WO1998001560A1 - Modified protein g and fragments thereof - Google Patents

Abstract

A polypeptide sequence which is a modified form of protein G or fragment thereof which selectively binds to one of the Fab or Fc fragments of an antibody relative to the other as compared to the corresponding selectivity of binding of unmodified protein G.

Description

Modified Protein C and Fragments Thereof

The present invention telates to piotein G and moic particularK to a modified tor of protein G and fragments theieol

Protein G is a member of a iaige and diveise gioup ol cell suilacc-associated i munoglobulin-binding proteins found in Gram-positive bacteria (Boyle. 1990) These proteins have attracted particular interest due to their possible role in enhancing microbial virulence, their commercial applications in the labelling and purification of antibodies and also as a means of studying protein/protein interactions Protein G originates from Goward group G Sli eptococci and binds to the Fc (fragment crystalhsable) and Fab (fiagment antigen binding) portions ot IgG with high affinity comparable with the magnitude ol antigen/antibody mtei actions (A erstrom et al . 1987. Sjobnng et al , 1991. Tahnstoc et al . 1990) The IgG-binding domains have been identified as three highly homologous sequences approximately 60 residues in length, which retain high affinity ioi IgG when expressed individually (Fahnstock et al., 1990, Guss et al , 1986, Sjobnng et al . 1991 ) In this lespect protein G is similar to protein A, from

aw ens. which also contains small IgG-binding domains (Uhlen et al , 1984)

The structure of protein G and its interactions with IgG have been the subject of extensive studies bv X-ray civstallography and NMR All IgG-binding domains have the same arrangement ot secondaty structure, consisting ol a central α-hehλ packed against a four stranded antiparallel-parallel-antiparallel β-sheet (Lian et al , 1991. 1993, Gronenborn et al . 1991 , Achan et al 1992, Gallaghei et al.. 1994) The crystal structures of the complexes ot a single domain ot piotein G with Fab and Fc have been determined independently (Deinck & Wigley. 1992, 1 94 Sauereπksson et al , 1995) In each case the mode of association of the piotein G domain with the lmmunoglobuhn is different piotein ϋ binds to Fab bv an antipaiallel alignment between the second β-strand in piotein G and the final -strand in the CH I domain In contrast to this 'edge-on type ol tei action, piotein G binds to Fc via the central α-hehx in the protein which binds in the cleft between the C112 and CHI domains (Sauereiksson et al . 1995) Recognition of Fab or Fc bv piotein G therefore utilises two different sets of amino acid residues on the surface of the protein. Solution studies by heteronuclear NMR have confirmed that these interactions persist in solution (Gore and Gronenborn. 1 93: Lian et al.. 1994; ato et al.. 1 95).

It is a classical result that antibodies can be cleaved by proteolysis into Fab and Fc portions, with the Fab portion of the antibody mediating antigen recognition and the Fc portion responsible for complement activation. The Fab fragments are then purified away from the Fc fragments and can be used in a variety of immunotherapeutic applications where the presence of the Fc portion is undesirable. The separation of Fab and Fc fragments is not necessarily a straightforward matter; the sequence hypervariability of antibodies complicates this process, and although traditional ion-exchange methods can be employed under these circumstances, they can be time-consuming and troublesome to implement.

It is therefore an object of the present invention to obviate or mitigate the above mentioned disadvantage.

According to a first aspect of the present invention there is provided a polypeptide sequence which is a modified form of protein G or fragment thereof which selectively binds to one of the Fab or Fc fragments of an antibody relative to the other as compared to the corresponding selectivity of binding ol unmodified protein G.

The invention has been based on the realisation that it is possible to produce modified forms of protein G and fragments thereof which are capable of selectively binding to either the Fab or Fc fragments of an antibody. Thus one embodiment of the first aspect of the invention is a modified form of protein G, or fragment thereof, which is capable of selectively binding to the Fab fragment of an antibody but not the Fc fragment. A further embodiment of the first aspect of the invention is a modified form of protein G. or fragment thereof, which is capable of selectively binding to the Fc fragment of an antibody but not the Fab fragment.

Depending on its selectivity ( i.e. Fab or Fc binding) a modified protein G, or fragment thereof, in accordance with the first aspect of the invention, is useful for effecting separation of the Fab or Fc fragment of an antibody from a mixture thereof. The selectivity of binding displayed by the polypeptide sequences of the invention (as compared to unmodified protein G) may be expressed m terms o :

(a) the ratio oϊ binding affinity (1C ₀') of the polypeptide sequence of the invention for Fab fragments to the binding affinity (1C\₀") of unmodified protein G for such fragments, and

(b) the ratio of binding affinity (IC₅₍₎ ^m) of the polypeptide sequence of the invention for Fc fragments to the binding affinity (IC ₎ ^IV) of unmodified protein G for such fragments, and

Depending on whether the sequence is selective for Fab or Fc. the value of one of (IC₅₀' IC₅₀") or (IC_3ϋ"7lC₅₍₎ ) is preferably less than 2 (more preferably less than 1.5) and the other is greater than 3 (preferably greater than 5 and most preferably greater than 7).

Thus for a polypeptide sequence in accordance with the invention which is more selective for Fab than Fc (as compared to unmodified protein G) it is preferred that (IC₅₀'/IC_S0") is less than 2 and that (IC _l"7lC₅ '^v) is greater than 3.

In accordance with the second aspect of the invention there is provided a method of effecting separation of a mixture of Fab and Fc fragments of an antibody from a mixture thereof, the method comprising treating said mixture with a polypeptide sequence in accordance with the first aspect of the invention which selectively binds to a predetermined one of the Fab and Fc fragments to form a complex therewith, and separating the unbound fragment from said complex.

The method in accordance with the second aspect of the invention is preferably an affinity separation in which the protein G. or fragment thereof, is immobilised on a support which is then treated with the mixture of Fab and Fc fragments to be separated. As a result, an immobilised complex is formed between the modified protein G. or fragment thereof, and the antibody fragment for which it is selective. The other fragment is not retained by the support and may be collected separately The bound fragment may then be i cleased horn the complex ( using standard techniques) for collection

The invention will be further descnbed by way of example only with reference to the accompanying drawings, in which

Fig. 1 shows the amino acid sequences of the thice igG binding domains of protein G(see also Seq ID Nos 1 -3) and associated hnkcis (see also Seq ID Nos 4,5),

Fig. 2 illustrates the overall procedure for producing the N40A mutation of Domain II as a protein G fragment in accordance with the fust aspect of the invention,

Fig 3 shows the coding sequence and complementai v sequences of pnmers JPD1 1 and JPD12 (see also Seq ID Nos 6 and 7 respectively).

Fig 4 illustrates the nucleotide sequence between the Bam HI and Hind III sites of mpl 8'JPDl shown in Fig 2 (see also Seq ID No 8).

Fig. 5 illustrates the binding of piotein G (WT) and the N40A mutation to Fab fragments from a murine IgGl monoclonal antibody;

Fig. 6 illustrates the binding of protein G (WT) and the N40A mutation to Fc fragments from a murine IgG2b monoclonal antibody.

Figs 7 to 1 1 illustrate results obtained using the N40A mutation of Domain II,

Figs 1 1 and 12 illustrate lesults obtained using the T23C mutation of Domain II; and

Fig. 13 illustrates further lesults obtained using the N40A mutation of Domain II.

