vectors is dependent on the integration of the desired gene sequences into the host cell chromosome. Cells that have stably integrated the introduced DNA into their chromosomes can also be selected by introducing drug resistance genes, such as E. coli gpt (Mulligan, RC and Berg, P., Proc. Natl. Acad. Sci., USA, 78: 2072 (1981)) or Tn5 neo (Southern, P.J. and Berg, P., J. Mol. Appl. Genet., 1: 327 (1982)).

The selectable marker gene can either directly bind to the DNA gene sequences to be expressed, or be introduced into the same cell by cotransfection (Wigler, M. et al., 20 Cell, 16:77 (1979)).

Since an immunoglobulin cDNA is composed only of sequences representing the fully developed mRNA encoding an antibody protein or its precursor, additional gene expression elements regulating transcription of the gene and processing of the RNA are required for optimal synthesis of immunoglobulin mRNA. These elements may include splicing signals as well as transcription promoters comprising inducible promoters, promoters, and termination signals. cDNA expression vectors which have incorporated such elements include those described by Okayama, H. and Berg, P., Mol. Cell Biol, 3: 280 (1983), Cepko, C. L. et al., Cell, 37: 30 1053 (1984) and Kaufman, R. J, Proc. Natl. Acad. Sci, USA, 82: 689 (1985).

31 DK 175581 B1

One way of assessing optimal vectors for the expression of immunoglobulin cDNA in mammalian cells involves first bringing the immunoglobulin DNA sequences into vectors that are stable to integrate into the cell genome or autonomously replicate as an extrachromosomal plasmid. The vectors can be used to judge 5 different gene expression elements for optimal immunoglobulin synthesis.

A further advantage of mammalian cells as hosts is their ability to express chimeric immunoglobulin genes derived from ffa genomic sequences. Thus, mammalian cells may express chimeric immunoglobulin genes composed of a variable 10 region cDNA module as well as a constant region composed wholly or partially of genomic sequences. Several human constant region genomic clones have been described (Ellison, J. W. et al., Nucl. Acids Res., 10: 4071 (1982) or Max, E. et al., Cell, 29: 691 (1982)). The use of such genomic sequences may be appropriate for the simultaneous introduction of immunoglobulin promoters, splicing signals, and transcription termination signals together with the constant region gene segment.

Various methods can be followed to obtain whole H2L2 antibodies.

First, the heavy and light chains can be separately expressed, followed by in vitro collection of purified light and heavy chains for whole H2L2 IgG antibodies. The pathways used to form whole H2L2 IgG molecules in cells have been extensively investigated (see, e.g., Scharff, M., Harvey Lectures, 69: 125 (1974)).

In vitro response parameters for the generation of reduced light and heavy chain IgG antibodies have been defined by Beychok, S., Cells of immunoglobulin Synthesis, Academic Press, New York, pp. 69, 1979.

Next, it is possible to co-express light and heavy chains in the same cells to achieve intercellular assembly and binding of heavy and light chains to whole H2L2 IgG antibodies. The co-expression may occur using either the same or different plasmids in the same host.

In a preferred embodiment, the co-expression may occur by secretion signals useful in yeast or bacteria. Under such conditions, fully folded and total H2Lrimmunoglobulins can be obtained.

The preparation of chimeric Fab fragments can also be carried out by the methods of the invention.

The methods described herein may also be used to switch classes of any antibody of a given specificity and class to an antibody of the same specificity, but of a different class, whether human or non-human. Eg. For example, human IgM antibodies can be transmuted to human IgG antibodies by producing constructs containing human constant IgG cDNA or genome sequences linked to variable human cDNA sequences obtained from a cell producing the original IgM antibody. These constructs are then introduced into appropriate hosts and expressed.

polypeptide

The invention provides "chimeric" immunoglobulin chains, either heavy or light. A chimeric chain contains a constant region substantially similar to that present in the heavy chain of a naturally occurring human immunoglobulin and a variable region of any desired antigenic specificity. The variable region is either of human or non-human origin.

The invention also provides heavy and light chain immunoglobulin molecules linked so that the overall molecule exhibits desired binding and recognition properties. Different types of immunoglobulin molecules are provided: monovalent, divalent, disspecific (i.e., with different variable regions), molecules with chimeric heavy chains and non-chimeric light chains, or molecules with variable binding regions linked to peptide moieties which carry desired functions.

33 DK 175581 B1

Chimeric heavy chain antibodies having the same or different variable region binding specificity and non-chimeric (i.e., fully human or completely non-human) light chains may be prepared by appropriate assembly of the necessary polypeptide chains.

These chains are manufactured individually by the modular assembly methods of the invention.

Chimeric Fab fragments are also part of the invention.

APPLICATIONS

10

The antibodies of the invention with human constant region can be utilized for passive immunization, especially in humans, without adverse immune reactions such as serum sickness or anaphylactic shock. The antibodies can, of course, also be used in known immunodiagnostic assays and kits in labeled form for in vitro imaging, where the label may be a radioactive emitter, or an NMR contrast agent such as a carbon 13 core, or an X-ray contrast agent such as a heavy metal core. . The antibodies can also be used for in vitro localization of antigens by appropriate labeling.

The antibodies may be used for therapeutic purposes per se in the complement mediated lysis or may be coupled to toxins or other therapeutic moieties.

Class change of antibodies is useful when it is desired to alter the association, aggregation or other properties of antibodies obtained by cell fusion or hybridoma techniques. Eg. are most human / human monoclonal antibodies of the 25 IgM class, which are known to be easily reduced and aggregate. Thus, changing such antibodies to other antibody types such as IgG, IgA or IgE is of great importance.

Mixed antibody / enzyme molecules can be used for immunoassay methods such as 30 ELISA. Mixed antibody-peptide effector conjugates can be used for targeted delivery of the effector moiety with a high efficiency and specificity.

34 DK 175581 B1

Having now generally described the invention, it will be further understood by reference to certain specific examples which are included herein for purposes of illustration only and are not intended to be limiting, unless otherwise stated.

5 EXPERIMENTS Materials and Methods Tissue Culture Cell Lines 10 The human cell lines GM2146 and GM1500 were delivered from the Human Mutant Cell Repository (Camden, New Jersey) and grown in RPM11640 as well as 10% fetal bovine serum (M. A. Bioproducts). Cell lines Sp2 / 0 and CRL 8017 were delivered from the American Type Culture Collection and grown in Dulbecco's Modified Eagle Medium (DMEM) as well as 4.5 g / l glucose (Μ. A. Bioproducts) and 10% fetal bovine serum (Hyclone, Sterile 15 Systems, Logan, Utah). Media were supplemented with penicillin / streptomycin (Irvine Scientific, Irvine, CA).

Recombined plasmid and bacteriophage DNA

The plasmids pBR322, pL1 and pUC12 were delivered from Pharmacia P-L-Biochemicals (Milwaukee, Wisconsin). The plasmids pSV2-neo and pSV2 * gpt were delivered from BRL (Gaithersburg, Maryland) and were available from the American Type Culture Collection (Rockville, Maryland). pHu-gamma-1 is a subclone of the 8.3 kb HindIII to BamHI fragment of the human IgG1 chromosome gene. A separate isolation of the human IgG1 chromosome gene is described by Ellison, J. W. et al., Nucl. Acids Res., 10: 4071 (1982). M8oRX12 contains the 0.7 kb XbaI to EcoRI fragment containing the heavy mouse chain effector ffa J-C intron region of the M603 chromosome gene (Davis, M. et al., Nature, 283: 733) inserted into M13mp10. G-tailed pUC9 was delivered from Pharmacia P-L. DNA manipulations involving purification of plasmid DNA by buoyancy density centrifugation, restriction endonuclease digestion, purification of DNA fragments by agarose gel electrophoresis, ligation and transformation of E.

Coli was performed as described by Maniatis, T. et al., Molecular Cloning, A Laboratory Manual, (1982). Restriction endonucleases and other DNA / RNA modifying enzymes were provided from Boehringer-Mannheim (Indianapolis, Indiana), BRL, New England Biolabs (Beverly, Massachusetts) and Pharmacia P-L.

5

oligonucleotide preparation

Oligonucleotides were either synthesized by the triester method described by Ito et al. (Nucl. Acids Res., 10: 1755 (1982)), or were delivered from ELESEN, Los Angeles, California, Tritylated, de-blocked oligonucleotides were purified on Sephadex. -G50, followed by reverse phase HPLC with a 0-25% gradient of acetonitrile in 10 mM triethylamine acetic acid, pH 7.2, on a Cl 8 uBondapak column (Waters Associates). Detritylation was in 80% acetic acid for 30 minutes, followed by evaporation three times. Oligonucleotides were labeled with [gamma-32P] ATP as well as T4 polynucleotide kinase.

RNA preparation and analysis

Total cellular RNA was prepared from tissue culture cells by the method of Auffray, C. and Rougeon, F. (Eur. J. Biochem., 107: 303 (1980)) or Chirgwin, J.M. et al (Biochemistry, 18: 5294 (1979)). Preparation of poly (A) + RNA, methylmercury silver agarose gel electrophoresis, and "Northem" transfer to nitrocellulose was performed as described by Maniatis, T. et al., Supra. Total cellular RNA or poly (A) + RNA was directly bound to nitrocellulose by first treating the RNA with formaldehyde (White, 25 B. A. and Bancroft, F. C., J. Biol. Chem. 257: 8569 (1982)). Hybridization to filter-bound RNA was with notch-translated DNA fragments using conditions as described by Margulies, DH et al. (Nature, 295: 168 (1982)) or with 32 P-labeled oligonucleotide using 4xSSC, 10X Denhardt's, 100 µg / ml salmon sperm DNA at 37 ° C overnight followed by washing in 4xSSC at 37 ° C.

cDNA preparation and cloning 30 36 DK 175581 B1

Oligo-dT primer-treated cDNA libraries were prepared from poly (A) + RNA from GM1500 and GM2146 cells using the methods described by Land, H. et al.

(Nucl. Acids Res., 9: 2251 (1981) and Gubier, V. and Hoffman, B. J., Gene, 25: 263, 81983). The cDNA libraries were screened by in situ hybridization (Maniatis T., supra) with 32 P-labeled oligonucleotides using the conditions described above or with notch-translated DNA fragments using the conditions described by de Lange et al. (Cell, 34: 891 (1983)).

10 Oligonucleotide Primer Extension and Cloning

Poly (A) + RNA (20 µg) was mixed with 1.2 µg of primer in 40 µl of 64 mM KCl. After denaturing at 90 ° C for 5 minutes and then cooling in ice, three units of human placental tribonuclease inhibitor (BRL) were added in 3 µl of 1M Tris-HCl, pH 8.3. The oily 15 gonucleotide was joined with the RNA at 42 ° C for 15 minutes, after which 12 µl of 0.05 M DTT, 0.05 M MgCl 2 and 1 mM of each dATP, dTTP, dCTP and dGTP were added.

2 µl of α-32P-dATP (400 Ci / mmol, New England Nuclear) was added followed by 3 μΐ AMV reverse transcriptase (19 units // µl, Life Sciences).

After incubation at 42 ° C for 105 minutes, 2 μ \ 0.5 M EDTA and 50 μΐ 10 mM Tris, 1 mM EDTA, pH 7.6 were added. Unincorporated nucleotides were removed by Sephadex G-50 vertebral column chromatography, and the RNA-DNA hybrid was extracted with phenol, then with chloroform and precipitated with ethanol. Second strand synthesis, homopolymer tail formation with dGTP or dCTP, and insertion into vectors with 25 homopolymer tail were performed as described by Gubier and Hoffman, supra.

Site-directed mutagenesis

Single-stranded M13 subclone DNA (1 µg) was combined with 20 ng of oligonucleotide primer 30 in 12.5 μΐ Hin buffer (7 mM Tris-HCl, pH 7.6, 7 mM MgCl2, 50 mM NaCl).

After heating to 95 ° C in a closed tube, the primer was joined with the template 37 by slow cooling from 70 ° C to 37 ° C for 90 minutes. 2 μΐ dNTPs (1 mM each), 1 μΐ 32P-dATP (10 μΟ), 1 μ] DTT (0.1 M) and 0.4 μΐ Klenow DNA Poll (2 μ Boehringer Mannheim) were added and chains extended at 37 ° C for 30 minutes. To this was added 1 μ \ (10 ng) of M13 reverse primer (New England Biolabs), and the heating / co-joining and chain extension steps were repeated. The reaction was quenched with 2 μΐ 0.5 M EDTA, pH 8 and 80 μΐ 10 mM Tris-HCl, pH 7.6, 1 mM EDTA. The products were phenol extracted and purified by Sephadex G-50 vortex column chromatography and ethanol precipitated prior to restriction enzyme digestion and ligation to the appropriate vector.

10

Transfection of mveloma tissue culture cells

A variation of the method described by Ochi, A. et al. (Nature, 302: 340 (1983)) was used for protoplast fusion. 50 ml of bacteria at A60o of 0.7 was converted to 15 protoplasts by the method described by Sandri-Goldin, R. M. et al. (Mol.

Cell. Biol., 1: 743 (1981)), then diluted with 20 ml of DMEM plus 10% FBS (final volume is 25 ml). Sp2 / 0 cells were collected, pelleted at 2,200 x g, washed, repelleted, and resuspended in DMEM at 2-5x106 / ml. Bacterial protoplasts (10 ml) were mixed with 10x10 6 Sp2 / 0 cells and pelleted by centrifugation at 4000 x g At 22 ° C for 20 minutes. After pipetting off the supernatant, the pellet was suspended in the remaining medium drop by snapping to the tube. 2 ml of 10% DMSO, 37% (w / v) PEG6000 (Kodak) in DMEM was added dropwise with mixing over 45 seconds. After 15 seconds, 2 ml of 42% PEG6000 in DMEM was added over 45 seconds. Fully DMEM (45 ml) was slowly added with mixing. Cells 25 were pelleted at 2500 x g, then washed and pelleted three times.

The electroporation method described by Potter, H. et al. (Proc. Natl. Acad. Sci., USA, 81: 7161 (1984)) was used. After transfection, cells were allowed to recover in complete DMEM for 48-72 hours and then seeded at 10,000 to 50,000 cells per well in 96 well culture plates in the presence of selective medium. G418 (GIBCO) selection was at a concentration of 0.8 mg / ml, mycophenolic acid (Calbiochem) was at a concentration of 6 µg / ml plus 0.25 mg / ml xanthine, and HAT (Sigma) was at the standard concentration.

Assays for immunoglobulin synthesis and secretion 5

Secreted immunoglobulin was measured directly from tissue culture cell supernatants. Cytoplasmic protein extract was prepared by vortexing 1x10 6 cells in 160 μΐ 1% NP40, 0.15 M NaCl, 10 mM Tris, 1 mM EDTA, pH 7.6 at 0 ° C, 15 minutes followed by centrifugation at 10,000 xg removal of insoluble material.

10

Double antibody sandwich ELISA (Voller, A. et al., In Manual of Clinical Immunology, 2nd ed., Editors Rose, N. and Friedman, H., pages 359-371, 1980) using affinity purified antisera was used to to detect specific immunoglobulins.

For detection of human IgG, plate-bound antiserum goat anti-human IgG (KPL, Gaithersburg, Maryland) is at a dilution of 1/1000, while the peroxidase-bound antiserum is goat anti-human IgG (KPL or Tago, Burlingame). in a dilution of 1/4000. For detection of human immunoglobulin kappa, the plate-bound antiserum is goat anti-human kappa (Tago) at a dilution of 1/500, while the peroxidase-bound antiserum is goat anti-human kappa (Cappel) at a dilution of 1/1000. .

20

Antibodies that bind hepatitis B surface antigen were detected using a commercial assay (Abbott, AYSAB).

EXAMPLES

25

The following examples show the preparation of chimeric antibodies, each having a human constant region and a non-human variable region. These examples show the step-by-step method of producing the chimeric antibodies.

EXAMPLE I: Human antibody constant region gene modules and cDNA expression vectors 39 (1) Preparation of cDNA clones and vehicles containing the same to human heavy chain constant region.

The cell line GM2146 was used as a source for mRNA synthesis and cDNA cloning. This cell line secretes IgG1 (Simmons, J. G. et al., Scand. J. Immunol., 14: 1-13, 1981). Tests of this cell line showed that it secretes IgA as well as IgG.

