-/-
RECOMBINOMAS
BACKGROUND
Immunogenicity and inefficient interaction with human effector functions such as complement activation or macrophage stimulation limit the utility of murine monoclonal antibodies (MABs) for human diagnosis and therapy. This problem has been alleviated by chimeric antibodies formed by joining murine antigen specific variable regions to human constant regions.
Two ways have been described for the generation of chimeric antibodies in transfectomas. Conven¬ tional procedure entails in vitro replacement of the constant region genes of a mouse MAB by human constant region genes. These DNA constructs are then transfected into a nonproducing cell line and screened for expression of the chimeric immuno- globulin. In another approach the codons for the hypervariable loops of a non-human, e.g., a mouse or rat MAB are grafted xn vitro onto the backbone of a cloned human antibody. The chimeric gene is then equipped with an immunoglobulin transcription enhancer and the DNA construct is again transfected into a nonproducer. Either procedure yields chimeric antibodies effectively expressed in mammalian cells with retention of the original MAB antigen specificity.
These methods are each attended by two major drawbacks. First, each hybridoma MAB of potential value for human xn vivo application must be cloned and genetically engineered into a chimeric MAB before
its final value for in vivo diagnosis and therapy can be assessed. The hybridoma cell lines which provide the immunoglobulin genes incorporated into the chimeric constructs possess undesirable features such as highly altered and sometimes unstable karyotypes. In particular, hybridomas are known to possess a number of nonfunctional, aberrantly rearranged immunoglobulin genes that can actively be transcribed or even translated. These genes are contributed at least in part by the immortalizing fusion partner cell lines (e.g., SP2/0-Agl4) utilized in the production of hybridomas. On the geno ic level amplification of these nonfunctional antibody genes interferes with the successful cloning. Second, secretion of antibody molecules by a transfectoma is generally too low for commercial practicality.
SUMMARY OF THE INVENTION
This invention invokes homologous recombination to provide chimeric MABs of any specificity without prior cloning procedures and subsequent in vitro construction of the chimeric gene.
Pursuant to the invention, mouse, rat and other non-human immunoglobulins are changed to im unoglobulins chimeric for any human constant region domain by use of a limited number of generally designed DNA vector constructs. Each vector construct is applicable to any mouse or similar hybridoma and will effect, in vivo by homologous recombination, a subclass switch corresponding to its specific design. The mechanism is such that the human constant region sequences are integrated into a genomic locus where the transcription unit of the non-human, e.g., mouse MAB resides. Because the genomic environment is unchanged, no difference is
expected in the level of expression of the antibody which has undergone homologous recombination
(hereinafter reco binoma) as compared to the non-human MAB in the hybridoma.
DETAILED DESCRIPTION OF THE INVENTION
Mechanism of Integration of Transfected DNA Into The Genome of Mammalian Cells
Transfer of recombinant DNA plasmids allows the expression of genes in eucaryotic cells, if proper regulatory sequences are supplied. The promoters that activate the genes may be part of the inserted DNA or reside on the vector itself, 5' to the DNA insert.
The methods of transfection most commonly used are the calcium phosphate precipitation method, protoplast fusion, and electroporation or lipofection. A more sophisticated method is the icroinjection of the DNA directly into the cellular nucleus. Trans¬ fection efficiency differs with the method selected. In methods other than microinjection the transfecting DNA must pass through the surrounding cell components to reach the nucleus. DNA may be altered or degraded during the passage.
The DNA that reaches the nucleus is integrated into the host's genome in stable transfectoma cells (transfected eucaryotes) . Genetic markers, e.g., neomycin resistance, included in the transfected DNA, if expressed, permit selection of successfully transfected cells. The ultimate actual number of successfully transfected cells may be higher than the number of transfectomas observed in the selective medium, since the plasmid DNA may be integrated into genetically inactive parts of the genome, with consequent non-expression of the marker. Since
marker expression is a prerequisite for the clone to survive the selection process, a non-expressor clone will be lost. The percentages of these events are not known. Also unknown are the machinery that allows the cell to integrate DNA and the mechanism by which the integration is carried out. The vast majority of integrations are random. Non-Random Integration by Homologous Recombination
Homologous recombination is a naturally occurring physiological phenomenon by which mammalian and other cells can exchange homologous sequences resulting in so-called gene replacement. In typical- homologous recombination experiments, sequences on the incoming plasmid DNA and homologous target sequences on the genome undergo recombination inter se. The result is a number of specific changes that can be observed in the genomic locus and the plasmid.