Reference is made firstly to Fig. 1 which illustrates the three highly homologous IgG-binding domains in protein G derived Irom Streptococcus strain G148. The Domains are identified as Domains I-III (and as Seq ID Nos 1 -3 respectively) and are associated with Linkers 1 and II (Seq Id Nos 5 and 5 respectively) as shown The sequences aie as shown by Goward et al ( 1990) and the illustrated numbering scheme is that used in the present application The amino acids of the domains and linkers are given according to the standard single letter code The linear arrangement of these sequences in the Protein G gene is as follows Domain I-Linker 1-Domaιn l i-Linkei II-Domam III As shown (and as mentioned above) the three sequences are highly homologous and differences in the sequences are highlighted in bold. Each of the domains I. II and III is. in isolation from the other domains, capable of binding both the Fab and Fc fragments of an antibody.

In a preferred embodiment of the invention, the selectively binding polypeptide is a protein G fragment which comprises the sequence of Domain I. II or III (or a modification thereof, part thereof, or modified part thereof) which has a mutation such that the sequence is capable of selectively binding to one of the Fab or Fc fragments of an antibody.

For a polypeptide which is capable of selectively binding the Fab fragment. the preferred mutation is at position 40 (according to the nomenclature of Fig. 1 ) of domain I, II or III, the most preferred mutation being the N40A mutation (i.e. replacement of asparagine at position 40 by alanine).

Alternative mutations for producing a sequence which selectively binds the Fab fragment are E32A, K.33A. Q37A and E47A. These mutations may be made individually but are more effective if used in combination. Mutation to small apolar amino acids other than alanine may also be used.

For a sequence which selectively binds to the Fc fragment the preferred mutations are T16A. T21A, T22A. Y38F and N42A. These mutations may be made individually but are more effective if used in combination. Mutation to other small apolar amino acid residues may also be used

The selective polypeptide binding sequences in accordance with the invention may be produced by expressing, in a host, a construct comprising a vector that contains a coding sequence for the desired polypeptide. A particularly convenient method of producing the construct ( which for convenience is referred to herein as the "mutation construct") starts with a vector incorporating the wild type gene coding sequence (the "wild type construct") which is replicated in the presence of deoxyuridine so as to produce a construct which misincorporates tleoxyuridinef or convenience referred to as the ^"'deoxyuridine construct^"). The ^"deoxyuridine construct" is then subjected to mismatch site directed mutagenesis. copying and selection to produce the ^"mutation consti uct" in which (i) the legion coding for the selective polypeptide is a mutation ot the wild Ivpe gene coding sequence and (n) there is no deoxyuridine misincoi poiation Moic paiticulai l) , the site duected mismatch mutagenesis may be etfected using an oligonucleotide which is substantially complementary to a portion of the wild type gene coding sequence but which incorporates at least one base change (the "mismatch") lelative to the wild type gene coding sequence The oligonucleotide is hybridised to the wild type gene coding sequence of the "deoxyuiidine construct^" and then used as a pπmei fiom which replication occurs so as to pioduce the mutation construct^" which incorporates the aforementioned oligonucleotide (so that a mutated coding sequence is present in the "mutation construct") but which does not have deoxyui idine misincoiporation The "mutation construct" may then be separated horn "deoxyuiidine construct" by translection of the constructs into an E toll strain that is capable of repairing DNA with deoxyuridine misincoiporation Giowth of the transfected bacterium results in the selective survival of the desired constiuct As indicated above, the "'mutation construct" may then be expressed in a host to pioduce the desned pioduct

The invention is illustrated

the following non-limiting Lxamplcs

Example 1

Preparation of the N40A mutation of Domain 11

1. Starting Materials and General Procedures

1 1 Vectors

M13 mp 18 and pUC 18 (Yanish-Peiron et al 1989)

1.2 DNA Modifying Enzymes

Restriction enzymes. l aq poiymeiase and T4 ligase weie obtained fi om New England Biolabs (Boston USA) oι horn Boehπngei Mannheim (Mannheim. Germany) and used in accordance w ith the manutactuier insti uctions 1.3 E. coli strains

JM101 F ^'traDid lacl(/D(lacZ)M15proΛ +β+ sιtpE tin ϋ(lac-proAB) (Vanish- Perron et all. 1985)

RZ1032 HFrKLJό Po/45 Zhd-27 Tnlϋ LysΛ lhi-1 relA l spoTl SupE4 dul- 1 ung-1 (Kiinkel et al. 19 7)

\ A Isolation of M13 double stranded DNA

This was carried out using the method described by Sambrook et al. ( 1989) for small scale preparations of plasmid DNA from E coli. with the following modifications:

1. 50 mM sucrose was used instead of 50 mM glucose in the lysis medium.

2. 0.1% lysozyme was included in the lysis medium.

3. A phenol extraction was included.

1.5 Isolation of M13 single stranded DNA

This was carried out using the method described by Sambrook et al. ( 1989) for small scale preparations of Ml 3 single stranded DNA

1.6 Dideoxy DNA sequencing

This was carried out using the Sequence kit (Version 2.0: Cambridge Bioscience) kit, according to the manufacturer^'s instructions.

1 .7 Culture Media

2YT medium was prepared as described by Sambrook et al. ( 1 89). 1 .8 Buffers

Phage buffer was prepaicd ti oni a stei ilc solution ol 3g/lιtre KI Ii O_j. 7 /htre Na₂HP0₄ and 5g/htre NaCI MgSO , was then added to I mM. CaCU to 0.1 mM and gelatin to 0.001 % (w/v) final concentrations Other buffers were prepared as described in the references given

2. Procedure

2.1 Cloning of streptococcal DNA coding for a single protein G IgG-binding domain into mpl8.

The DNA coding lor the second domain from piotein G (domain II ) was replicated from pSPG29 (Gowaid et al 1990) by PCR using the oligonucleotide primers JPD-1 1 (Seq ID No. 6) and JPD- 12 (Seq ID No 7). and transferred into mpl8:IM3:ClaI (Murray et al . 1988) with deoxyuridine misincoiporation by sticky foot mutagenesis (Clackson & Wintei . 1989. see Lian et ai . 1993 )

The domain II coding sequence was designed to express a 64 residue protein which includes some of the "linker" sequence on either side of the IgG-binding domain (see Fig. 1 ).

The overall procedure is illustrated in fig 2 and utilised primers JPD- 1 1 and JPD-12 which are shown in Figure 3. Each primer incorporated a sequence complementary to the DNA sequences at the 5^" and 3^' ends of the coding sequence for domain II in order that they could act as primers for amplification of the domain II coding sequence by PCR The pnmeis also incorporated a sticky foot^' sequence that was designed to be complementary to the lelevant portions of the mpl 8:IM3 Gal vector and permit annealing to this vector in the second part of the 'sticky foot¹ mutagenesis procedure (see below) T he primer JPD-1 1 also included a single mismatched thymine residue in the sticky foot^' sequence that was designed to modify the Gal site in mpl 8: IM3 Clal to permit digestion in methylated DNA by removing the consensus sequence foi the dam methylase. The primer JPD- 1 incorporated a 'hinge^' region in the centre w hich intioduced a single tianslation stop codon and also inserted an EcoRI site tot use in fuithei cloning prυceduies

The 'sticky foot' mutagenesis expeinnent substituted the C A I ^{■ • •} gene in mpl IM3 Clal with the gene foi domain II oi piotein G and intioduced a Clal site and an EcoRI site in a single step, to generate mp l S 1PD 1 (Figuie 2 and Seq ID No 8)