The cell line was cloned and results showed that 5 out of 6 subclones only secreted IgG, 10 while one in 6 subclones secreted only IgA. Poly (A) + RNA was prepared from the cell line, and a cDNA library was prepared from the poly (A) + RNA by the method described by Gubier, U. and Hoffman, B. I, Gene, 25: 263-269 (1983). An initial plating of the cDNA transformed into E. coli strains HB10I and RR1 yielded a total of 1500 colonies which were screened by hybridization to a HindIII to 15 BamHI fragment of a human IgG1 genome clone (pHU gamma). 1). Four positive clones were found. A fragment containing the CH3 coding region of one of these clones, pGMH-3 (Fig. 4) was used to re-screen the original library as well as a novel transformation of ca. 5000 colonies. Two of the largest clones, pGMH-6 and pGMH-15, were analyzed by restriction enzyme digestion (Fig. 4). Both clones 20 contained the entire constant region of human IgG1, although it was found that pGMH-6 had deleted ca. 1500 base pairs of pBR322 DNA apparently without affecting the IgG1 cDNA sequences.

Clone pGMH-6 provided the IgG1 constant region module in the construction of 25 heavy chain variable region cloning vectors.

(2) Preparation of cDNA clones and vehicles containing these for human light chain constant region.

A human cell line (GM1500) producing IgG2K was selected for the initial cloning phase. Poly (A) + RNA prepared from GM1500 is active by in vitro translation using rabbit reticulocyte extracts. A cDNA library was prepared from this RNA using the method described by Land et al., Nucl.

Acids Res., 9: 2251-2266 (1981) using KpnI digested and dG tail-formed pQ23 as the cloning vector (Fig. 5). This vector contains BglII, Kpnl, and 5 SstI sites inserted between the BamHI and SalI sites of pBR322.

To identify the cDNA clones formed from GM1500 RNA corresponding to light chain mRNA, a DNA probe, UIG-HuK, was synthesized and purified. The UIG-HuK oligonucleotide had the sequence 5'-AGCCACAGTTCGTTT-3 'and is designed to hybridize to all functional human kappa mRNA species at the J-C junction. This probe was used to prime cDNA synthesis on GM1500 RNA in the presence of dideoxynucleotides and reverse transcriptase. From 1.2 µg of total GM1500 poly (A) + RNA used in this experiment, the entire J sequence and some of the V region were read, showing that (1) the GM1500 RNA is intact, (2) the 15 kappa probe has the correct sequence and (3) GM1500 light chain mRNA contains JK4 sequences.

cDNA clones positive for hybridization to the light chain probe were selected. As the probe hybridizes to the J-C compound, the most important point was to determine whether the clones had completely constant region sequence beyond the J region.

Insertion sizes for the two largest kappa cDNA clones were 0.6 and 0.9 kb. The restriction enzyme mapping revealed that the entire constant region coding sequence was present in both clones (Fig. 6). The human kappa cDNA clone pK2-3 was used to prepare the light chain constant region vector pING2001 by inserting the Sau3A fragment comprising the human kappa constant and J regions into the BclI site of pBR325 (Fig. 6B). .

A variant of the human kappa cDNA clone was prepared by placing a HindIII site in the J region. This was done by in vitro mutagenesis using a JKHINDIII oligonucleotide primer (Fig. 7C). The resulting plasmid is pGML60.

41 DK 175581 B1

A vector, pING2003, was constructed for transfer and expression of cDNA sequences to mammalian cells (Fig. 10). This vector was constructed from pUC12 and two plasmids containing SV40 sequences. PLI provides an SV40 early 5 region promoter and an SV40 late region placement sequence. pSV2 neo sequences provide a selectable marker for mammalian cell transformation and SV40 polyadenylation signal sequences. pUC12 provides a multiple cloning site for cDNA insertion.

The pING2003 vector had several suitable modification restriction sites. These include a HindIII site suitable for insertion of promoter sequences and a HindIII to XhoI fragment suitable for the insertion of alternating promoter sequences. This vector is useful in the expression of cDNA genes in mammalian cells.

15 Addition of promoter element to pING2003

Immunoglobulin promoter elements have been shown to promote transcription of genes in their vicinity in stably transformed mouse myeloma cells by several hundred times (Gillies, SD et al., Cell, 33: 717, 1983; and Baneiji, J. et al., Cell. 33: 729 , 1983). To facilitate expression of mouse / human immunoglobulin genes in mouse myeloma cells, mouse immunoglobulin heavy chain ffem element was added to the cDNA expression vector pING-2003 (Fig. 10). The mouse heavy chain promoter region DNA was isolated from an M13 subclone of mouse heavy chain genome DNA (M8-a-RX12, Deans, R. J., unpublished). DNA isolated from a SalI and EcoRI digest of this subclone was modified with HindIII couplers and inserted into the HindIII site of pING2003, resulting in the new cDNA expression vector pING2003E. This vector is useful in the efficient expression of cDNA genes in mammalian cells, especially mouse myeloma or hybridoma cell lines.

Example II: Human / mouse chimeric anti-HBsAG antibody chain 42 (17) Preparation of cDNA clones and vehicles containing them for mouse heavy chain anti-HBsAg variable region.

The cell line CRL8017 was delivered from ATCC and subcloned. Subclones were cultured and assayed for mouse IgG anti-hepatitis B binding activity using a commercially available anti-HBsAg detection kit. Three positive subclones were found. Poly (A) + RNA was prepared from one of these subclones and fractionated on a methylmercury agarose gel. The RNA contained intact light chain and heavy chain mRNAs derived from specific hybridization to kappa UIG-MJK primer 10 and to mouse heavy chain UIG-MJH3 probe (see Figure 7). Furthermore, the UIG-MJK primer was used for specific priming of anti-HBsAg poly (A) + RNA by a dideoxy sequence reaction. Sufficient sequence was read to show that a substantial kappa RNA of the anti-HBsAg cell line contains the JK2 sequence.

The conditions for variable region cDNA synthesis were optimized using heavy and light chain UIG primers on anti-HBsAg poly (A) + RNA. Dideoxy chain extension experiments showed that the mouse UIG-MJK primer and the UIG-JH3 primer correctly primed kappa and heavy chain RNAs. When reverse transcription was performed in the absence of dideoxynucleotides, the main product using the 20 kappa UIG-MJK primer was a 410 ± 20 nucleotide fragment, while the main product using the heavy chain UIG-JH3 primer was a 430 + 30 nucleotide fragment.

These correspond to the expected lengths of the variable and 5 'untranslated regions of kappa and heavy chain immunoglobulin mRNAs. The conditions for the optimal priming of poly (A) + RNA from CRL8017 cells should work well for the poly (A) + RNA isolated from any cell line producing a monoclonal antibody.

After determining optimal conditions for priming hybridoma mRNA with oligonucleotide primers, two oligonucleotides were designed and used for heavy chain V region cDNA synthesis. These two oligonucleotides are UIG-MJHBSTEII (13) and 30 UIG-MJH3 (Figs. 7 and 8). It should be noted that the primer sequence was designed to introduce a BstEII recognition site (GGTGACC) into the clone so that it could be connected at that site to the human IgG 1 constant module at the analog position at the latter's J region. . In this case, the primer had a single incorrect match between G and U with the mouse mRNA sequence which uses the JH3 coding sequence. The UIG-MJHBSTEIl (13) primer was 13 bases long, and the non-matching residue was flanked by 7 matches 5 'and 5 matches 3' for it. This was the 13-mer BstEII primer.

To assess the priming efficiency of the 13-mer BstEII oligonucleotide, a 21-mer primer specific for mouse JH3 (UIG-MJH3) was used. This primer matched perfectly in terms of the 17 nucleotides on its 3 'end.

These two primers and the JH3 coding sequences are shown in FIG. 8. The first strand cDNA products produced via the 13-mer BstEII and 21-mer JH3 primers included bands of ca. 430 nucleotides representing the entire VH region. Under the standard priming conditions used, the priming efficiency of 13-mer BstEII was much less than that of 21-mer JH3. Therefore, a cDNA library was generated from 15 the first strand synthesis from each of these primers using the method described by Gubier and Hoffman, supra.

First, the 21-mer JH3 library was screened with the 21-mer JH3 oligonucleotide. Filter hybridization was performed at 30 ° C overnight following the method described by Lange, 20 T. et al., Cell, 34: 891-900 (1983). The filters were then washed at 51 ° in 6 x SSC, 0.1% SDS. Five colonies were selected. The largest one had a deposit of approx. 460 bp. More significantly, it contains three restriction sites predicted from the known JH3 sequence present upstream of the primer sequence. This clone pJ3-1 was sequenced using the JH3 primer by chain termination method 25 (Wallace, R.B. et al., Gene, 16: 21-26 (1981)). The obtained sequence had the remaining JH3 coding segment. Just upstream, a 13-nucleotide segment matched with a published D-segment sequence (Dsp 2.2) (Kurosawa, Y. et al., J. Exp. Med., 155: 201 (1982) and Tonegawa, S., Nature, 302: 575 (1983)). A nonapeptide predicted from this range showed characteristic homology to the 30 published mouse heavy chain V subgroups at amino acid residues 86 to 94 comprising FR3 of heavy chain molecules. Plasmid pJ3-1 represented a rearranged 44 apparently 171 VDJ sequence and apparently contained the anti-hepatitis VH sequence produced by this cell line.

In order to isolate a VH region cDNA clone which had a BstEII site in the 5 J region, a 265 nucleotide long Alul to Sau96I probe from pJ3-11 was then used to screen the cDNA library. which was formed from the 13-mer BstEII primer. Six positive clones were isolated. The largest, pBsl3-1, was further analyzed. The insert was 280 nucleotides long, and its restriction map was consistent with that of pJ3-1, except for the inserted BstEII site. In FIG. 9 10 illustrates how these two insertions were recombined to form pMVHCa-13, a VH clone with the module joining BstEII site. Three additional VH cDNA clones were isolated from a cDNA library formed from 21-mer oligonucleotide UIG-MJH3BSTEII primer containing a BstEII site. These clones may provide alternating VH cDNA sequences for joining to human 15 Cn sequences.

(2) Preparation of cDNA clones and vehicles containing the same to variable light chain mouse anti-HBsAg region.

Since the JK2 sequence is present in mRNA prepared from the anti-hepatitis hybridoma cell line, the oligonucleotide UIG-JK2BGLII (Fig. 7B) was designed to introduce a BglII site into the JK2 region. Digestion with BglH would then allow direct insertion of a VK cDNA coding region into the BclI site of the previously mentioned human Ck * vector, pING2001. This insertion would result in the precise joining of a variable mouse region segment (including the J region) into a human, constant kappa region segment, each in the correct coding frame without any change in amino acid sequence for either the variable mouse region or constant human region.

The JK2BGLI1 oligonucleotide was used to prime anti-HBsAg mRNA to form a heavy chain cDNA library, supra, in pUC9. cDNA was selected in size 45 DK by pre-cloning polyacrylamide gel electrophoresis, and 80% of the cDNA clones were shown to have insert sizes between 300 and 750 nucleotides in length. Copy filters of this library were screened with two oligonucleotides, the original primer and a second probe complementary to JK2 sequence 5 'to the original primer.

It was discovered that the monoclonal anti-hepatitis B cell line CRL 8017 secretes immunoglobulins with at least two different light chains. One of them is derived from ffa myeloma NS-1, which was used as a fusion partner in forming the anti-hepatitis 10 B cell line. Because NS-1 is derived from myeloma MOPC21, the possibility was investigated that MOPC21 VK mRNA may be present in the VK cDNA library from the monoclonal antihepatitis cell line. Indeed, one of the analyzed cDNA clones (p6D4B) had an identical restriction enzyme map to that of the MOPC21 Vk cDNA except for the inserted BglH site.

15

Two conclusions can be drawn from these results. The first is that it is possible to effectively use an oligonucleotide to introduce a restriction enzyme site while cloning a VK region ffa a hybridoma cell line. The second is that one must carefully monitor hybridoma cell lines for the presence of multiple V region sequences, 20 of which only one has the desired sequence.

To further characterize kappa chain J regions present in the cell line mRNA, poly (A) + RNA was bound to nitrocellulose by the "Dot blot" formaldehyde procedure of White and Bancroft, J. Biol. Chem., 257: 8569 (1982). RNA was hybridized to 32 P-labeled oligonucleotide probes specific to each functional kappa J region. These probes are shown in FIG. 7B as the UIG probes 5JK1, MJK, 5JK4 and 5JK5. The results showed that mRNA hybridized strongly to both the MJK and 5JK4 oligonucleotide probes, indicating that both JK2 and JK4 sequences were present. Since JK2 mRNA had previously been identified as that derived from parental hybridoma partner NS-1, it was concluded that Ji <4 mRNA encoded anti-hepatitis binding specificity of the CRL 8017 cells.

46 DK 175581 B1

Two different cDNA libraries were screened for isolation of V region clones encoding JK4 sequences. The first was primed by JK2BGLII, supra. The other was prepared using the oligonucleotide primer JK4BGLII, which is specific for 5 JK4 mRNA and introduces a BglII site into the J region of cloned V regions. The JK4BGLII primer was used to prime synthesis of first cDNA strand to build a cDNA library by the same method used to construct a JK2BGLII primed cDNA library, except that cDNA was not selected after size prior to cloning.

10 In FIG. 7B lists the imperfections that each primer has with other functional kappa J mouse region sequences. It should be noted that JK4 has five imperfections in 21 nucleotides when compared to the JK2BGLII primer and three in 23 when compared to the JK4BGLII primer.

15

Both libraries were screened for V region clones containing JK4 sequences by hybridization to an oligonucleotide probe specific for JK4 sequences (5JK4). The results of this screening are shown in Table 1.

Table 1 1

Library Probe specificity

Jk.2 Jk4 25 JK2BGLII 2% (30/1500) 0.15% (2/1500) JK4BGLII N / D 3.5% (31/875)

Percentage of clones containing JK2 or JK4 sequence plus a V region. The probes used were the oligonucleotide 5JK4 (JK4 specificity, Fig. 7) and p6D4B, which contains the NS-1 (MOPC21) V region sequence. N / D, not made.

47 DK 175581 B1

Several JK4-V region cDNA clones isolated from both libraries were characterized.

These clones have identical restriction enzyme maps, including the genetically engineered BglII site resulting from the oligonucleotide-primed cDNA cloning procedure. The restriction map and sequence of a clone, pV17, showed that pV17 contains 5 V region gene sequences.

These results indicate that the JK2BGLII primer could prime JK4 mRNA sequences correctly, albeit ineffectively. Since the JK2BGLII primer had fewer imperfections with any other J1 region mRNA than with JK4 mRNA (Fig. 7B), it is expected that the other JK mRNAs could be primed at the correct site by improved efficiency using the JK2BGLII primer. Thus, efficient cDNA cloning of any functional kappa V mouse region can be obtained using a mixture of the JK2BGLII and JK4BGLII primes.

Placement of a BglII site in the J region during cDNA cloning of the V regions enables joining of the cloned V region mouse gene module to the human constant kappa region gene module (Fig. 9B).

After the aforementioned experiments were performed, it was found that the cDNA clone pV17 20 lacked a complete 5 'coding region. Nucleotide sequencing showed that A in the initiator codon ATG was not copied in pV17. This was not a random cDNA cloning artifact because two other cDNA clones had the same deficiency. Two methods were devised to obtain a light chain gene with a complete 5 'coding region.

First, a new cDNA library was constructed by first priming with an oligonucleotide (5'-ATATTTGCTGATGCTCT-3 ') complementary to pV17 sequences 155 bases from the 5' end. From this library, clones that hybridize to a p V17 DNA fragment probe were selected and some of these new cDNA clones had initiator ATG plus ca. 20 30 nucleotides of untranslated 5 'region. One of these clones, p2-12, provides an untranslated 5 'region of 23 nucleotides and a complete ATG initiator codon. When 48 p172 was combined with pV17-derived sequences, a variable region with a complete 5 'end was formed (pING2013E).

Next, a site-directed mutagenesis on the existing light chain clone was used to simultaneously remove the poly-G region and apply a ribosome recognition sequence adjacent to initiator ATG. The pst1 fragment from pV17 was subcloned into M13-mp18.