Depending on the detailed arrangement of the incoming DNA sequence relative to the target, the recombination either introduces new sequences into the targeted locus by a single cross-over recombination event or substitutes sequences on the target by a double cross-over, resulting in gene conversion or so-called gene replacement.
Homologous recombination is thought to be part of a repair mechanism by which the highly complex genome of a mammalian cell is enabled to repair its DNA during a relatively long life span. Recombination also enables certain cells to generate an active gene from inactive germline configurations. This is especially true for cells of the immune system which, during maturation, extensively recombine parts of their germline antibody genes using recombination signals that flank the different exon cassettes.
Homologous recombination can also be observed in the interaction between E. coli and bacteriophages that become lysogenic. Here the integration is executed via very short stretches of homology on the host genome (attB) and on the phage genome (attP) . In the case of the participating host sequence the so-called core region is only 15 base pairs (bp) long. The mechanism in at least this recombination system seems to follow what has become known as the Holliday junction model of homologous recombination (Holliday, R. , Genet. Res. 5:282-304 (1964)) . According to this model the exchange between the sequences is made after forming and resolving an intermediate structure (the Holliday junction) . Neither the machinery involved, nor the enzymes are known . It is not clear whether the homologous recombination follows the same model in mammalian cells.
The post-transfection probability that a homologous recombination event will occur is low as compared to the probability of a non-homologous event. Therefore, factors to enhance the probability of homologous recombination are important to any technology based upon it. Such factors are to be considered in the design of homologous recombination vectors and in the selection and definition of transfection procedures.
The length of homologous sequences between the involved DNA molecules is the major factor that influences the likelihood that homologous recombin¬ ation will occur. Homologous recombination frequency as a function of the length of shared sequence homology has been reported to be 0.8 x 10~3/Kb of
homologous sequence for plasmids in mammalian cells (D. Ayares, et al.. Genetics 111:375-388 (1985)).
Another factor reported to increase the number of homologous recombination events in a transfectoma population is the pretreatment of the introduced plasmid DNA molecule. Linearized plasmids are between 5 to 80 times superior to supercoiled plasmids in homologous recombination. A linearized DNA molecule normally will facilitate end to end joining with a double stranded gap in the genome, thus increasing the overall transfection efficiency. Since the helical structure of such a plasmid is relaxed as compared to a supercoiled one, the homologous sequences may also be more accessible for involvement in a homologous recombination event. The site at which the plasmid is linearized also influences the rate of homologous recombination.
If the ends of a linear plasmid are treated with dideoxynucleotides (as they are used for chain termination reactions in DNA sequencing) , the integration rate of the plasmid ends into a double strand gap of a random genomic locus can be decreased without decreasing the number of homologous recombinants. The rate shift in favor of homologous recombination is between 3 and 8-fold. Alternatively, nucleotide analogs that possess thiophosphates instead of phosphates can be used.
Substances that are mitogenic will influence the number of homologous recombination events. As a side effect, targeting of an incoming sequence can be increased.
The stage of the cell cycle at the time of transfection also influences the likelihood of homologous recombination events. Cells that are in
synchronized cycles show an up to 15-fold increase of homologous recombination in the early to mid-S phase as compared to other phases.
To facilitate screening for homologously recombined cells, it is important to optimize the homologous recombination rate.
