CA2177367A1 - Recombinant binding proteins and peptides - Google Patents

Abstract

DNA constructs comprise a first exon sequence of nucleotides encoding a first peptide or polypeptide, a second exon sequence of nucleotides encoding a sec-ond peptide or polypeptide and a third sequence of nucleotides between the first and second sequences encoding a heterol-ogous intron, for example that of Tetrahy-mena thermophila nuclear pre-rRNA, be-tween RNA splice sites and a site-specific recombination sequence, such as loxP, within the intron, the exons together en-coding a product peptide or polypeptide.
Such constructs are of use in methods of production of peptides or polypeptides, transcription leading to splicing out of the intron enabling translation of a single chain product peptide or polypeptide. Iso-lated nucleic acid constructs consisting es-sentially of a sequence of nucleotides en-coding a self-splicing intron with a site-specific recombination sequence within the intron, for use in creation of constructs for expression of peptides or polypeptides, are also provided.

Description

WO95/15388 21 77367 .~Il. l.''~2''~
l7~rf~MRT~NT sINDING ~k~ N~ AND ~;~LlV~iS
The E)resent invention relates to polypeptides which comprise two or more component polypeptides or peptides, methods fo_ making them and DNA constructs for the use in thi5 making. In particular, it relates to the provision of repertoire5 of such polypeptides and encoding nucleic acid therefor.
In this application, we describe the generation of binding protein8 and peptides using nucleic acid ~0 c~ntA;nin~ intron5 with RNA splice 6ites such as se`f-splicing introns, preferably in conjunction with a site-specific rP~- ' ;n~tion system, such as lox P
~Ioess et al Proc. Natl. Acad. Sci. USA 79 3398-3402, 1982; Sternberg et al ~. Mol. Biol. 150 467-486, 1981).
The site-specific recombination allows two sequences of nucleic acid to be cloned separately as libraries and be brought together sub6equently by a r/~ n~tion event (Waterhouse et al Nucleic Acid6 Re6. 21 2265-2266, 1993; A.D. Griffith6 et al. E~lBO J. in press; WO
92/20791~ WO 93/19172. One library of sequence is cloned into a first replicon and a second library of sequences into a second replicon. R~c~ ` in;ltion between the 6ite6 bring5 together libraries o~ both sequences on the 6ame replicon. This recombination can be performed in vivo e.g. by Pl infection or by using a recombina6e encoded by a plasmid in E. coli or in vi tro using soluble recombinase. For lox P, the recombinase i~ Cre. This allows a large library to be made where Wo 9S/15388 P~~ . 7----7 the l~mitation i5 not the cloning efficiency but rather the number of cells which can be grown. Thus the method is particularly powerful in combination with phage display technology which allows the selection of 5 proteins with des.red binding properties from a large library of displayed proteins (WO 92/01047; WO
92/20791; WO 93/06213; W0 93/11236; Wo 93/19172;
PCT/G~393/02492). The size of the library is significant for ability to select antibodies or other 10 binding proteins of appropriate af f inity and specif icity .

2 describes recombining two libraries of nucleic acid using a site-specific e.g. lox P, system mainly to code for heterodimeric proteins in 15 which two chains encoded by distinct (separate) nucleic acid sequences associate to form a functional binding site. Also described is the bringing together of two polypeptides for r~ntln~ us open reading frames.
Xowever, this imposes the use of an amino acid sequence 20 encoded in the site-specific recombination sequence at the junction between the two parts of the sequence, for instance the linker in single chain Fv molecules. A
problem with this is that there is only one open reading f rame in the lox P sequence and the amino acids 25 encoded by this may be incompatible with the expression of many proteins in functional form. If alternative lox P sites to the wild-type are used (eg see Figure wo 95/15388 2 1~ 7 7 3 6 7 r~l 1 r 6?
4), further different amino acid sequences may be generated, but the pos6ibilities are still restricted.
For in8tance, functional single chain Fv molecules can be ccnstructed with 15 amino acid linker3 5 encoded in part by the loxP recombination site. The length of the loxP site (34bp) however means that a minimum of ll heterologous ("foreign") amino acid3 must be incorporated into the final expressed protein. This makes the incorporation of a loxP site into a 10 r-nntinll~ug reading frame unsuitable for the construction of a diabody repertoire and also leaves little scope for the modification of~ scFv linkers to enhance expression.
The present invention involves RNA splicing, 15 particularly the use of self-splicing introns. This allows the recombination site to be inserted within the intron 50 that amino acids encoded by nucleotides which are spliced out are not incorporated into the final expressed protein. In such circumstances, the only 20 "foreign" amino-acids which need be incorporated are those derived from the sequences at either end of the self-splicing intron. (Note: the amino acid composition and sequence of the product can be engineered with precision and amino acids inserted, 25 substituted or deleted according to choice and using techniques known in the art. ) ~ hen a self-splicing intron is used, the amino acids that are incorporated derive from the Pl sequence WO 95/15388 2 1 7 7 3 6 7 ~ /S'?''~ --at the 5' splice site (5'SS) and the P10 sequence ~t the 3 ' splice site ~3 ~ SS) . These pair with the ;nt~rnill guiding se~uence of the intron to form hairpin loops (Figure l) and splicing then occurs as indicated.
The use of Yel:E-splicing intron8 allows the use of recombination by lox P to be extended to construction of large libraries of contiguous polypeptide chains where the two parts of the chain separated by the intron are varied.
In the application EP 93303614.7, pri~rity from which is claimed by PCT/G;393/02492, an example i9 given of use of a loxP site in8erted within a self-splicing intron with a bivalent or bispecific ~diabody~. A
~diabody'~ is a multivalent or multispecific multimer 15 (e.g. bivalent or bispecific dimer) of polypeptides wherein each polypeptide in the multimers comprises a f irst domain comprising a binding portion of an ~ l obulin heavy chain variable region linked to a second domain which comprises a binding protein of an 20 immunoglobulin light chain variable region such that the domain of a given polypeptide cannot associate with each other to form an antigen binding site. Antigen binding sites are formed -from an antigen binding site.
Antigen binding sites are formed by multimerisation 25 (e.g. dimerisation) of the polypeptides.
The expression of bivaIent diabodies from DNA~
cont~;n;n~ a self-splicing intron is shown in Figures 1 and 2. Application EP 93303614.7 also shows the use of Wo 95/1~388 2 1 7 7 3 6 7 P~
this system for chain-shuffling. (See al~o ~igure 3 . ) PCT/GB93/02492 describes sp~icing out a lox P ~ite using a self-3plicing intron for a bispecific diabody (Example l of t~is application). In these two earlier applications the use of 3elf-splicing introns was described for splicing only between the two domains of diabodies. The use of self-splicing introns to bring together two portions of polypeptide chain however has general applicability and can equally well be applied to single chain Fv fragments, peptide libraries or indeed any polypeptide sequence.
The use of systems such as lox P which promote recombination allows one polypeptide sequence to be replaced by another one with a similar or different function, originally encoded on another replicon. This is particularly useful with polypeptide chain3 such as single chain Fvs which have two or more domains which contribute to function. The invention allows the use o~ two repertoires of nucleic acid, with a splice site between the two repertoires and proteins or peptides thus encoded 3elected. In one embodiment, termed "chain ~hllffl;n~", one nucleic acid sequence is kept constant and the library of other chains reco-~hl n~ at the lox P site in the intron.
Self-splicing intron8 have been shown to be f~ln~-tir7n~l in E. coli using a system in which the Tetrahymena intervening sequence (a group I self-3plicing intron) was inserted into the gene ~n(-o~l; n~

WO 95/1~388 ~ IA~7 2~ 77367 6 the ~-~peptide of $-galactosidase ~J.V. Price & T.R.
Cech Science 228 719-722, 1985; R.B. Waring et al Cell 40 371-380, 1985; M.D. Been & T.R. Cech Cell 47 207-216, 1986). The presence of blue colonies indicated 5 that self -slicing was functional in E. coli ., becau3e the ~-peptide complemented the ~-galactosidase enzyme acceptor. This system has been used in diagnosis of the intron sequences which are compatible with self-splicing .
Although self-splicing introns have been inserted into functional proteins as above splicing introns have not been used for protein .~n~;n~ r;n~ strategies or for processes which involve the recombination o~ two repertories of nucleic acid.
The present invention provides a DNA construct comprising a f irst sequence of nucleotides encoding a first peptide or polypeptide, a second sequence of nucleotides encoding a second peptide or polypeptide and a third sequence of nucleotides between the first 20 and second sequences encoding a heterologous intron between RNA splice sites and a site-specif ic recombination sequence within the intron. The presence and position of the RNA splice sites render the intron operable for splicing out of nucleotides from between 25 the first and second sequences upon transcription o~
the DNA construct into RNA, which may result in splicing together of the first and second sequences.
Depending on the intron used, one or more nucleotides ~I wo 95/1~388 7 7 3 6 7 ~ 7 7 may remain between the f irst and second sequences in transcribed RNA following splicing, resulting in one or more amino acids between the first and second peptides or polypeptides in the product of translation of the 5 RNA. However, those skilled in the art will recognise that the first and second sequences may be termed " exon ~ se4uences .
The term "heterologous" (or "foreign") indicates that the lntron is one not found naturally between the 10 first and second sequences in a position operable for removal of nucleotides from between the first and second se~uences upon transcription. DNA constructs according to the present invention are "artificial~' in the sense that they do not occur naturally, ie without 15 human intervention by means of r~ro~'-; nAnt DNA
technology .
The first and second peptides or polypeptides may be any sequence of amino acids. Preferably, the first and second polypeptides together f orm a member of a 20 specific binding pair (sbp~, such as the antigen binding site of an immunoglobulin (antibody or antibody fragment). Thus, the rr~ nAtion of first and second polypeptides may form a polypeptide sbp member which is a scFv antibody fragment consisting of a VH domain 25 linked to a V~ domain by a peptide linker which allows the VH and VL domains of the sbp member to associate with one another to form an antigen binding site (Bird et al, Science, 242, 42~-426, 1988; Huston et al, PNAS
_ _ _ _ , . . _ _ _ _ , . . _ _ . _ Wo 95/15388 ~ 'A?''~

21773~67 8 USA,-85, 5879-5883, 1988) Ill such a case, the DNA
construct comprises a first sequence of ~ucleotides ~nro~l;n~ a VH or VL domain, a second sequence of nucleotides Pn~~orlin~ a counterpart VL or VH domain and 5 a third sequence of nucleotide9, between the f irst and second sequences, comprising a heterologous intron.
Upon transcription of the DNA construct into RNA and splicing out of nucleotides of the third sequence, nucleotides of the third sequence " ~ni n J in the RNA
10 encode, and are tr~n~l~t~hle into, the peptide linker of the scFv antibody f ragment .
This principle, with nucleotides of the third sequence Pn~o~;n~ and being tr~n~l~t~hle into amino acids of a linker joining the first and second peptides 15 or polypeptide chains, may be used or any peptides or polypeptides, for example in the creation of peptide libraries .
In pref erred ~ t ~ of the present invention, the first and second sequences encode 2~ peptides or polypeptides which are not linked in any naturally occuring polypeptide. The peptides or polypeptides may be derived f rom the same naturally occuring molecule but not linked directly by a peptide bond, ie they may be two parts of a polypeptide 25 naturally separated by one or more intervening amino acids. One or both of the first and second peptides or polypeptides may be an antibody fragment, for example VH, VL, CH, CL, VH-CH or VL-CL. The peptide or .. . . . . , , , _ _ _ ~ WO 95/15388 P~,l,~_, 1.'^7''?

polypeptide need not be a complete domain. One or both of the first and second peptides or polypeptides may be encoded by a synthetic nucleotide sequence, eg one created randomly. Thus, a random sequence peptide or 5 polypeptide library may be created f or example by expression from a repertoire or population of DNA
constructs, as disclosed, wherein the first and second exon sequences comprise randomly-generated nucleotide sequences .
The DNA construct may be transcribable into RNA
which, following splicing, encodes a "diabody~' polypeptide, ie a polypeptide comprising a first domain which comprises a binding region of an; n~l obulin heavy chain variable region and a second domain which 15 comprises a binding region of an ' n~l nh~l; n light chain variable region, the domains being linked (eg by a peptide bond or peptide linker) but; n~ hle of as~ociating with each other to form an antigen binding site. Where the domains are linked by a peptide 20 linker, the linker may, for instance, be 10 amino acids or fewer in length. See ~Iolliger et al, PNAS USA 90:
6444-6448 (1993) and PCT/US93/02492. Polypeptides of this kind are able to associate with one another to form multivalent or multispecific binding proteins 25 DNA constructs which can be transcribed into RNA which, following splicing, encodes such a "diabody~
polypeptide may, however, be excluded from the present invention .