An oligonucleotide (V17-IVM; 5-GTGTCGACTCAGCATGAGG-TTCCAGGTTC-3 ') was then used as a primer to mutate the pV17 sequence to include a SalI site and an initiator ATG in the pV 17 sequence. The resulting plasmid pV17-IVM 10 provided a changing variable mouse region for joining to human constant human modulus.

The complete nucleotide sequence of the pV17 variable region was then determined. The sequence shows that pV17 contains a VK-JK junction region containing 15 several conserved amino acids and the JK2 / JK4 hybrid region formed by priming J] <4 RNA with the UIG-JK2BGLII oligonucleotide. However, the VK region in pV17 is non-functional because the VK and JK regions are not in the same coding frame. Thus, translation of the pV17-V region would result in an abnormal immunoglobulin light chain, where the J region is translated into an incorrect frame. This defect may be caused by aberrant V-J association, resulting in a non-functional kappa mRNA, as has been observed by D. E. Kelley et al., Mol.

Cell. Biol., 5: 1660-1675 (1985).

Since the pV17-V region encodes an abnormal immunoglobulin, this light chain is highly unlikely to be part of a functional anti-hepatitis antibody molecule. These results demonstrate the importance of monitoring hybridoma cells for the presence of multiple RNA species encoding V regions, only one of which has the desired sequence.

Further screening of CRL 8017 cDNA libraries was performed to search for VK cDNA clones that are not from either of the two VK cDNA classes found until now (M0PC21-p6D4B, pV17). First, an oligo-dT-primed cDNA library prepared from CRL 8017 RNA was screened with a DNA fragment probe specific for this constant kappar region and separately with probes specific for MOPC21 and pV17-VK regions. A cDNA clone (plE9L-81) containing the constant 5 kappa region but having a VK region different from that of MOPC21 and pV17 was detected. This method of screening oligo-dT-primed cDNA libraries is a useful alternative for oligonucleotide screening of cDNA libraries because notch-translated probes with high specific activity are used. This method also enables the simultaneous isolation of several classes of V region clones, such as all VK clones, at 10 appropriate probe choices. Second, the UIG-JK2BGLII primed cDNA library was prepared from CRL 8017 RNA screened with the UIG-5JK2 oligonucleotide probe (see Figure 7). A new class of VK cDNA clones were found whose constituents are homologous to pI9L-81 and hybridize to the UIG-5JK2 probe but not to a MOPC21 VK probe. The restriction endonuclease site maps and nucleotide sequences for these clones also differ from MOPC21 homologous VK cDNA clones from CRL 8017 cells. However, these clones have an aberrant V-J association, resulting in a non-functional mRNA, and appear to be identical to that described by Cabilly and Riggs (Gene, 40: 157 (1985)).

Therefore, it was concluded that the anti-hepatitis B cell line CRL 8017 has at least three classes of VK mRNA corresponding to the cDNA clones described above, p6D4B (MOPC21), plE9L and pV17. The pIE9L and pV17 clones are derived from mRNA from aberrantly arranged kappa genes, while the p6D4B clone is from the parent hybridoma partner NS-1. None of these clones appear to encode the desired 25 light anti-hepatitis chain.

(3) Production and expression of heavy chain containing constant human / variable mouse regions.

The 30 V region sequences of pMVHCa-13 were linked to the human IgG1 constant (C) region clone pGMH-6. Due to the presence of another BstEII site within 50 DK 175581 B1

The IgG1 CH1 region of pGMH-6 was a multistage ligation required. First, the 220 nucleotide long BstEII fragment from the J-CH1 region of pGMH-6 was ligated to the 100 nucleotide long IgG region BstEII for BamHI fragment of pGMH-6. In a separate ligation, the 420 nucleotide BstEII to BamHI fragment of pMVHCa-13, which fragment comprises the V mouse region, was joined to a calf intestinal phosphatase-treated BamHI plasmid vector. The two ligations were then combined, ligase added, and the products transformed into HB101, resulting in the chimeric mouse V-human C clone pMVHCc-24 (Fig. 9A).

The 10 V region of the heavy hybrid chain gene in pMVHCc-24 was further analyzed by partial sequence analysis. This analysis showed that the cloned V region contained a D sequence that matches a known D sequence, DSP2.2 (Kurosawa and Tonegawa, supra). The sequence also predicted a 19 amino acid leader peptide corresponding to known heavy V mouse chain leader peptide sequences and an untranslated 5 'region of at least 3 nucleotides.

The BamHI phage fragment containing the heavy mouse human hybrid chain gene of pMVHCc-24 was cloned into BamHI digested pING2003E vector, resulting in expression plasmid piNG2006E (Fig. 11). The pING2006E plasmid should have an increased probability of effective expression of the chimeric mouse human immunoglobulin in B lymphoid cells due to the presence of the heavy mouse chain enhancer region.

A modification of the chimeric heavy chain gene present in pMVHCc-24 was made to provide an alternating heavy chain gene lacking the oligo-dC region preceding initiator ATG. The pING2032E and pING2006E vectors are identical except for the immediately preceding ATG nucleotides, as shown in FIG. 12th

Bacteria harboring the pING2006E and pSV2 neo plasmids were converted to 30 protoplasts by the method of R. M. Sdr-Goldin et al., Mol. Cell. Biol., 1: 743 (1981). The protoplasts were then separately joined to SP2 / O-Agl4 hybridoma cells (ATCC CRL 1581) by treatment with polyethylene glycol (A. Ochi et al., Nature, 302: 340, 1983). The pooled cells were allowed to recover for 72 hours in complete medium before plating at 10,000 or 50,000 cells per well in a 96 well tissue culture plate. The cells were selected with G418 at 0.8 mg / ml for two 5 weeks as growth in some wells was quite evident. Under these selection conditions, SP2 / 0 cells were completely killed within 4-7 days of G418. Only cells that have integrated and expressed the neo gene present in the vectors will grow during G418 selection.

The number of wells positive for growth by these integrative transfectants is shown in Table 2.

52 DK 175581 B1

Table 2 *

Strain / 10,000 50,000 5 plasmid cells / well cells / well MC1061 / pING2006E 3 (13%) 12 (50%) MC1061 / pSV2-neo 7 (29%) 4 (17%) MC1061 / none 0 0 10 _ * Percentage of wells that show positive growth out of 24 wells.

Cells transfected with pING2006E and pSV2 neo were examined for immunoglobulin gene expression at the RNA and protein levels. Total cell RNA was prepared from transfected cells, bound to nitrocellulose and hybridized to notch-translated probes specific for the heavy mouse human hybrid chain gene. Two clones were found that had a strong signal, representing expression of the gene at the RNA level. The amount of total cellular RNA hybridizing to the mouse human probe appeared to be approx. 1/10 of the level of heavy 20 RNA chain in the original hybridoma cells. This probably represented approx.

1% of the total mRNA of the transfected cell.

The transfected mouse cells were also examined for production of cytoplasmic heavy human chain protein by an ELISA assay. It was found that three out of 7 25 pING2006E transfected cell lines produced detectable levels of heavy human chain protein. The mouse cell transformant, which produced the most heavy mouse human chain protein, gave a signal in the ELISA assay comparable to that of a 1/100 dilution of a human B cell line producing intact human immunoglobulin IgG1. This modest level of detected heavy mouse human chain protein may be due to several factors, including heavy chain instability in the absence of light chains in hybridoma cells or incorrect reprocessing of the chimeric gene transcript.

(4) Re-amplification of the integrated chimeric gene.

5

Southem epitope analysis showed that multiple copies of the pING2006E DNA sequences were integrated sequentially in the mouse genome. Restriction enzymes Apal and Bglll both cleave pING2006E individually. In the transformant, 2AE9, a band from an Apal or BglII digest of the expected size (8.2 kb) was found to hybridize to the human C-gamma 1 sequences (data not shown). A BamHI band of the correct size (1.6 kb) was found to hybridize to human as well as 1E9 VH sequences. Gene copy titration experiments (Fig. 14) indicated that there are approx. 5 copies of pING2006E in the 2AE9 genome. The fact that only a single band was detected in the Apal or BglII path indicates that these individual copies are arranged one after another in an orderly arrangement. A set of double digests showed that pING2006E sequences did not undergo any rearrangement upon entry into mouse DNA (data not shown).

Next, the 2AE9 cells were transfected with a plasmid containing a different selectable marker, the gpt gene, and selected clones growing into DMEM-HAT. A 20 clone, 2BH10, has approx. 38 ng of soluble human gamma 1 protein per 106 cells. Southem analysis showed that 2BH10 has approx. 30 copies of pING2006E (Fig. 14). They were propagated from the 5 copies in 2AE9 without rearranging the DNA sequences. (Compare 2AE9 panel with 2BH10). SI data (data not shown) revealed that this increase in template led to a higher amount of IgG gene transcripts. It is believed that these 25 sequences were co-amplified with adjacent cellular sequences as a result of the second selection.

Example III: A human mouse chimeric antibody with cancer antigen specificity.

(1) Antibody L6.

54 DK 175581 B1

Monoclonal L6 antibody (MAb) was obtained from a mouse that had been immunized with human lung carcinoma cells and then spleen cells were hybridized with NS-1 mouse myeloma cells. The antibody binds to a previously unidentified carbohydrate antigen which is expressed in large quantities at the surface of the cells of most 5 human carcinomas, including lung carcinomas (adeno, squamous cell), breast carcinomas, colon carcinomas and ovarian carcinomas, the antigen being present only. in trace amounts in normal cells of the adult host. MAb L6 is an IgG2a and can mediate antibody-dependent cellular cytotoxicity, ADCC, in the presence of human peripheral blood leukocytes as a source of effector cells to lyse L6-positive tumor cells, and it can lyse L6-positive tumor cells in the presence of human serum. a source of complement; the light is detected as the release of 51 Cr from labeled cells over a 4 hour incubation period. MAb L6 can localize to L6-positive tumors that are xenotransplanted on exposed mice and can inhibit the growth of such tumors. MAb L6 is described in Cancer Res. 46: 3917-3923, 1986 (on 15 MAb specificity) and in Proc. Natl. Acad. Sci. 83: 7059-7063, 1986 (on MAb function) .p3 (2) Identification of J sequences in immunoglobulin mRNA of L6.

Frozen cells were thawed on ice for 10 min and then at room temperature. The suspension was diluted with 15 ml of PBS and the cells were centrifuged. The 20 were resuspended, after washing in PBS, in 16 ml of 3M LiCl, 6M urea and destroyed in a polytron displacer. The preparation of mRNA and the selection of the poly (A) + fraction were performed according to Auffray, C. and Rougeon, F., Eur. J. Biochem. 107: 303, 1980.

Poly (A) + RNA from L6 was hybridized individually with labeled JH1, JH2, JH3 and 25 JH4 oligonucleotides under conditions described by Nobrega et al. Anal. Biochem. 131: 141, 1983. The products were then electrophoresed in 1.7% agarose TBE gel. The gel was fixed in 10% TCA, pressed dry and subjected to autoradiography. The result showed that L6 VH contains JH2 sequences.

For the analysis of VK mRNA, the dot imprinting method of White and Bancroft, J.

Biol. Chem. 257: 8569, (1982) used. Poly (A) + RNA was immobilized on 55 nitrocellulose filters and hybridized to labeled probe oligonucleotides at 40 ° in 4xSSC. These experiments show that L6 contains JK5 sequences. A weak hybridization to JK2 was observed.

5 (3) V-region cDNA clones

A library primed by oligo (dT) on L6 poly (A) + RNA was screened for kappa clones with a CK mouse region probe. Several clones were isolated from the L6 library. Another screening with. a JK5-specific 5 'probe identified the light 10 L6 (JK5) chain clones. Heavy chain clones of L6 were isolated by screening with the JH2 oligonucleotide.

The heavy and light chain genes or gene fragments ffa cDNA clones, pH3-6a and pL3-12a, were inserted into M13 bacteriophage vectors for nucleotide sequence analysis. The 15 complete nucleotide sequences of the variable region of these clones were determined (Figs.

15 and 16) by the dideoxy chain termination method. These sequences predict V-region amino acid compositions that are well consistent with the observed compositions and predict peptide sequences that have been verified by direct amino acid sequencing of portions of the V regions.

20

The nucleotide sequences of the cDNA clones show that they are immunoglobulin V region clones, containing amino acid residues diagnostic for V regions (Kabat et al., Sequences of Proteins of Immunological Interest; U.S. Dept. of HHS, 1983).

25 L6-Vh belongs to subgroup II. cDNA predicts an N-terminal sequence of 24 amino acid residues identical to that of a known VH (45-165 CRI; Margolies et al. Mol. Immunol. 18: 1065, 1981). L6-Vh has the JH2 sequence. L6-V1 is from the VK-KpnI family (Nishi et al., Proc. Nat. Acd. USA 82: 6399, 1985) and uses JK5. The cloned L6-V1 predicts an amino acid sequence which was determined by amino acid sequencing of 30 light L6 chain peptides corresponding to residues 18-40 and 80-96.

56 DK 175581 B1 (4) In vitro mutagenesis for genetic engineering of restriction enzyme sites in the J region for association to a human C module and for removal of oligo (dC) sequences 5 'to the V modules.

Both clones formed from priming with oligo (dT) L6-VK and L6-VH were necessary to modify. For L6-VK, the J region mutagenesis primer JKHindIII was used, as shown in FIG. 17B. A human CK module derived from a cDNA clone was mutagenized to contain the HindIII sequence (see Fig. 17A). The mutagenesis reaction was performed on M13 subclones of these genes. The frequency of mutant clones ranged from 0.5 to 10% by weight of the plaques obtained.

It was previously observed that the oligo (dC) sequence upstream of the AUG codon of a chimeric VH gene interferes with proper splicing in a particular gene construct. It was estimated that perhaps as much as 70% of RNA transcripts 15 had undergone misplacement in which a cryptic 3-splice acceptor in the leader sequence was used. Therefore, the upstream oligo (dC) sequence upstream of the initiator AUG was removed in all the clones.

In one method, an oligonucleotide was used, which oligonucleotide contains a 20 SalI restriction site to mutagenize the N6-VK clone. The primer used for this oligonucleotide-directed mutagenesis is a 22-mer introducing a SalI site between oligo (dC) and the initiator met codon (Fig. 19).

In another method, the BAL-31 nuclease was used to chew oligo (dC) into the L6-VH clone pH3-6a. The size of the deletion in two of the obtained mutants was determined by nucleotide sequencing and is shown in FIG. 17. In both of these mutants (delta 4 and delta 21) all oligo (dC) 5 'to the coding region was deleted.

These clones were then modified by oligonucleotide-directed mutagenesis with the 30 MJH2 ApaI primer (Fig. 17). This 31 base primer introduces an Apal site into the mouse Cn gene at a position analogous to an existing Apal site in human C-ganima 1 cDNA gene module. The primer introduces the relevant codons for the human C-gamma l gene. Thus, the chimeric heavy chain gene produced by joining the mutagenized mouse VH gene module to a human Cn module encodes a chimeric protein that does not contain human amino acids for the entire VH region.

5

The human C-gamma 1 gene module is a cDNA derived from GM2146 cells (Human Genetic Mutant Cell Repository, Newark, New Jersey). This C-gamma 1 gene module was previously combined with a mouse VH gene module to form the chimeric expression plasmid pING2012E.

(5) Chimeric L6 expression plasmids.

Chimeric heavy L6 chain expression plasmids were derived from replacing the VH module pING2012E with the VH modules of mutants delta 21 and delta 4 to obtain 15 expression plasmids pING2111 and pING2112 (Fig. 17). These plasmids direct the synthesis of chimeric heavy L6 chain by transfection into mammalian cells.

For the light chimeric L6 chain gene, the Sall to HindIII phage fragment of the mouse VK module was joined to the human CK module by the method shown in general in FIG. 18, to form pING2119. Replacement of the neo sequence by the E. coli gpt gene derived from pSV2-gpt resulted in pING2120, which expressed chimeric light L6 chain and provides mycophenolic acid resistance upon transfection into mammalian cells.