RECOMBINOMA VECTORS
For Generation of Chimeric Antibodies in Non-Human Hybridomas
Recombinoma vectors comprise a portion of a non-human, typically murine immunoglobulin gene and a portion of a human immunoglobulin constant region gene. The purpose of the non-human immunoglobulin gene portion is to provide a DNA sequence for homologous recombination involving the natural gene locus of a productive hybridoma antibody gene. The human sequences are provided to replace, with respect to the final immunoglobulin gene transcript the target non-human constant region gene portion by in situ homologous recombination in its natural gene locus.
Constant regions of antibody genes consist of coding sequences (exons) and surrounding non-coding sequences (introns) , both of which are highly conserved in each species. With respect to size the introns represent the major parts of an antibody gene (K. Marcu, Cell 29:719-721 (1982)). Because these conserved intron sequences are also part of active genes in each hybridoma, a common vector comprising these gene portions is useful to target the natural gene locus and become integrated into the genome by homologous recombination in any hybridoma at the site defined by the homologous sequences.
Like homologous recombination events occur for analogous reasons with vectors that include a human exon cassette(s) such as a human light chain constant region gene or a human heavy chain constant region gene next to the non-human homologous portion.
Depending on the design of the vector, a single cross-over (O-type geometry, R.G. Gregg and O. Smithies in Cold Spring Harbor Sy p., Vol. 51, 1093-1113 (1986) ) or a double cross-over (Omega-type geometry, Gregg, supra) with its own transcriptional stop signal will be placed in front of the non-human constant region exon(s) , thereby excluding the non-human constant region from the transcription unit.
The Omega-type recombination vector differs from the above O-type vector. In addition to the homologous sequences provided by the O-type vector, a second non-human constant region gene portion is provided. This new portion represents the sequences that flank the particular non-human constant region gene exon(s) at their 3' ends. Upon homologous recombination the recombinoma vector will be integrated such that the original non-human constant region exon(s) is physically removed from the non-human genome and replaced by the human constant region exon(s) .
Thus, typically one vector is required for making the change from any non-human kappa light chain to the non-human/human kappa light chain and only one vector is required to make the change from any non-human lambda light chain to the respective chimeric light chain.
The same rationale is applicable to the heavy chain genes with the exception that, if Omega-type vectors are required, the total number of vectors
needed to make any possible change in the hybridoma cell, will be higher than the number of O-type vectors for the same number of changes.
In terms of O-type vectors, the stretch of homology to the non-human genome comprises the sequences common to all heavy chain constant region genes in the non-human genome regardless of its antibody class. Such general vector can have the exons of either human heavy chain constant region gene (i.e., IgGl, IgG2, IgG3 , IgG4 , IgE, IgA, IgD) at the 3' end of the sequences of homology. Therefore, only seven recombination vectors are needed, each having one human constant region gene at the 3' end of the non-human heavy chain constant region intron sequences. For example, use of a recombination vector including the human IgGl constant region exons will yield an IgGl chimeric antibody, regardless of the preexisting antibody class that the hybridoma was expressing. The original non-human constant region gene exons will be excluded from the transcription unit by the gene insertion.
Choice of a recombination vector that carries the appropriate human constant region exons permits preselection of the human antibody class expressed by the recombinoma upon homologous recombination. To make, e.g., chimeric F(ab)2 fragments or monovalent antibodies, or any change that eventually is reflected in the gene product, appropriate mutations are introduced only into the human part of the respective recombination vector construct.
Thus, a total of 21 non-human O-type recombination vectors will produce either complete chimeras, chimeric F(ab)2 fragments or monovalent chimeras of seven different human classes from all
non-human hybridomas for which these vectors were designed. Consequently, 21 vectors can be designed for homologous recombination in murine genome, 21 for homologous recombination in rat genome and so on.
For the removal of the targeted gene portion from the genome of the hybridoma, homologous recombination must occur as a double cross-over event on both sides of the constant region gene to be removed (Omega-type geometry) .