In detail, the procedures weie as follows

2 1 1 5' phosphorylation of JPD-12

50 pmol of JPD-12 was dried undei vacuum in a 0 5ml disposable plastic tube and subsequently redissolved bv addition of a solution containing 1 μl ot a solution of 1 M TπsHCl (pH 8 3) plus 0 1 M MgCU (kinase buffei ), 1 μl of a solution of 1 mM ATP,

1 μl of a solution of 10 mM DTT and 7 μl of watei All solutions were pre-steπlised either by autoclaving (kinase buffei and watei ) oi by filtration (ATP and DTT) Phosphorylation of JPD-12 was initiated bv addition ot 4 units of T4 polvnucleotide kinase and the reaction left to pioceed foi 30 minutes at 37 "C beloie tianstei to an oven at 65 °C for 15 minutes to stop the leaction

2 1 2 PCR amplification of tit e coding l on foi piotein G domain II The following solutions were prepared

2X PCR Buffer 20 mM rusHCl (pH 8 3) 100 mM KG 3 mM MgC and 0 2 mg/ml gelatine The solution was filtered and autoclaved

2 mM dNTPs A solution of 2 mM deow ATP. 2 mM deow GTP 2 mM deoxy CTP and 2mM deoxy TTP The solution was sterilised bv filtiation

The PCR incubation was prepared as follows

2X PCR Bufrei 40 μl

2 mM dNTPs 8 μl phosphorylated JPD- 12 (from

e ) 8 μl (40 pmol)

JPD- 1 1 8 μl (40 pmol )

Taq Polymerase 1 μl (5 units) pSPG plasmid 16 μl ( 1 6 ng)

The incubation was subsequenth divided into 20 μl ahquots between iour sterile plastic 0 5 ml disposable tubes and the liquid in each tube was covered with 20 μl of Nujol Mineral Oil PCR was earned out with 25 cvcles of the following temperature programme 94 °C, 1 minute, 37 "C 2 minutes. 74 ' C 3 minutes I he pioducts of all four tubes were then pooled and subiected to electiophoiesis on a 1 5% (w/v) agaiose gei, with 40 mM Tπs-acetate, 1 mM EDTA as running buffei A single product was clearly visible within the anticipated molecular weight lange this band was excised and the double stranded DNA pioduct was electroeluted into 200 μl of 3 M sodium acetate The salt was then removed from the DNA by centπfugation through a 1 ml Sephadex G-25 size exclusion column The DNA was then piecipitated in 0 3M NaCI and 70% ethanol at 0 "C and sedimented by centi if ugation at 1 3.000 g_ιU foi 10 minutes The supernatant was lemoved and the DNA dned undei vacuum before being redissolved in 15 μl of stenle watei

2 1 3 Preparation of mpl8:IM3:ClaI single stranded DNΛ with deoxyuridine mis in corporation

This procedure was applied to produce mpl 8 single slianded DNA with deoxvundine misincoiporation for mpl 8 1M3 Clal (construction described by Murray et al . 1988) E coli JM101 was transfoimed by a standard CaCl2 method (Maniatis el al 1982) with mpl 8 IM3 Clal double stianded DNA and plated out foi single plaques Following overnight growth at 37°C a single plaque was picked into 1 5ml of sterile 2YT medium, also containing an inoculum of IM 1 01 and grown overnight at 37°C Bacterial cells were sedimented ccnlπfugation in a minituge foi 1 0 minutes Bacteπophage particles horn the supernatant weie l ecoveied

addition of 0 625M NaCI and 2 5% PEG 8000 ( final oncentiations) and sedimentation of the precipitated bacteriophage by centrifugauon in a minifuge for 10 minutes. Bacteriophage particles were redissolved in phage buffer and used to inoculate a culture of E coli RZ1032 in 2YT medium, containing 12 5 μg/ml tetracycline and 0 25 μg/ml deoxyuridine The culture was grown overnight at 30°C with vigorous shaking in a baffled flask. Bacterial cells were sedimented by centrifugauon at 1 .200 χ_av tor 10 minutes and the supernatant recovered. Bacteriophage particles were isolated from the supernatant by precipitation as described above, and single stranded DNA isolated from the Ml 3 phage by a scaled-up version of the method described by Sambrook et al ( 1989) This procedure was used to yield mpl 8 IM3 Clal with deow undine misincoiporation. which is depicted in Figure 2 as a dashed line

2 1 4 Preparation of mp 18: JPD1 (double stranded vector (using "sticky foot" mutagenesis procedure)

0 5 pmol of mpl 8:IM3:GaI, prepared with deoxyuridine misincoiporation. was dried under vacuum in a 0.5 mi disposable sterile plastic tube 6 μl of the double stranded DNA PCR product, prepared as described in the section 2.1.2, was then added and annealed to the deoxyuridine template by use of the following temperature ramp programme: 92 °C. 1 minute. 67 "C. 2 minutes. 37 "C. 2 minutes, with a 1 minute linear temperature gradient between each temperature

The following solution was prepared separately

0.2 M TπsHCl (pH 7 5) plus 0 1 M MgCb 4 0 μl lO mM ATP 2.0 μl l OO mM DTT 4 0 μl

DNA hgase 4 0 μl (400 units)

Klenow fragment DNA polymerase 1 5 μl ( 1.5 units)

Steπle water 24 5 μl (The TπsHCl/MgCh solution was sterilised by autoclaving and the ATP and DTT solutions were sterilised by filtration ) 10 μl of the solution was then added to the deoxyuπdine-enπched mpl 8 template and domain II PCR product, and incubated for 15 hours at 15 °C 80 μl of stenle watei was added to the incubation and 5 μl was used to transform E coli JM 101 by the CaCh method (Ma atis el al 1982) Positive clones (referred in fig 2 as mpl 8 JPD 1 (double stranded vectoi )) weie identified by dideoxy DNA sequencing^* the sequence between the Bam HI and Hind III sites of mpl 8 JPD1 is shown in Figure 4 which shows between the locations identified as X and Y the sequence coding foi domain II (modified with a N-terminal methiomne residue The codon (ACC) foi aspaiagine at position 40 ot domain II is highlighted within the coding sequence Fig 4 illustiates the JPD 1 1 and JPD 12 denved sequences as well as the BamH l , Clal , EcoRI and Hindlll lestπction sites

2 1 5 Preparation of mpl8: JPD1 (single stranded vector with deoxyuridine misincorporation).

The procedure ot Section 2 1 3 was repeated but using mp l 8 JPD 1 (double stranded vectoi ) instead ot mpl 8 1 M Clal

2.2 Site-directed mutagenesis of streptococcal DNA

Oligonucleotide-directed mismatch mutagenesis was employed using the deoxyuridine selection protocol described by Kunkel et al ( 1987) The mutagenic oligonucleotide used was 5' CACCGTTGTCGGCAGCG I AT I GTTTGAA (Seq ID No 9) (with the mismatched nucleotides shown in bold) The legion of mpl S JDl which the mutagenic oligonucleotide bound to is show n between the locations identified as a and b in Pig 4 T he piocedures used were as follows 2 2 1 5' phosphorylation of the mutagenic oligonucleotide

SOpmol of the mutagenic oligonucleotide was dned undei vacuum in a 0 5ml disposable plastic tube and subsequentK ledissolved b\ addition of 1 μl of a solution of 1 M TπsHCl (pH 8 3) plus 0 1 M MgCU (kinase buffei ) 1 μl ot a solution of 10 mM ATP, 1 μl of a solution of 10 mM DTT and 7 μl of watei All solutions were pre-steπhsed either by autoclavmg (kinase buffei and watei ) oi by filtiation ( ATP and DTT) Phosphorylation was initiated by addition ol 4 units of T4 polynucleotide kinase and the reaction left to pioceed for 30 minutes at 37 "C, before tiansfer to an oven at 65 °C for 15 minutes to stop the leaction