Inclusion of both heavy and light chimeric chain genes in the same plasmid allows insertion into transfected cells into a 1: 1 gene ratio between heavy and light chain genes, leading to balanced gene dosing. This can improve expression and reduce manipulations of transfected cells for optimal chimeric antibody expression. To this end, the DNA fragments derived from the chimeric heavy and light chain genes of pING2111 and pING2119 were combined in the expression plasmid pING2114 (Fig. 19).

This expression plasmid contains a selectable neoR marker and separate transcription units for each chimeric gene, each including a heavy mu sequence enhancer.

The modifications and V-C association regions of the chimeric L6 genes are briefly shown in FIG. 20th

(6) Stable transfection of mouse lymphoid cells to produce chimeric antibody.

Electroporation was used (Potter et al., Supra; Toneguzzo et al. Mol. Cell. Biol.

10 6: 703, 1986) for insertion of chimeric L6 expression plasmid DNA into Sp2 / 0 mouse cells.

The electroporation technique yielded a transfection frequency of 1-10 x 10 s for the Sp2 / 0 cell bodies.

The train expression plasmid pING2114 was linearized by digestion with 15 Aat11 restriction endonuclease and transfected into Sp2 / 0 cells to obtain ca. 50 G418 resistant clones screened for human heavy and light chain synthesis. The levels of chimeric antibody chain synthesis from the two manufacturers, D7 and 3E3, are shown in Table 3. Chimeric L6 antibody was prepared by culturing D7 transfectant cells for 24 hours at 2x106 cells / ml in 5 L DMEM supplemented with HEPES buffer. and penicillin and streptomycin. The supernatant was concentrated over an "Amicon YM30" membrane in 10 mM sodium phosphate buffer, pH 8.0. The preparation was loaded onto a DEAE cellulose column which separated the immunoglobulin into unbound and bound fractions.

Samples from the DEAE-unbound preparations, DEAE-bound preparations and the PREDAE preparations (from 1.6 µl of medium) were separately purified by affinity chromatography on a pr otei η - A-seph arosis column 1 e eluting with 0.1 M sodium citrate, pH 3.5. The eluted antibody was neutralized and concentrated by "Amicon centricon" filtration in phosphate buffered salt. The yields for the three preparations were 12 µg (DEAE-unbound), 6 µg (DEAE-bound) and 9 µg (pre-DEAE column). Western analysis of the antibody chains indicated that they were combined in an H2L2 tetramer just like naturally occurring immunoglobulins.

59 DK 175581 B1 (7) Another purification for chimeric L6 antibody secreted into tissue culture.

Sp2 / 0.pING2114.D7 cells were grown in culture medium [DMEM (Gibco # 320-1965) supplemented with 10% fetal bovine serum (Hyclone # A-1111-D), 10 mM

5 HEPES, 1 x Glutamine Pen-Strap (Irvine Scientific # 9316) to 1 x 10 6 cells / ml.

b. The cells were then centrifuged at 400 x g and resuspended in serum-free culture medium at 2 x 10 6 cells / ml for 18-24 hours.

The medium was centrifuged at 4000 rpm in a JS-4.2 rotor (3000 x g) for 15 minutes.

d. 1.6 liters of supernatant was then filtered through a 0.45 µm filter and then concentrated over a YM30 filter (Amicon Corp.) to 25 ml.

E. The conductivity of this concentrated supernatant was adjusted to 5.7-5.6 mS / cm and the pH was adjusted to 8.0.

f. The supernatant was centrifuged at 2000 x g for 5 minutes and then loaded onto a 40 ml DEAE column which had been previously equilibrated with 10 mM sodium phosphate, 20 pH 8.0.

g. The flow fraction was collected and loaded onto a 1 ml protein A-Sepharose column (Sigma) pre-equilibrated with 10 mM sodium phosphate, pH 8.0.

The column was washed first with 6 ml of 10 mM sodium phosphate buffer, pH 8.0, followed by 8 ml of 0.1 M sodium citrate, pH 3.5, and then with 6 ml of 0.1 M citric acid (pH value 2.2). 0.5 ml fractions were collected in tubes containing 50 µl of 2 M Trisbase (Sigma).

Most of the IgG was in the eluent at pH 3.5 and collected and concentrated over Centricon 30 (Amicon Corp.) to ca. 0.06 ml.

j. The buffer was changed to PBS (10 mM sodium phosphate, pH 7.4, 0.15 M

5 NaCl) in Centricon 30 by repeated dilution with PBS and reconcentration.

k. The IgG solution was then adjusted to 0.10 ml and bovine serum albumin (Fraction V, U.S. Biochemicals) was added to a concentration of 1.0% as a stabilizing reagent.

(8) Preparation and Purification of Chimeric L6 Antibody Secreted in Bug Water Sewage.

a. The bugwater soot liquid was first centrifuged at 2000 x g for 10 minutes.

15 b. The conductivity of the supernatant was adjusted to 5.7-5.6 mS / cm and its pH was adjusted to 8.0.

c. The supernatant was then loaded onto a 40 ml DEAE cellulose column pre-equilibrated with 10 mM Na 2 PO 4 H, pH 8.0.

20 d. The flow fraction ffa DEAE column was collected and its pH was adjusted to 7.4 and then loaded onto a 1.0 ml goat anti-human IgG (H + L) sepharose column.

E. The column was washed, first with 6 ml of 10 mM sodium phosphate, 0.5 M sodium chloride, followed by 8 ml 0.5 M NH 4 OH and 3 M sodium thiocyanate.

f. The sodium thiocyanate eluate was collected and dialyzed against 2 L of PBS overnight.

The antibody may be further concentrated at steps j. and k. in the former method.

61 DK 175581 B1 TABLE 3

Levels of secreted chimeric L6 chains from Sp2 / O transfectants 5

Sp2 / 0.D7 Sp2 / 0.3E3

Culture conditions FBS Kappab Gammac Kappab Gammac 1.20 ml, 2d, + 17 77 100 700 graft @ 2xl05 / ml 10 2,200 ml, 2d, + 0.9 6 80 215 graft @ 2.5x105 / ml 3,200 ml, ld, - 1.9 3.8 97 221 15 graft @ 2x106 / ml

4. Balb / c bug water soot - 5,160 19,170 ND ND

ascites 20 α - Sp2 / 0 cells transfected by electroporation with pING2114 (pL6HL) b - p.g / 1 measured by ELISA specific for human kappa - human Bence-Jones protein standard.

c - μg / I measured by ELISA specific for human gamma - human IgG standard.

ND - not determined.

25 FBS: fetal bovine serum.

(9) Studies on the chimeric L6 antibody.

The samples were first assayed by a binding assay in which cells of both a L6 antigen positive and L6 antigen negative cell line were incubated with standard monoclonal mouse antibody L6, chimeric L6 antibody derived from the cell culture supernatants, and 62 L 17 ~ chimeric L6 antibody. derived ffa bug water soot (ascites) (as previously described) followed by another reagent, fluorescein isothiocyanate (FITC) -conjugated goat antibodies for human (or mouse, by standard) immunoglobulin.

Since the binding assay showed strong reactivity of the chimeric L6 on the L6 antigen-positive cell line and total lack of reactivity on the negative cell line, the next step was to examine the ability of the chimeric L6 to inhibit the binding of mouse L6 to antigen-positive cells. cells; such inhibition assays are routinely used to establish the identity of two antibodies recognizing antigen. These 10 data are discussed below ("binding inhibition"). As part of these studies, a thorough assessment of antibody zeal was made.

Finally, two aspects of antibody function were investigated, the ability to mediate ADCC in the presence of human peripheral blood leukocytes and the ability to kill L6-positive 15 tumor cells in the presence of human serum as a complement source (see "Functional Assays" below).

Bindingsanalvser. Cells from a human colon carcinoma line, 3347, which had previously been shown to express ca. 5 x 10 5 molecules of the L6 antigen at the cell surface, 20 were used as target cells. Cells from the T cell line HSB2 were used as a negative control since, according to a previous study, they do not express detectable amounts of the L6 antigen. The target cells were first incubated for 30 min at 4 ° C with either the chimeric L6 or with the L6 mouse standard which had been purified from mouse bug water soot. This was followed by incubation with another FITC-labeled reagent, which for the chimeric antibody was goat anti-human immunoglobulin, delivered from TAGO (Burlingame, CA) and used at a dilution of 1:50. For the mouse standard, it was goat anti-mouse immunoglobulin, also delivered from TAGO and used at a dilution of 1:50. Antibody binding to the cell surface was determined using a "Coulter Model EPIC-C" cell sorting apparatus.

30 63 DK 175581 B1

As shown in Tables 4 and 4A, both the chimeric L6 and mouse standard L6 bound significantly and ca. similar to the L6-positive 3347 line. They did not tie in background to the L6 negative HSB2 line.

5 In view of the fact that the three different chimeric L6 samples presented in Table 4 behaved similarly in the binding assays, they were pooled for the inhibition studies described below. The same inhibition studies were carried out for chimeric L6 derived from the aqueous effluent presented in Table 4A.

10 Inhibition of binding. As the next step, the extent to which graduated doses of the chimeric L6 antibody or standard mouse L6 could inhibit the binding of a FITC-labeled mouse L6 to the surface of antigen-positive 3347 colon carcinoma cells was investigated.

Both the chimeric L6 and mouse standard L6 inhibited the binding of the directly labeled L6 antibody, the binding curves being parallel. The chimeric antibody was slightly less effective than the standard, as indicated by the results, which showed that 3.4 µg / ml of the chimeric L6 MAb chimeric compared to 2.0 µg / ml of the standard mouse L6 MAb was required. to 50% inhibition of the binding, and that 5.5 µg / ml 20 of the chimeric L6 (derived ffa abdominal water soot) compared to 2.7 / tg / ml of standard mouse L6 MAb was required for 50% inhibition of binding.

As part of these studies, a rough estimate of antibody zeal was made.

The udder of the standard mouse L6 was previously determined to be approx. The data 25 indicated that there were no significant differences in zeal between the chimeric L6 and mouse L6.

Functional analyzes. A comparison was made between the ability of the chimeric L6 and standard mouse L6 to lyse L6 antigen-positive cells in the presence of human peripheral blood leukocytes as a source of effector cells (mediating antibody-dependent cellular cytotoxicity, ADCC) or human serum as a complement source (mediating complement-dependent cytolysis, CDC).

As shown in Table 5 and Tables 5A-5D, the chimeric L6 was superior to the 5 co-examined mouse L6 sample to cause ADCC, as measured by a 4 hour 51 Cr release test.

Tables 6 and 6A-6B show the data ffa studies of complement mediated target cell lysis. In this case, a high cytolytic activity was observed with both 10 mouse L6 antibodies and chimeric L6 antibodies.

Conclusions.

The above results show a number of important unexpected qualities of the 15 chimeric L6 chimeric antibody of the invention. First, the chimeric L6 antibody binds to L6 antigen-positive tumor cells for ca. same size as mouse standard L6 and with approx. the same zeal. This is significant for the following reasons. The L6 antibody defines (a) a carbohydrate surface antigen and (b) a protein antigen of ca. 20,000 daltons, both characteristic of non-small cell lung carcinoma (NSCLC) and certain other human carcinomas. It is worth noting that the L6 antibody does not detectably bind to normal cells, such as fibroblasts, endothelial cells, or epithelial cells of the major organs. Thus, the L6 chimeric monoclonal antibody defines an antigen that is specific to carcinoma cells and not normal cells.

25

In addition to the ability of the L6 chimeric monoclonal antibodies of the present invention to specifically bind to malignant cells and localize tumors, the chimeric L6 exhibits profound biological effects upon binding to its targets, making the chimeric antibody an excellent candidate for tumor immunotherapy. The 30 results presented herein show that chimeric L6 is capable of binding to tumor cells and upon binding kills the tumor cells either at ADCC or CDC. Such tumor-killing activity was detected using concentrations of chimeric L6 antibody as low as 0.01 pg / ml (10 ng / ml).

Although the prospect of trying tumor therapy using monoclonal antibodies is attractive, with some partial tumor relapse reported, such monoclonal antibody therapy has so far been met with limited success (Houghton, February 1985, Proc. Natl. Acad. Sci. 82 : 1242-1246). The therapeutic efficacy of mouse monoclonal antibodies (which have been tried so far) seems to be too low for most practical purposes. The discovery of the excellent biological activity of chimeric L6 associated with its specificity for a carcinoma antigen makes the chimeric L6 antibody a well-chosen therapeutic agent for the treatment of tumors in vivo. Furthermore, because of the "human" properties that will make the chimeric L6 chimeric antibodies more resistant to clearance in vivo, the chimeric L6 chimeric antibodies will advantageously not only be used for therapy with 15 unmodified chimeric antibodies, but also for the development of various immunoconjugates with drugs, toxins, immunomodulators, isotopes, etc., as well as for diagnostic purposes such as in vivo imaging of tumors using appropriately labeled chimeric L6 antibodies. Such immunoconjugation techniques are known to those skilled in the art and can be used to modify the chimeric L6 antibody molecules of the present invention.

Two illustrative cell lines secreting chimeric L6 antibody were deposited prior to the filing date of the present application at ATCC, Rockville, Maryland. These are transfected hybridoma C255 (corresponding to 3E3 cells, supra), ATCC HB 9240 and transfected hybridoma C256 (C7 cells, supra), ATCC HB 9241.

The present invention is not to be limited in scope to these deposited cell lines, since the deposited embodiment is intended as a single illustration of one aspect of the invention and all cell lines which are functionally equivalent are within the scope of the invention. Various modifications of the invention in addition to those shown in the art from the foregoing description and accompanying drawings are intended to be within the scope of the appended claims.

TABLE 4 Binding Assays of Chimeric L6 Antibody and Monoclonal L6 Mouse Antibody on an L6 Antigen-Positive and L6 Antigen-Negative Cell Line.

Binding conditions of 1 H3347 cells (L6 +)

10 Antibody Batch GAM GAH

Standard L6 56.6 4.2

Chimeric L6 a 1.3 110.3 b 1.3 110.3 c 1.3 110.3

Binding Ratio of1 HSB-2 Cells (L6

20 GAM GAH

Standard L6 1.1 1.1

Chimeric L6 a 1.0 1.0 b 1.0 1.1 25 c 1.0 1.1

All assays were performed using an antibody concentration of 10 µg / ml. The binding ratio is the number of times a test sample is more clear than a control sample treated with GAM (FITC-conjugated goat anti-mouse) or GAH (FITC-conjugated goat anti-human) alone. A ratio of 1 means that the test sample is as clear as the control; a ratio of 2 means that the test sample is twice as clear as the control, etc.

67 DK 175581 B1

TABLE 4A

Binding assays of chimeric L6 antibody and mouse monoclonal antibody on an L6 antigen positive and L6 antigen negative cell line.

Antibody Binding Ratio of 1 Concentration H3347 Cells (L6 +)

Antibody (qg / mO GAM GAH

Standard L6 30 38 4 10 49 4 10 3 40 3

Chimeric L6 30 2 108 (Bugwattersot) 10 2 108 3 1 42 15 Chimeric L6 30 1 105 (Cell Culture) 10 1 86 3 1 44

Binding Ratio of 20 HSB-2 Cells fL6 -)

GAM GAH

Standard L6 10 1 1

Chimeric L6 10 1 1 (Bugwattersot) 25 Chimeric L6 10 1 1 (Cell Culture)

The binding ratio is the number of times a test sample is more clear than a control sample 2 treated with GAM (FITC-conjugated goat anti-human) alone. A ratio of 1 30 means that the test sample is as clear as the control; a ratio of 2 means that the test sample is twice as clear as the control, etc.

TABLE 5 ADCC of chimeric L6 antibodies and standard (mouse) L6 antibodies on colon carcinoma cell line 3347.

68 DK 175581 B1 5 Antibody concentration PBL per%

Antibody µg / ml) target cell Cytolvse1

Chimeric L6 10 100 64 10 5 100 70 10 0 2

Standard L6 10 100 24 5 100 17 15 10 0 2

None 0 100 1 The target cells had been labeled with 51 Cr and were affected for 4 hours by a combination of MAb and human peripheral blood leukocytes (PBL), and the release of 51 Cr was measured 20 successively. The release of 51 Cr (after corrections for values of spontaneous release from untreated cells) is a measure of the percentage cytolysis.