WO95115388 2 1 7 7 3 6 7 1~l . , ?''~ --Other examples of first and second peptides or polypeptides include any polypeptide comprising binding regions of immunoglobulin heavy and light chain variable domains; V~Y/VI~ domains of T cell receptors; T
5 cell receptor/antibody tfragment) fusions; peptides, for example or epitope mapping of an antibody, receptor binding peptides, enzyme, eg protease, inhibitors; mutagenesis libraries of any multiple domain protein, for example nucleotide dehydrogenases l0 which have nucleotide binding domains and substrate binding domains, ~ hPq;nrl molecules such as ICAM-l, receptors such as PDGF-receptor which have a ligand binding domain and a kinase domain, transcription f actors which have a DNA binding domain and a second 15 domain which interacts with a ligand - such as the g1~ cnrticoid receptor. For a review of multiple domain proteins see Branden :and Tooze, "Introduction to Protein Structure~, Garland l99l.
The intron may be a self-splicing group I intron 20 such as ICEl0 from Tetrahymena (T.R. Cech Ann. Rev.
Biochem. 59 5~3=568, l990). Splicing out of the intron occurs at the RNA level leaving behind seo.,uences at the 5 ~ and 3 ~ splice sites, which would encode three amino acids between the two peptide or polypeptide components 25 o the product polypeptide. The self-splicing may be designed so that the number of amino acids . . i n; ng is dif f erent .

2 1 773~7 WO 95/15388 P~ '?~

other group I introns or group II self-splicing introns may be used. There are at least 149 self-splicing group I introns known, including: Tetrahymena thermophila rRNA intron, N~:ur V~JOLd CraBsa cytochrome b gene intron 1, Neurospora crassa mitochondrial rRNA, ~Drr crassa cytochrome oxidase subunit 1 gene oxi3 intron, phage T4 thymidylate synthase intron, Clamydomonas reinhardtii 23S rRNA Cr.~SU intron, phage T4 nrdi3 intron, ~n;lh~n.q pre tRNA(~eu) intron. Group 1~ II self-splicing introns include yeast mitochondrial oxi3 gene intron57 and Podospora anserina cytochrome c oxidase I gene.
Self-splicing introns may be used in cf~ in~ n with recombination, ~or example, at a lox P site, in the construction of molecules. For example, a lox P
site may be lnrl~ i in a self-splicing intron between the two domains (eg V~l and VL) of a polypeptide chain.
This may, for example, be recombined at t~e DNA level through a lox P site on another replicon carrying another variable domain gene and the appropriate region of a self-splicing intron. Self-splicing at the RNA
level following transcription will now lead to a product polypeptide chain with a new combination of f irst and second polypeptides .
In one aspect of the present invention the third serluence of nucleotides in the DNA construct, the intron, comprises a sequence for s~te-specific recombination. The sequence may be suitable for site-_ Wo 95/ls388 2 1 7 7 3 6 7 PCT/GB94102662 ~1 speci~ic recombination in vivo and/or in vitro. It may be the lox P site, a 34bp site at which recombination is catalysed by the protein Cre (~oess et al., PNAS USA
79: 3398-3402, 1982, and Sternberg et al., J. Miol.
~iol.; 150: 467-48~, 1981). The 34bp of the lox P
site consists of two 13bp inverted repeats separated by an 8bp non-symmetrical core ~see Figure 4) .
In order to provide more controlled recombination between two sequences leading to the resultant recombinant vectors desired, each vector may include two site-specific recombination sequences each of which is different from the other. The sequences should then be such that recombination will take place between like sequences on different vector8 but not between the different sequences on the same vector. The use of site - specif ic recombination allows f irst and second nucleic acid sequences originally on different (first and second) vectors/replicons to be brought together onto a single recombinant vector/replicon.
Each of the first vectors and each of the second vectors may include a first site-specific recombination sequence and a second site-specific recombination s~ n~-e different~ from the first, site-~pecific recombination taking place between first site-specific recombination se~uences on different vectors and between second site-specific recombination sequences on different vectors but not between a first site-specific wo 95/15388 r~ A?''7 recomoination sequence and a second site-specific recombination sequence on the same vector.
The first qite-specific r~- n--~in~ti,~n sequence may be lox P obt~-nilhle from coliphage P1 and the second site-.qpecific re~ ;n~tion sequence a mutant lox P sequence, or vice versa. Potentially, both the first and second site-speci~ic recombination sequences may be mutants, as long as the f irst sequence will not r~o~, ` ;nP with the each other and second sequences will recombine with each other.
A suitable mutant lox P se~[uence is lox P 511.
See Figure 4.
The f irst vectors may be phages or phagemids and the second vectors plasmids, or the f irst vectors may be plasmids and the second vectors phages or phagemids.
This system (ie employing site-specific recombination but not intron splicing) has been used in the preparation of ,~ntlh~ q displayed on phage (P.
Waterhouse et al., Nuc. Aci~ ~esearch 21: 2265-2266, 1993; and WO93/19172).
In one ' ~ , the recombination is intrac~ r and takes place in a bacterial host which replicates the r~ ' ;n~nt vector preferentially over the f irst vectors and the second vectors . This may be used to enrich selection of successful recombination events. The intracellular recombination may take place in a bacterial host which replicates plasmids preferentially over phages or phagemids, or which -w09slls388 21 77367 14 replicates phages or phagemids preferentially over plasmids. For instance, the bacterial host may be a polA strain of E. coli or of ~another gram-negative bacterium. PolA cells are unable to support 5 replication of plasmids, but can support replication of ~' 1 ous phage ~and phagemids (plasmids rnnt:~in;n~
f; 1 0~ 1 phage intergenic regions) . So, for instance, if the first vectors are plasmidg ~nntr~;n;n~
a first marker gene, and the second vectors are phage 10 or phagemids ~nnt~;nln~ a second marker gene, selection f or both markers will yield recombinant vectors which are the product of a success~ul recombination event, since recombination transferring the first marker from plasmid must take place in order for that marker to be 15 replicated and expressed.
The bringing together of ~ucleic acid f or two components or subunits of a product polypeptide, initially present on two separate replicons enables favourable combinations of subunit genes to be isolated 20 directly without recourse to extensive recloning, e.g.
using phage display. This may be achieved by re~ h; n~tion between the replicons once they have bee~
introduced into the same cell . In a pref erred configuration, recombination events are ef~ected such a5 that the genes for one of the component is recombined onto a recipient replicon which ~nnt~;nc the gene for a partner component. Preferably, the recipient replicon is capable of being packaged into a bacteriophage .. , . _ .. , . , . , _ _ _ _ _ _ _ ~ WO95/15388 2 ' 7 ~ 3 67 P~ 7 ~
parti`cle. Most preferably, the genes f~nrorl;nq one or more of the subunits is fused to a capsid gene such as gIII in order that the functional multimer can be displayed on the surf aee of the rgdp .
A variety of recombina~ion systems are known, and many of these could be harnessed in such as way as to effect recombination between replieons.
One of the most fully understood site-specific recombination sy9tems is that used in integration and excision of bacteriophage lambda (In "Escherichia coli and S~l m~nrl l a typhimurium. C~ and Molecular Biology. ~ (1987) . pplO54-1060. Neidhart, F.C. 13ditor in Chief. ~meriean Soeiety for Microbiology~. This bact-~- rrhAr,e can follow two devel~ 1 pathways onee inside the cell; lysis or lysogeny. The lysogenic pathway involves integration of the lambda genome into the chromosome of the infeeted baeterium; integration is the result of a site-specific recombination between a ea. 240bp sequenee in the baeteriophage ealled att P
and a 25bp site in the baeterial el~ srm~ ealled att B. The integration event is eatalysed by a host encoded faetor called IHF and a phage eneoded enzyme called Int recombinase, which rl~ro~n~ Rf~ a 15bp region common to the two att sites. The integrated DNA is flariked by ser~uences derived from att B and att P, and these are called att I, and att R. The integration event is reversible and is catalysed by Int, IXF and a second baeteriophage eneoded enzyme, Xis. It is Wo 95/15388 r~ s envis'aged that this system could be used for sequence transfer between replicons within E. coli. For example, the donor gene could be flanked by att L and att R sites such that when Int and Xis proteinæ are 5 provided in host cell, recombination between att L and att R site8 would create a circular DNA se~ment rf~n~in;n~ the donor gene and a recreated att B site.
This circular segment could then recombine with an att P site ~n~; n~P~-ed into the recipient plasmid.
For the work described in this application, the lox P/Cre system was chosen of the possibilities available because the recombination is highly sequence-specif ic, very ef f icient and occurs at a short target site that is readily incorporated into cloning vectors.
15 However, other site-speci~ic recombination systems may be used, for instance: flp recombinase (A. I-andy, Cur~. Opinion Genetics Devel. 3 699-707, 1993) .
A way of enriching for productive recombination events is to employ mutant sites. Several mutants of 20 the lox P sequence are known, and these are ~ I ~ed with respect to their ability to recombine with each other and the wild-type lox P sequence (EIoess, R.H., Wierzbicki, A. and Abremski, K. (1986) Nucl. Acids Res.
14, 2287-2300) . For example, lox P 511 has a G-,A
25 point mutation in the central 8bp segment, with the result that it will only r-~ ' in~ with other lox P 511 sites, but not the wild-type lox P sequence (Hoess, R.H. Wierzbicki, A. and Abremski, K. (1986) et supra.) .

.. . ... . . .

`1 WO9S/15388 2 ~ ~7~ 7 Placernent of wild-type and mutant lox P sequence comhinations can direct which recomhination events are possible The sites loxPl, loxP2, loxP3 and loxP4 (Figure 4) can he used in a similar way to loxP511.
5 These sites do not recombine significantly with loxP511. There is in some cases a degree of recombination between the loxPWT site and these mutant sites, derived from it. For instance, in one experiment 5~ rf~l ' ;n~tion was observedbetween loxP3 10 and loxPWT sites. All of these new loxP sites recombine efficiently with identical sites, ie like sites, eg one loxP4 site with another loxP4 site, and show strong preference for this over recomhination with a different site.
Provision of further different mutant loxP sites permits even greater control over the occurrence of recombination events leading to more complex, controllable and ~ff;c-i~n~ recomoination strategies being possible. The avA; l ~h; 1; ty of these loxP sites 20 has allowed the construction of a vector system including 3 loxP sites as in Example 6 This 310xP
system offers two additional features compared with the systems cnnt~;n;n~ two loxP sites:
(a) It should facilitate chain shuffling of light 25 and heavy chain genes for affinity maturation of antibody fragments (see Marks et al (1992), Bio/Technol~gy 10, 779-783 ) since one variable domain .

~ 2~ ~73~7 . -may be kept constant and a library of VH and VL genes recombined with it using an appropriate donor vector.
For example, a clone=specific for an antigen may be isolated where the gene for a VH domain of a scFv 5 fragment is located between loxPS11 and loxP wt of a vector c~nt~;n;n~ 3 loxP sites, such as fd310x. A
library of VL domains may then be shuffled with the VH
domain gene kept constant by recombining the clon in the 3 loxP site vector with a library of Vl genes on a 10 donor vector such as pUC19 which are located between the loxP~ site and the loxP 511 site The library of VL aomain genes is now encoded in the 3 lox site vector and scFv fragments, eg with improved affinity, may be selected from the phage displayed sc~v fragment 15 repertoire Although chain shuffling may be performed in 210xP systems, this 310xP system gives more ~lexibility, particularly to the nature of the replicon, phage or plasmid, where the reshuffled 20 repertoire is expressed, since both repertoires are flanked by loxP sites.
Example 6 and Figure 12 show the use of a loxP
eystem in model experiments for the construction of a diabody or single chain Fv repertoire where the VE~ and 25 Vl, genes are separated by a self-splicing intron containing a loxP site. The design of the system will faciliate chain shuffling as above.