2 2 2 Preparation ofmpl8: JPD1:N40A (Double-Stranded DNA Ml 3 Vector)

0 5 pmol of mpl 8 JPD l fiom section 2 1 5 above piepaied with deow uiidine misincoiporation as descnbed in section 2 1 5. was dued undei vacuum in a 0 5 ml disposable sterile plastic tube To this tube were added

0.2 M TrisHG (pH 7 5). 0 1 M MgCh. 1 μl

0 5 M NaCl and lO M DTT kinased mutagenic oligonucleotide 2 μl ( 10 pmol)

(from previous section)

Sterile water 7 μl

The solution was incubated at 65 °C foi 30 minutes, followed by 30 minutes incubation at 20 °C

A separate solution was piepaied containing the following

0 2 M TπsHCl (pH 7 5) 0 1 M MgCh 1 μl

0 1 M DTT

10 mM ATP 1 μl

2 mM dNTPs (as descnbed in section 2 2 above) 4 μl

Klenow fragment DNA polymerase I 2 units

T4 DNA ligase 3 units

Sterile water to give 10 μl total volume

The TrisHCl/MgCl2 solution was steiilised by autoclaving and DTT added subsequently. The ATP and dNTP solutions were steiilised by filtration. 10 μl of the polymerase/ligase solution was added to the annealed template and primer (to give 20 μl total volume) and incubated at 1 5 °C toi 15 houis 80 μl ol sienle water was added to the incubation and 5 μl was used to tianstorm E coli .IM1 1 b\ the CaCb method (Maniatis et al 1982) l iansfoimants containing mpl 8 JP l with the appropriate N40A mutation were identified by dideoxy DNA sequencing of the sequence between the BamHI and Hind III sites of mpl 8 JPD l . (The numbering of the amino acids in the sequence for protein G domain II is the same as that described by Lian et al (1993)) This procedure yielded Ml 3 double stranded DNA containing mpl 8 JPDl with the N40A mutation (mpl 8 IPD 1 N40A (double stranded DNA M 13 vector))

2 2 3 Preparation of expression vector pLIC18:JPDl

Ml 3 doubled stranded DNA was isolated from JM 101 transformed with mp l 8 JPD l as described in section 1 4 above The rephcative (double stranded) lorm of mpl δ JPDl from JM101 grown to stationary phase in 3 ml of 2YT medium was dissolved in 20 μl of stenle watei and cleaved with BamHI and Hind III according to the manufacturer's instructions The BamHI/Hindlll fragment was ligated into pUC 18 as described by Maniatis et al 1982. using pUC1 8 pieviously digested with BamHI/Hindlll and phosphatased to pievent relegation Ligation was allowed to proceed for 2 hours at 37 °C. 80 μl of sterile water was added and 5 μl was used to transform E coli JM 101 by the CaCh method (Maniatis et al 1982) Transformants were selected b\ plating onto nutnent agar containing 50 μg/ml ampicilhn The expression of domain II fiom IM 101 in 2YT medium containing 50 μg/ml ampicilhn was constitutive, as is the case foi the equivalent construct containing CATllI (Murray et al , 1988) Between 25 and 50mg of domain II weie pioduced fiom 1 litre of culture at 37°C

2.2.4 Preparation of pUC18:JPDl :N40A (double stranded expression vector) mpl 8:JPDl :N40A double stranded M13 DNA vector was isolated from JM101 transformed with the same vectoi (section 2.2.2.) as described in section 1 4 A BamHI Hindlll fragment containing the protein G gene with the N40A mutation was excised by digestion with BamHI and Hindlll as described by Maniatis et al (1982) PUC18:JPD1 (section 2.2.3) was also digested with BamHI and Hindlll to remove the wind type protein G gene and phosphatased to prevent lehgation. as described in section 2.2.3. The protein G gene containing the N40A mutation was then heated into the BamHI/Hindlll cleaved and phosphatased pUC18 JPDl expressed vectoi. to generate a pUCl 8:JPD1.N40A double stranded expression vector

2.2.5 Preparation of E.coii JM101 transformant containing pUC18:JPDl :N40A (not illustrated in Fig. 2)

PUC18:JPD1 :N40A was prepared as described in section 2.2 4 and transformed into E.coii using the standard CaCl₂ method (Maniatis et al , 1982)

2.3 Expression and purification of mutant (N40A) protein G domain II

All manipulations were carried out at room temperature, except where stated otherwise. A colony originating from a single E coli JM 101 transformant containing pUC18:JPDl with the N40A mutation described in section 2 2 was picked into a starter culture of 10 ml of 2YT medium plus 50 μg/ml ampicilhn and grown by constant agitation at 37 °C until the absorbance ot the cell suspension at 600 nm reached 4 0. The startei culture was then used to inoculate two 2 litre baffled flasks each containing 500 ml of sterile 2YT and ampicilhn at 50 μg/ml The bacterial cells were grown into stationary phase bv constant agitation at 37 °C . expression of protein G domain II and its mutants were constitutive in this vectoi and no induction was lequired Cells were hai vested b\ centntugation at 8.900 χ_av loi 10 minutes, lesuspended in 20 mM T sHCl (pl l 7 5). 1 mM EDTA. and disi uptcd by sonication, cell debris being removed by centiiiugation at 1 3.200 i>₍,v lot 45 minutes The soluble extract was heated at 70°C foi 10 minutes, and insoluble material then removed by centrifugation at 13,200 g₍ιv lot 10 minutes The supernatant was adμisted to pH 3 0 with glacial acetic acid and stuied foi 30 minutes Insoluble matenal was removed by centrifugation as before, and the pH of the supernatant adjusted to 7 0 with I πs base The preparation was dialysed overnight at 4"C against 2 litres of 20 mM TnsHCl (pH 7.5) Ammonium sulphate was added to 80% saturated and the precipitated protein recovered by centrifugation The piecipitate was washed once with 100% saturated ammonium sulphate, redissolved in 5- 10 ml of 10 mM Tns Hcl (pH 7 5) and dialysed overnight at 4°C against 2 lilies of the same bullei The pieparation was applied to a 40ml column of DEAE Sephacyl (Sigma), equilibrated in 20 mM I s HG (pH 8.3) and washed with several column volumes of the same buffei The N40A mutant was then eluted by application of equilibration buffei (20 mM Tns HCL pH 8 3 plus 170 mM NaCI) and all eluted fi actions containing more than 0 3 mg/ml protein were pooled and lyophilised overnight I^'he sample was ledissolved in distilled water and dialysed overnight against 1 litie ot 0 1 M NaHCO*/0 5M NaCI (coupling buffer)

3 Assessment of wild type and mutant (N40A) protein G binding affinity for mouse IgG Fab and Fc fragments.

Experiments were conducted to assess the binding affinity of the protein G mutants of the invention

