The invention relates to the preparation and use of gene banks of synthetic human antibodies (huAb) or parts of antibodies which contain th e antigen-binding domain. Starting from a huAb framework in a suitable vector, the hypervariable regions of the antibody cDNA are formed by almost "randomly" combined oligonucleotides. Relatively conserved amino acids in the hypervariable regions have here been taken account of in the choice of appropriate nucleotides during the oligonucleotide synthesis and the ratio of the nucleotides used is likewise chosen such that a nonsense codon is to be expected at most in every 89th position. Expression of this synthetic huAb cDNA in microbial expression systems, e.g. in E. coli in the vector pFMT which is describe d below, thus makes a synthetic huAb library with a comprehensive repertoire for screening using selected antigens available in vitro.
The human or mammalian immune system comprises an estimated number of between 10.sup.6 and 10.sup.8 different antibodies. This number of antibodies seems to be sufficient to cause an immune reaction of the body both against all naturally occurring antigens and against artificial antigens. If it is furthermore taken into account that often different antibodies react with the same antigen, the repertoire of antibodies that are really different would rather be in the region from 10.sup.6 to 10.sup.7.
Up to now specific antibodies have always been obtained starting from an immunization with the particular antigen, for example injection of the antigen into the body or in vitro incubation of spleen cells with this antigen. In the case of polyclonal antibodies, the immunoglobulins can then be isolated from the serum and the specific antibodies can be isolated therefrom, e.g. by absorption methods. Monoclonal antibodies are isolated from the cell supernatants or from the cell lysate of spleen tumor cells (hybridoma cells) which have been fused with individual B lymphocytes and cloned. The above mentioned methods are unsuitable in particular for the preparation of specific human antibodies or human monoclonal antibodies.
The present invention therefore has the object of developing a generally usable method for generating specific human monoclonal antibodies (huMAbs) or parts of antibodies, which contain synthetic hypervariable domains.
It has now been found that by using almost randomly synthesized oligonucleotides coding for the three hypervariable regions of each variable part of heavy or light chains (called CDR1, 2 and 3, CDR meaning complementary determining region) synthetic human gene banks can be generated. The synthesized antibody DNA was then preferably ligated into an antibody expression vector especially constructed for this purpose, namely the vector pFMT, preferably after amplification using the polymerase chain reaction (PCR).
The oligonucleotides which are used for the synthesis of the variable domains of heavy (H) and light (L) chains are compiled in Tab. 1 (SEQ ID NOS:1-20). Set A (SEQ ID NOS:1-19) here contains fewer limitations than set B (SEQ ID NOS:1-2, 5, 7-9, 12 and 14-20). The randomness of the synthesis of the hypervariable regions (see CDR regions in Table 4, SEQ ID NOS:35 & 37) was restricted by limitations (a) to (f) (see below) concerning H3, H4, H6, L2, L3 and L5 in set A. The randomness was restricted in order to
(1) allow for positions in the sequence for certain conserved amino acids;
(2) reduce the number of stop codons; and
(3) incorporate a new restriction site.
(a) In order to reduce the probability of the stop codon occurring, only half the amount of the three other nucleotides was allowed for T at the first position of each codon and A was omitted at the third position of each codon, in each case. As a statistical average, only every 89th codon will thus be a stop codon.
(b) For the 2nd codon in the CDR1 region of the light chain (SEQ ID NO:37), only those nucleotides were allowed which code for the amino acids V, A or G.
(c) Likewise, only those combinations coding for V, I or M were allowed for codon No. 10 in the CDR1 region of the light chain (SEQ ID NO:37) and for codon No. 4 in the CDR1 region of the heavy chains (SEQ ID NO:35).
(d) In the CDR3 region of the light chain (SEQ ID NO:37), only those nucleotides coding for the amino acid glutamine were allowed for codon No. 1.
(e) In the CDR2 region of the heavy chain (SEQ ID NO:35), only those nucleotides coding for the amino acid tyrosine were allowed for codon No. 11.
(f) An A was advantageously incorporated at the third position of the last codon in the CDR2 region of the heavy chain (SEQ ID NO:35) in order to introduce a restriction site for MluI.
The random nature of these oligonucleotides was preferably limited even further in those positions where predominantly one or few amino acids occur (set B in Tab. 1 (SEQ ID NOS:1-2, 5, 7-9, 12 and 14-20), the limitations here are based on the tables by Kabat et al. (1987), Sequences of Proteins of Immunological Interest-U.S. Dept. of Health and Human Services, U.S. Government Printing Offices). A list of the corresponding nucleotides and brief notes on the codon combinations are compiled in Tab. 1 (SEQ ID NOS:1-20) and in the notes for Tab. 1.