~INC~D SHffT

21 ~7367 .
19 ' ' (b) It ~acilitates the trans~er of light and heavy chain gene pairs which have been selected on the surface o_ f;l -ntous bacteriophage for binding to antigen into a soluble expression vector for expression 5 of e.g. soluble scFv fragments, which at present needs to be done by cloning using restriction enzymes. The transfer by recombination could be achieved by creating an expression vector .nnt~;n;n~ a new mutant loxP site such as loxP4 and the ~T site and by recombination 10 between these two sites and the corresponding sites on the fd310x ve tor. Model experiments for this are described in example 6 and Figure 12.
The use of three dif_erent loxP sites also allows, for example, the recombination of three 15 sequences in order. One sequence to be recombined could be flanked by loxP and loxP511, a second sequence by loxP511 and loxP3. These sequences may then be recombined into a third replicon ~-~,nt~;n;n~ a third DNA
sequence and three loxP sites. The location of 2 loxP
20 sites within dif ferent self splicing introns allows the three sequences to be expressed continuously as shown in Figures 7 and 8.
Selection of productive arrangements may be facilitated by use of a polA strain of bacteria, 25 preferably E. coli or other gram negative bacterium.
These cells are ~f;~ nt i~ DNA polymerase I and are unable to support replication of plasmids (Johnston, S.
and R, D.S. 1984, supra. ) . ~owever, they are able to AAIE~ED SK ET

W0 9~/15388 1 ~
2~ 77367 suppo~t replication of f i 1 il~^ntnus phage and plasmids r~nt~;n1n~ f;li~ontous phage intergenic regions. If Cre-catalysed recombination is performed in polA
bacteria, by selecting for the presence of both 5 selectable markers in the same ~?olA cell successful recombination events are on7-; chot~, since rP~ ~ in~tion must take place f or t~e second marker gene to be replicated and expressed. The resulting cells then contain the complete repertoire and can be propagated lO as cells and infected with helper phage to produced phagemids cnnt~;n~n~ the genes for both chains and expressing them on their surface.
The invention also provides a vector comprising a DNA construct as discloaed. Generally, the vector 15 comprises nucleic acid necessary for expression. The vector may comprise nucleic acid f or secretion of the product polypeptide upon expression.
The present invention also provides a method of producing a polypeptide product which comprises a 20 combination of a first peptide or polypeptide component and a second peptide or polypeptide component, the method comprising:
providing a DNA construct comprising a first sequence of nucleotides encoding a first peptide or 25 polypeptide, a second sequence of nucleotides encoding a second peptide or polypeptide and a third sequence of nucleotides between the first and second sequences Wo 95115388 1 ~l,. L A?~--?

encoding a heterologou5 intron with a site-specific recombination sequence within the intron;
transcribing DNA of the construct into RNA;
causing or allowing splicing of nucleotides of the third sequence to produce an RNA molecule encoding the polypeptide product;
translating the R~A molecule into the polypeptide product .
The transcription, splicing and translation steps may take pl ce in in vi tro or i71 vivo systems .
CQnveniently, and particularly preferably for the construction of repertoires, these steps are performed in vivo, eg in E~. coli. Splicing may also be accomplished, less preferably, using in introns which are not self-splicing, by introducing the components of the splicing apparatus of eukaryotic cells, which promote splicing (J.A. Wise Scie~ce 262 1978-1979, 1993; A.J. Lamond, BioEssays 15 595-603, 1993), into eg E. coli.
The DNA construct provided may be any as discussed above. Suitable vectors for expression (transcription) can be chosen or constructed, cnnt~in;n~ d~J~LUpLiate regulatory sequences, including promoter seque~ces, terminator fragments, polyadenylation sequences, enhancer sequences, marker genes and other sequences as d~LUI!Liate, as is well known to those skilled in the art. For further details see, for example, Molecu7ar Cloning: a Laooratory , _ _ WO9S/15388 r~.. t.'r''7 --Manual: 2nd edition, Sambrook et al, 1989, Cold Spring Harbor Laboratory Press. Transformation procedures depend on the host used, but are well known.
Preferably, a phage or phagemid vector is used 5 and the vector, with the DNA construct, packaged into a bacteriophage particle. Advantageously, the polypeptide product comprises a domain which is a surf ace component of an organi9m such as a bacteriophage, for example a f;li ~us bacteriophage 10 such as fd or M13 . Preferably, the surface c ~ nn~nt is GIII of bacteriophage fd or the equivalent from another f; 1 i t~us ~ phage . Suitable technology is described in WO92/01047, WO92/20791, WO93/06213, WO93/11236, WO93/19172 and PCT/GB93/02492. Thus, the 15 provided DNA construct i9 packaged into a particle which displays on its surface the polypeptide product of expression from the construct, including the splicing step. III this way, polypeptide product with binding affinity or enzymatic eg catalytic affinity for 20 a target can be extracted from medium or selected from a mixture of dif f erent polypeptide products without such binding affinity or enzymatic activity, by contact with target eg using a chromatographic technique.
Where the polypeptide product is a sbp member, 25 selection may be on the basis of binding affinity for complementary sbp member: eg an immunoglobulin binding domain ( such as scFv f ragment ) can be selected on the basis of binding affinity for antigen.

_, , . , . . . _ . . , _ _ _ _ _ wo 95JI5388 2 ~ 7 7 3 5 7 . ~1 ~, n? ?

~ The step of provision of a Dr~A construct may actually involve the provision of a plurality, eg a repertoire, of constructs with different nucleic acid sequences. The term ~'repertoire~ is used to indicate 5 genetic diversity, ie variety in nucleotide sequence, and generally implies a large number of different sequences, perhaps of the order of m;ll;~nc (eg 107-lOg-10l'-101~. Highly diverse repertoires may be created when a sequence for site-specific re~, ' ;n~t;on, (aE3 10 discussed, eg lox P wild-type or mutant~, is included within the third sequence in the D~A construct at a site compatible with splicing upon transcription. The 6ize of a library generated by r~cn~; n~tion between one library and another is limited only by the 15 transfection efficiency. In principle, if each library 1nt~7;n.q, 107 clones, each recombination may introduce a further level of diversity of 107, thus recombination between a first repertoire encoding 10' different VH
domains with a second repertoire encoding 107 different 20 VL domains yields a re: ` ;n~nt repertoire encoding 101' different product polypeptides. Similarly, two libraries of 103 clones can be recombined to give a library of 1o6 clones.
For example, a first repertoire of replicons 25 comprising nucleic acid ~nf n~,;n~ a repertoire of first peptide or polypeptide component may contain part of a self-splicing intron, while a second repertoire of replicons comprising nucleic acid oncn~7;n~ a repertoire Wo95/15388 2 1 7 73 67 P~ 7 of ~e~cond peptide or polypeptide ro~n~nPnt contains a complement part of the self-splicing intron. The replicons in each of the f irst and 3econd repertoires of replicons each comprise a se~uence for site-specific 5 recombination, suitably positioned such that recombination of a replicon from the first repertoire of replicons with a replicon from the second repertoire of replicons results in frrr-t;r,n of the self-splicing intron in the resultant recombinant replicon.
lO Alternatively, replicons in either or both of the irst and second repertoires may contain a complete self-splicing intron.
The f irst and second repertoires of replicons may be recombi~ed ( "crossed" ), eg at a site-8pecific 15 recombination seriuence, to produce a third repertoire of (re,- i n~nt ) replicons which i nr~ nucleic acid f~nro~l;nrj a plurality of different ~ ~ in~tion~ of first and second peptide or polypeptide ~ , ^nt, with a self-splicing intron between the nucleic acid encoding 20 the first and second peptide or polypeptide ~ ^nt~
on each replicon. The recombination may take place in vivo in bacterial ho6t cells following transfection with the first repertoire of replicons and subseguent transfection with the second repertoire of replicons.
25 If the seSIuences for site-specific L~ ~in~tion are lox P, the recombination may be catalysed by Cre recombinase .

wo 95/15388 2; 7 7 3 b 7 r~ ,'^?''7 Tran6cription of nucleic acid in the third repertoire of replicons into RNA is followed by splicing out of the intron ~rmt~;ning the seguence for 6ite-specific recombination, leaving mRNA coding for 5 polypeptide product which can be tr~nql ~te-l into the polypeptide product. The production of a repertoire of polypeptide products comprising different combinations of first and second peptide or polypeptide co~nn~nts may be followed by a step of selection of products of lO interest, such as those with a particular binding specificity or enzymatic activity.
Each replicon in the third repertoire of replicons may comprise a seguence enabling packaging of the replicon into a bact~or; f rh~e particle, and the 15 polypeptide product may comprise a surface ~ ^nt of a bacteriophage, as discussed. Then, particles may be 6elected from a repertoire of particles by their display of polypeptide product with a binding specificity or enzymatic activity of interest. Each 20 selected particle then c~mt~in~ DNA encoding that polypeptide product.
Figure 5. demonstrates the principle for use in production of a scFv repertoire. There the "first polypeptide compone~t" of the polypeptide product is a 25 V~I domain and the "second polypeptide component ~ of the polypeptide product is a Vl domain. A lox P site is included within a Class I self-splicing intron. The peptide linker of each scFv fragment in the product Wo95/15388 2177367 ~ .'^7~7--reper~oire is formed, at least in part, by remnants of the splice sites left after splicing out of the intron between the ~/H and V~ domains upon transcription.
Instead of using two repertoires in the 5 generation of a recombinant repertoire for expression, a single first or second peptide or polypeptide -nt may be '~chain shuffled" against a repertoire of corresponding second or first peptide or polypeptide ~ rn~nt. Thus, in the generation of a repertoire of lO scFv ~L~- rq to be used in selection for a scFv fragment able to bind to an antigen of interest, either a VX or a VL domain known to be able (with complementary VL or V~I domain) to bind to the antigen may be ~ ' ~in~o~ with a repertoire of complementary VL
15 or VH domain to produce a repertoire for expression f ollowed by selection on the antigen ~or palrings able to bind.
A further aspect of the present invention provides nucleic acid comprising a sequence of 20 nucleotides encoding self-splicing intron with a site-specific recombination seriuence, such as a loxP site or a mutant or derivative thereof, within the intron.
Preferably such nucleic acid consists essentially of a seriuence of nucleotides .-nrr~l;ng self-splicing intron 25 with a site-specific recombination sequence within the intron. Such nucleic acid may be isolated and is suitable for use in creation of constructs for use in a method as herein disclosed. ~Preferably, the nucleic . . . ~

WO 9SI1~i388 2 1 7 7 3 6 7 P~ I

acid comprises restriction sites flanking the i~tron, for ligation of nucleic acid ~n~n~l;n~ or peptides. The nucleic acid may be incorporated in a vector operably linked, ie under the control o, a promoter for 5 expression. Other preferred features are as disclosed herein with ref erence to the methods and the DNA
constructs. In particular, the 3ite-specific recombination sequence within the intron is preferably heterologous, as discussed.

10 SEIIF-SPLICING 1'0 FORI~ DIAi30DIES OR SINGLE C~AIN Fv FRAG~¢ENTS
A recombination site (eg. lox P) may be nrl 71~Pt~
in a self-splicing intron between the two antibody domains of the polypeptide chain. This may, for 15 example, be r~r~ hinP~ at the DNA level through a lox P
site on another replicon carrying another variable domain gene and the appropriate region of a self-splicing intron. Self-splicing at the RNA level following transcription will now lead to a diabody 20 polypeptide chain with a new combination of variable domains or a single chain Fv polypeptide, depending on the length of the linker region encoded. In PCT/GB93/02492 the splicing of an intron from RNA
Pn,n~;ng a diabody polypeptide is described. This can 25 readily be /~tPn~P~l to single chain Fv fragments by introducing the sequence encoding the extra amino acids ,.

2l 77367 on either side of the RNA splice sites encoding the appropriate length of linker.
Chain shuffling can be performed for bivalent or bispecific diabodies or for single chain Fv fragments using the systems described in Figures 3 and 5. As noted above, a further level of control may be e9tAhl ;~h~1 by the use of a system with 3 loxP sites, as shown in Figure 12. The expression of diabody and single chain Fv molecules from clones rnni~i~;ninr~ loxP
sites within self splicing introns is demonstrated in mrl I~C 1, 2 and ~ . Example 3 demonstrates the feasibility of making a large library which recombines two exons into a longer rnntinllrus ser~uence. This methodology for making a repertoire can be applied to other molecule9 such as single chain Fv fragments and diabodies where the VE~ and VL genes replace the peptide seriuences. Example 6 describes model experiments which demonstrate that recombination can be performed between loxP sites configured for the construction of diabody or single chain Fv repertoires. It is concluded that this methodology is suitable for the librari~s described in example 3 and Griffiths et al (199~, supra) and that libraries of more than 1012 independent scFv or diabody clones are feasible.
As discu9sed further herein, introns with splice sites, such as self-splicing introns, rrnt~in;n~ an int~rn~l lox P site may be applied to any other system where two functional domains come together, for ~IEN~D St~

~ W095/15388 21773~7 I~ . tlr--~
instar;ce T cell receptors or two domain proteins. In addition to proteins with natural variants such as antibodies, for any two domain proteins mutagenesis libraries can be made for the two domains and then S C~ ` in~d using the lox P system.
In addition to splicing together libraries of domains, such as V~I and VL domains, parts of domains may be spliced together, eg using a self-splicing intron. For instance, the use of a self-splicing 10 intron ~-nntAln;ng a recombination site such as lox P in f ramework 3 of V domains allows recombination of fragments c~)n~A;n;ns CDRs 1 and 2 with fragments c~7ntAin;n~ CR3, eg in 3R3 shuffling SPLICTNG INT~ONS/R~ rNZrION IN THE~ c L1Nsl~tu~lON OF
15 PEPTIDE r.Tr~RA1~Tr.'~
Libraries may be made where two se~uence ~nrnf~; n~
peptides are encoded separated by a self-splicing intron c~ntA;n;n~ a rec ' ;n~tion, eg lox P, site. For instance, two separate libraries of ten amino acid 2 0 peptides can be cloned and then recombined via the lox P 511 and lox P sites as is shown in Figure 6. The amino acids encoded by the region of the 5 ' and 3 ' splice sites make this into a total 25 amino acid peptide with 5 constant amino acids in the centre. The 25 peptide library can then be used for a number of purposes, for instance the epitope mapping of antibody binding sites or to derive new molecules such as W0 95/15388 r~ ^?~?
2 ~ 773~7 receptor binding proteins, protease inhibitors or Gubstrates .
Example 3 shows that a large phage display library of ca. 5 x 101 recombined 25 amino acid 5 peptides may be constructed using recombination between loxP sites c-~nt=;n~-l in a self splicing intron and peptides ,-~t=in;n~ the epitope recognised by an anti-pS3 antibody selected. Constrained peptide libraries may be made by incorporating a cysteine residue in each lO of the 10 amino acid peptides to be recombined 80 that a disulphide bond is formed and the peptides between the cysteine would form a loop. The five amino acid linker may be varied in length and amino acid sequence by varying the 5 ' and 3 ' splice sites and the reading 15 frame. The number of random amino acids may also be varied and need not be the same on either side of the linker. This example demonstrates the feasibility of making a large library which recombines two exons into a longer continuous sequence.