TABLE 5A

ADCC of chimeric L6 antibodies and standard (mouse) L6 antibodies on colon carcinoma cell line 3347.

69 DK 175581 B1 5 Antibody concentration PBL per%

Antibody (qg / mO target cell Cytolvse *

Chimeric L6 20 100 80 (Bugwattersot) 10 100 74 10 5 100 71 2.5 100 71 20 0 0

Chimeric L6 10 100 84 (cell culture) 5 100 74 15 2.5 100 67 10 0 3

Standard L6 20 100 32 10 100 26 20 0 0 20 _ * The target cells had been labeled with 51 Cr and were affected for 4 hours by a combination of MAb and human peripheral blood leukocytes (PBL) and the release of 51 Cr was measured subsequently. The release of 51 Cr (after corrections for values of spontaneous release from untreated cells) is a measure of the percentage cytolysis.

25

TABLE 5B

ADCC of chimeric L6 antibodies and standard (mouse) L6 antibodies on colon carcinoma cell line 3347.

70 DK 175581 B1 5 Antibody concentration PBL per%

Antibody (qg / ml) Target Cell Cytolvse *

Chimeric L6 5 100 84 (Bugwatersot) 2.5 100 78 10 1.25 100 85 0.63 100 81 0.31 100 80 0.16 100 71 0.08 100 65 15 5 0 0

Standard L6 5 100 32 5 0 0

None The target cells had been labeled with 51 Cr and were affected for 4 hours by a combination of MAb and human peripheral blood leukocytes (PBL), and the release of 5 Cr was measured subsequently. The release of 51 Cr (after corrections for values of spontaneous release from untreated cells) is a measure of the percentage cytolysis.

25

TABLE 5C

ADCC of chimeric L6 antibodies and standard (mouse) L6 antibodies on lung carcinoma cell line H2669.

71 DK 175581 B1 5 Antibody concentration PBL per%

Antibody (tcg / ml) Target Cell Cytolvse *

Chimeric L6 10 100 35 · (Bugwatersot) 1 100 31 10 0.1 100 27 0.01 100 15 0.001 100 13 0.0001 0 15 15 Standard L6 10 100 9 1 100 15

None 0 100 9 20 Chimeric L6 10 10 19 (Bug water soot) l 10 15 0.1 10 11 0.01 10 13 0.001 10 22 25 0.0001 10 11

Standard L6 10 10 7 1 10 6

None 0 10 S

Chimeric L6 10 0 4 30 (Bugwattersot)

Standard L6 10 0 9 72 DK 175581 B1 * The target cells had been labeled with 51 Cr and were affected for 4 hours by a combination of MAb and human peripheral blood leukocytes (PBL) and the release of 51 Cr was measured subsequently. The release of 51 Cr (after corrections for spontaneous release 5 values from untreated cells) is a measure of the percentage cytolysis.

TABLE 5D

ADCC of chimeric L6 antibodies and standard (mouse) L6 antibodies on colon carcinoma cell line H3347.

10 _

Antibody concentration PBL per%

Antibody µg / ml) Target Cell Cytolysis1

Chimeric L6 10 100 62 15 (Bug water soot) 1 100 66 0.1 100 69 0.01 100 26 0.001 100 8 0.0001 0 3 20 10 0 0

Standard L6 10 100 19 1 100 24 0 0 25

None 0 100 8 The target cells had been labeled with 51 Cr and were affected for 4 hours by a combination of MAb and human peripheral blood leukocytes (PBL), the release of 51 Cr (after corrections 30 for values for spontaneous release from untreated cells) is a target of the percent cytolysis.

73 DK 175581 B1 TABLE 6

Complement-dependent cytotoxic effect of chimeric L6 and standard (mouse) -L6 on colon carcinoma cells ffa line 3347 as measured by a 4-hour 5lCr release assay.

5 Human serum from a healthy individual was used as a complement source.

Antibody Human Complement% Cytolysis 10 L6 Standard 10 µg / ml Yes 90 L6 Chimeric 10 µg / ml Yes 89 L6 Standard 10 µg / ml No 0 L6 Chimeric 10 µg / ml No 1 15 _ 74 DK 175581 B1

TABLE 6A

Complement-dependent cytotoxic effect of chimeric L6-ten substances and $ tandard (mouse) -L6 antibodies on colon carcinoma cell line 3347 5 Antibody concentration PBL per%

Antibody µg / ml) target cell Cytolvse *

Chimeric L6 20 + 29 10 (Bug water soot) 10 + 23 5 +18 2.5 + 8 20 Inactivated 0 10 0 0 15

Chimeric L6 20 + 29 (Cell Culture) 5 +26 2.5 + 18 20 +4 20 10 0 4

Default L6 20 + 55 10 + 37 20 Disabled 0 25 20 0 1

No 0 + 0 * Complement-mediated cytolysis was measured by a 4 hour 51 Cr release assay.

30 Human serum from a healthy individual was used as a complement source.

75 DK 175581 B1

TABLE 6B

Complement-dependent cytotoxic effect of chimeric L6 antibodies and standard (mouse) L6 antibodies on colon carcinoma cell line 3347 5 Antibody concentration PBL per%

Antibody (µg / ml) target cell Cytolvse *

Chimeric L6 10 + 209 10 (Bug water soot) 5 + 155 2.5 + 166 1.25 + 114 0.6 + 63 0.3 +17 15 10 0 0

Standard L6 10 + 96 5 +83 2.5 + 48 20 1.25 + 18 0.6 + 7 0.3 + 4 10 0 2 25 None 0 +0 * Complement-mediated cytolysis was measured by a 4 hour 51 Cr release assay .

Human serum from a healthy individual was used as a complement source.

Example IV: A Chimeric Human Mouse Antibody with Specificity for Human B Cell Antigen 76 DK 175581 B1

The 2H7 monoclonal mouse antibody (gamma 2i> K) recognizes a human B cell surface antigen, Bp35 (Clark, E. A., et al., Proc. Nat. Acad. Sci. USA, 82: 1766 (1985)). The Bp35 molecule plays a role in B cell activation. mRNA was prepared from the 2H7 cell line. Two cDNA libraries were generated using 5 of the heavy UIG-H chain primer and the other, oligo (dT). A VH clone, pH2-1, was isolated after screening with the same UIG-H oligonucleotide. To isolate the light chain clone, a kappa-specific DNA mouse fragment was used to screen the oligo (dT) library. Candidate clones were further screened with JK5 mouse sequences. Thus, a VK clone, pL2-12, was isolated. The light chain UIG-K was then used to genetically engineer a restriction enzyme site in the J region.

The two cDNA clones were also modified at the 5 'end to remove the artificial oligo d [C] sequence. In pH2-1 this was done using the restriction enzyme NcoI, which cuts a 5 'nucleotide residue for the ATG initiator codon. In pL2-12, this was achieved by an oligonucleotide in vitro mutagenesis using a 22-mer containing a SalI site.

The DNA sequences of these two clones are shown in FIG. 21 and 22. To construct the chimeric heavy chain plasmid, the VH module was joined to the human C-gamma 1 module 20 (pGMH6) at the JH-BstEII site, and the chimeric light chain, the VK module, was joined to it. human Cjc module (pGML60) at the JK-HindIII site. The expression vector sequences were derived from pING2012 neo as well as pING2016 gpt. The plasmids constructed are pING2101 (VHC-gamma 1-neo), pING2106 (VKCK-neo) and pING2107 (Vi <.CK-gpt). pING2101 and pING2106 were also used to generate plasmids containing both genes. These are pHL2-1 and pHL2-26. Furthermore, pING2106 and pING2014 were combined into a plasmid containing two light chains, pLL2-25, to compensate for the poorer (in comparison with heavy chain) steady-state light chain protein accumulation in transfected cells (see Fig. 23). . In FIG.

24 the changes are made to the variable region sequences during construction.

30 77 DK 175581 B1

The plasmid, pHL2-1, was linearized by Aat11; and the DNA was used to transfect Sp2 / 0 cells by electroporation. Transformants were selected in G418-DMEM. A transformant, 1C9, produces 9.3 ng / ml chimeric kappa protein and 33-72 ng / ml chimeric gamma 1 protein, as determined by ELISA. Southem analysis of 5C9 DNA revealed that a copy of the plasmid integrated into the Sp2 / 0 genome.

EXAMPLE V: Secretion of a Functional Chimeric Antibody from Yeast (1) Association of fully developed light and heavy chimeric L6 chain genes to the yeast invertase signal sequence and truncated phosphoglycerate kinase (PGK promoter).

Yeast cells are capable of recognizing mammalian secretory signal sequences and directing secretion of mammalian proteins (Hitzman et al., Supra). However, there is evidence to suggest that certain naturally occurring yeast signal sequences are more effective than mammalian signal sequences to direct secretion of some mammalian proteins ffa yeast (Smith et al, Science 229: 1219 (1985)). An example is the yeast invertase gene signal sequence. To improve efficiency of derived and heavy chain secretion, the fully developed light and heavy chain sequences were joined to the yeast invertase signal sequence and placed under transcription control by the truncated PGK promoter (US Patent Application No. 797,477) using the strategies outlined in Figs. 25 and 26. An important element of these constructs is the use of in vitro mutagenesis to introduce a restriction site at the signal sequence processing site of both the invertase signal sequence (see U.S. Patent Application No. 797,477) and the light and heavy chain genes. These restriction sites are positioned such that a blunt-ended ligation of restriction enzyme-digested T-4 DNA polymerase-treated DNA results in in-phase translation associations of the 5 'end of the fully developed immunoglobulin chains with the 3' end of the yeast invertase signal sequence. . Such genes, when expressed in a yeast cell, can direct the synthesis, reprocessing and secretion of deleterious and heavy chimeric chains with the same primary peptide sequence as chimeric light and heavy chains secreted from transfected mouse Sp2 / 0 cells. The DNA sequences of the mutagenesis primers used for light and heavy chain genes, as well as the corresponding non-mutagenized sequences, are shown in Figs. 25B and 26B. Using this method, the light and heavy chimeric L6 chains were joined to the yeast invertase signal sequence and truncated PGK promoter, resulting in plasmids pING 1407-7 and pING1415 (Figures 25C and 26C).

5 (2) Removal of non-yeast 3 'untranslated DNA.

Recent studies on expression of hepatitis B surface antigen in yeast have shown that removal of non-yeast 3 'and 5' untranslated sequences may result in increased 10 levels of heterologous gene expression in yeast (Knieskin et al., Gene, 46 : 135 (1986)). The light chain sequence of chimeric L6 antibody in pING 1407-7 (Fig. 25C) contains approx. 200 bp 3 'untranslated DNA followed by 70 bp poly A sequence and 20 bp poly G sequence. An initial treatment of the chimeric light L6 chain DNA with double-stranded exonuclease Bal31 removed the poly A and poly G sequences and all, except 90 bp 3 'untranslated DNA, to generate the plasmid pING2121b (FIG.

27). A restriction fragment from pING2121b, which contained only C1, was cloned into a derivative of pBR322 to form pING1419 (Fig. 27). Another Bal31 digest was then used to remove everything except 13 bp non-yeast 3 'untranslated DNA to generate plasmid pING 1431 (Fig. 27). The chimeric L6 heavy chain gene of pING1415 20 (Fig. 26) also contains extensive 3 'untranslated sequence which includes 80 bp poly A. All but 11 bp of the 3' untranslated DNA were removed using the embodiment of FIG. 28 showed the strategy for forming the plasmid pING1429.

Site-directed in vitro mutagenesis can introduce undesirable base-pair changes at regions of DNA outside the region being mutagenized at a low frequency. To ensure that such mutations were not present in the light and heavy chimeric L6 chain sequences that had been cloned into M13 and subjected to site-directed mutagenesis, light and heavy chain genes were constructed to combine with the invertase signal sequence and the truncated PGK gene. promoter, which consisted of coding sequences that were either determined by DNA sequence analysis or had been shown to be functional by virtue of their expression in transfected mouse Sp2 / 0 cells to produce functional chi antibody. The plasmids pING1439 (light chain, Fig. 27) and pING1444 (heavy chain, Fig. 28) were formed by these constructs.

(3) Construction of yeast expression plasmids containing light and heavy chimeric 5 L6 chain genes from pING1439 and pING1436, respectively, joined to the PGK polyadenylation signal.

In order for yeast to produce an intact functional antibody molecule, a balanced synthesis of both light and heavy chain protein within the host cell is preferred. One approach is to apply the light and heavy chain genes to separate expression vectors, each containing a different selective marker. A yeast strain deficient in selective markers found on the plasmids can then be transformed either simultaneously or sequentially with these plasmids.

The light and heavy chimeric L6 chain genes from pING1439 (Fig. 27) and pING1436 (Fig. 28) were cloned as BglII-XhoI and BamHI-XhoI phagem, respectively, in two different medium copy numbers (about 20 copies / cell) - expression vectors (yeast E. coli pendulum). One of these, pING804CVS, contains the complete yeast 2-μτη ring, the PGK transcription termination and polyadenylation signals, and the leu2 gene as the selective marker.

Another vector, pING1150, contains yeast replication origins, oriY, a cis-acting sequence (REP3) from the endogenous 2-μτη yeast plasmid, the PGK transcription termination and polyadenylation signals, and the ura3 gene as the selective marker. Both plasmids also contain the B-lactamase gene (bia) for ampicillin resistance and bacterial replication origin (oriB) of pBR322 for selection and propagation in bacteria.

Four plasmids resulted from these constructs: pING 1441 light chain, leu2 and pING 1443 light chain, ura3 (Fig. 29); pING1440 heavy chain, leu2 and pING1442 heavy chain, ura3 (Fig. 30).

(4) Secretion of chimeric L6 antibody from transformed yeast cells.

30 DK 175581 B1

LAKE

Two separate transformation experiments were performed in an attempt to achieve both light and heavy chain synthesis in yeast cells. Four µg each of pING1444 and pING1444 and separately of pING1442 and pING1444 were cotransformed into Saccharomyces cerevisiae strains, BB331C (MATa, ura3, leu2) by selection for growth on SD agar (2% 5 glucose, 0.67% gemrogen base). , 2% agar). Ura + Leu + transformants were obtained by 2-3 days incubation at 30 ° C. Ca. 100 transformants were obtained for pING1440 plus pING1443; and only 15 transformants were obtained for pING1442 plus pING1441. Ten colonies were seeded from each plate in 5 ml SD decoction supplemented with 50 mM sodium succinate, pH 5.5, and grown for 65 hours at 30 ° C. The cells were removed by centrifugation and the culture supernatants analyzed by ELISA for light chain and heavy chain levels and for the degree of association of the secreted light and heavy chains.

The latter was assessed using goat anti-human kappa antiserum to coat the microtiter wells and a peroxidase-labeled goat anti-human gum ten serum to detect the level of heavy chain bound to the anti-kappa coating.

The results of these analyzes (Table 7) revealed that all the culture supernatants of the cells transformed with pING1440 (heavy chain, leu2) plus pING1443 (light chain, ura3) contained a disproportionately high level of light chain protein relative to the levels of heavy chain protein and none. signs (at least as determined by ELISA) on total light and heavy chains. On the other hand, the supernatants from the 20 cells transformed with pING1442 (heavy chain, ura3) + pING1441 (light chain, leu2) contained a more balanced production of light and heavy chain proteins, and eight out of ten isolates were found to contain some total light and heavy chains as determined by ELISA. Two of these isolates, # 1 and # 5, produced a significant proportion of total light and heavy chain.

81 DK 175581 B1 TABLE 7 LEVELS OF SECURED LIGHT AND HEAVY CHEMICAL L6 CHAIN USED BY YEAR TRANSFORMERS8

Kappa / 5 Pladmiderb Isolate No. Kappac Gammad Gamma6 pING1440 + 1 284 39 0 pING1443 2 324 33 0 3 473 52 0 4 387 40 0 10 5 316 34 0 6 188 28 0 7 381 45 0 8 455 45 0 9 380 26 0 15 10 579 32 0 pING1441 + 1 128 79 35 pING1442 2 150 30 1 3 124 29 0 20 4 185 55 5 5 114 52 35 6 139 23 5 7 149 34 5 8 245 57 12 25 9 202 26 11 10 157 19 7 a. S. cerevisiae strain BB331C (MATa, leu2, ura3) transformed into Ura + Leu + with plasmids carrying ura3 and leu2 with light or heavy chains.