The immediate flanking sequences of the non-human constant region gene exons are specific for the respective constant region on the genome. Therefore, upon constant region gene class switch that occurs in the B-cell, those sequences will be present in the active hybridoma heavy chain gene. The Omega-type recombination vector, in addition to the homologous sequences comprised in the O-type vector, has to contain these flanking sequences, at least the ones to the 3' end of the non-human exon cassettes.
For example, mouse hybridomas can express specific MABs in four different subclasses of IgG (IgGl, IgG2a, IgG2b, IgG3) . Since the regions flanking the IgG subclass genes are specific for the respective subclass, four vectors, comprising the 3' flanking regions of these constant region genes, are necessary for the Omega-type homologous recombination to these four murine immunoglobulin classes.
Choice of the appropriate Omega-type vector is given by the IgG subclass expressed in the hybridoma.
To generate recombinomas of all possible human i.e. IgG subtypes (IgGl, IgG2, IgG3, IgG4) from all possible i.e. murine IgG subtype monoclonal antibodies by an Omega-type geometry recombination
event, a total of 16 vectors will be needed theoretically.
For Generation of Chimeric Antibodies in Human Antibody Producing Cells
Chimeric human/non-human immunoglobulins with the specificity of the non-human hybridoma can also be generated by in situ homologous recombination to the genome of a human myeloma that produces an antibody, regardless of its specificity.
The design of such vector can be generally O-type or Omega-type, comprising sequences of the human constant region JC-intron (the intron between J-region minigenes and human constant region gene exon(s)). For the Omega-type vector, additional sequences 5' of the human variable region gene are also appropriate (see Figure 5d) .
In the case of O-type geometry recombination vector the cloned heavy chain and light chain sequences of the respective variable region genes of a non-human MAB are positioned in the vector 5' to the human homologous sequences with respect to their transcriptional orientation. The non-human segment has the same orientation as the human region. The genetic marker gene and most of the vector sequences are placed 3' of the homologous region.
Successful targeting of the myeloma antibody genes on the genome will result in a chimeric immunoglobulin comprising the desired antigen specificity of the non-human MAB together with the human constant region genes expressed by the myeloma cell.
DESCRIPTION OF THE FIGURES
The design of and procedures for the construction of recombinant vectors are further elucidated by reference to the Figures 1 to 6 and the accompanying explanatory legends.
Figures la and lb illustrate murine light chain gene rearrangement as it occurs in the cell during the maturation and the constant region gene sequences useful for homologous recombination.
Referring to Figure la, a schematic of the murine light chain gene cluster is shown. The "V" symbols refer to the unknown number of variable region genes in the repertoire of the mouse.
In Figure lb, the V-region participating in the rearrangement may be any of the V-regions indicated in Figure la. The rearrangement described and illustrated is accomplished in known manner. J.G. Seidman and Ph. Leder have provided early evidence that rearrangement of light chain gene results in the joining of a variable region gene with one of several Joining-region minigenes (Nature 276:790-795 (1978)). That rearrangement of antibody genes is not seen in sperm, embryonic and somatic tissue and therefore is a tissue-specific function of cells of the immune system, has been shown by Joho et al. PNAS 7:1106-1110 (1980). The complete nucleotide sequence of the J-gene region has been reported by Sakano et al. Nature 280:288-294 (1979). The mechanism by which the B-cell joins a V-region and a J-region, has been recently identified. Akira et al. show the role of short conserved sequences flanking the J-regions in the process of the VJ-joining (Science 238:1134-1138 (1987)). The transcription activity of the so rearranged light chain gene is
caused by an immunoglobulin promoter 5' of the V-region, which has, by the rearrangement, come under the influence of the immunoglobulin transcription enhancer. Such promoter sequences have been suggested recently following functional studies (T.G. Parslow et al. PNAS £51:2650-2654 (1984) and Y. Bergman et al. PNAS 81:7041-7045 (1984)). The transcription enhancer resides n the JC-intron of the light chain gene (published in GENbank entry MUSIGKJC2) .
Figures 2a, 2b and 2c illustrate heavy chain gene rearrangement and the constant regions that are common for any hybridoma heavy chain gene and can be used for homologous recombination in the mouse genome.