After ligation of equimolar amounts of the oligonucleotides H1 to H8 (set a SEQ ID NOS:1-7 or set b SEQ ID NOS:1-2, 15-16, 17 and 18) and L1 to L6 (set a SEQ ID NOS:9-13 or set b SEQ ID NOS:9, 18-19, 12 and 20), these are ligated into the pretreated expression vector pFMT. Preferably, a PCR step using the primers H1 (SEQ ID NO:1) and H8 (SEQ ID NO:6), or L1 (SEQ ID NO:9) and L6 (SEQ ID NO:14) should be carried out beforehand in order to amplify the amount of DNA. After producing suitable restriction sites at the ends of the antibody DNA using appropriate restriction enzymes, the DNA is ligated into the antibody expression vector pFMT in the same manner as above (see examples).
The expression pFMT makes possible the expression of antibody cDNA and the subsequent secretion of the expression products in bacteria (E. coli). The antibody operon of the plasmid contains the sequences of the variable parts of both the heavy and light chain of an antibody. Suitable leader sequences from the amino terminal part of a bacterial protein makes secretion of the antibody parts possible. The leader sequences are cleaved off by a bacterial enzyme during the secretion. During the secretion of the antibody cDNA products, the light and heavy chains of the antibody (with or without an adjacent constant domain) become associated. This results in the formation of an antibody or antibody fragment which, in either case, contains a functional antigen binding site. Similar constructs for individual antibodies have also been described by other authors (Better et al. (1988), Science 240, 1041, and Skerras & Pluckthun (1988), Science 240, 1038).
In the synthetic human-antibody library formed by the expression in, for example, E. coli, the desired human antibodies or antibody parts are found by screening bacterial clones using the selected antigen. In a preferred embodiment, an additional sequence which codes for a marker peptide, for example a TAG sequence, is incorporated so that the expression products can be detected in a simple way using established monoclonal antibodies against the marker peptide (Wehland et al. (1984), EMBO J. 3, 1295).
The above mentioned exemplary formulations and the examples below shall be understood as illustrating but not restricting the invention.
The invention therefore relates to gene banks of synthetic huAb or antigen-binding parts thereof, obtained by means of (1) cDNA for the hypervariable regions generated on a random basis, where the random sequences are limited by (a) to (e) set A the limitations set forth (SEQ ID NOS:1-14) described above in or in accordance with Tab. 1, set B (SEQ ID NOS:1-2, 5, 7-9, 12 and 14-20), (2) preferably a subsequent amplification step of these random sequences and (3) ligation of the said cDNA into a suitable expression vector, preferably pFMT, an additional coding sequence for a marker peptide being incorporated in a preferred embodiment.
The invention also relates to a process for the separation of the above mentioned banks, and the process and the use thereof for the isolation of clones which secrete specific antibodies or antigen-binding parts thereof.
Finally, the invention is explained in detail in the examples and in the patent claims.
EXAMPLES
Example 1
Preparation of an antibody expression vector
The plasmid pKK233-2 (Amann and Brosius, (1985) Gene 40, and Straus and Gilbert (1985) Proc. Natl. Acad. Sci. 82, 2014) was chosen as base vector for the construction of the antibody expression vector (FIG. 1).
Before the incorporation of the antibody operon, the plasmid was cut with SalI and BamHI, the ends were filled in with Klenow polymerase and ligated. By doing so, the two restriction sites and the DNA between them were deleted.
Additionally, the plasmid was cleaved with HindIII, the ends were filled in with Klenow polymerase and ligated using BamHI linkers. By this procedure, the HindIII restriction site was removed and a BamHI site inserted. The antibody DNA was inserted into this modified plasma. A simplified structure of the antibody operon coding for a dicistronic antibody mRNA is shown in Tab. 2. In order to make possible the secretion of the antibody, the leader sequence of the bacterial enzyme pectate lyase was used. The leader sequence of this enzyme has already been used for the expression and secretion of a chimeric murine/human antibody (Fab fragment, Better et al., loc. cit.), and of the variable region of a "humanized" antibody (Ward et al., loc. cit.; Huse et al., loc. cit.). DNA for the first leader sequence (P.sub.1 upstream of the heavy chain) SEQ ID NO:21), and the sequence for a second ribosome binding site (RBS) and a second leader sequence (P.sub.2 upstream of the light chain) (SEQ ID NO:23) were synthesized from several oligonucleotides (SEQ ID NOS:25-32) (Tab. 3).