Plasmids: pING1440 = heavy chain + leu2; 82 DK 175581 B1 pING1443 = light chain + ura3; pING1442 = heavy chain + ura3; pING 1441 = light chain + leu2.

5 cng / ml as measured by ELISA specific for human kappa with human Bence Jones protein as standard.

d. ng / ml measured by ELISA specifically for human gamma with human IgG standard.

10 µg / ml target by ELISA using anti-human kappa as coating antibody and anti-human gamma as other antibody with human IgG standard.

Further analysis was performed to determine if this compound was the result of the synthesis of a H2L2 size protein. The culture supernatants from isolates # 1 and 5 as well as from isolate # 8, which contained a much lower level of apparent light and heavy chain association, were concentrated by ultrafiltration on a "Centricon 30" filter (Amicon Corp.). The concentrated supernatants were run on a 7% polyacrylamide gel under non-reducing conditions, imprinted with nitrocellulose and probed with goat anti-human kappa antiserum followed by peroxidase-labeled rabbit anti-goat antiserum. 1 and 5, but not from # 8, contained a single immunoreactive band that migrated with the purified chimeric L6 antibody from transfected Sp2 / 0 cells, these results suggest that isolates # 1 and 5 synthesized and secreted total chimeric L6 antibody.

(5) Purification of chimeric L6 antibody from yeast culture supernatant.

In order to further characterize the H2L2 size protein secreted by the yeast and determine if this was a chimeric L6 antibody overall, a sufficient amount of yeast-produced material was purified to allow for the performance of various binding assays and functional assays. The pING1442 + 1441 transformant isolate # 5 was grown for 58 hours at 30 ° C in a 10 liter fermentor using a synthetic medium (Table 8). The cells were initially cultured in 9 liters of the column A medium until the glucose level dropped below 1 g / l, at which time they were fed a total volume of 2.5 liters of medium from column B. Glucose levels were maintained at 0 , 5 g / l during the remaining course of the fermentation. The cells were removed by centrifugation and the 5 culture supernatants were analyzed by ELISA for the presence of light and heavy chain proteins and for association of the heavy and light chains. The supernatant contained approx.

250 μg / l light chain, 240 μg / l heavy chain and 130 μg / 1 heavy chain united with light chain. The culture supernatants were then concentrated by ultrafiltration over a D.C.

1 unit (Amicon Corp.), filtered through a 0.45 μη \ filter and concentrated over a 10 YM30 filter (Amicon Corp.) to 250 ml. The concentrated supernatant was adjusted to a pH of 7.4 with KOH, brought to 500 ml with PBS (10 mM, pH 7.4, 150 mM sodium chloride) and loaded onto a 1 ml protein A Sepharose column ( Sigma) which was previously equilibrated with PBS. The column was washed first with 20 ml of PBS followed by 10 ml of 0.1 M sodium citrate, pH 3.5 and then with 10 ml of 0.1 M citric acid, pH 2.2. The eluates at pH 3.5 and 2.2 were each collected in a tube containing 1 ml of 2 M Tris base (Sigma). The majority of the light and heavy immunoreactive chain proteins were in the eluate at pH 3.5, which was then concentrated over Centricon 30 (Amicon Corp.) to a final volume of 106 μΐ. Analysis of this protein on non-reducing polyacrylamide gels using coomassie blue staining and immuno-imprinting with anti-human kappa antiserum (Sigma) to visualize the proteins revealed a H2L2 size protein band of 150 kilodaltons. This protein was purified away from other proteins by HPLC using ABX 5-μιη column equilibrated with Buffer A (10 mM KPO4, pH 6.8). After applying the sample to the column, the column was washed with 25 buffer A for 10 minutes (flow rate = 1 ml / minute) and subjected to a linear gradient of 0% to 50% buffer B (250 mM KPO4, pH 6.8) in over 50 minutes at 1 ml / minute.

84 DK 175581 B1 TABLE 8 MEDIUM USED FOR Yeast fermentation for the preparation of SECERED CHEMICAL L6 ANTIBODY 2 Components Ab Bc 5 1. Cerelose (Glucose) 119 g / 1,538 g / 1 2. (NH4) 2S04 13.9 g / 1 83, 3 g / l 3. Thiamine HCl 0.011 g / l 0.05 g / l 4. Biotin 0.00011 g / l 0.005 g / l 5. Pantothenic acid 0.002 g / l 0.009 g / l 6. Inositol 0.194 g / l 0.875 g / l 7. H3 PO4 5.67 ml / l 25.5 ml / l 8. KH2 PO4 5.78 g / l 26.0 g / l 9. MgSO4.7H20 3.33 g / l 15.2 g / l 1 10. CaCl2.2H2 0.33 g / l 1.5 g / l 11. FeSO4.7H2O 0.072 g / l 0.34 g / l 12. ZnSO4.7H20 0.022 g / l 0.104 g / l 13. MnCl2 .4H20 0.0039 g / 1 0.018 g / 1 14. CuS04.5H20 0.0067 g / 1 0.031 g / 1 15. Wife. H 2 SO 4 0.0056 ml / l 0.026 ml / l 20 a. Fermentation was carried out as described in the text.

b. Ingredients of 9-liter starter batch.

c. Components of 2.5-liter food batch.

The bulk of the protein was dissolved in a single large wide peak between 20 and 50 minutes as determined by absorbance at 280 nm. Another smaller peak was observed at 52-56 minutes, which corresponded to the normal elution position of chimeric L6 antibody from transfected Sp2 / 0 cells. ELISA analysis of the column fractions showed a larger heavy + light chain cross-reactive peak corresponding to the UV absorbance peak at 52-56 30 minutes. Analysis of the 52-56 minute fractions on non-reducing SDS-polyacrylic amide gels using coomassie blue staining and immuno-imprinting showed a predominantly pure protein that comigrated with chimeric L6 antibody purified from transfected Sp2 / 0 cells. .

(6) Studies on the chimeric L6 antibody 5 secreted by yeast.

The purified yeast-derived antibody was evaluated for function in several ways. First, the purified antibody was tested for its ability to bind directly to an L6 anti-gene positive cell line. Second, the antibody was tested for its ability to inhibit binding of mouse L6 antibody to antigen-positive cells. Finally, the purified antibody was tested for two aspects of antibody function - the ability to mediate ADCC in the presence of human peripheral blood leukocytes and the ability to kill L6-positive tumor cells in the presence of human complement.

15 Direct binding analysis. Cells from a human colon carcinoma line, 3347, which express approx. 5 x 10 5 molecules of the L6 antigen per cell on the cell surface were used as target cells. Cells from the T cell line, T51, were used as a negative control since, according to previous study, they do not express detectable amounts of the L6 antigen. The target cells were first incubated for 30 min at 4 ° C with either 20 Sp2 / 0 cell or yeast derived chimeric L6 antibody or with mouse L6 antibody standard purified from mouse bug water soot. This was followed by incubation with FITC-labeled goat anti-human immunoglobulin for the chimeric antibodies or with FITC-labeled goat anti-mouse immunoglobulin for the mouse standard. Both labeled antibodies were delivered from TAGO (Burlingame, CA) and used at a dilution of 1:50.

Antibody binding to the cell surface was determined using a Coulter Model EPIC-C cell sorting apparatus.

As shown in Table 9, both the mammalian and yeast-derived chimeric L6 antibodies bound significantly and ca. similar to the L6-positive 3347 line. They did not tie beyond 30 backgrounds to the L6 negative T51 line.

86 DK 175581 B1

Inhibition of binding. As the next step, the chimeric L6 yeast antibody and the Sp2 / 0 cell-derived chimeric L6 antibody were examined for their ability to inhibit the binding of a FITC-labeled mouse L6 antibody to the surface of antigen-positive 3347 colon carcinoma clay.

5 Both the yeast-derived and Sp2 / 0-derived chimeric L6 antibodies inhibited the binding of labeled mouse L6 antibody and the binding curves were parallel. Based on the results of these studies, a rough estimate of antibody levels was made. The iver of the Sp2 / 0 cell-derived chimeric L6 was previously determined to be about The data 10 indicated that there were no significant differences between the yeast of yeast derived chimeric L6 antibodies and Sp2 / 0 cell derived chimeric L6 antibodies for the L6 antigen.

Functional analyzes. A comparison was made between the ability of the yeast-derived chimeric L6 antibodies, the Sp2 / 0 cell-derived chimeric L6 antibodies and the standard mouse L6 antibodies to lyse L6 antigen-positive cells in the presence of human peripheral blood leukocytes. as a source of effector cells mediating antibody-dependent cellular cytotoxicity (ADCC). As shown in Table 10, the yeast chimeric L6 was slightly better than Sp2 / 0 cell-derived chimeric L6, and as previously observed, both were superior to standard mouse L6 to cause ADCC, as measured by a 4 20 hour 5 Cr -frigivelsestest.

A comparison was then made between yeast-derived chimeric L6 antibodies, Sp2 / 0 cell-derived chimeric L6 antibodies and standard mouse L6 antibodies for their ability to lyse L6 antigen-positive cells by complement-dependent cytolysis (CDC) when human serum is reached. was used as a complement source. The results of this comparison (Table 11) showed that while both the Sp2 / 0 cell-derived chimeric L6 antibodies and the standard mouse L6 antibodies showed high cytolytic activity, the yeast-derived L6 antibody could not induce any cytolysis even at the highest antibody concentration. These results were unexpected and show that the yeast-derived antibody 30 has new and unique properties.

87 DK 175581 B1 (7) Conclusions

A method is described whereby yeast can be genetically engineered to secrete functional antibodies. The yeast-derived chimeric antibody in this example binds to the target target antigen by ca. the same zeal as the chimeric antibody produced by lymph (Sp2 / 0) cells. The yeast-derived antibody also exhibits similar ADCC activity as Sp2 / 0-derived antibody. In contrast to the Sp2 / 0 cell-derived antibody, the yeast-derived antibody did not exhibit any CDC activity, thus demonstrating the novel and unique properties of this yeast-derived antibody. The method should be practically useful for the preparation of a wide variety of monoclonal antibodies and chimeric antibodies which carry selected antigen binding regions bound to a selected constant region isotype. Genetic antibodies and derivatives thereof produced in yeast will also exhibit novel functional properties, e.g. the ability to selectively mediate target killing at ADCC without any detectable CDC activity. The technology 15 described herein may also be suitable for the production of various other heterologous multimeric secreted proteins by genetically engineered yeast.

TABLE 9

BINDING ANALYSIS OF CHIMARIUS L6 ANTIBODY PRODUCED BY Yeast OR 20 MUSE-SP2 / 0 CELLS ON AN L6 ANTIGEN POSITIVE AND ONE

L6-ANTIGEN-NEGATIVE CELL LINE Binding conditions for: H3347 cells T51 cells

Antibody fL6 +) (L60 Standard mouse L6 95 1.0

Sp2 / 0 chimeric L6 116 1.0 Yeast chimeric L6 116 1.0 a. All antibodies were used at a concentration of 10 µg / ml.

30 b. The binding ratio is the number of times a test sample is more clear than a control sample treated with FITC-conjugated second antibody. Goat anti-mouse antibody was used as the second monoclonal mouse L6 standard antibody. Goat antihuman antibody was used as the second chimeric yeast L6 antibody and chimeric Sp2 / O-L6 antibody.

5 TABLE 10 ADCC OF CHEMICAL L6-ΑΝΤΙ SUBSTANCE DERIVED FROM Yeast OR Sp2 / 0 Cells AND STANDARD (MUSE) -L6 ANTIBODY ON THICKNESS CARCINOM CELL LINE 3347 10 Antibody concentration

Antibody µg / ml)% cvtolvse1

Standard mouse L6 5.0 42 15 1.0 48

Sp2 / 0 chimeric L6 1.0 96 0.1 71 0.01 54 0.001 37 20 Yeast chimeric L6 1.0 114 0.1 108 0.01 76 0.001 60

None The target cells had been labeled with 51Cr and were exposed to a combination of MAb and human peripheral blood leukocytes at 100 per target cell for four hours, and the release of 51Cr was measured subsequently. The release of 5 Cr (after corrections for values of 30 spontaneous release from untreated cells) is a measure of the percentage cytolysis.

89 DK 175581 B1 TABLE 11 HUMAN COMPLEMENT-DEPENDENT CYTOTOXIC EFFECTS OF CHIMARIUS L6 ANTIBODY PRODUCED BY Yeast OR MUSEUM Sp2 / 0 CELLS ON THE THICKNESS ARCINCOM CELL CELL NON 3347

Antibody concentration Complement8 Percent Antibody (tte / mB 0 or -3 cvtolvse

Standard mouse L6 5 + 122 10 1 + 53 5 - 1

Sp2 / 0-chimeric-L6 5 + 73 1 + 22 0.1 + 5 15 5 - 2 Yeast chimeric-L6 5 + 3 1 + 2 0.1 + 4 5 - 2 20 _ a Human serum of a healthy individual was used as a source of complement. b Complement-mediated cytolysis was measured by a four-hour 51 Cr release assay.

EXAMPLE VI: Secretion of Functional Chimeric Fab from Yeast

The Fab portion of IgG consists of a single light chain molecule linked by a disulfide bridge to a single truncated heavy chain molecule consisting of the variable region and CH1 (Figs.

31). This heavy chain fragment is known as Fd. Fabs are potentially useful for a variety of therapeutic and diagnostic procedures. Furthermore, they are capable of being produced by microbial fermentation.

90 DK 175581 B1

The usual method of producing Fab involves digesting intact IgG with papain (see Fig. 31), followed by purification of Fab away from the Fc fragments formed in the digested material. Although this procedure is relatively straightforward and can result in 5 high yields of Fab, it is somewhat time consuming as it requires first production and purification of whole antibody followed by formation and finally purification of Fab. Furthermore, one third of the entire antibody molecule - the Fc portion (Fig. 31) - is not used.

Recent advances in gene cloning technology and site-specific mutagenesis technology allow for a more direct and simple alternative method for producing Fab molecules. In this method, a stop codon is introduced into the heavy chain gene within the hinge region at ca. the codon for the amino acid at which papain digestion occurs. Fab is then produced directly by simultaneous expression of both the light chain and the Fd genes to produce their respective proteins, which are collected and secreted from the cell.

(1) Insertion of a stop codon into the hinge region of chimeric heavy L6 chain.

The strategy for introducing a stop codon into the hinge region of chimeric heavy L6 chain is outlined in FIG. 32A. The location of the stop codon within the hinge region and of the DNA sequence of the mutagenesis primer are shown in FIG. 32B. The stop codon location corresponds to amino acid 226 of FIG. 31. In this procedure, plasmid pING1402 was generated containing an Fd gene encoding a protein of 228 amino acids and extending six amino acids beyond cysteine to which the light chain is linked. At the mutagenesis, a unique BclI site is also introduced at the stop codon, which can be readily used for subsequent manipulations of the 3 'end of Fd. This includes, but is not necessarily limited to, removal of heavy chain 3 'untranslated DNA as well as splicing of various types of modifications of Fd including addition of coding sequences for specific amino acids and production of association proteins.

91 DK 175581 B1 (2) Fusion of the fully developed Fd gene into yeast invertase signal sequence and truncated PGK promoter.

The strategy for associating the Fd gene with the yeast invertase signal sequence is outlined 5 in FIG. 33. In this approach, the prior yeast invertase signal sequence construct - fully developed heavy L6 chain fusion (Fig. 26) was used, and a unique Apal site in the J region of the L6 chimeric heavy chain was used. to replace the constant region of pING1415 consisting of CH1, CH2 and CH3 with the constant region of pING1412 containing the stop codon in the hinge region. In this method 10, plasmid pING1418 was generated.

(3) Removal of non-yeast 3 'untranslated DNA.

Insertion of a unique BclI site at the stop codon of the Fd chain provided a convenient method for retrieving all non-yeast 3'-untranslated DNA. This was accomplished using the one shown in FIG. 34, and plasmid pING1428 was generated.