ao USE OF TWO OR ~ORE INTRONS IN C'(JN~ U~lON. OF
REr~ rNA NT ANTIBODIES
Two or more ~ splicing introns may be used to link together three or more nucleic acid sequences encoding polypeptides. This may be particularly advantageous in 2s constructing libraries where V-D-J recombination (for the antibody heavy chain) occurs in E. coli. The use of site-specific r~ ' ;n=tion sequences (e.g. lox P) W0 95/15388 2 ~ 7 7 3 6 7 P~ n7~7 withi`n the introns (e g. using the scheme in Figure 7~
allows this V-D-J recombination of VH domains to occur in E. coli in the presence of recombinase (Cre for lox P~. The VH, DH and JH regions may be natural V, D and S J genomic segments regions or derived from synthetic oligonucleotide sequences, perhaps of different lengths, ~Are~ y for the D region, so that the range of CDR3 lenghts generated by the re~~ ' in;~tion may reflect the same (or a modified) distribution of 10 natural CDR3 lengths and the presence or absence of base addition. Figure 7 shows the use of lox P to achieve V-D-J recombination to obtain a 6ingle chain molecule and Figure 8 shows the expression of this molecule. The introns and splice donor and acceptor 15 sites need to be designed to ensure that splicing does not cut out the exon 9ited between the two introns.
The introduction of a fourth intron c~nt~ini"~ a different r~o~in~t1on site would allow the linking of different CH1 domains to the J region.
An analogous system may be used for T cell receptors a similar system may be used for r~hllffl in~
V and J regions of light chains.
SELECTION OF ~'~ Uh'NO'h'S FOR T~lE 5 ' AND 3 ' SE'LICE SITES
When an intron is deleted by a self-splicing 25 process, a residue of the intro~ is left behind within the coding region of the polypeptide, due to the 5 ' and 3~ splice sites. Example 1 shows two different amino , _ , WO 95115388 r~
21 773~7 3-2 acid sequences incorporated into a diabody due to this residue of the intron, with variation in expression occurring . There are likely to be dif f erences in the expression of a number of proteins depending on the 5 nature of the ~l and PlO se~uences. Therefore, there may be a need in certain cases to identify amino acids which are compatible with successful splicing of the intron and expression of protein.
Identif ication of suitable amino acids lO incorporated due to the bases at the 5 ' and 3 ' splice sites may be done by mutating bases (eg randomly) in the region of the internal guiding sequence with complementary bases which form the Pl hairpin loop of the intron. If the intron is now inserted, between the 15 nucleic acid encoding the first and second peptides or polypeptides, for instance between the VH and Vl domains of antibody ~L~y~ such that efficiently spliced polypeptide product is produced and may be displayed on phage and selected by binding to target, 20 those sequences ,~ ,~r;hle with efficient splicing can be selected. Similarly, sequences of the 3 ' splice site can be varied together with those of the ' nt~rn~l guiding se~uence and those which are ef f iciently spliced selected by the expression of the polypeptide 25 sequence.
The above procedures apply when the bases of the internal guiding sequence that are to be changed only participate in one of the Pl and PlO hairpin loops. It WO95/15388 2 ~ 7 7 3 6 7 ~ ,'^7''7 can be seen f rom Figure l that the central ~ases of the internal guiding sequences participate in both the Pl and P10 hairpin loops. Thus for these bases it is n~r~qq~ry to mutate the bases of both the 5' and 3' 5 splice sites as well as the ;n~PrnAl guiding sequence in order to --;nt;l;n complementarily and self splicing.
Example 4 shows that mutations may be made at the 3' splice site and ;n~rn~l guiding sequence of the self splicing intron to allow the ~nt~n~; n~ of amino-10 acids compatible with higher expression, after selfsplicing of R~A, of both dia} ody and single chain Fv antibody fragments. This directed mutation procedure may be applicable to other sites of the self splicing intron .
Whe~ repertoires are to be made, the GLSSG
sequence used in Example 1 may be used as the irst trial sequence for the sequence linking the two polypeptides following splicing out of the intron.
Further sequences identified, eg using a mutation 20 process as described in Example 4, may be used as alternatives .
To select the sequences of the splice site at the 5 ' end of the exon which are retained in the mature protein after splicing of the pre-mR~lA that are 25 compatible with self-splicing, the sequences of known self -splicing introns may be P~ m; n~cl (F . Michel and E. Westhof J. Mol. Biol. 216 581-606, 1990; F. Lisacek et al .J. Mol. i3iol. 235 1206-1217, 1994) . Sequences Wo9SI15388 21 77367 r~,l,~..,,~,'A?''7 --compatible with self-6plicing leading to the incorporation of f avourable amino acids may then be chosen .
CONAI'ROL OF SELF-SPLICING USING ST~BPAl'OMY IN
Streptomycin prevents ~elf-splicing. Thus the use of streptomycin in Str-R B. coli will prevent splicing occurring in transcribed RNA. The removal of streptomycin will aloow the generation of a spliced RNA
product, leading to, on translation, a protein product which is only generated on splicing. Thus, one could have a cloned gene which does not express an active protein in the presence of streptomycin in the growth medium, but does 80 in its absence. This may be useful for expressing proteins which are toxic or reduce growth in E. coli, for example ~n~;hn~l;es directed against E. coli proteins or inhibitors of E. col i enzymes, where expression o~ the toxic protein can be switched of f until re~uired.:~
The present invention will now be illustrated further by way of example. Modifications and variations within the scope of the present invention will be apparent to those skilled in the art.
All documents mentioned in the text are incorporated herein by reference.

WO ~5115388 r ~. .. 1 . -'7 Figure 1 shows a schematic of a self-splicing intron, including the Pl and P10 helices and the lnt~rn~l guiding sequence. The splice sites are marked by arrows.
Figure 2 illustrates the expression of a single chain Fv or diabody polypeptide from DNA rr,nt:lln;nrg a self splicing intron . The sequences f ~ ;Ink; ng the self splicing intron will determine the length of the peptide linker. Ribosome binding sites are indicated by open circles, Lg3 is the leader sequence for phage fd gene III.
Figure 3 illustrates chain qhl]~fl ;nrJ of a diabody (or a single chain Fv) molecule. It shows the replicons generated by Cre-mediated rt-~ ' ;niltion ~5 between the acceptor phage vector fdDOG-2dialoxsplice (A) and the donor pla5mid vector pUC19-2dialoxsplice (B). A is based on fd-tet-DOG1, with the chain VHA-VLB
in one cistron under control of the gene III promoter.
Between V~A and VLB is inserted the self-splicing intron from Tetrah~nnena rnnt~in;n~ the lox P 511 recombination site inserted at a site compatible with sel~-splicing activity. B is based on pUC19 and rr~nt~;nq lox P 511, the distal part of the self-splicing intron ~rom Terahymena, VLA, and the lox P
wild type sequence in the same a~ rd~ -nt as A.
~ithin E. coli an equilibrium between the six replicons develops due to the reversible nature of recombination in th lox-Cre system. The same scheme will apply to Wo 95/15388 r~ A?~7 2~ 77367 36 both single chain Fv and diabody molecules, ~ r.-n~1;
on the length of the linker peptide between the variable domains. Product E would express fd phage displaying a single chain Fv or a diabody depending on 5 the linker length used.
A and B can cointegrate by recombination between either mutant or wild-type loxP sites to create chimaeric plasmids C and D respectively. Further recombination can then occur between the two wild-type l0 or the two mutant loxP sites, to generate the original vectors (A and B) or two new vectors (E and F) . The light chains of A and B are therefore exchanged, and product E now encodes fd phage displaying a single chain Fv or a diabody ~r~n~ll ns on the linker length 15 used. Product F contains the VL originally in A.
Within E~. coli an equilibrium between the six replicons develops due to the reversibel nature of rec1 ' ;n~t;on in the lox-Cre system.
Figure 4 shows the sequence of wild type and 20 mutant lox P sites.
Figure 5 illustrates the g~n.~r~t; ~n of a single . chain Fv repertoire by recombination between repertoires of VH and VL domains.
Figure 6 illustrates the generation of a peptide 25 library by recombination between two replicons (a) pUCl9-P~P and (b) fdDOG-PEP. rbs represents ribosome binding sites; LpelB is the leader peptide se~auence;
gIII is fd phage gene III; l0aa is a random :

Wo 95/15388 2 1 7 7 3 6 7 i ~1 ~ r --~

oligonucleotide (NNK) ~0 f~nt o~l; n~ ten amino- acid residues (K is an equimolar mixture of G and T); * iY an ochre stop codon. The expressed 8equence is:
aal -aa2 -aa3 -aa4 -aa5 -aa6 -aa7 -aa8 -aa9 -aalO -A-I.-~-R-Y-aall-aal2-aal3-aal4-aal5-aal6-aal7-aal8-aala-aa20.
Figure 7 illustrates the recombination of V, D
and J regions using recombination between lox P sites within self splicing introns. The VH, DH and JH regions may be natural VH, DH and JH regions or derived from synthetic oligonucleotides se5Iuences, perhaps of different lengths, especially for the D region, 80 that the range of CDR3 lengths generated by the recombination, reflects the same (or a modified)-distribution of natural CDR3 lengths. The scheme is ~5 shown for a single chain Fv molecule with the V~ domain fused to gene III protein. lxl~ lx2 and lx3 are 3 different lox P sites e.g. wild type lox P, lox P511 and lox P3. in2 and in3 are the two introns which contain lx2 and lx3 site3 such as the Tetrahymena rRNA
and the T~ sunY intron. (a) Acceptor vector; (b) donor vector l; (c) donor vector 2; (d) recombined fd phage.
Figure 8 shows the transcription, splicing and expression of a single chain Fv molecule constructed as in Figure 7, c~)nt~;n;n~ recombined V, D and J
regions, fused to gene III protein. The nucleic acid regions encoding the amino acids of the final product are shown as Expressed scPv-gene III fusion. (a) DNA;
~, 2 ~ 77367 -. .

(b) Primary tranacript; (c) Spliced transcript; (d~
Expressed scFv-geneIII fusion.
Figure 9 shows an alternative final product from recombination which mimics VH, DH and ~I recombination 5 in vi tro to generate a new VH domain . Two separate libraries of sequences of random nucleotides (x and y) which encode O to 15 amino acids are made and recombined using the lox/Cre system. lxl and lx2 are two distinct lox sites such as lox P5 511 and lox P
10 (wild type) . This scheme requires only one self-splicing intron and two dif~erent lox P sequences.
Figure 10 shows the construction of the vector fdDOG-PEP. (a) p~C19 NQ10 K; (b) fd DOG-FCK; (c) INT~ON_LoxP (wt); (d) fd DOG-PEP; r.b.s . - ribosome 15 binding sites; LpelB - leader peptide sequence; gIII -fd phage gene III (gIII); 10 aa - random oligonucleotide; * - OCHRE stop codon.

Figure 11 shows construct fdDWT/3 and three different linkers formed on expression from constructs 20 described in example 4. Sequence A is derived ~rom the unmutated self splicing intron. Sequence B is derived from the self splicing intron mutated at the 3 ' splice AtiEN[)ED S7~

site and in the i nt~rn;l~ guiding sequence . Sequence C
shows the sequence derived from the single chain Fv fragment. Bases contributing to the P1 and P10 hairpin loops are underlined. Restriction site bases are 5 outlined. The diagonal slashed line shows the bases between which the self-splicing intron i9 spliced out.
T7 is the promoter for T7 RNA polymerase. Fx is a site f or Factor X protease . Part D shows the schematic of the self splicing intron highlighting the bases which 10 are mutated (G to C in the P10 hairpin loop and its complementary base in the ;nt~orn~1 guiding sequence).
Figure 12A shows the fd phage acceptor vector, -fdDWT/4 cr~nt~;n;n~ 3 lox sites is shown. It (~nnr~;nq the V~I and Vl genes of the anti-NIP clone G6 (Gri~iths et al, 1994 supra). The sites loxP511 and loxPWT flank the VH gene and the sites loxPWT and loxP4 f lank the VL
gene. The loxPWT site is in the self splicing intron and the lo~P4 site sits between the VL gene and gene III. The diabody or 3ingle chain Fv polypeptide chain 20 encoded is expressed as a fusion with the gene III
protein. A site for the factor X protease is included between the Vl gene and gene III to allow the possibility of the elution by proteolysis of phage ~rom the antigen during selection procedures. Alternative 25 ver3ions of fdDWT/4 were also made with the site loxP4 replaced with loxP3 and loxPl respectively. The donor vector PDN8 cnn~ ;n~l the VH-D10 gene flanked by loxP511 and loxPWT sites. The donor vector pRWT/4 contains the ~ ENDED SHE~T

VL-Dlo gene ~lanked by loxPWT and loxP4 sites. In the donor vectors pRWT/3 or pWT/1 the loxP4 site of pRWT/4 is replaced by the loxP3 or loxP1 site respectively.
The expresaion vector pEX511/4 contains the S12 gene, 5 which confers streptomycin sensitivity on bacteria, flanked by loxP511 and loxP4 sites.
Figure 12B summarises the recombination ef f iciencies obtained in the experiments described in example 6. The left hand loxP site is loxP511, the 10 middle loxP site is the loxP site within the self splicing i~tron and the right hand loxP site is the loxP site between the VL gene and gene III.