Figure 2a schematically shows the chromosomal configuration of the germline (unrearranged) murine immunoglobulin heavy chain gene region cluster. It illustrates the relative positions of Diversity-genes (D) , Joining-genes (J) , the switch region for the heavy chain class switch as well as the colinear arrangement of all the heavy chain constant regions in the murine genomic repertoire. The V-regions are not shown. Murine J-region genes have been characterized by Ravetch et al. Cell 27:583-591 (1981) . The nature of the switch region for the different heavy chain genes has been investigated by a number of groups (Sakano et al. Nature 286: 676-683 (1980) ; Honjo et al. Cell 18:559-568 (1979) ; Kataoka et al. PNAS 72:919-923 (1980); Szurek et al. J. Immunol. 135:620-626 (1985; T.H. Rabbitts et al. Nucl. Acids Res. 2:4509-4525 (1981)) . It has been shown in these reports that the sequences 3 ' of the switch region in a rearranged heavy chain gene also flank the respective genes in the germline
configuration and are quasi imported into the rearranged heavy chain gene by the switch recombination event. The enhancer region has been characterized (S. Gillies et al. Cell 33:717-728 (1983)). Recent studies have dealt with the identification of heavy chain promoter sequences (S. Eaton and K. Calame, PNAS 84:7634-7638 (1987)). It is well known from all these reports that, upon successful rearrangement of the antibody genes by the B-cell, an IgM is produced as first immune response towards the antigen (Figure 2b) . The rearrangement follows the same rationale as the rearrangement in the light chain genes with the exception that the variable region genes are first joined to one of the Diversity-genes before this VD-complex eventually joins up with one of the J-genes. By this additional joining process the cell is able to even increase the number of possibilities for antibody diversity. With maturation to a plasma cell, the B-cell stops making IgM but produces an i.e. IgG2a (as is the example in Figure 2a) that has the same antigen specificity as the IgM. This implies that the change to IgG2a only involves the constant region gene of the immunoglobulin. This process is known as class switch and is thought to occur by homologous recombination either between sister chromatids or by deletion of the gene sequences between the IgM and the newly imported constant region gene. The new constant region gene is now included into the transcription unit and therefore expressed together with the already well established variable region gene to produce an IgG2a. The sequences between the rearranged J-gene and the switch recombination site are not affected by any switch even and therefore can
be found in every murine hybridoma heavy chain gene. It is significant that sequences between switch site and constant region exons and the sequences 3' of the constant region exons are specific for the respective constant region gene. The 3' sequences can be included into a recombination vector to generate a double cross-over-Omega-type recombination event.
Figures 3a, 3b and 3c illustrate the construction of a recombination vector expressing a marker gene, the targeting of an active murine Ig gene with the vector to yield a reco binoma by a single cross-over event (O-type geometry) .
Figure 3a diagrams the construction of a recombination vector expressing a neomycin (neo) marker gene. The Figure includes a reproduction of Figure 2c with the region between the residual J's and the switch region isolated for incorporation into the vector as shown.
Figure 3b illustrates the targeting of an active murine Ig gene with the chimeric recombination vector illustrated in Figure 3a. In Figure 3b, the expression "O-type geometry" means that "the recombination introduces new seguences into the recipient chromosome by a single cross-over [event]", R.G. Gregg and O. Smithies in Cold Spring Harbor Symp., Vol. 51, 1093-1113 (1986) . The restriction site for linearization of the incoming plasmid lies at the 5' end of the homology with the genome or within this region.
Figure 3c illustrates the completed recombinoma.
Figures 4a and 4b illustrate the construction of a vector for initiation of a double cross-over, homologous recombination event (Omega type geometry) and the targeting of the mouse genome by the
linearized vector to yield a recombinoma and the replacement of a portion of the original mouse genome by the human region of the vector.