Since the stop codon was introduced into the hinge region by site-specific mutagenesis of a heavy chain fragment cloned into M13, the possibility existed that unwanted mutations could have been introduced during the mutagenesis step. To ensure that such mutations were not present, an Fd gene was linked to the invertase signal sequence and truncated PGK promoter and consisting of known coding sequences constructed using the one in FIG. 34 outlined strategy to generate the plasmid pING 1444.

(4) Construction of yeast expression plasmids containing the pING1444 chimeric L6-Fd gene associated with the PGK polyadenylation signal.

In order for yeast to produce an intact, functional Fab molecule, a balanced synthesis of both light chain proteins and Fd chain proteins must occur simultaneously within the cell. As described in Example V, one method is to place the light chain and Fd genes on separate shuttle vectors containing separate selective markers and to transform these vectors into a yeast strain deficient in both selective markers.

5

The Fd gene from pING1444 (Fig. 34) was cloned as a BamHI-XhoI phagem into two medium copy numbers of yeast-E. coli pendulum vectors containing yeast replication sequences and PGK polyadenylation / transcription termination signal: pING804CVS for leu2 selection and pINGH50 for ura3 selection (see Figs. 29 and 30). The two 10 plasmids resulting from these constructs - pING1445 (ura3) and pING14446 (leu2) - are shown in FIG. 35th

(5) Secretion of chimeric L6-Fab from transformed yeast cells.

Two separate transformation experiments were performed in an attempt to achieve both light chain synthesis and Fd chain synthesis in yeast cells. Four µg each of pING1444 (Figure 35) and pING1441 (Figure 30) and separately by pING1444 (Figure 35) and pING1442 (Figure 30) were cotransformed into S. cerevisiae strain BB331c (MATa, ura3, leu2). by selection for growth on SD agar (2% glucose, 0.67% gem nitrogen base, 2% agar). Ura + 20 Leu + transformants appeared after 2 to 3 days of incubation at 30 ° C.

Five colonies were seeded from each plate in 6 ml SD decoction supplemented with 50 mM sodium succinate, pH 5.5, and grown for 65 hours at 30 ° C. The cells were removed by centrifugation and analyzed by ELISA for light chain levels.

The results of these analyzes revealed that the light chain levels in the culture supernatants of the pING1446 + pING1443 transformants were three to six times higher than the levels in the culture supernatants of the pING1445 + pING1441 transformants. The culture supernatants for each group of transformants were then concentrated by ultrafiltration on a "Centricon 30" filter (Amicon Corp.) 30 and run on a 10% polyacrylamide gel under non-reducing conditions. The proteins were imprinted on nitrocellulose paper and probed with goat anti-human kappa anti-serum 93 followed by peroxidase-labeled rabbit and ten goat antiserum. The concentrated supernatant from the pING1446 and pING1443 transformants contained a significant anti-kappa cross-reactive smear over much of the imprint with only a weak cross-reactive band at the position expected for the Fab protein.

By comparison, the concentrated supernatants from the pING1445 + pING1441 transformants contained relatively slightly elongated antihuman kappa cross-reactive protein on the imprint. Furthermore, one of the five samples (# 4) contained a particularly intense, distinct anti-kappa cross-responsive band that migrated at the position expected for a Fab protein.

10 (6) Purification of chimeric L6-Fab from yeast culture supernatant.

To establish that Fab-sized anti-kappa cross-reactive protein secreted by yeast is actually chimeric L6 Fab protein, purification of sufficient amounts was required to perform binding assays. The pING1441 + pING1445 transformant isolate # 4 was therefore grown in 1 liter SD meat decoction supplemented with 50 mM sodium succinate, pH 5.5, for 95 hours at 30 ° C. The cells were removed by centrifugation and the culture supernatant analyzed with ELISA for light chain protein level. The supernatant contained approx. 130 μΐ / ΐ light chain protein. The culture supernatant was then concentrated by ultrafiltration over an "Amicon YM30" filter to 20 ml. The concentrated supernatant was washed with 130 ml of 10 mM potassium phosphate, pH 7.5 (buffer A) and re-concentrated over the "YM30" filter to 12.5 ml. The concentrated supernatant was then brought to 54 ml with Buffer A and placed on a 1.5 ml S-Sepharose column equilibrated with Buffer A. The column was washed with 20 ml of Buffer A and subjected to a linear gradient of 0 to 200 mM sodium chloride in buffer. A (total volume 40 ml). ELISA analysis of the column effects showed a large anti-kappa cross-reactive peak between fractions 8 and 21, corresponding to a salt concentration of ca. 60 mM. These fractions were pooled, concentrated on "Amicon YM10" and "Centricon-10" filters (Amicon Corp.) to 51 μΐ and analyzed on 30 non-reducing and reducing polyacrylamide gels using coomassie blue staining and Westem imprinting with anti-human kappa and anti-human Fab anti-94 DK 175581 B1 serum. These assays showed a predominantly pure protein, which migrated at ca. 46 kd on the non-reducing gel and dissolved in two bands which ran at approx. 23 and 24.5 kd on the reducing gel, which corresponds to the predicted (based on amino acid sequence) molecular weights for light chain protein and Fd chain protein, respectively. The smallest of the two bands reacted strongly with anti-human kappa antiserum to the Western imprint. Both protein bands responded with anti-human Fab anti-serum to the Western imprint.

(7) Studies on chimeric L6-Fab secreted by yeast.

The primary activity of a Fab molecule is its ability to bind to the target antigen. Therefore, the yeast-derived chimeric Fab was examined for its ability to bind directly to an L6 anti-gene positive cell line and for its ability to inhibit use of L6 antibody to antigen positive cells.

15 Direct Bond Analysis. Cells from the human colon carcinoma cell line 3347 containing the L6 antigen on the cell surface were used as target cells. Cells from the antigen negative cell line T51 were used as a negative control. The target cells were first incubated for 30 min at 4 ° C with either yeast derived chimeric L6 Fab, Sp2 / 0 cell derived chimeric L6 antibody or with mouse L6 antibody. This was followed by incubation with FITC-labeled goat anti-human kappa immunoglobulin for the chimeric Fab, FITC-labeled goat anti-human IgG for chimeric antibody, or with FITC-labeled goat anti-mouse immunoglobulin for the mouse antibody. Both labeled antibodies were provided by TAGO (Burlingame, CA) and used at a dilution of 1:50. Antibody binding to the cell surface was determined using a Coulter 25 Model EPIC-C cell sorting apparatus.

As shown in Table 12, the yeast-derived chimeric L6-Fab bound to the L6-positive 3347 line.

The yeast-derived chimeric L6-Fab did not bind over background to the L6-negative T51 line.

30 Inhibition of binding. As the next step, the extent to which graduated doses of the yeast-derived chimeric L6-Fab or the Sp2 / 0 cell-derived chimeric 95 L 17 antibody could inhibit binding of a FITC-labeled mouse L6 antibody to the surface of the antigen was investigated. -positive colon carcinoma cells 3347.

The yeast-derived chimeric L6-Fab inhibited the binding of the directly labeled 5 mouse L6 antibody. However, a higher concentration of yeast L6 Fab was required to achieve 50% inhibition of mouse L6 antibody binding to the target cells than was necessary for the same degree of binding inhibition by Sp2 / O cell-derived chimeric L6 antibody.

10 (8) Conclusions

A method is described whereby yeast can be genetically engineered to secrete functional Fab regions of immunoglobulins. The yeast-derived chimeric Fab in this example binds to that target antigen. Such Fab molecules provide 15 convenient target agents for a wide variety of diagnostic and therapeutic applications. This method also shows the feasibility of secreting heterologous heterodimeric molecules from yeast.

TABLE 12

20 CHINARY L6-FAB BINDING ANALYSIS PRODUCED BY Yeast ON A

L6-ANTIGEN-POSITIVE AND A L6-ANTIGEN-NEGATIVE CELL LINE

Binding ratio of: 3347 cells T51 cells 25 Antibody (L6 + 1 (L6-)

Sp2 / 0-chimeric-L6 103 1 Yeast chimeric-L6-Fab 32 1 30 a. All antibodies were used at a concentration of 10 µg / ml.

96 DK 175581 B1 b. The binding ratio is the number of times a test sample is more clear than a control sample treated with FITC-conjugated second antibody. Goat anti-human antibody was used as the second antibody for the chimeric Sp2 / 0-L6 antibody, and goat anti-human cap-pa antibody was used as the second antibody for yeast Fab.

EXAMPLE VII: Secretion of Functional Chimeric Fab Molecules from Bacteria

Bacteria are suitable for producing chimeric antibodies expressed from mammalian cDNA, since whole coding sequences can be expressed from finely characterized promoters.

Escherichia coli is one of the many useful bacterial species for the production of foreign proteins (Holland, I.B., et al., BioTechnology, 4: 427 (1986)), since a wealth of genetic information is available to optimize its gene expression. E. coli can be used for the production of foreign proteins internally or for secretion of proteins from the cytoplasm, where they are mostly accumulated in the periplasmic compartments (Gray et al., 15 Gene 39: 247 (1985); Oka et al., Proc Natl Acad Sci USA 82: 7212 (1985)). Secretion from the E. coli cytoplasm has been observed for many proteins and requires a signal sequence. Proteins produced internally in bacteria are often not folded correctly and precipitates in subcellular particles termed inclusion bodies (Schoner et al. BioTechnology 3: 151 (1985)). However, protein secreted from bacteria is often folded correctly and assumes naturally occurring secondary and tertiary structures (Hsiung et al., BioTechnology 4: 991 (1986)). Although immunoglobulin peptides have been synthesized in genetically engineered E. coli (Cabilly et al., Proc. Natl. Acad. Sci. USA 81: 3273 (1984); Liu et al., Proc. Natl. Acad. Sci. USA 81: 5369 ( 1984); Boss et al., Nucl. Acids Res. 12: 3791 (1984)), there are no reports of secretion of these E. coli peptides in the form of functional antibodies or antibody fragments.

A Fab molecule consists of two non-identical protein chains bound by a single disulfide bridge. These two chains are the intact light antibody chain and the V, J and CH1 moiety of the heavy antibody chain, Fd. The correct cDNA clones for chimeric light L6 gene and Fd gene 30 have already been identified. In this example, these cDNA clones were organized into a single bacterial operon (a dicistron message) in the form of gene associations to the Erwinia carotovora (Lei et al.) J. Bacteriol. (1987)) and expressed from one of two strong, regulated promoters. The result is a system for simultaneous expression of two E. coli protein chains and secretion of immunologically active, properly assembled Fab of chimeric L6 antibody in the culture growth medium.

5 A. Construction of E. coli expression systems for chimeric L6-Fab.

1. Collection of pelB leader sequence cassette.

Erwinia carotovora EC encodes multiple pectal lases (polygalacturonic acid transliminase) (Lei et al., Gene 35: 63 (1985)). Three pectatease genes had been cloned and the DNA sequence of these genes determined. By cloning into E. coli under the control of a strong promoter, the pelB gene is expressed and large amounts of pectalasease accumulate in the periplasmic compartment. The pelB signal sequence functions efficiently in E. coli and 15 was used as a secretion signal for antibody genes in this example. The nucleotide sequence enclosing the signal sequence of the pelB gene is shown in FIG. 36a.

The pelB signal sequence contains a HaIII restriction site at amino acid 22 adjacent to the signal peptidase cleavage site: ala-ala. Plasmid pSS1004 (Lei et al., 20 J. Bacteriol., Reprint (1987)) containing the pelB gene in plasmid vector pUC8 (Vieirra and Messing, Gene, 19: 259 (1982)) was digested with Haelll and EcoRJ. This DNA was ligated with an 8 base pair long Sstl coupler to Sspl, and EcoRl cut pBR322.

The resulting plasmid contained a 300 bp fragment which included the 22 amino acid leader sequence of pelB and ca. 230 bp upstream of E. caratovora DNA. This plasmid pING173 contains an insert which, after digestion with Sstl and treatment with T4 DNA polymerase, can be directly ligated to a DNA fragment flanked by the first amino acid of a fully developed coding sequence for any gene to form a gene. protein association containing a functional bacterial leader sequence in the frame of the incoming gene. Sstl to the EcoRI restriction fragment of pING173 was cloned into pUC18 (Yanich-Perron et al., Gene 33: 103 (1985)) to form pRR175, which contains the pelB leader sequence and neighboring upstream coding sequence (including a ribosome binding site) downstream of the lac promoter. The construction of pRR175 is outlined in FIG. 36b.

2. Preparation of chimeric light L6 gene for bacterial expression.

5

The intact chimeric light L6 chain gene containing an AatII restriction site at the signal sequence process site and a unique BglII site downstream of the gene was excised from the yeast expression plasmid pENG1298 (Fig. 25a) as a 1200 bp DNA fragment. This fragment was inserted into plasmid pRR175. The resulting plasmid pRR177-8 contained an in-frame association of the pelB leader and the light L6 chain downstream of the lac promoter located in the parent plasmid. A number of derivatives of this plasmid were constructed for deletion of non-coding sequences from both the 5 'and 3 ends of the pelBr.let chain gene association in pRJR.177-8. Upstream noncoding sequences were deleted using an Ndel restriction site at -48 bp from the pelB leader sequence-15 initiation codon (Fig. 36) to generate pRR180-2. The non-coding 3 'sequences were eliminated by substituting a fragment from the plasmid optimized for light L6 chain expression in yeast, pING1431 (see Fig. 27a) in pRR179 to form pRR191.

Another plasmid, pRR190, is similar to pRR191 but contains 90 bp of non-coding eukaryotic DNA at the 3 'end of the light chain gene. These constructions are shown in FIG. 37th

3. Preparation of chimeric L6-Fd gene for bacterial expression.

The intact chimeric L6-Fd gene containing an Sstl restriction site at the signal sequence process site, a BclI site introduced by site-directed mutagenesis (Figs. 32a, b) and forming a termination codon at amino acid 226 and a unique BamHI restriction site downstream of the gene was excised from plasmid pING1406 (Fig. 33) as an 880 bp DNA fragment. This DNA fragment was inserted into plasmid pRR175 to form an in-frame association of the pelB-1 gene sequence and the L6-Fd gene downstream of the lac promoter pRR178-5. A number of derivatives were constructed for deletion of 30 non-coding sequences from both the 5 'and 3' ends of the sequence contained in pRR178-5.

The non-coding 3 'sequences were eliminated by substituting a restriction fragment 99 from the piasmid optimized for L6-Fd expression in yeast, pING1428 (Fig. 34), which contains an XhoI coupler immediately after the termination codon of Fd. the gene to generate plasmid pRR186. Removal of E. caratovora DNA sequences upstream of the Ndel site at -48 from the leader sequence generated plasmid pRR196. The construction of these plasmids is shown in FIG. 38th

4. Multicistron light chain and Fd gene expression system.

For the production of bacterially derived Fab it was necessary that both light chain and Fd were produced simultaneously within the cell. Using the plasmids constructed with each of these genes separately, a series of expression vectors were constructed which contained both genes aligned, so that transcription from a single promoter would specify both genes. This was done in a way that minimized the non-coding DNA between the two genes to 60 bp. Each gene has a ribosome binding site 15 required for translation initiation and the identical DNA sequence from -48 to the pelB leader :: antibody gene association. Several cloning steps were needed to put the two genes in a row together. A portion of the light chain gene bound to the pelB leader in pRR180-2 was cloned downstream of the Fd gene in pRR186 to form pFK10OO. The remainder of the light chain gene was subcloned into pFK100 from pRRI77-8 to form 20 pFK10O. Similarly, DNA fragments containing 3 'deletions of eukaryotic sequences from pRR190 and pRR.191 were cloned into pFK101 to form pFK103 and pFK102, respectively. DNA fragments from pRR192 and pFK101 were ligated to generate pFK104, which contains a deletion of sequences upstream of -48 bp from the Fd gene.

Maps of the Fd cassette and the light chain gene cassette of these plasmids are shown in FIG. 39th

25 5. Placing the dicistrone light chain message and Fd under the control of inducible promoters.

Plasmids, pFK101, pFK102, pFK103 and pFK104, contain Fd and light chain genes sequentially cloned under the control of the lac promoter in vector pUC18 or pUC19. In E. coli strains, such as JM103 F'laciQ (Messing et al., Nucl. Acids. Res. 9: 309 (1981)), the amount of light chain accumulated in the periplasm is not affected by lac The promoter inducer isopropyl BD-thi ogal actopyranoside (IPTG), see Table 13. Furthermore, bacterial growth is slower (compared with cells containing pUC18) and bacterial colonies exhibit altered morphology, being small, dry and rough, suggesting , that substantial foreign gene expression is detrimental to cell growth. Two strategies were used to place this gene cassette under more tightly regulated promoters.