EXAMPLE 1: I:JSE OF SELF-SPLICING IDrrRONS IN 'rHE
~C~N~l~U~ ON OF DIA}30DY ~OLECULES
In the work described in this example, a self splicing intron was introduced between the VE~ and VL
domain genes o~ two antibodies cloned in the diabody ~ormat, NQ11 and Dl . 3 directed against 2-phenyloxazol -5-one and hen egg lysozyme respectively. This self 20 splicing intron was shown to be spliced out following expression, as t~Pt~rmi nP~ by the expres3ion of functional bivalent diabodies.

G~)n.qtruction of NOll ~n,l ~1.3 clone~q Cont~in;n~ a self-s~licina intron. ex~i.qed to leave a five ~m~n~) acid 25 7ink~=r between VH ~nr7 VL ~ ;n.q of bivi~7ent ~ hf~lie,q ~E~G~D q~

-Wo 95115388 2 1 7 7 3 6 7 r~

The self-splicing intron from Tetrahymena (T.R.
Cech Ann. Rev. Biochem. 59 543-568, l990) has been shoo1n to be able to splice in the E . coli cytoplasm .
Such a self-splicing intron, from clone IOElO (Ian 5 Eperon, University of Leicester) was inserted between the genes encoding the VH and VL domains of the antibodies Dl . 3 and NQll in such a way as to create upon splicing out an open reading frame Pn~~O~; n~ a diabody with linker VE-GLSSG-VL. Without splicing no lO functional diabody can be produced as the self splicing intron cn~ n~ several stop codons in 3 reading f rames .
A restriction site for BstEII was incorporated at the 5 ' end of the primer TlbaBstEII and a SacI
l5 restriction site introduced in the primer TlfoSac.
This allowed the self splicing intron fragment to be cloned in a 2-way ligation reaction into the expression vectors pUCl19Dl.3 (Pnr~;n~ the V domains of the Dl.3 anti-lysozyme antibody) or pUCl9NQll (encoding the V
20 domains of the anti-phOx antibody NQll) each cut with BstEII and SacI.
TlbaBstEII primes at the 5 ' end of the self splicing intron and conserves the ; nt-~n~l guidance se~uence (IGS) reguired for splicing activity and 25 inserts a extra glycine residue at the 3 ' end of the VH
domain . Tlf oSac primes at the 3 ' end of the self splicing intron and conserves the thymidine base just 3 ~ of the self splicing intron which, though not part Wo 95115388 r~ . 1, `?''7 1 ~2 of the intron, is present in Tetrahymena DNA. TlfoSac inserts a extra Gly and Ser residue at the 5 ' end of the VL creating a 5 amino acid linker.
The self splicing intron was amplified with the primers TlbaBstEII and TlfoSacI using standard conditions (see eg example 14 of PCT/GB93/024921. The product of the PCR reaction was digested with restriction enzymes SacI and BstEII and ligated into BstEII/SacI digested p~JC119D1.3 or pl:JC19NQ11 in a molar ratio 4:1 (SSI:pUC119D1.3 or pUC19NQll) and the resulting ligation mixes used to transform E. coli TG1 cells . Recombinants were screened f or inserts of correct size using primers specific for self splicing intron, TlfoSac and TlbaBstEII.
Soluble diabody was expressed by growth at 37C.
Cells in log phase growth in 2 mL 2YT/0.1~ gluco3e/100 g mL~I ampicillin were induced by adding IPTG to a final concentration of lmM IPTG and grown 3 hours 22C.
The cells were cPntr; fl~ed (lOOOg 10 minutes) and the cell pellet rPqllqpPn~lP~l in 100111 ice cold PBS/lmM EDTA
and left on ice, 60 minutes. The cell suspension was centrifuged (lOOOg for lO minutes) and the diabody-cnnt~;n;n~ supernatant used in ELISA on lysozyme and phox (as described in example 1 of PCT/GB93/02492).
The ELISA signal (absorbance at 405nm) was equivalent (greater than 1.0 after 10 min) for the spliced 5 amino acid linker D1. 3 diabody to that obtained with the 5 amino acid linker D1.3 diabody ~Vo sslls388 2 1 7 7 3 6 7 r~,~. 1 ~? 7 (constructed in example l of PCT~GB93/02492) . However for the spliced 5 amino acid linker NQ11 diabody the signal was much lower (0.2 compared to 2.0 after 20 min) when compared to the 5 amino acid linker diabody constructed in example 1 of PCT/GB93/G2492, There three possible P~l~n~t1~n~ for this:
- the NQ11 diabody is not functional with the GLSSG
linker ser~uence, although thi8 appears unlikely;
- self-splicing does not work properly in the case of the diabody NQ11 because the DNA secluence 3 ' of the intron (at the 5 ' of the VL domain) is not suitable for self splicing . Whereas the D1. 3 ser~uence at 5 ' end of the VL domain gene is Pff ~ riPn~ at allowing self splicing, the NQ11 secuence in this region is poor;
- there is a cryptic splice site in this construct.
tructiQn of ~0~1 ~nr7 Dl.~ clones Gon~i"i"r a self-spliri"n int~o~ inr7U~7inn a loxP 8ite. exGised to leave a Aix ~min~ ac~d 7 inkPr between V~ An~ VL r7r---inr o,f bi val en t ~7 i ;~ di es The primers Tlba2Bst7~II and Tl fo'~ rI were designed to introduce into the ~7QlI construct æecluences 3 ~ of the self splicing intron which should enable efficient self splicing at the RNA level.
The self splicing intron was amplified with Tlba2BstEII and Tlfo2SacI by PCR. This intron was inserted between the VH and VL domain genes of antibody NQll and creates upon splicing out an open reading WO 95/15388 2 1 7 7 3 6 7 ~ ''7--~ramè encoding a diabody with linker VH-GSLKVG-VL.
Without splicing no functional diabody can be produced as the self splicing intron~ contains several stop codons in 3 reading frames.
A restriction site for BstEII was incorporated at the 5 ' end of the primer Tlba2BstEII and a SacI
restriction site introduced in the primer Tlfo2Sac.
This allowed the self splicing intron fragment to be cloned in a 2-way ligation reaction into the expression vector pUCl9NQll cut with BstEII and SacI Tlba2BstEII
primes at the 5 ' end of the self splicing intron and con3erves the bases at the 5' splice site which pairs with the internal guidance se auence (IGS~ required for splicing activity and inserts a extra glycine residue at the 3 ' end of the VX . Tlf o2Sac primes at the 3 ' end of the self splicing intron and conserves the thymidine base just 3 ' of the self splicing intron which, though not part of the intron, i8 present in Tetrahymena DNA
and inserts a extra Gly and Ser residue at the N-terminal end of the VL domain.
The self splicing intron used in this case t-ont~in~d a lox P site inserted between bp 236 and 237.
It was amplified with the primers Tlba2BstEII and Tlfo2SacI using standard conditions. The product of the PCR reaction was digested with restriction enzymes SacI and BstEII and ligated into BstEII/SacI digested pUCl9NQll in a molar ratio 4:1 ~SSI:pUCl9NQll) and the wo 95115388 2 ~ 7 7 3 ~ 7 ~ ,N`?''?
resulting ligation mix used to transform E. coli T
cells .
Recombinants were screened for inserts of correct size using the primers specific for self splicing 5 intron, Tlfo2Sac and Tlba2BstEII.
Soluble diabody was expressed as above and assayed by E3-ISA. In this case an equivalent signal (greater than l. 0 after lO min) was obtained with the 6 amino acid linker NQll diabody formed by self splicing lO as for the 5 amino acid linker diabody constructed in example 1 of BCT/GB93/02492. Thus this strategy allows more efficient self splicing in the NQll construct.

EXAMPLE 2: USE OF SELF-SPLICING INTF~ONS IN THE
C~NS~U~ ON OF A SINGLE CHAIN FV CLOÆ
In the work described in this example, a self-splicing intron is introduced between the V~l and VL
domain genes of an antibody, Dl . 3, cloned in the single chain Fv format, directed against hen egg lysozyme.
This self-splicing intron i5 3hown to be spliced out 20 following transcription, as determined by the expression of a functional single chain Fv molecule with a 15 amino acid linker.

Orn~:truction of Dl.3 clorle Cont~ininr a self-3l~licinq intron, excised to leave a fifteen ;~m;nr, acid lirlker 25 between VH and ~ rl~m~in~ of a sinqle rh~in Fv molecule Wo 95115388 2 1 7 7 3 6 7 r~

The self-splicing intron from l'etrahymena (T.R.
Cech Ann. Rev. Biochem. 59 543-568, (1990) ) has been shown to be able to splice in the E. coli cytoplasm.
It is inserted between the genes encoding the VH and VI.
5 domains of the antibody Dl . 3 in such a way as to create upon splicing out an open reading frame encoding a scFv with linker VH~ il,SSG-V~ ithout splicing no functional scFv can be produced as the self-splicing intron contains several stop codons in three reading 10 f rames .
A restriction site for BstEII is incorporated at the 5 ' end of the primer TlbascFvBstEII and a SacI
restriction site is introduced in the primer TlfoSac.
This allows the self-splicing intron fragment to be 15 cloned in a 2-way ligation reaction into the expression vector pUC119Dl.3 (~nroriinr the V domains of the Dl.3 anti-lysozyme antibody: Holliger et al (1993) s~pra) each cut with BstEII and SacI.
TlbascFvBstEII primes at the 5 ' end of the self -20 splicing intron and conserves the se~uences at the 5splice site which pair with the internal guidance sequence tIGS) required for splicing activity, and inserts an extra 10 amino acia residues at the 3 ' end of the V~. TlfoSac primes at the 3 ~ end of the self -25 splicing intron which, though not part of the intron,is present in Tetrahymena DNA and inserts extra serine and glycine residues at the N-terminal end of the domain .

~ W095ll5388 2 ~ 773~ r~l ~ r "~

The sel~-splicing intron used in this case contained a lox P site inserted between bp 236 and 237.
It was amplified with the primers TlbascFvBstEII and TlfoSacI using standard conditions. The product of the 5 PCR reaction was digested with restriction enzymes SacI
and BstEII and ligated into BstEII/SacI digested pUC119D1.3 in a molar ratio 4:1 (SSI:pUC19NQ11) and the resulting ligation mix used to transform E. coli TG1 cells. Recombinants were screened for inserts of 10 correct size using the self-splicing intron specific primers TlfoSac and TlbascFvBstEII.
Soluble single chain Fv is expressed as in example 1 and assayed for ability to bind ly80zyme by ELISA. A signal of greateer than 1. 0 is obtained after 15 10 minutes Uence, self-splicing introns may be used in nucleic acid encoding single chain Fv molecules.