Figure 4a diagrams the assembly of a recombination vector for the initiation of a double cross-over homologous recombination event utilizing the rearranged heavy chain illustrated by Figure 2c. The term "Omega-type geometry" refers to the substitution of chromosomal sequences by double cross-over (R.G. Gregg and O. Smithies, supra) . As illustrated in Figures 4a, b and c, the two areas of homology involved in the double cross-over homologous recombination event are identified as 1 and 2.
Figures 5a, b and c illustrate the construction and use for single and double cross-over homologous recombination of vectors which comprise a non-human variable region gene including a promoter and sequences homologous to a human constant region gene.
Figure 5a illustrates the components required for the construction of a recombination vector to target the genome of a human myeloma cell and deliver a non-human variable region gene to be included into the resident chromosomal immunoglobulin gene. The non-human variable region gene with its promoter and the Leader peptide (L) which is necessary for the secretion of the molecule by the cell are shown schematically. The human sequences homologous to the heavy chain constant region gene (areas shown in bold) are indicated by ro an numbers I and II. Arrangement of the sequences in a vector comprising a marker gene (neo) to yield O-type and Omega-type vectors are shown.
Figures 5b and 5c show the recombination events carried out by the respective vectors.
An alternative design for an Omega-type vector is shown in Figure 5d for the heavy chain gene of a human myeloma. To increase the length of homology between the human genome of the myeloma and the length of homology between the human genome of the myeloma and the recombination vector, the vector comprises the region upstream of the resident active myeloma heavy chain variable region gene in addition to the sequences shown in Figure 5a. Arrangement of homologous sequences is shown by arrows. Upon double cross-over the resident human variable region gene will be removed from the genome by the same rationale as outlined before and replaced by the recombination vector comprising the non-human variable region gene. The vector is constructed such that the marker gene does not lie in the antibody transcription unit, meaning not between the introduced non-human variable region and the resident human constant region, but will be located upstream on the chromosome upon resolution of the recombination intermediate Holliday structure as shown at the bottom of the Figure.
Figure 6 shows the murine JC-intron of an active hybridoma heavy chain constant region gene (in this case IgGl) according to published information. The vertical bar indicates the boundary to the murine IgGl heavy chain constant region gene exons which have been replaced in the recombination vector by a 1957bp Stul restriction fragment comprising all the human IgGl constant region exons. Horizontal bars for which sequence data are published are shown in bold. For a stretch of around 2.3 kb no sequence information is available, but the region has been mapped with restriction enzymes. See, e.g. , Kataoka et al. PNAS 77:919-923 (1980). The stretch of
homology from the J-region to the human/murine boundary (vertical bar) can be between around 5.9 and 7.2 kb for VD-J4 and VD-J1 respectively. Relevant GENbank entries for murine and human sequences as indicated. As mentioned earlier, the sequences between J4 and the switch region are common for all of the murine heavy chain constant region genes. This region spans around 3kb. The sequences upstream of J4 may not be present in particular hybridomas because of J4 rearrangement and therefore are not common to all hybridomas. The sequences downstream of the switch site belong to the respective constant region genes after class switch and therefore also are not necessarily common to all hybridomas but can be used for construction of "class-specific" recombination vectors as outlined in the section "Recombinoma Vectors". The advantage for doing this lies in the increased region of homology facilitating the targeting probability of the genome. A recombination vector comprising all the sequences mentioned above could be preferably used for homologous recombination in a IgGl expressing hybridoma while a recombination vector comprising only the J4-switch region stretch of homology can be used for homologous recombination in all murine hybridomas regardless of the Ig class expressed by the particular cell.
Kataoaka et al. have reported a switch site of a murine gamma-1 heavy chain gene that is positioned around 500bp upstream of the murine constant region exons and around 6.3kb downstream of the heavy chain variable region gene in the mouse myeloma MC101. Similar numbers have been obtained by other authors
indicating that the numbers given in the Figure apply generally.