Antibody cDNAs which code for the variable regions of the heavy and light chains of a human antibody (HuVhlys or HuVllys; Riechmann et al., (1988) J. Mol. Biol. 203, 825) were obtained from Dr. G. Winter (Cambridge, UK). The restriction sites HindIII (HuVhlys) and EcoRV (HuVllys) were introduced to make possible the insertion of the antibody cDNA into the expression vector. Further restriction sites for BanII (HuVhlys) and BstEII or KpnI (HuVllys) were introduced to exchange hypervariable regions en bloc. At the end of the HuVhlys cDNA sequence a stop signal was incorporated. A BanII site in the light chain was removed. These alterations were carried out by means of site directed mutagenesis in the bacteriophage M13mp18 (Zoller and Smith, Meth. Enzymol. 100, 468-500). The sequence of the completed antibody DNA is shown in Tab. 4 (SEQ ID NOS:35-38).
For the insertion of the leader sequence P.sub.1 (SEQ ID NOS:21-22) (Tab. 3) the modified plasmid pKK233-2 was digested using the restriction enzymes NcoI and PstI, and P.sub.1 was inserted in between these sites (pKK233-2-P.sub.1). Further cloning steps, apart from the last step, were carried out using the plasmid pUC18. The reason is that the presence of individual parts of the antibody operon in the expression vector adversely influences the growth of the bacterial host.
Before the cloning in pUC18, its BamHI restriction site had to be removed. After digesting with BamHI, the single-stranded ends were filled in using the Klenow fragment and were relegated. This modified plasmid was then digested using PstI and HindIII, and P.sub.2 plus RBS (SEQ ID NOS:23-24) was ligated in between the restriction sites (pUC18-P.sub.2). During this process , the original HindIII e restriction site of the plasmid disappears and a new HindIII restriction site is incorporated. pUC18-P.sub.2 was then digested using pstI and HindlIl, and the DNA of the heavy chain (PstI-HindIII insert from M13) was ligated into these two sites (pUC18-HP.sub.2). This plasmid was then digested using EcoRV and BamHI, and the DNA of the light chain (EcoRV-BamHI insert from M13) was ligated in (pUC18-HP.sub.2 L).
The PstI-BamHI insert was then recloned in pUC18 after the restriction sites for HindIII, BanII and KpnI therein had previously been removed. The HindlII restriction site was removed as above for pKK233-2, the religation taking place without an insertion of BamHI linkers, however. Subsequently, the resulting plasmid was digested using SmaI and BanII, and, after filling in the protruding ends by means of T4 DNA polymerase, religated. The insertion of the PstI-BamHI restriction fragment results in pUC-HP.sub.2 L. In a preferred embodiment, a Tag sequence was additionally inserted in the BanII and HindIII restriction sites (Tab. 3) (SEQ ID NOS: The Tag sequence encodes the recognition sequence Glu-Gly-Glu-Glu-Phe of the monoclonal antibody Yl 1/2 (Wehland et al., (1984), EMBO J. 3, 1295). Because of this peptide marker the expression product of the resulting plasmid pUC-HTP.sub.2 L is readily detectable.
For the insertion of HP.sub.2 L or HTP.sub.2 L in the expression vector, the two plasmids were cut using PstI and BamHI, and the PstI-BamHI HP.sub.2 L insert from pUC-HP.sub.2 L or the HTP.sub.2 L insert from pUC-HTP.sub.2 L was ligated into the modified plasmid pKK233-2-P.sub.1 into these two restriction sites. A diagrammatic representation of the completed expression vector pFMT is shown in Tab. 5.
Example 2
Synthesis of antibody DNA containing random sequences in hypervariable regions
The synthesized oligonucleotides for the synthesis of the variable parts of antibody DNA are compiled in Tab. 1. For the synthesis of the hypervariable regions almost random nucleotide sequences were used. Limitations on the random nature are illustrated in Tab. 1 (SEQ ID NOS:1-20). Two different sets of oligonucleotides were synthesized. In set A (SEQ ID NOS:1-14) the hypervariable regions are predominantly random apart from those few positions where almost exclusively certain amino acids occur. In set B (SEQ ID NOS:1-2, 5, 7-9, 12 and 14-20), the random nature of the nucleotide sequences in those positions where predominantly one or few amino acids occur was additionally limited.