EXA~PLE 3: CclNslKuclloN OF A DIVERSE REPERTOIRE OF 2 A~INO ACID PEPTIDES (CONTAINING 20 VARIED RESIDUES) DISPI.AYED ON PEAGE USING LOX P ~lzcnM~N~rIoN SITES

In the work described this example a diverse repertoire of 25 amino acid peptides (consisting of two variable lO amino acid peptide sequences separated by five constant amino acids) displayed on bacteriophage 25 was prepared by the recombination o~ two separate repertoires of 10 amino acid peptides cloned in separate replicons. Recombination between the lox P

WO 95/15388 P~~

sites under the control of the Cre recombinase allows their sequenceE to be linked. The f iIlal repertoire thus prepared rr~l in~q the diversity of the two peptide libraries (Figuro ~ 6) .
Conqtruction of the vector fdDOG-PEP
The VHCX fr~,; t of the antibody NQ10/12 . 5 was amplified from the vector pUC19 NQ10 k using oligo 3249, which introduces the lox P 511 site upstream of the pelB leader ser~uence and an ApaLI restriction site (see Table l and Figure 10) and oligo LMB2. The résulting fragment was then cloned into fdDOG1. (T.
rkqnn et al, supra) cut with ApALI and NotI. The group I self-splicing intron from Tetrahymena (T.R.
Cech et al Structural }3iology 1 273-280, 1994) rr,nt~;n;n~ a wild type lox P site (between nucleotide 236 and 237) was amplified with oligo 3189 (which introduces a EcoRI restriction site) and oligo 3193 (which includes the random oligonucleotide (NNK) 10 and a NotI restriction site). The resulting fragment was then cloned into fdDOG-BLX cut with SfiI and NotI to create the vector fdDOG-PEP.
-r,nqtrUction of the vector ~UC19-PEP
The group I self-splicing intron from Tetrahymena cnnt~;n;nrg a wild type lox P site (between nucleotide 236 and 237) was amplified with oligo 3I94 (which introduces a EcoRI restriction site and includes the ~ 2 1 77367 .- -random nucleotide (NNK) 10 and oligo 3~ 98 (which introduce9 a SfiI restriction site. The resulting fragment was then cloned into pUC19-210x (P. Waterhouse et al, 1993 supra) cut with SfiI and EcoRI to create the vector pUC19-PEP.
Cc-hinAtoriAl infection An~l in vivo re~ ' ;nAtion To create a large combinatorial repertoire of 25 amino acid peptides (with 20 amino acids displayed varied) on a fd phage the strategy of combinatorial in_ection and in vivo recombination was used (P.
Waterhouse et al Nucleic Acids Res. 21. 2265-2266, 1993). This system uses the lox-Cre site-speciiic r~ mh;nAtion system to bring the two 10 amino acid repertoires together on the same replicon, separated by a self-splicing intron.
109 E. coli TG1, harbouring the library of 10 amino acid peptides in ~dDOGPEP was used to i~oculate 1 litre of 2xTY broth r~n~Ain;n~ 12.5~Lg/ml tetracycline (2xTY-TET) and the culture shakeu for 30 hours at 30CC in two 500ml aliguots in 2 litre ba~fled Erlenmayer flasks.
Phage were purified irom the supernatant by precipitation with polyethylene glycol (J. McCaf ferty et al, Nature 348 552-554, 1990), resuspended in PBS
(phosphate buffered saline (phosphate buffered saline:
25mM NaH2PO4, 125mM NACl, pE~7.0) . Phage were titred by infecting exponential phase E. coli TGl (30 min, 37CC) A~.iENDED Sl{~ET

WO g5115388 2 1 7 7 3 6 7 P~ n7-~7 and p;ating on TYE-TET Yields are typically 6 x iol3 t . u . per litre of culture .
2.~ x 103 E.coli harbouring the plasmid pACYCara/Cre (Example 4) and the library of 10 amino 5 acid peptides cloned in pUClS PEP we~e used to inoculate 200 ml of 2xTY c~n~;nin~ 100 ~g/ml carbenicillin, 25 /lg/ml chloramphenicol, 2g/l glycerol and 1~ glucose (2xTYCaChglyglc) and grown overnight at 37C with shaking. 10 ml aliquots of the overnight 10 culture were used to inoculate 10 x 1 litre culture of 2xTYCaChglyglc in 21 Er1~ y~L baffle flask and the culture grown with shaking at 37C to A6~0 of 0.4.
1.4 x 10~2 t.u. of fdDOG PEP library were added to each Erlenmeyer baffle flask and incubated for i0 mins 15 at 37C without shaking. The 2xTYCaChglyglc r~7n~lnins the infected cells were then filtered through a 0.45 ~m tangential flow filter (P13I,LICON cassette, MIL~IPORE), and resuspended in 10 x 1 litre 2xTY ~ n~;nin~ 100 I~g/ml carbenicillin, 25 ~Lg/ml chl-,l h~nicol~ 15 20 ~g/ml tetracyclin, 2g/l glycerol and 0 . 5 g/l IJ (+) arabinose (2xTYCaChTetglyara) in 2 litre ~rlenmeyer baffle flasks and the culture grown with shaking at 30C
for 36 hrs. A sample was taken before growth to determine the library size by plating on 2xTY agar 25 plates containing carbenicillin, chlc,L ~ h~ni ~rll and tetracyclin . There were 4 . 7 x 10l~ independent clones .
The culture was then filtered as before. The recombined phage, in the f iltrate, were precipitated ~ ~ .

W095/15388 21 77 3 67 P~ ,r, ~'7 using PEG/NaCl and resuspended in a final volume of 26 ml PBS . The phage were titred by inf ectirlg exponential phase E. coli (30 mins, 37C) and by plating on TYE-tet.
The yield obtained was 6.0 x 10~3 t.u. total (the fdDOG-5 REC library glycerol stock). To deten~ine thefrequency of recombination, a PCR screen was performed by amplifying D~A from individual colonies u3ing oligos 4226 and pelBBACK (Table 1). 13 clones out of 50 xcreened gave a band on electrophoresis on a 6~
10 polyacrylamide gel whose mobility correspondæ to a size of 314 base pair3 (the expected size from recombined phage) and the others a band whose mobility corresponds to a size of 284 base paire (the expected size from unr~rr~l i n~ phage) . The recombination freque~cy was 15 thus 26~6. As there are multiple copies of plasmid and phage replicons in each bacterial cell when Cre recombinase i5 induced to promote r~ ' ;n~tion~ and at least 60 phage are produced per bacterium after overnight growth, we believe that each bacterium 8hould 20 yield at least one phage cr~nr,~in;nq the peptide ~rom the donor vector and that the overall library size is 4 . 7 x 101 clones .

Pro~aqation of l~haqe from the recomb; nl~ l; hr~rv 10 litres of 2xTY-TET were inoculated with a 35ml 25 aliquot of the recombined fdDOG-REC library glycerol stock (2.4 x 101l c.f.u). The cultures were grown with shaking overnight at 30C in baf1ed flasks (1 litre Wo 95115388 2 1 7 7 3 6 7 P~ ?''7 mediurn per flask). The cultures were centrifuged at 5000g for 15 min at 4C, the fd phaye precipitated from the supernatant using polyethylene glycol and each repertoire resuspended in a final volume of 10 ml PBS.

5 Total phage yields (from 10 :litres) are typically around 101~ t.u.

In V7 tro s~licinq of the intron within the recnmhined ~haqe To test or~the splicing of the intron within t~e 10 rP~ ' inP~l phage, 5 clones out of 31 positive recomhined clones were amplified using oligo-3520 and fdS~Q1. The size of the product after PCR was 619 base pairs ~expected size for a re~ ' ~inPd phage) and 589 base pairs (expected size for a unrP~ ' inPd phage).
15 The in vitro transcription was performed on 5 clones using an in vitro tran9cription kit ~Promega, Riboprobe II core System T7 R~A Polymerase, cat.#P2590) according to the manufac~urer' s instructions (l unrecomhined and 4 rP~ ~i nPd) . The samples were boiled and 20 electrophoresed on a 6~ polyacrylamide gel.
All 4 recombined clones showed a band corresponding to the spliced exon (198bp); and the unrecomhined one gave a band whose mohility cJLL~ ds to 168bp (spliced exon). These results indicate that 25 the splicing reaction occurs in the unrecomhined phage as well as in the recomhined one.

95/15388 r~ A?~"

Selection of clones from the 1 ;hraX~
The peptide library displayed on phage was selected f or the ability to bind an anti-p53 antibody (Pab240) which recognize a linear epitope on the surface of the cell with the amino acid se~auence RE~SV
(C.W. Stephen & D.P. Lane ~. ~ol. Biol 1992 225 577-5 83 ) .
The selection was pf~rf~ -1 on Immunotubes (Nunc;
Maxisorp) coated with the anti-p53 antibody coated at lO~lg/ml using methodology as previou31y described (J.D.
Marks et al., ~. I~ol. 131ol., 222, 581-597. 1991; A.D.
Griffiths et al., (1993) E~MBO ,J., 12, 725-734) . Four rounds of growth and selection were performed for binding of peptides displayed on phage to the anti-p53 antibody on Immunotubes using methodology as described by A.D. Griffiths et al (1994) E~BO ,J., 13 3245-3260).
The ability of phage from single isolated clones to bind to anti-p53 antibody was assessed by ELISA on plates coated with antibody p53. Phage were prepared as described by McCafferty et al (~upra) and ELISA was performed as described by Griffiths et al, (1993 supra) except that the second antibody used was an anti-sheep antibody coupled to ~ l k~ 1 i n~ phosphatase .
31 clones giving positive ELISA signals were amplified by PCR using oligo 3870 and fd SEQ1 (Table l). Aliquots were analysed by electrophoresis on a 1~
agarose gel. The r~ ;nin~ product was puri~ied using Magic PCR Preps (Promega) and used in PCR cycle WO 95/1~388 ~ r7 sequencing reactions with fluorescent dideoxy chain terminators (Applied Bio9ystem) and oliyos 4445 and 3358 according to the manufacturer' s instructions. The sequences are shown on table 2.
To check that the selected clones were specified, the same phage from single isolated clones were assayed by ELISA for binding to i~nt;hotli~R with the same isotype as Pab240 (IgG1) and either lambda and kappa light chains (Fog-1 and Fog-B) . The ELISAg showed that none of the selected clones cross-reacted with thege antibodies .
It was concluded that the same epitope RHSV is selected as a consensus sequence selected f rom the phage peptide library as described by Steven & Lane (1992, supra). Of the 31 selected peptides displayed on phage, 8 included the seSruence RHSV, 4 KHSV and 5 (R
or K)HS(L or I) and 3 (R or K)HSX.
Thus a large phage display library of ca. 5 x 10l recombined 25 amino acid peptides may be constructed 20 using r~ ;nAtion between loxP sites ~r1nt~;n~rl in a self splicing intron. This method should be particularly valuable for selecting, for example, peptides involved in binding to receptors. Constrained peptide libraries could be made by incorporating a 25 cysteine residue in each of =the 10 amino acid peptides to be re~l ' ;n~d so that a disulphide bond is formed and the peptides between the cysteine would form a loop. The amino acid linker could be varied in length wo 95/15388 2 1 7 7 3 6 7 F~~ A?Sr?

and amino acids by varying the 5 ~ and 3 ' splice sites and the reading frame.
This example demonstrates the feasibility of making a large library which recombines two exons into 5 a longer cont;n~ us sequence. This methodology for making a repertoire may be applied to other moleculeæ, including, for example, single chain Fv f ,~ _ nt~l and diabodies .

EXAD~PLE 4: MUTATION OF THE 3' SPLICE AND rNTT;~RN~, 10 GUIDING ~'h'~Uh~lC'h' OF A SELF SPLICING INTRON CONTAINING A
L,OXP SITE TO ENCODE A NEW DIA130DY LINKER WHICH IS
CO~PATI.3LE WITH HIOEIER E,.'PRESSION.
To utilise recombination by loxP in the construction of antibody repertoires a loxP 3ite can be 15 included between the two antibody domains, VH and VL of a single chain Fv fragment, in a continuous open reading f rame, employing the amino acid sequences encoded by tho5e loxP sequences as a linker. In this case the choice of linker is dictated by the length and 20 sequence of the loxP sites used. An alternative strategy is to employ RNA splicing of a group I self splicing intron inserted between the V~ and V~. A
recombination site such as loxP may be inserted within the intron 50 that the amino acid sequence encoded by 25 the site is spliced out from the R~A after expression and is therefore not incorporated into the final expressed protein.

-~ 21 77367 .
When a group I intron is deleted by self-3plicing process, a residue of the intron, derived from the 5' and 3 ' splice sites (which pair with the internal guiding sequence in the Pl and P10 hairpin loops 5 respectively), remains within the coding region of the polypeptide. Successful splicing is dependent on base pairing in the Pl and P10 hairpin loops involving the ;ntPrn~l guiding sequence (IGS) .
This example demonstrates that the 3 ' splice site 10 and the internal guiding sequence ~ay be mutated so that following splicing the amino acids encoded by the RNA are altered. These amino acids contribute to a 7 amino acid (diabody) linker which is compatible with higher level expression. It is further shown that the 15 mutated 3 ' splice site can be used in the construction of a single chain Fv molecule c~nt:~;n;n~ a 15 amino acid linker. In this example, vectors encoding scFv ragments or diabodies directed against the hapten NIP
(3-iodo-4-hydroxy-5-nitrophenyl-acetate) are 2 0 constructed and expressed using self splicing introns which include loxPWT sites to link the VH and VL
domains .

1, r~n~truction An~ P~Lnression of ~nti-NIp rl; ~hody from i:ln P~nression vecto~ conts;n;n(~ l-n~P in a self 25 s~licinç~ ;ntron, ~n~l mutation of tllP P10 hairpin loo~.
A diabody expression vector c~nt~;n;ng loxP in its self splicing intron is shown in Figure 11.