EXAMPLE
The construction of the O-type vector shown by Figure 3 involves the cloning of portions of the JC-intron that connects variable and constant region genes in the heavy chain. The region of homology available for generating a general O-type vector will be about 3kb, the region available for construction of a subclass-specific recombination vector will be between 5.9 and 7.2kb as outlined in the legend to Figure 6. Partial digestion of the human IgGl constant region gene yields a 1957bp blunt end restriction fragment comprising the transcriptional stop signal. See GENbank entry J00228. The Stul fragment will be placed to the 3 ' end of the murine JC-intron in the appropriate orientation.
The so constructed immunoglobulin heavy chain gene will thus be a chimeric IgGl gene lacking a variable region gene and will be recloned into a eucaryotic expression vector, i.e., pSV2 neo. Alternatively, the vector pcD neo will be used to obtain superior transfection efficiencies.
The plasmid will then be linearized and treated with dideoxy nucleotides prior to transfection. DNA transfer will be carried out using synchronized hybridoma cells secreting a monoclonal antibody (here Anti-CEA MAB T84.66-A3.1H11) in the Mid-S phase of the cell cycle. The transfection can be carried out by standard methods to introduce the linearized vector into the hybridoma cell.
The target locus is the heavy chain gene region cluster. Three different homologous recombination events can be postulated only one of which will yield the desired chimeric T84.66 antibody.
In the first case, homologous recombination will generate chimeric T84.66 that can be detected by screening with anti-human IgG. Antigen specificity will be the same as that of the T84.66 MAB.
In the second case, homologous recombination yields a non-rearranged immunoglobulin locus. The non-rearranged heavy chain gene cluster has virtually no transcription rate. Accordingly the non- rearranged chimeric cluster will be transcriptionally silent, since no variable region gene promoter is provided by recombination vector. In the third case homologous recombination may occur with aberrantly rearranged T84.66 genes that either stem from the unsuccessful VDJ-joining of a murine B-cell on its first trial to form an active antibody gene. Since in the diploid set of chromosomes in a normal mouse B-cell allows a second try, a mature, productive plasma cell can, besides its active antibody genes, either have an aberrrantly or unrearranged second allele depending on whether the immunoglobulin variable region rearrangement was successful on the first or second trial. This applies for both light and heavy chain rearrangement. The other possibility for the presence of aberrantly arranged genes is the nature of the fusion partner, used to generate the T84.66 hybridoma (in this case SP2/0) . The fusion partner cell, which is itself a non-secreting hybridoma cell possesses a highly altered haryotype. Note that the definition non-secretor does not exclude the possibility that there are active antibody genes in the SP2/0 and therefore most likely also in T84.66. One can find a number of transcripts that can be assigned to the SP2/0 cell line. They are just not translated properly to functional
immunoglobulins and therefore not secreted. These genes can be amplified. Therefore, homologous recombination with these genes would yield a chimeric aberrantly rearranged Ig gene, that could not be translated, because the misalignment was a defect during generation of the variable gene (light or heavy chain) . This defect is not repaired by the homologous recombination.
Identification of transfectomas which have undergone homologous recombination (recombinomas) is carried out using immunological methods. Specifi¬ cally, the cultured cells are incubated with goat anti-human IgG polyclonal antibodies that are (FITC) labeled (Zymed, San Francisco, CA) and sorted by (FACS) . Cells showing fluorescence are further cultured. Untransfected T84.66-A3.1H11 cells or the human lymphoblastoma cell line IM-9 (ATCC:CLL 159) that is secreting a human IgG as FACs experiment controls. The controls are incubated with FITC labeled anti-human IgG or anti-mouse IgGl antibodies (Zymed) . The mouse hybridoma should be negative with anti-human antibodies and positive with the anti-mouse antibodies, while IM-9 should be positive only with the goat anti-human IgG. Recombinomas should be only positive with anti-human IgG and the antibodies secreted by these cells should have the specificity and affinity of T84.66-A3.1H11 for the carcinoembryonic antigen. Evaluation of these parameters and rates of secretion of the chimeric MAB by the recombinomas are assessed by standard ELISA techniques.