The oligonucleotides were purified by HPLC chromatography or polyacrylamide gel electrophoresis, and then 5'-phosphorylated.
Example 3
Ligation of the synthetic oligonucleotides
The oligonucleotides in Tab. 1 (SEQ ID NOS:1-20) were ligated together stepwise on an antibody DNA template. For this purpose, large amounts (about 1 mg) of single-stranded M13mp=18 DNA containing the antibody DNA inserts were isolated. In order to separate the antibody DNA from the vector, the inserts were made double-stranded on the two ends using two appropriate oligonucleotides and were digested using the enzymes PstI and HindIII (heavy chain) or using EcoRV and BamHI (light chain). The antibody DNA was then purified using agar gel electrophoresis.
On these DNA templates, first only three oligonucleotides were ligated: H1, pH2 and pH3 (heavy chain (SEQ ID NOS:1-3 and 15)), and L1, pL2 and pL3 (light chain (SEQ ID NOS:9-11 and 18-19)), H1 and L1 having been marked first with .sup.32 P at their 5' end ("p" designates 5'-phosphorylated). Amounts of 100 pmol of each oligonucleotide were used. The hydridized oligonucleotides were purified on 2% agarose gels and analyzed on a sequencing gel. The amount was determined by a radioactivity measurement. Equimolar amounts of pH4 and pH5 (SEQ ID NOS:4-5 and 16) (heavy chain (SEQ ID NOS:4-5)), and pL4 and pL5 (SEQ ID NOS:12-13 and 20) (light chain) were then ligated onto the already ligated oligonucleotides on each particular template. These DNAs were then purified as in the preceding step and the procedure was repeated up to the purification step, using equimolar amounts of pH6 and pH7(SEQ ID NOS:6-7 and 17). Finally, the ligated oligonucleotides were purified by means of a denaturing polyacrylamide gel and preferably amplified using the polymerase chain reaction (PCR). Alternatively or in order to avoid losses caused by the last purification step, the oligonucleotides were amplified using PCR directly after the last ligation step. The primers H1 (SEQ ID NO:1) and H8 (SEQ ID NO:8) (heavy chain), and L1 (SEQ ID NO:9) and L6 (SEQ ID NO:14) (light chain) were used under standard conditions for the PCR. Amplified template DNA was digested selectively using KpnI (light chain) or using AluI (heavy chain). Where appropriate, a second amplification step using the PCR was subsequently carried out.
Example 4
Insertion of the antibody DNA into the expression plasmid
The synthesized antibody DNA was cut using the restriction enzymes PstI and BanIII (heavy chain), and BstEII and KpnI (light chain). The bands having the expected molecular weight were purified by agar gel electrophoresis, precipitated using ethanol and then, in two steps (first the DNA of the light chain and then the DNA of the heavy chain), ligated into the pUC-HP.sub.2 L (see above) which had been cut and purified in the same way. The HP.sub.2 L insert was then ligated into the restriction sites PstI and BamHI of the plasmids pKK233-2-P.sub.1 (see Example 1). An analogous way was used for the HTP.sub.2 L fragment. The antibody library is therefore established in the antibody expression plasmid (Tab. 6). The reason for intermediate cloning in pUC is that the presence of individual parts of the antibody operon in the expression vector has an adverse influence on the growth of the bacterial host (see above also).
Example 5
Expression and screening of antibodies in E. coli
Competent E. coli are transfected with pFMT plasmids containing the inserted antibody-DNA library, grown on agarose plates and then incubated using nitrocellulose filter s coated with the desired antigen. After removing non-specifically bound antibodies, the active clones are identified with a labeled antibody against the human immunoglobulins secreted from E. coli. In the preferred embodiment, the antibody YL 1/2 which is directed against the Tag sequence is used for this purpose.
Legend for FIG. 1:
Restriction map of the expression vector pKK233-2 (Amann and Brosius, loc. cit.).
Ptrc denotes hybrid tryptophan lac promoter
RBS denotes ribosome binding site
rrnB denotes ribosomal RNA B operon
5S denotes gene for 5S RNA
Before cloning antibody DNA in the expression vector, the following alterations were carried out:
1) The SalI and EcoRI restriction sites were removed together with the DNA between them.
2) The HindIII restriction site was converted to a BamHI restriction site.