~IENDED S~ET

Wo 95/15388 2 1 7 7 3 6 7 p~

Salient featureæ of the construction of this vector are given below. The intron was amplified by PCR from the vector pUS19Tet-intron-loxP (which rrn~7~;n~ t~e loxPWT
sequence in~erted between bp236-237 of the Tetrahymena ICE10 intron sequence) using #3312 intron-lox-back and 3463 intron-for-2 oligos (Table l) which contain the sequences of the 5' 8plice site and the 7n7norn;~1 guiding sequence of the pl hairpin loop f lanked by a XhoI and NcoI sites at the 5 ' end, and 3 ' splice site of the P10 hairpin loop flanked by an ApaLI site and NotI at the 3 ' end respectively. The amplified product was cloned as a NcoI-EcoRI fragment into pUC19-210x (Waterhouse et al, 1993 supra). The intron is flanked by XhoI and ApaLI sites.
For the experiments de8cribed in this example, the VH and VL genes originate from the Fab fragment clone G6 (anti-NIP; A.D. Griffiths et al EM30 J. 13 3245-3250, 1994) . The VH gene was cloned into tne pUC
vector derivative as a NcoI-XhoI fragment. Promoter ser1uences for T7 RNA polymerase were introduced into the HindIII site and were flanked by SalI and HindIII
sites. The salI-NotI fragment rrnt~7nln~ the VH-NIP, self-splicing intron, loxP sites and T7 polymerase promoter was now sllhrl r~n~.7 from the pUC vector derivative into fd-DOG1 (~l Ark~rn et al Nature 352 624-628, 1991) which had its ApaLI site convertea to a SalI
site. The VL gene of G6 was cloned in as a ApaLI-NotI
fragment. An AscI site was subsequently introduced at ~ ~ 7 ' 2 1 773-g7 the 3 ~ end of the VL gene with a loxP3 site and a Factor X protease cleavage site between this AscI site and a NotI site at the 5 ' end of gene III_ The resulting construct fdDWT/3 is shown in Figure 11.
After splicing the RNA transcribed from fdDWT/3 encodes the polypeptide chain of an anti-NIP diabody with a seven amino acid linker S~K~ISA~ (Figure lla).
TGl cells were transformed with fdDNA encoding the construct and phage were prepared as-described in A.D.
10 Griffiths et al (1994, supra). This diabody was poorly expressed. The phage titer was lower than 107 TU/ml (at least a hundred fold lower than woul~ be normally expected). There was no detectable signal in the phage E~ISA for binding to NIP-BSA performed as described by 15 Griffiths et al. (1994, supra) . However, the intron was shown to be spliced correctly as shown by serluencing of the cDNA made from the spliced transcript .
To test for intron splicing the vector was first 20 amplified by PCR with the primers fd-PCg-Back and BamHI-for to prepare the DNA template rnnt~;n;nr, the T7 promoter ser~uence (Table 1) . From this template RNA
was prepared using an i~ vi tro transcription kit (Promega, Riboprobe II core System T7 RNA Polymerase, 25 cat.#P2590) . The original DNA template was first removed by digestion with DNaseI, and cDNA was then prepared using the First-Strand cDNA Synthesis Kit (Amersham). The cDNA was amplified by PCR with A~iENDED SH~ET

59 .
VH3BackSfi and .JK-FOR primers (J.D. Marks et al, ~J.
Mol. Biol.222 581-597, 1991), and was sequenced using the same primers. The sequence obtained demonstrated accurate splicing resulting in an S~KVSAI linker in the 5 expressed diabody product.
To improve expression, amino acids more compatible as linkers for the expression of diabody may be identified and used to alter the bases of the splice sites. To this end a second anti NIP diabody was 10 constructed in which the first G within the 3 ' splicing signal (P10) was mutated to C. To enable perfect pairing with the IGS the corresponding C within the IGS
was changed to G (Figure lld) . The intron of the vector pUC19Tet-iIltron-loxP was amplified by PCR with a second set of primers, #3877 encoding P1 and the mutated C to G in the IGS, and #3878 encoding P10 having a G to C mutation (Table 1) . The intron was cloned as above to give an analogous fdDWT/3 construct, but in this case after splicing of the intron, the 20 resulting RNA encodes the linker VX-SLNVSAL-VL (Figure llb) . The splicing of the mutated intron was tested by the cDNA sequencing of expressed RNA as above.
The ti~t j t~n of K to N in the diabody linker dramatically improved the expression of the diabody, 25 displayed on phage fd, resulting in a phage titer in the range of 5xl0a-109TU/ml and a phage ELISA signal for binding to NIP-BSA in the range of 1 absorbance unit.

lDED SIIEET

2~77~7 - .

Since diabodies require two polypeptide chains to form the antigen binding site (P. Holliger et al 1993, supra) and the only diabody polypeptide chains present are fused to gene III-protein, the E~ISA signal 5 indicated that some diabody polypeptide chains are cleaved from the fusion and combine with the gIII-diabody polypeptide fusion retained on the surface cf the phage, to form a flln~tif~n~l bivalent diabody which can bind to a NIP. Western blots were performed of 10 phage proteins with detection with an antibody directed against gene 3 protein as described by J. McCafferty et al (Protein ~ngineering 4 955-961, ~ 9gl) . This gave the relative proportions of gIII protein-diabody polypeptide fusion to cleaved fusion migrating at the position of native gIII protein to be 40g~ and 609~, respectively .

2. ~ression of ~nti-NIP scFv from ~n ~ression vector conti~;ning 1~ P in a self splicin~ ;ntron i~n~1 a m-ltation of the P10 h~ ~};)in loo~.
20 Since the SIN~SAII linker, derived by splicing of the mutated P10 hairpin loop, wa3 compatible with the (high) expression of a diaboody, a single chain F~
construct was made with a 15 amino acid linker which utilises the same mutated P10 hairpin loop for the self-splicing intron. The self-splicing intron cont~;n;n~ the loxP site is spliced out to give the amino acid sequence GGGGSL~IVGGGGSAL (Figure llc) .
~IEND~D S11ET

Wo95/15388 21 773~7 r~ ?

The OEelf splicing intron was amplified by PCR
from the vector pUCl9Tet-intron-loxP using the oligonucleotides 4243 and 4244 (Table l). These contain bases encoding a stretch of four glycine S residues f~l ~nki n~ the 5 ' and 3 ' splice sites respectively. Oligonucleotide 4243 cnnt~;nR the At;r~n of the int~rn;ll guiding sequence and oligonucleotide 4244 and the mutation of the 3 ' splice site, to effect the K to N ~t;r~n as above. The lO intron i3 spliced out after transcription and there is fllnct;~n;:~l display of anti-NIP scFv fragments on the surface of phage fd as determined by phage E~ISA on NIP-BSA with an absorbance of l. 0 . Further, the phage titre was in the range o~ 5 x lO8 - l x lO9 TU/ml, 15 indicating that these phage fd clones grew well.
Thus, mutations may be made at the 3 ' splice site and ;nt~rn~l guiding sequence of the self splicing intron to allow the e~coding of amino acids compatible with higher expression on self-splicing. Depending on 20 the amino acid which it is desired to alter, it may be necessary to mutate the bases of the Pl hairpin loop as ~ell as the PlO hairpin loop or the Pl hairpin loop only .

EXAMPLE 5: CV~/N~ u~ ON OF THE PLASI!~ID pACYCar~Cre 2 5 EXP~ESSING CRE F~ECf)M~ TNA SF~ UNDE~ THE CONTR.OL OF AN
A~ABINOSE P~201WTE~;!

Wo95/15388 21 77367 r~ (`?~7 ~

In the work described in this example, a plasmid was constructed in which Cre recombinase is expressed under the control of a promoter ;nr~ ;hle by arabinose.
The origin used pl5A makes it suitable for use in 5 combination with plasmids with ColEl origin and with phage or . phagemids with phage origins .
A fragment wa& amplified by PCR from pUC119 ~Vieira, J. and Messing, J. ~1987). ~ethods in Enzymol. 153, 3-11) using the primers lacfor2 and 10 l ~rh~k'~ . This fragment ~ t~n~ A from within the lacI
gene fragment ~inactive) to the polylinker of pUC119 and the primers incorporate a series of restriction sites at both ends of the fragment.
This PCR LLd~ was cut with PvuII and ~'asI and 15 re-cloned into pUC119 digested with the same enzymes to generate pUC119lacipoly.
pARA14 ~Cagnon, C., Valverde, V. and Masson, J.-M. ~1991) . Protein E~gin~, rin~ 4, 843-847) was digested with SacI and NcoI to release a fragment ~nt~;n;n~ the 20 araC gene and the promoter-operator region of araB.
This fragment was ligated into pUC119lacipoly cut with the same enzymes to generate pUC119ara.
The Cre re~ _ ' ;n~ce gene was amplified by PCR
from bacteriophage PlCm cl.100 r~m~ ~Yarmolinsky, M.B., 25 Hansen, E.B., Jafri, S. and Chattoraj, D.K. (1989) . J.
Bacteriol., 171, 4785-4791) using the primers crefor and creback . Af ter digestion with ~saI and ~pnI this Wo 9SI15388 2 1 7 7 3 6 7 p ~ 1 . , 7 7 fragment was ligated into pUC119ara cut with NcoI and ~pnI to gener~te pUC119araCre.
Finally, the PvuII-HindIII fragment of pUC119araCre ront;7;n;nrJ the araC gene and the Cre 5 reco~7~; n;7R~ gene under the control of the promoter-operator region o~ araB was s1lhrl~-n~d into pACYC184 (Chang, A.C.Y. and Cohen, S.N. (1978). ,J. Bacteriol., 134, 1141-1156) cut with BsaBI and HindIII, thereby replacing the tetracycline resistance gene of pACYC184.
10 The plasmid produced (pACYCaraCre) thus contains the an arabi~ose ; n~.7~lr; hl o Cre gene on a plasmid with a pl5A
origin of repIication. This plasmid can co-exist in E.
coli with both the heavy chain donor vector (which has a ColEl origin) and with the acceptor vector (which has 15 a f; 1: ntol7R phage origin) and is useful for the generation of a large phage display library in the lox P f ormat .

EXa~PLE 6: MODEL EXPERI~ENTS FOR THE ~/~Ns~u~llON OF A
DIABODY REPERTOIRE USI~G T~E FD3LOX SYSTE~, USIN~ A
2 0 LOXP SITE WIT~IN A SELF-SPLICING INTRON
In this example, model experiments are described which demonstrate that the loxP site within the self-splicing intron may be used in the construction of a diabody or single chain Fv repertoire by recombination 25 of VH and VII gene repertoires. To this end model experiments are described using a fd phage acceptor vector rrn~;7in;n~ 3 lox sites encoding an anti-7~IP

diabody molecule, where recombination is pe--__r;ned with donor vectors encoding VH or VL domains. Re-^mbination o~ the diabody ca3sette with an expreqsion ve--or is also demonstrated. The methods of this exam~_e are 5 equally applicable to the construction of s--c~' e chain Fv repertoires using 15 amino acid linkers as lescribed in Example 4, and other polypeptides_ The fd phage acceptor vector, fdDWT/4 ~-ntaining 3 lox sites is shown in Figure 13. It C'~71t- -q the VH
10 and VL genes of the anti-~IP clone G6 (Grif_~=is et al, 1994 supra). The sites loxP511 and loxPWT _a.~c the V~
gene and the sites loxPW'r and loxP4 flank t:^e ~L gene.
The loxPWT site is in the self 9plicing int-^~ and the loxP4 site sits between the VI, gene and gene __~ The 15 diabody or single chain Fv polypeptide chai-. elcoded is expressed a5 a fusion with the gene III pro~e- 1. A
site for the factor X protease is; nf~ d ._e=-.~reen the ~L gene and gene III to allow the possibili=-,- of the elution by proteolysis of phage from antige- :uring 20 selection procedures. Alternative versions ~- fdDWT/4 were also made with the site loxP4 replaced -~-~h loxP3 and loxP1 respectively (see Figure 12).
If, for example, a VL gene repertoire -s first cloned irLto fdDWT/4 as ApaLI-AscI fragments, a VH gene 25 repertoire may then be introduced by recom~ =ion with a donor vector rf nt~;n;n~ the VH gene repe_=- -e, flanked by loxP511 and loxPWT sites.