First, a PstI to EcoRI ffagment ffa pFK104 was ligated to pIT206 to place the 10 Fd gene cassette and the light chain gene cassette under direct control of the Salmonella typhimurium araB promoter, a well characterized strong promoter in E. coli. A restriction map of pIT206 and construction of pIT104 are shown in FIG. 40. Use of the araB promoter and its regulatory protein araC for expression of bacterial genes is disclosed in U.S. Patent Application Nos. 695,309, filed January 28, 1985 and no.

15,797,472, filed November 13, 1985. As can be seen in Table 14, the resulting plasmid, pIT104, is now regulated for light chain synthesis by the addition of arabinose to the culture growth medium. An at least 10-fold induction is done by arabinose addition.

Although Fab secreted into the growth medium increases more than 10-fold, cell growth after induction with arabinosis stops. This confirms that high-level expression of the 20 Fab genes is detrimental to cell growth. Bacterial colonies harboring pIT104 are phenotypically indistinguishable from E. coli harboring pIT206 when grown in the absence of arabinose.

Next, a DNA fragment containing the laci gene, a repressor of the lac promoter, was cloned into the high-copy expression vector pFK102. Expression of laci from a high copy number vector is useful for regulating expression of the lac promoter on a high copy number vector (Russel et al., Plasmid, in print (1987); Hsuing et al., Biotechnology 4: 991 (1986)) . A 1.7 kb EcoRI fragment containing the laci gene on pMC9 (Calos et al., Proc. Natl. Acad. Sci. USA 80: 3015 (1983)) was excised, filled with T4 polymerase for blunt ends, ligated with PstI couplers and cloned into the unique PstI site of pFK102 to form pFK102 laci. The map of pFK1021aci is shown in FIG.

DK 175581 B1 ΙΟΙ 40b. The selection procedure used to identify the correct clone ensured that the resulting plasmid, pFK102laci, contained a functionally suppressed lac promoter. All white or slightly pink colonies on MacConkey lactose plates contained plasmids with laci insertions, while transformants containing pFK102 alone were red, indicating the functional suppression of the lac promoter by the high copy number laci gene. Table 14 shows that expression of bacterial Fab ffa cells containing pFK1021aci corresponds to expression of pFK102. In contrast to cells containing pFK102, which formed aberrant colonies and grew slowly in meat decoction culture, cells containing pFK1021aci were similar to those containing 10 PUC18.

B. Expression, SDS-PAGE, and Purification of Bacterially Produced Fab.

1. Growth of E. coli house cloned antibody genes.

15

Plasmid DNA was transformed into either E. coli JM103 or -MC1061 by standard E. coli transformation procedures. Bacterial cultures were grown in TYE (tryptone 1.5%, yeast extract 1.0% and NaCl 0.5%) supplemented with the antibiotics concerned (penicillin 250 µg / ml, tetracycline 15 µg / ml). Bacterial cultures were grown in 20 volumes of 5 ml to 1 L at 37 ° C to an optical density OD 600 = 0.8 (about 4 x 10 8 cells / ml) and samples were induced with IPTG (0.2 mM), lactose ( 1.0%) or arabinose (1.0%). Cultures were grown for an additional period of 4 to 21 hours. Portions of each culture were analyzed for light chain production. Protein was released from the periplasmic compartment of E. coli cells by osmotic shock as described by Yanagida et al., J. Bacteriol. 166: 937 (1986). Alternatively, culture supernatants were analyzed directly for the presence of antibody chains.

Quantitative determination of light L6 * chain was performed by ELISA by goat antihu-man-kappa light chain antibody (Cappel, Malvern, PA). Fd could be detected by ELISA with mouse anti-human Fd antibody monoclonal (Calbiochem, San Diego, CA).

Table 13 shows representative data for expression of light chain-responsive material 102 in periplasmic E. coli extracts. Light chain is secreted from the bacterial cytoplasm into periplasms. Antibody chains are also released from the bacteria to the culture supernatant.

This is an unusual finding and may be a unique feature of L6-Fab among eukaryotic proteins expressed in E. coli. However, under certain conditions, 5 bacterial proteins are known to be released from E. coli. (Abrahamsen et al., Nucl. Acids Res. 14: 7487 (1986); Pages et al., J. Bacteriol. 169: 1386 (1986)). In Table 14, the amount of chain chain secreted in the periplasm is compared with the amount secreted in the culture supernatant. Light chain responsive material is present in plasmid containing cultures which house cloned light chain alone or light chain plus Fd. The top 10 manufacturers of Fab (pFK102, pFK104 and pFK1021aci) typically secrete 300-1000 ng / ml of ELISA-responsive light chain in the culture media. A separate construct was made in which the light chain gene is followed by the Fd gene (pFK107).

This construct directs the synthesis and secretion of Fab at approximately the same levels as the constructs with the genes in reverse order. Thus, gene order 15 is not critical for the secretion of Fab.

2. SDS-PAGE of bacterially produced chimeric light L6 chain and Fd.

Bacterially produced antibody chains were analyzed by polyacrylamide gel electrophoresis under reducing and non-reducing conditions. Protein extracts intercepted whole bacterial cells, protein released from the periplasmic compartment by osmotic shock, and protein secreted into the culture supernatant were electrophoretically analyzed. Transfer of gel-separated protein under fully reducing conditions to nitrocellulose and immunological imprinting with goat anti-human light chain antibody by Western analysis showed that a protein of the same molecular weight as authentic chimeric light L6 chain was present (about 23 K d). Analysis of protein samples by SDS-PAGE under non-reducing conditions showed that extracts from cells harboring a light chain gene plasmid alone (pRR191 or pRR190) contained a large proportion of the light chain-responsive material united in a higher molecular weight form. Much of this material ran at 30 approx. 46 Kd, which is probably a light chain dimer. Light chain dimers have been observed from myeloma cells that produce only light chain. There are also other immunoreactive protein bands that may represent nonspecific light chain disulfide formation and E. coli proteins. Protein samples (periplasmic extracts or culture supernatants) from E. coli cells harboring both light chain gene and Fd gene contain a light chain responsive band at ca. 48 Kd upon separation under 5 non-reducing gel conditions, which runs at a slightly higher molecular weight than the bacterial light chain dimer. This material is bacterially produced L6 chimeric Fab. In E. coli harboring pFK1021aci, pFK101, pFK102, pFK103 or pFK104, the 48 Kd band observed on an SDS gel run under non-reducing conditions is the most conspicuous immunoreactive species. Furthermore, the background stain or smear of immunoreactive proteins seen in extracts containing only the light chain is greatly reduced in extracts from cells containing both light chain and Fd.

3. Purification of bacterially produced chimeric L6-Fab.

15

Immunologically and functionally active (see below) bacterial Fab was purified from either culture supernatants or periplasmic protein extracts of E. coli pFK1021aci or pIT104. For purification of periplasmic material, the periplasmic fraction from 1 liter of cells induced for 4 hours was released in 50 ml of distilled 20 water. This material was centrifuged for 20 minutes at 5000 g and filtered through a 0.45 μτη filter. The periplasmic extract was then concentrated over a "ΥΜΙΟ" membrane (Amicon) to about 5 ml. This material was diluted 8 times in starting buffer (10 mM K 2 HPO 4, pH 7.5) and applied to a 1 ml S Sepharose column at a flow rate of 1.0 ml / min. The column was washed with 25 ml of starting buffer and eluted with a 0 to 200 mM NaCl gradient in starting buffer (total volume 200 ml). The immunoreactive gradient peak was collected (elution occurred). at about 100 mM) and concentrated on Centricon 10. Purified Fab was stored in PBS + 2.0% BSA.

For purification of secreted Fab from 1 liter of bacterial culture supernatant, cells were removed by centrifugation after growth for 21 hours with inducers and the supernatant was filtered through a 0.45 μτη filter. The media was concentrated over DK 175581 B1104 a "ca." membrane (Amicon) to about 16 ml and then diluted with 10 mM K 2 HPO 4 to 105 ml. This material was applied to a 1.6 ml S-Sepharose column. and eluted with a 0 to 200 mM NaCl gradient in 40 ml. Fab recovered from S-Sepharose chromatography was more than 70% pure as determined by densitometry detection of a non-reducing, commassion-colored 10% acrylamide gel. Fab purified from bacterial culture supernatants were dissolved in 2 major protein bands of approximately 23 Kd and 24.5 Kd on a 15% reducing gel The molecular weight of Fd and light chain based on the DNA sequence is 24.5 Kd and 23 Kd, which corresponds well to the observed protein sizes. The lesser of the two bands reacted strongly with goat anti-human kappa chain antiserum 10. to a Western imprint Bacteria Fab purified from either the periplasmic compartment or the bacterial culture supernatants cannot be distinguished by all the analytical assays tested here. criteria.

4. Functional binding activity of bacterially produced chimeric L6-Fab to the L6 anti-15 gene.

Bacterially produced Fab purified by S-Sepharose chromatography was examined for binding to L6 antigen-containing cells. As shown in Table 15, bacterial Fab specifically binds to human colon carcinoma cell line 3347. T cell line T51 cells were used as a negative control. Target cells were incubated for 30 min at 4 ° C with bacterially produced L6 chimeric Fab, intact chimeric L6 antibody produced in Sp2 / 0 cells, or mouse L6 antibody purified from mouse bug water serotonin. This was followed by an incubation with FITC-tagged goat anti-human light chain antibody for Fab detection, FITC-tagged goat anti-human immunoglobulin for chimeric antibody detection, or FITC-tagged goat anti-detection. murine immunoglobulin for mouse antibody detection. Antibody binding to the cell surface was determined using a Coulter Model EP1C-C cell sorting apparatus.

Bacterially produced Fab also exhibits characteristic binding inhibition of 30 FITC-labeled mouse L6 antibody to the surface of antigen-positive 3347 colon carcinoma cells. Bacterially produced Fab and Sp2 / 0-derived chimeric L6 have similar binding inhibition profiles, thus suggesting that the ester of bacterially produced Fab and Sp2 / 0-derived chimeric L6 is about the same.

Conclusions 5

A novel method is described in which E. coli has been used as a host for the production of functionally active Fab regions of immunoglobulins and to secrete these into the periplasmic space and also in the culture medium. This molecule exhibits binding properties expected from a properly assembled antibody recognition site. This technology can be used to express antibody genes with other binding specificities in E. coli.

1. Proteins encoded by modified antibody cDNA clones can be secreted from bacteria using a signal sequence.

2. Two antibody genes can be expressed from a single bacterial promoter as a dicistron message.

3. Two foreign proteins (in this example light chain antibody and Fd antibody) can be assembled correctly, i.e. assume proper secondary, tertiary and quaternary structure when secreted from bacteria.

4. At least two and probably many bacterial promoters can be used to express antibody genes.

25 5. This example is a general method whereby genes encoding other antibody chains can be expressed together as a dicistron message. These include either light chain genes and Fd genes or light chain genes and intact heavy chain genes.

106 DK 175581 B1 6. The sequence of the promoter is not important for the ability of E. coli to produce Fab. A construct of the Fd gene followed by the light chain works just as well as the genes, which are organized in the reverse order.

7. Fab can be secreted from E. coli in the culture supernatant where it is stable and can be purified. Most are Fab chains that pass through the cytoplasmic membrane secreted into the culture supernatant.

Microorganism deposits 10

Saccharomyces cerevisiae BB331C (41 / 42-5), GI 87 was deposited with ATCC on July 9, 1987 and is assigned accession number 20856. Eschericia coli JM 103 (pFK1021 aci), G186 was also deposited there on the same date and assigned accession number 67457 Both deposits were made under the Budapest Treaty.

15 TABLE 13 QUANTITATIVE DETERMINATION OF EIGHT CHAIN FROM E. COLI PERIPLASMA Ng / ml of culture ng / ml of culture plasmid - + plasmid - + 20 _______ pRR175 0 0 pFKIOl 36 28 pRR177-8 8.5 11 PFK102 68 55 pRR180 399 412 pFK103 38 45 pRR190 200 241 pFK104 91 68 25 pRR199 463 772 E. coli-JM103 or MCI061 (approximately the same results) were transformed with each plasmid. Fresh transformants were grown in TYE at 37 ° C to an OD 600 = 0.8. Cultures were divided and the inducer (IPTG) was added to 0.2 mM for a sample (-30 or + IPTG). Cells were grown at 37 ° C for 4 hours. Periplasmic protein extracts were prepared and samples were examined for light chain by ELISA with 107 goat anti-human kappa antibody. Each value is of the average of at least two separate trials. Removal of non-coding sequences both 5 'and 3' to the antibody gene caused an increase in chain-derived accumulation in the periplasm.

TABLE 14

COLLECTION OF LIGHT CHAIN IN THE SUPERNATANT AND THE PERIPLASMA AFTER

INDUCTION

Induction Supernatant Periplasm 10 Plasmid agent 4 hours 21 hours 4 hours 21 hours pRR 190 - 0 nd 200 nd pRR190 + 5 188 241 nd 15 pFK102 - 12 nd 68 nd pFK102 + 57 828 55 40 pFK104 - 13 nd 91 nd pFK104 + 150 290 68 35 20 pFK1021aci - 25 360 50 100 pFK102aci + 72 606 37 40 pIT104 - 13 nd 10 nd 25 pIT104 + 150 216 19 35

Plasmid containing E. coli strains were cultured, prepared and tested as described in Table 13. For pRR190, pFK102, pFK104 and pFK1021acia, cells were induced with 0.2 mM IPTG; pIT104 was induced with 1% arabinose. Each value is the average of at least two separate trials. For analysis of E. coli culture supernatants, bacteria were removed by centrifugation and culture supernatants were passed through a 0.45 µm filter. Values are expressed in ng / ml of culture.

nd - not determined 5 TABLE 15 BACKGROUND ANALYSIS-Fab Binding Ratio * 3347 cells T51 cells 10 Antibody L6 + L6-

Standard mouse L6 95 1

Sp2 / 0-chimeric-L6 116 1

Bacteria-L6-Fab 54 1 15 Standard L6-Fab 16 1 * The binding ratio is the number of times a test sample is more clear than a control sample treated with FITC-conjugated antibody.

20

Standard L6 Fab was prepared by enzymatic digestion of mouse L6 antibody.

Claims (8)

109 DK 175581 B1 A method of producing an immunoglobulin molecule with heavy and light chains or fragments thereof, which are linked such that the parent molecule exhibits desired binding and recognition properties, the method comprising (a) growing a transformed or transfected prokaryotic host. under appropriate conditions such that heavy chain immunoglobulin and light chain immunoglobulin or fragments thereof are expressed as a result of transcription of polynucleotide sequences in the host, wherein the polynucleotide sequences encode at least one functionally active region of a variable region of an antibody, and a prokaryotic secretion signal peptide, wherein the variable region under expression is operably linked at its N end to a prokaryotic secretion signal peptide to enable secretion from the host; and (b) obtaining the immunoglobulin chain or fragment of an immunoglobulin chain which, prior to secretion from the host, was thus initially bound to the signal peptide in the form of a heavy and light chain immunoglobulin molecule or fragments thereof that the parent molecule exhibits. desired binding and recognition properties. The method of claim 1, wherein the secretion signal peptide is a pectal lyase secretion signal. 25 The method of claim 1 or claim 2, wherein the immunoglobulin molecule is isolated from the periplasmic compartment. The method of claim 1 or claim 2, wherein the immunoglobulin chain or fragment of an immunoglobulin chain is isolated from the culture growth medium. 110 DK 175581 B1 A method according to any one of the preceding claims, wherein the immunoglobulin molecule is an Fd phagem. A method according to any one of the preceding claims, wherein the immunoglobulin molecule comprises a chimeric immunoglobulin chain. The method of any of claims 1-4, wherein the variable region of antibody is operably linked to a sequence encoding a polypeptide other than an immunoglobulin chain. 10 The method of claim 7, wherein the polypeptide is an enzyme. 15