3 1 Assay Method

A microtitre plate-based assay was used to measuie the lelative binding affinities of the wild type and mutant piotein G domains toi Fab and fc fragments A 96 well plastic microtitre plate was coated overnight with Fab or Fc fragment in 15mM Na₂C0₃/35mM NaHCO , at loom temperatuie For the results presented in f igures 10 and 1 1 , piotein concentiations υj 10μg/ml lor Fab and 35μg'ml foi l c fragments were used. Each well was then washed five times with l OOμl of phosphate buffered saline solution (PBS) plus 0.05% Tween-20. PBS contains 0.01 M NaH₂P0₄ NaOH (pH 7.4), 0.138M NaCI and 2.7mM KG. The relative binding affinity of each protein G domain was assessed by competition for adhesion to the plate with an alkaline phosphatase-protein G conjugate, prepared as described below. Typically in an experiment, the alkaline phosphatase-protein G conjugate was diluted 1000-fold into PBS/0.05% Tween-20 to give a final protein concentration of approximately lOμg/ml, and a variable amount of competing protein G domain was added (generally to a final concentration between 5mg/ml and 5ug/ml). The protein G domain (wild type or mutant) was purified as previously described. The alkaline phosphatase-protein G conjugate was then incubated within the microtitre plate well for 90 minutes, during which time it bound to Fab or Fc components that were immobilised on the plate. The presence of competing protein G domain suppressed the binding of alkaline phosphatase-protein G conjugate to the Fab or Fc fragment, and hence to the plate. Following the incubation, each well was washed five times with l OOμl of PBS/0.05% Tween-20. and alkaline phosphatase activity detected by incubation with lOOμl of l mg/ml p-nitrophenyl phosphate in 0.1 M glycine/HCl (pH 10.4). plus ImM MgCU and I M ZnCU. Alkaline phosphatase activity was measured by recording the rate of change of absorbance of each well at 415nm in a microtitre plate reader after incubation for a fixed period of time (typically, time points were taken every 20 minutes for one hour).

Preparation of alkaline phosphatase conjugated protein G

Recombinant protein G was prepared as described by Goward et al. ( 1990). 20mg of the recombinant protein in 1.28ml was added to l Omg of alkaline phosphatase in 0.49ml (ALPI-XG; Biozyme) and dialysed extensively against 0.1 M Na₂HP0₄/NaH₂P0 (pH7.0) at 4°C. The protein solution was transferred to a glass test tube and lOOμl of a 1% (v/v) glutaraldehyde solution was added, with continuous stirring of the solution. Cross-linking was left to proceed for three hours at room temperature. To stop the reaction. 2ϋϋul of 1 M ethanolamine (pH 7.0) was added.and the solution left for a luithei two houi s at om tempeiatuie 1 tnall\ , the protein solution was dialysed veisus PBS buffet overnight Aftei dialysis paniculate matter was removed by centrifugation at 22 00ϋ _u foi one houi and glyceiol added to 50% (v/v) ZnCl₂ to I mM, MgS0₄ to I mM and NaN₃ to 0 05% (w/v) The solution was stored at 4°C

3 2 Selectivity of the mutant N40A for binding to Fc fragment.

The assay procedure described in 3 1 was used to compare the relative binding affinities of wild type and N40A mutant piotein G domains loi mouse Fab and Fc fragments To quantify the binding affinity of different piotem G domains for Fab or Fc we define the concentiation ol piotem G domain lequiied to

e 50% inhibition of maximal binding of the piotem G-alkalme phosphatase conjugate to the microtitre plate as the IC-₀, using the expei itnenlal piotocol descnbed above in section 3 1 Wild type protein G domain competed foi binding to Fab fiagment. with 50% inhibition (IC,o) occurring at 23μg/ml (Figure 5) Tor N40A mutant the value was 32μg/ml For binding to Fc fragment. 50% inhibition occurred at 370μg/ml for wild type protein G Howevei in experiments measuring the binding oi the N40A mutant 50% inhibition occurred at 3350μg/ml oi highei giving a 9-fold difference factor (Figure 6)

4 Use of mutant CN40A^ protein G

4 1 Immobilisation of mutant (N40A) protein G domain onto Sepharose

2 5 g (dry weight) of CNBi-activated Sephaiose (Phaimacia AB) was washed twice with 40 ml of ImM Hcl and once with 40 ml ot distilled watei 8 5 mg of N40A domain II protein horn section 2 4 above were added in 10ml ot coupling buffer to the resin, and coupling lelt to pioceed at 7"C foi 3 5 houi s with continuous agitation Finally, the Sepharose beads were washed thiee times w ith 30 ml ot 0 1 M TnsHCl (pH 8 0)/0 5 M NaCI followed bv three times washing with 30 ml of 0 1 M sodium acetate/acetic acid (pH 4 0)/0 5 M NaCI Finally, the beads weie resuspended in 0 1 M sodium acetate/acetic acid (pH 5 0) befoie use

4 2 Use of Immobilised N40A in the separation of Fab and Fc fragments.

Mouse IgG (previously punfied fiom ascites fluid oi concent! ated cell culture supernatant) was digested with papain or pepsin to gcneiate Fab oi F(ab')₂ fiagments The digest was then applied duect on to the N40A protein G-Sepharose matrix and washed with equilibration buffer (see section 4 1 ) until the absorbance at 280 nm had returned to zero At this point the Fc fragments are not retained bv the column Fab fragments are then eluted by application of the elution bulfer (0 1 M glycine/HCL pH 2 8)

T he value of this method lies in its extensive applications to a range of monoclonal IgG antibodies fiom difleient subclasses and with different antigen binding specificities To illustrate this point, results aie given toi the separation of Fab from Fc fragments for mouse IgGs derived fiom all toui γ chain subclasses (Results 1 to 4 below) Fab fragments fiom any mouse IgG monoclonal antibody will therefore bind and can be eluted fiom the N40A piotein G-Sepharose matrix by this procedure

Result 1 Separation of Fab and Fc fragments denved from mouse IgG l (Figure 7) Result 2 Separation of Fab and Fc fiagments derived fiom mouse IgG2a (Figure 8) Result 3 Separation of Fab and Fc fragments derived from mouse IgG2b (Figure 9) Result 4 Separation of Fab and Fc fragments derived from mouse IgG3 (Figure 10)

F(ab')₂ will also bind to the N40A protein G-Sepharose matrix, and can be purified in exactly the same way

Result 5 separation of I (ab )-> and Tc liagments denved horn mouse IgG l (Figure 1 1 ) 5 Analysis of the binding affinities of selected Protein G mutants for Human IgG

Experiments were conducted to measure the binding affinity of selected protein G mutants for human IgG using a fiuoiescence binding assay A single cysteinc residue was introduced into domain II (w ild type domain II contains no cystemes). and used to attach the lluoiescent reportet group 5-((((2- ιodoacetyl)amino)ethyl)amιno)naphthalene- l -sulphonιc acid ( 1 ,5-lAEDANS) The change in the environment of the lluoiophore on binding ot the T23C-1AEDANS mutant to IgG results in a change in its fluorescence emission characteristics This change in fluorescence can be used to determine the equilibrium binding constant for the T23C-IAEDANS mutant to IgG, and the equilibrium binding constants for non- fluorescent protein G domain II species can also be determined by competition. In this way, the binding affinities ot domain II and selected domain II mutants for human IgG4 can be measured. The results show that the N40A mutation has a profound effect on binding to human IgG4. and that the mutations K33A and Q37A can also have a minor effect. Because the binding affinity of protein G for the Fc portion of human IgG is much greater than for the Fab portion (Fahnestock et al., 1990). we infer that these changes correspond to changes in binding affinity of protein G for human Fc.