~IENDED Sl',ET

~ 2~77367 ' ~ dDWT/4 was recombined with the donor vector pDN8 containing the VH-D10 gene flanked by loxP511 and loxPWT sites. This was performed by transforming E~.coli TG1 pACYCaraCre (Example 5) with pND8 donor 5 vector cnnt~ln;n~ VH-D10 and then infecting with fdDWT/4 phage c~nt~;n;n~ the genes encoding the variable dcmains, VH-G6 and VL-G6. Recombination was allowed to ~f~nt;n~ at 30~C overnight. Recombined phage from the bacterial supernatant were used to infect TG-1. ~8 a result of recombination between the loxP511 sites of donor and acceptor and between t~e loxPWT
sites oi the donor and acceptor, the recombined fd phage ~-,mt~;nc VH-D10 while keeping the original VL-G6.
Successful r~ '; n~tion was analysed by PCR
15 screening of indiYidual fd phage clone colonies by amplif ication using oligonucleotides that prime speciiically on the se~uences encoding the VL-G6 and VH-D10 CDR3s present in the donor vectors. Thus a PCR
product i9 only obserYed when recombination has 20 occurred. The reCOmh;nAtion efficiency was 75~.
Similar experiments recombining fdDWT/3 or fdDWT/l with pDN8 gave similar efficiencies (Figure 12~.
Alternatively, a VH gene repertoire may be cloned between the NcoI and XhoI sites of fdWT/4 and a VL
25 repertoire, flanked by loxPWT and loxP4 sites.
fdDWT/4 was recombined with the donor vector pRWT/4 ~-~nt~;n;n~ the VI.-D10 gene flanked by loxPWT and lox~4 sites. This was performed by transforming TG1 ~IENGED S~EET

W0 95/15388 . ~ '?

pACYCaraCre (Example 5) with pRWT/4 donor vector containing V~-P10 and then infecting with fdDWT/4 phage c~rltA;ninr~ the genes ~nro~l;nr~ the variable domains, VH-G6 and VL-G6. ~ ' ;n~t;on was allow~d to c~nt;n~ at 5 30C overnight. Recombined phage from the bacterial supernatant were u9ed to infect TG-1. As a result of recombination between the loxP4 5ites of donor and acceptor and between the loxPWT 5ites of the donor and acceptor, the re~ ' ;n~ fd phage ~r~nt;~;nC VL-DlO while 10 keeping the original V~-G6.
Successful recombination was analysed by PCR
screening of individual fd E~hage clone colonies by amplification using oligonucleotides that prime specifically on the sequences encoding the VX-G6 and 15 VL-D10 CDR3s present in the donor vectors. Thus a PCR
product is only observed when r~l hin~tion has occurred. The recombination efficiency was less than 10~6. Similar experiments recombining fdDWTt3 or fdDWT/1 with the donor vectors pRWT/3 or pWT/1 (where 20 the loxP4 site of pRWT/4 is replaced by the loxP3 or loxP1 site respectively) gave ef ficiencies of 0~ and 9 6 96 respectively .
Since in f dDWT/4 the diabody polypeptide is only made as a fusion with gene III, phage displayed 25 bivalent diabody results from the association of free diabody polypeptide, cleaved from gene III protein, with diabody polypeptide gene III fusion. It is 21 77367 -.

desirable to express the bivalent diabody directly as a soluble molecule.
To test the feasibility of subcloning directly into an expression vector by recombination using the 5 loxP sites, the expression vector pEX511/4 was constructed (Figure 12) . This contains the S12 gene, which confers s~reptomycin sensitivity on bacteria, flanked by loxP511 and loxP4 sites . E. coli TG1 pACYCaraCre (Example 5) were transformed with pEX511/4 10 and then infected with fdDST/4 r~nt;~;n;nr the genes ~nr~ n~ the variable domains, ~ G6 and VL-G6.
Recombination is allowed to crnt;n~ at 30C
overnight and the cells were replica plated on 2xYT
agar with or without streptomycin. If recombinatior 15 has occurred the genes encoding the diabody polypeptide will have replaced the streptomycin sensitivity gene n pEX511/4. This will make the bacteria streptomycin resistant .
The recombination was shown to have taken place with an efficiency of 40 to 70~. Similar experiments were performed where fdDWT/3 or fdWT/1 were recombined with pEX511/3 or pEX511/1 (where the loxP4 site of pEX511/3 was replaced with the loxP3 or loxP1 site respectively). No recombination was observed.
Thus it is demonstrated that recombination ca-. be performed between loxP sites configured for the construction of diabody or single chain Fv repertoi-e3.
~lENDED S1~T

W095ll5388 2 1 7 7 3 67 A preferred approach to construction of a repertoire may be f irst to clone the VL genes as ApaLI -AscI fragment6 into fdDWT/4 and then recombine ~ith a V~I repertoire a6 NcoI-XhoI fragments in pDN8. Thi6 5 would generate a diabody repertoire (or 6ingle chain Fv repertoire, i modified 61ightly) 6uitable for phage di6play. Following 6l~1ect;tn of diabodie6, individual or pooled clone6 could be 6ubcloned into pEX511/4 for 60luble expre66ion. In conjunction with the re6ult6 lO from the peptide display chain Fv expression from clone6 cr~t~;n;n~ self-6plicing intron6 with loxP 6ites in examples l, 2 and 4, it is rrnr~ o~ that this methodology is suitable for making large diabody and single chain Fv repertoires of the order of lol to lol7 15 or more independent clones.

a ~ t ' ,"

~ WO 95115388 2 1 7 7 3 6 7 Z
U ~ Z
~ Z Z
U ~ C, ~ ~ C Z U
E~ U ~ ~ ~ Z
V ~ ~
U ~ Z U
V V V U E~ _ ~ V E~
U ~: U ~ ~ ~ ~ V E~ z ~ z ~
E~ U U U U ~ ~ ~ U Z ~: U ~ ,¢
E~ U ~ E U ~ ~ U
U U U ~ ~U ~ ~ ~ r7 U 2 ~ V ~ '¢
3 ~ ~ ~ U ~ ~u u~ Uu U U E-l EU-' E--~ ~ U ~ U ~ V U
U V EU~ V ~ U ~ C~ ~ U V U r-~ U
~U ~ V ~ ~ ~ V ~ ~ V U ~U
,~ ~, ~ ~ ~ ~ ~ ~ 3 ,~ ~ v 3 ~ u 3 1 u ~ u ~ u v ~ ~ u ~ ~ ~ u u~ U ul E~ rn E~ rn ~ V ~) V rn m V m U rn 2z C
'~ CA~
_~ r I ~ ~ C~ CA, ~, r ~, ~ rr~ (~ ~
O I I O f~ O O O O
E l r E- E- E- E- O ._1 0 0 0 0 WO 95115388 . ~ 7 ~

E~ V rl E~ E~
E~ V
V ~ ~ V
~ V ~ V V E~ V g ~
~8 ~ 8~ ~ ~ gv ~ g g v v ~ v V ~ 8~ g ~ "
v ~ ~ 8 V E~ v E~ V ~ E~ V ~C V
V ~ ~ V ~ ~ ~
g ~ V
V V V ~ ~ ~ 8 v u vg ~ v v ~, V ~ ~ 8 ~ v v v 8.,~
E~ ~ V ~ ~ V 1' v : g g 8 ~ V v E~
V ~ ~ V V ~
g 8 g v ~ ~ v v E~ 8 v ~ ~v w E8~ v E8-. v ~Z; v g ~ E~ v v vg v v H V~ V ~ H ~ E~ V ~ V
W ~ 8 ~ 8 ~v v ~ V ~ ~ ~ V
o ~8 ~ 8 v v ~ ~ v ~ E~ v v V EV V EV '' E~ ~V EV., tV~ VV V V V
V V E~
V u g~ ~ 7 ~
W ~ m ~ V U~ U~ V ~ g v8 p p ~ ~ ~ z v g C~ ' O h ~
c la C ~a h ; ~ r~ X
rxX X X ~ ~ ~ U h O ~O O r i ~ V V ~ p~
# ~: # # :t~ # ~ ~ ~1 0 4 SUBSTITUIE SHEET (RULE 26) ~ WO95/15388 2 1 77 3 67 r ~ 5~

V
V V
'J D rl V
V ~
a~ ~ Et V = V ~
V V
~ E~
, U E~
E~ ~
V V
S V E~
V V
Wo ~ V
U
V
E~ V
g V
V
~ U
S . E ~ ~1 ` V
V ' C~ V , V
O ~ ~ V
~ ' g g ~ .. -¢ .. .. .. .. .. , 1 U ' h E-~ " V h V h h h h I .. h ;
- ~ e ~ u 8 v 8 E 8 8 h 8 -S ~,~ U V ~: ~ V Ul 5, u~ ~ I o _ E
~,~ u g e ~ o~ v ~
~" rt t U t '~ ' t I ~` rt t ~ rt r~ 1 V V ~ N V~ ~Vt E~ ~Vt I Vt I l ovt R C ~0 ` E-' ~i ` ~ ` - ` ` ~ `
SUBSTITUTE SHEET (RULE 26~

WO 95/15388 ~ r '''~ ~

E~
r d o . I
., 1.. 1.. ;
~ E ~ E
r ~ r ,1 ~ _ ~ rl ~ ' ,~ I I H I E
o m Z
SUBSTITUTE SHEET (RVLE 26)

Claims (34)

73

1. A DNA construct comprising a first exon sequence of nucleotides encoding a first peptide or polypeptide, a second exon sequence of nucleotides encoding a second peptide or polypeptide and a third sequence of nucleotides between the first and second sequences encoding a heterologous intron between RNA splice sites and a site-specific recombination sequence within the intron, the exons together encoding a product peptide or polypeptide.

2. A DNA construct according to claim 1 wherein the product peptide or polypeptide comprises a member of a specific binding pair (sbp).

3. A DNA construct according to claim 2 wherein the sbp member comprises a binding domain able to bind complementary sbp member.

4. A DNA construct according to claim 3 wherein the binding domain is an immunoglobulin antigen-binding site .

5. A DNA construct according to claim 4 wherein the product peptide or polypeptide is a scFv antibody fragment which comprises a VH domain linked to a VL
domain via a peptide linker which allows the VH and VL
domains to associate to form the antigen-binding site.

6. A DNA construct according to claim 5 wherein transcription of said construct leads to mRNA having nucleotides corresponding to nucleotides of said RNA
splice sites which encode and are translatable into amino acids of the peptide linker of said scFv antibody fragment.

7. A DNA construct according to claim 1 wherein the product peptide or polypeptide comprises a T cell receptor V.alpha. domain and a T cell receptor V.beta. domain, or a T cell receptor/antibody fusion, or a T cell receptor/antibody fragment fusion, or a receptor binding peptide, or an enzyme, or a multiple domain protein, or an amino acid sequence variant or derivative of any of these.

8. A DNA construct according to claim 1 wherein the first and second peptides or polypeptides are not linked together in any naturally occurring polypeptide.

9. A DNA construct according to claim 8 wherein one of said first and second peptides or polypeptides comprises an antibody fragment.

10. A DNA construct according to claim 9 wherein the antibody fragment is selected from the group consisting of VH, VL, CH, CL, VH-CH and VL-CL.

11. A DNA construct according to claim 8 wherein the first and/or second peptides or polypeptides comprise an amino acid sequence encoded by a synthetic nucleotide sequence.

12. A DNA construct according to any one of the preceding claims wherein the intron is a self-splicing group I or group II intron.

13. A DNA construct according to claim 12 wherein the self-splicing intron is obtainable from Tetrahymena thermophila nuclear pre-rRNA.

14. A DNA construct according to any one of the preceding claims wherein the site-specific recombination sequence is the loxP sequence obtainable from coliphage P1, or a mutant or derivative thereof.

15. A DNA construct according to any one of the preceding claims wherein the product peptide or polypeptide comprises a surface componant of an organism .

16. A DNA construct according to claim 15 wherein the organism is a bacteriophage.

17. A DNA construct according to claim 16 wherein the bacteriophage is fd or M13.

18. A DNA construct according to claim 17 wherein the surface component is the gene III product.

19. A DNA construct according to any one of the preceding claims wherein the product peptide or polypeptide does not comprise a polypeptide comprising a domain (a) which comprises a binding region of an immunoglobulin heavy chain variable region and a domain (b) which comprises a binding region of an immunoglobulin light chain variable region, the domains (a) and (b) of the polypeptide being linked and capable of inter-molecular association in a multimer to form an antigen binding site but incapable of intra-molecular association to form an antigen binding site.

20. A DNA construct according to any one of the preceding claims which is a vector further comprising nucleic acid for expression of the product peptide or polypeptide.

21. A DNA construct according to claim 20 further comprising nucleic acid for secretion of the product peptide or polypeptide.

22. A DNA construct according to claim 20 or claim 21 wherein said vector is a plasmid, a phage or a phagemid vector.

23. A host cell comprising a DNA construct according to any one of the preceding claims.

24. A plurality of DNA constructs according to any one of claims 1 to 22 collectively encoding a repertoire of product peptide or polypeptides wherein each product peptide or polypeptide in the repertoire has a different amino acid sequence.

25. A population of host cells comprising a plurality of DNA constructs according to claim 23.

26. A method of producing a product peptide or polypeptide which comprises a combination of a first peptide or polypeptide component and a second peptide or polypeptide component, the method comprising:
providing a DNA construct according to any one of claims 20 to 22;
transcribing DNA of the construct into RNA;
causing or allowing splicing of nucleotides of the third sequence to produce an RNA molecule encoding the product peptide or polypeptide;
translating the RNA molecule into the product peptide or polypeptide.

27. A method according to claim 26 wherein transcription, splicing and translation take place in vitro.

28. A method according to claim 27 wherein transcription, splicing and translation take place in vivo .

29. A method according to claim 28 wherein transcription, splicing and translation take place in E. coli cells.

30. A method according to any one of claims 26 to 29 wherein a plurality of DNA constructs is provided for transcription, splicing and translation.

31. A method according to any one of claims 26 to 30 wherein, following said translation, product peptide or polypeptide of interest is selected or isolated from other peptides or polypeptides present.

32 . An isolated nucleic acid construct consisting essentially of a sequence of nucleotides encoding a self-splicing intron with a site-specific recombination sequence within the intron.

33. A nucleic acid construct according to claim 32 wherein the site-specific recombination sequence is the loxP sequence obtainable from coliphage 1, or a mutant or derivative thereof.

34. A nucleic acid construct according to claim 32 or claim 33 wherein the self-splicing intron is obtainable from Tetrahymena thermophila nuclear pre-rRNA.