5.1 Preparation of site-specific mutants

Site-specific mutants of protein G were constructed using the oligonucleotide mismatch mutagenesis method described in relation to Fig. 2 above The mutagenic ohgonucleotides used were:

For the T23C mutant: 5^" ATCAACAGCTTCACAAGTTGTTTCGCC 3^" (SEQ ID NO. 10)

For the K33A mutant: 5^' TTGTTTGAAGACTGCTTCTGCAGTAGC 3 ^' (SEQ ID NO. 1 1 ) For the Q37A mutant 5^' GTCGTTΛGCG TATGCTTTGAAGACTTTTTC 3^' (SEQ ID NO 12)

(Mismatched bases arc highlighted in bold)

5.2 Production of T23C-AEDANS Coupled Protein

The T23C mutant was expressed in iccombinant toim in E coli and puiified as described in section 2 3 above, with the exception that 1 mM dithiothreitol (DTT) was included in the 20mM TnsHCl (pH 7 5) I mM FDTA buffei that was used for resuspending the bacterial cells altei harvesting, and in all punfication buffers used after that Following the DEAE Sephacyl chromatography step, a final FPLC chromatography step was introduced to improve the purity of the protein The sample collected from the DEAE Sephacyl column was dialysed overnight against 2 litres of 20 mM Tris HC1 (pH 7 5), 1 mM DTT and then applied to a Mono Q 10/10 FPLC column, pre-equilibrated in the same buffei The column was pumped with a further 20ml of 20 mM Tns HC1 (pH 7 5). I mM DTT [Buffei A] at a flow rate of 4 ml/min. and elution of the protein was initiated by application of a lineal salt giadient at a rate of 3mM NaCl/min in Buffer A Clution of the protein was monitored fiom the absorbance of the eluate at 280 nm Two forms of protein G eluted from the column a minor component, peak 1 , elutes first at 60 mM NaCI, and corresponds to partial proteolytic digest removing the fust two ammo acids in the protein (the N terminus starts PAV ) The second, majoi . peak elutes at 75 mM NaCI and coiresponds to the full-length domain II ol piotein G The presence of a minoi proportion of proteolytically digested protein (generally less than 20%) was common to w ild type and mutant protein G preparations I ractions corresponding to the ma|oi peak oi the lull length \ ersion, of the T 23C mutant were pooled and concentiated

tιeeze-dιγιng overnight The dned piotein was then redissolved in 2-3ml of water and the salt and DTT components removed by gel filtration over a 50 mi Sephadex G-25 column, pie-equilibrated in 20 mM 1 i isHG (pH8 4), 0 2mM EDTA The eluted piotein was collected fieeze-dned and le- dissolved in 1 ml of water to give a solution containing 0 5 ImM mutant T 23C, 160mM TnsHCl and 1 6 mM EDTA at a final pH ol 8 4 Solid 1 ,5-IAEDANS was dissolved into this solution to give a final concentiation of 37 mM and incubated overnight in the dark at 4 °C Du ng the couise of the incubation the iodine atom within the 1 ,5-IAEDANS moiecule is displaced by nucieophihc attack from the -SH group of cysteιne-23, generating a stable tluoethei bond that attaches the fluoiescent IAEDANS group to the piotein G I Jnieacted 1 5-1AEDANS was lemoved by gel filtration on a 50 ml Sephadcx column into 20 mM TnsHCl (pH 7 5), plus 0 05% NaN Unreacted T23C domain 11 was lemoved fiom T 23C-AEDANS by ion exchange chromatography using a Mono Q 10/10 anion exchange column . undei the same conditions described above, but omitting the DTT Unreacted T23C elutes at a NaCI concentration of 75 mM, and mutant T23C-ALDANS at 1 30 M NaCI The peak corresponding to T23C-ALDANS which was clearlv identified fiom its characteristic absorbance spectrum, was pooled, treeze-dπed, redissolved in l -2ml water and transferred by gel filtiation into 20 mM l iisHG (pH 7 5), 0 05% NaN, The eluted T23C-AEDANS was then fieeze-dried again and redissolved to produce a solution of protein in 100 mM TnsHCl (pH 7 5). 0 05% aN₃

5 3 Measurement of T23C-AEDANS Binding to Human IgG by Fluorescence

A Perkin-Elmer LS-5B spectrofluonmetei was used foi all measurements, although an instrument of equivalent specification could be used instead A 30 nM solution of T23C-AEDANS in lOOmM sodium acetete/acetic acid (pH 4 8) was placed in a 3 ml cuvette in a thermostatted cuvette holdei maintained at 298K b\ a uiculat g watei bath The cuvette was stirred continuously and maintained undei an argon atmosphere to remove oxygen Using excitation light of wavelength 337 nm a slit width of 20nm and a tempeiature ol 298K an emission scan reveals a peak emission wavelength of 490nm (see lowei cuive. Figure 12) On addition of human IgG4 to a concentration of 439 nM, the peak emission wavelength shifts to 478 nm. and is accompanied by a 22% enhancement of fluorescence (see uppei cuive. Figure 12) In Fig 12, the fluorescence (y-axis) is 0 5 x the actual fluoiescence value Note, the fluorescence background oi human l gϋ4 is not accounted foi in Fig 12

The enhancement of fluoiescence that occurs on binding of T23C-AEDANS to human IgG4 can be used to quantitatively lecoid the formation of the T23C-AEDANS IgG4 binary complex as a function of antibody concentration (see Figure 1 ) Anaerobic conditions were used foi this assay The excitation wavelength was 337 nm. the emission wavelength was 480 nm The solution used in the sample and control cuvettes was 100 mM sodium acetate pH 4 8 The sample cuvette contained 30 nM FPLC pure T23C-AEDANS whilst the contiol cuvette contained 30 nM glutathione- AEDANS Human IgG4 was then added in increasing concentrations to the sample and control cuvettes The fluoiescence change (y-axis) lepresents the change in fluorescence in the sample cuvette after removal of the background fluorescence of human IgG4 The (+) lepresents a point lemoved fiom the cuive fit The data can be fitted using a suitable sottwaie piogiamme a dissociation constant ot 13 nM for the dissociation constant ol the T23C-ALDANS-human IgG4 complex gives good agreement with the recorded data (Figure 1 )

5 4 Measurement of equilibrium binding constants for protein G domain II and mutants for binding to human IgG4 by competition with T23C-AEDANS binding

The fluorescence enhancement that occurs on binding ot T23C-AEDANS to IgG can be used to indirectly deteimine the dissociation constants foi non-fluoiescent domain II protein G wild type and mutant proteins by competition Titiation ot protein G domain II into a cuvette containing the T23C-AEDANS IgG complex will displace T23C-AEDANS from its bindinu site on f c and hence lesult in a leduction in fluorescence. An experiment was therefore conducted using the mutant domain N40A described above to show that this is indeed the case (Figure 14). In Fig. 14, the jointed up dots (•) represent the best fit curve, the diamonds (0) represent experimental data. The slitwidth of excitation used was 15 nm. the slitwidth of emission 10 nm. the excitation wavelength 337 nm. the emission wavelength 480 nm and the temperature 298 K. The concentration of human IgG was 62.9 nM and T23C- AEDANS was 131.5 nM. As with the data presented in section 5.3. it is possible to fit the data to a theoretical model by assuming competition between T23C-AEDANS and the N40A mutant, using the K_c, value for T23C-AEDANS determined previously in section 5.3 and varying the K₍, value for N40A until an optimum fit is obtained. As can be seen from Figure 14, there is good agreement between the predicted and observed values at the end of the fitting procedure. The best fit resulted in an equilibrium dissociation constant (K₍|) for the N40A:lgG complex of 1515 nM. This experiment was carried out using wild type protein G and different three mutants: the results are tabulated below:

*K33A and Q37A were expressed and purified as described in section 2.3 above for

T23C, except that DTT was not included in any of the purification buffers.

These results demonstrate that:

1. The N40A mutation has a potent effect on the binding affinity of protein G for human IgG, reducing it by 63-fold. This is an even bigger effect than the binding to mouse IgG, and provides strong evidence that the mutation described above could also be effective in separating Fab and Fc fragments from human antibodies. 2. The mutations K33A and Q37Λ al.so have significant effects on Fc binding, and could be incorporated into a multiple mutant that has a greatly reduced binding affinity for human Fab fragments. The calculated combined effect of all three mutations is 658-fold.

REFERENCES

Achari. A., Hale. S.P., Howard. A.. I.. Gore. G.M.. Gronenborn. A.M., Hardman. K.D.

& Whitlow, M. ( 1992). Biochemistry 31 , 10449. Akerstrom, B., Nielsen, E. & Bjorck. L. ( 1987) .1. Biol. Chem. 262, 13388. Boyle. M.D.P. (1990) 'Bacterial Immunoglobulin-Binding Proteins'. Academic Press,

New York. Derrick, J.P. & Wigley, D.B. ( 1992) Nature 359. 752. Derrick. J.P. & Wigley. D.B. ( 1994) .1. Mol. Biol. 243. 906. Fahnestock, S.R., Alexander. P.. Filpula, D. & Nagle, J. ( 1990) in 'Bacterial

Immunogiobulin-Binding Proteins' (Ed. M.D.P. Boyle) Structure and evolution of the streptococcal genes encoding protein G, Academic Press. New York. Gallagher, T., Alexander. P.. Bryan. P. & Gilliland. G.L. ( 1994). Biochemistry 33,

4721. Goward, C.R., Murphy. J.P., Atkinson. T. & Barstow. D.A. ( 1990) Bioche . J. 267,

171. Gronenborn, A.M., Filpula. D.R.. Essig, N.Z.. Achari. A.. Whitlow. M.. Wingfield,

P.T. & Gore, G.M. ( 1991 ). Science 253, 657. Gronenborn, A.M. & Gore. G.M. ( 1993). J. Mol. Biol. 233. 331 . Guss. B., Eliasson. M.. Olsson. A.. Uhlen. M., Frej. A-K.. Jornvall. H.. Flock. J-1. and

Lindberg, M. ( 1986). EMBO J. 5. 1567. Kato. K., Lian, L-Y., Barsukov, I.L.. Derrick. J.P.. Kim, H., Tanaka. H.. Yoshino. A.,

Shiraishi, M.. Shamada. I., Arata. Y. & Roberts. G.C.K. ( 1995) Structure 3. 79. Kunkel, T.A., Roberts, J.D. & Zakour. R.A. ( 1987) Methods Enzymol. 154 . 367. Lian. L-Y., Yang, J.C, Derrick. J.P.. Sutcliffe. M.S.. Murphy. J.P.. Goward. C.R. &

Atkinson. T. ( 1991 ). Biochemistry 30, 5335. Lian. L.-Y., Derrick. J.P.. Sutcliffe. M.J.. Yang. J.C. & Roberts. G.C.K. ( 1993) J.

Mol. Biol. 228. 1219. Lian. L-Y.. Barsukov. I.L.. Derrick. J.P. &. Roberts. G.C.K. ( 1994). Nature Structural

Biology 1. 355-357. Sambrook, J, Fritsch, FF & Maniatis T (1989) 'Moleculai cloning- a laboratory manual' 2nd Edition. Cold Sp ng Haibot Laboratory Piess USA Sauereπksson. A E , Klcywegt, G 1. Uhl M & lones FA ( 1995) Stiucture 3.265 Sjobnng, U , Bjorck, L & Kastein. W (1991) J Biol Chem 266,399 Uhlen, M , Guss, B , Nilsson. B , Gatenbeck. S , Phihpson. L & Lindberg, M (1984)

Yanish-Perron, C . Vieia. J & Messing. I (1985) Gene 33.103

S EQUENCE LISTING

GENERAL INFORMATION

( l ) APPLICANT

(A) NAME THE UNIVERSITY OF MANCHESTER INSTITUTE Of

SCIENCE & TECHNOLOGY

(B) STREET P ϋ BOX 88

(C) CITY MANCHESTER

(E) COUNTRY UNITED KINGDOM

(F) POSTAL CODE (ZIP) MGO 1QD

(A) NAME THE UNIVERSITY OF LEICESTER

(B) STREET UNIVERSITY ROAD

(C) CITY LEICESTER

(D) STATE LEISTERSHIRt

(E) COUNTRY UNITED KINGDOM

(F) POSTAL CODE (ZIP) 1E1 7PH

(A) NAME JEREMY DERRICK

(B) STREET 135 FOG LANE, DIDSBURY

(C) CITY MANCHESTER

(E) COUNTRY UNITED KINGDOM

(F) POSTAL CODE (ZIP) M20 6ED

(A) NAME GORDON ROBERTS

(B) STREET THE OLD STABLES

(C) CITY BRUNTINGTHORPE

(D) STATE LEICESTERSHIRE

(E) COUNTRY UNITED KINGDOM

(F) POSTAL CODE (ZIP) LE17 SQ7

(ll) TITLE Or INVENTION MODIFIED PROTEIN G AND FRAGMENTS THEREOF

(ill) NUMBER OF SEQUENCES :

l IV) COMPUTER READABLE FORM

(A) MEDIUM TYPE Floppv disk

(B) COMPUTER IBM PC comDacible ( C) OPERATING SYSTEM DΓ-DOS/MS-DOS

(D) SOFTWARE Patent Ir Release 01 0, V.rsion Ml 30 IEPO)

(2) INFORMATION FOR SEQ ID NO l

(l) SEQUENCE CHARACTERISTICS-

(A) LENGTH "57 ammo acidu

(B) TYPE amino acid

(C) STRANDEDNESS single

(D) TOPOLOGY linear

(ii) MOLECULE TYPE pepciαe

(ill) HYPOTHETICAL NO

(vi) ORIGINAL SOURCE

(A) ORGANISM STREPTOCOCCUS

(B) STRAIN G148

(XI ) SEQUENCE DESCRIPTION SEQ ID NO 1

Thr Asp Thr Tyr Lys Leu lie Leu Asn Gly Ly.s Thr Leu ι_ys Gly Glu l b 10 lb

Thr Thr Thr Glu Ala Val Asp Ala Ala Thr Ala Glu Lys Val Phe Lys 20 25 30

Gin Tyr Ala Asn Asp Asn Glv Val Asp Gly Glu Tip Thr Tyr Asp ASD 35 40 45

Ala Thr Lys Thr Phe Thr Val I'hi Glu 50 55

(?) INFORMATION FOP SEQ ID NO

(l) SEQUENCE CHARACTEP T T^* 5

(A) LENGTH bl amino _cιds (B) TYPE: amino acid

(C) STRANDEDNESS . uinq.

(D) TOPOLOGY lineal

(11) MOLECULE TYPE peptlde

(in) HYPOTHETICAL: NO

(vi ) ORIGINAL SOURCE.

(A) ORGANISM. STREPTOCOCCUS

(B) STRAIN G148

(XI ) SEQUENCE DESCRIPTION SEQ ID NO- 2.

Leu Thr Pro Ala Val Thr Thr Tyr Lys Leu Val He Asn Gly Lys Thr 1 5 10 15

Leu Lys Gly Glu Thr Thr Thr Glu Ala Val Asp Ala Ala Thr Ala Glu 20 25 30

Lys Val Phe Lys Gin Tyr Ala Asn Asp Asn Gly Val Asp Gly Glu Trp 35 40 45

Thr Tyr Asp Asp Ala Thr Lys Thr Pne Thr Val Thr Glu 50 55 60

(2) INFORMATION FOR SEQ ID NO: ϊ

(l) SEQUENCE CHARACTERISTICS

(A) LENGTH: 61 ammo acids

(B) TYPE: amino acid

(C) STRANDEDNESS: s nqle

(D) TOPOLOGY: linear

iπl MOLECULE TYPE pepcide

(ill) HYPOTHETICAL. NO vi ) ORIGINAL SOURCE.

(A) ORGANISM STREPTOCOCCUS

(B) STRAIN G148

xi) SEQUENCE DESCRIPTION. SEQ ID NO. 3.

Leu Thr Pro Ala Val Thr Thr Tyr Lys Leu Val He Asn Gly Lys Thr 1 5 10 15

Leu Lys Gly Glu Thr Thr Thi Lvs Ala Val Asp Ala Glu Thr Ala Glu

20 25 30

Lys Ala Phe Lys Gin lyr Ala Asn Asp Asn Gly Val Asp Giy Val Trp 35 40 45

Thr Tyr Asp Asp Ala Thr Lys Thr Phe Thr Val Thr Glu 50 55 60

(2) INFORMATION FOR SEQ ID NO: 4

(1) SEQUENCE CHARACTERISTICS

(A) LENGTH. 9 amino acids

(C) STRANDEDNESS. single

(D) TOPOLOGY: linear

(ii) MOLECULE TYPE, peptide

(ill) HYPOTHETICAL. NO

ivi) ORIGINAL SOURCE

(A) ORGANISM- STREPTOCOCCUS

(B) STRAIN G148

xi ) SEQUENCE DESCRIPTION SEQ ID N Lys Pro Glu Val He A D Ala S->r (,]ι

1

( ) INFORMATION FOR SEQ ID NO _,

(l) SEQUENCE CHARACTERISTICS

(A) LENGTH y ammo acids

(C) STRANDEDNESS single

(D) TOPOLOGY linear

,11) MOLECULE TYPE peptide

(m) HYPOTHETICAL NO

(vi) ORIGINAL SOURCE

(A) ORGANISM STREPTOCOCCUS

(B) STRAIN G148

(XI ) SEQUENCE DESCRIPTION SEQ ID NO 5

Lys Pro Glu Val He Asp Ala Sei blu 1 5

(2) INFORMATION FOR SEQ ID NO 6

(l) SEQUENCE CHARACTERISTICS

(A) LENGTH 36 base pairs

(B) TYPE nucleic acid

(C) STRANDEDNESS single

(D) TOPOLOGY linear

(n) MOLECULE TYPE DNA (genomic

( i) SEQUENCE DESCRIPTION I'J NO ( TTCAGAAGGT ATCGATTΛTG ACACCAGCCG TGACAA

(2) INFORMATION FOR SEQ ID NO 7

(l) SEQUENCE CHARACTERISTICS

(A) LENGTH: 51 base pairs

(B) TYPE: nucleic acid

(C) STRANDEDNESS: single

(D) TOPOLOGY, linear

ui) MOLECULE TYPE DMA (yenomi i

(xi) SEQUENCE DESCRIPTION: SEQ ID NO: 7:

CAATGACTTT TTGGTCTTAT TCGGACTTAA GAGACAAAGA TCCGTAAAAT A

(2) INFORMATION FOR SEQ ID NO: 8

(i) SEQUENCE CHARACTERISTICS

(A) LENGTH: 39B base pairs

(B) TYPE: nucleic acid

(C) STRANDEDNESS: both

(D) TOPOLOGY- both

(n) MOLECULE TYPE: DNA igenomic

(xi ) SEQUENCE DESCRIPTION- SEQ ID NO: 8:

GGATCCATCG CTACCGTCGT GTCCTGACTC GTTGGGAGAA AAAGGTCGAA AATTACGAGG 60

CAATGCTGCA TCTTGCCTGT GC ΛTCATTG TCTGGAATAA AATCCTTTTG GGATAGGTTC

TTAGTAAATA AAAATTCAGA ΛGGTATCGAT TΛTGACACCA GCCGTGACAA CTTACAAACT TGTTATTAAT GGTAAAACAT TGAAAGGCGA AACAACTACT GAAGCTGTTG ATGCTGCTAC 240

TGCAGAAAAA GTCTTCAAAC AATACGCTAA CGACAACGGT GTTGACGGTG AATGGACTTA 300

CGACGATGCG ACTAAGACCT TTACAGTTAC TGAAAAACCA GAATAAGCCT GAATTCTCTG 360

TTTCTAGGCA TTTTATGATA ATTTTTTAAA GGTAAGCT 398

(2) INFORMATION FOR SEQ ID NO- 9

(l) SEQUENCE CHARACTERISTICS

(A) LENGTH: 28 base pans

(B) TYPE, nucleic ac d

(C) STRANDEDNESS single

(D) TOPOLOGY: linear

(ll) MOLECULE TYPE: DNA (genomic)

(X ) SEQUENCE DESCRIPTION SEQ ID NO

CACCGTTGTC GGCAGCGTAT TGTTTGAA

(2) INFORMATION FOR SEQ ID NO. 10

(l) SEQUENCE CHARACTERISTICS

(A) LENGTH. 27 base pairs

(B) TYPE: nucleic acid

(C) STRANDEDNESS: sinqlt

(D) TOPOLOGY, lineal

(ll) MOLECULE TYPE- DNA (genomic)

( i ) SEQUENCE DESCRIPTION SEQ ID NO 10 ^" .-.

ATCAACAGCT TCACAAGTTG TTTCGCC 27

(2) INFORMATION FOR SEQ ID NO 1.

( ) SEQUENCE CHARACTERISTICS

(A) LENGTH 27 base pans

(B) TYPE nucleic acid

(C) STRANDEDNESS single

(D) TOPOLOGY lineai

(ll) MOLECULE TYPE DNA (genomic

(xi) SEQUENCE DESCRIPTION SEQ ID NO 11

TTGTTTGAAG ACTGCTTCTG CAGTAGC

(2) 'INFORMATION FOR SEQ ID NO 1

(l) SEQUENCE CHARACTERISTICS

(A) LENGTH 30 base pairs

(B) TYPE, nucleic acid

(C) STRANDEDNESS single

(D) TOPOLOGY linear

(ll) MOLECULE TYPE DNA (genomic

(xi) SEQUENCE DESCRIPTION SEQ I^"3 0 1?

GTCGTTAGCG TATGCTTTGA AGACTTTTTC 30

Claims

1 . A polypeptide sequence which is a modified form oi^' protein G or fragment thereof which selectively binds lo one of the Fab or Fc fiagments of an antibody relative to the other as compared to the corresponding selectivity of binding of unmodified protein G.

2. A polypeptide sequence as claimed in claim 1 which is a Protein G fragment comprising the sequence of Domain 1. 11 or III (or a modification thereof, part thereof, or modified part thereof) as shown in Fig. 1 of the drawings which has a mutation such that the sequence is capable of selectively binding to one of the Fab or Fc fragments of an antibody but not the other.

3. A polypeptide sequence as claimed in claims 1 or 2 wherein the selectivity expressed in terms of:

(a) the ratio of binding affinity (IC₅ ') of the polypeptide sequence of the invention for Fab fragments to the binding affinity (IC^,") of unmodified protein G for such fragments, and

(b) the ratio of binding affinity ( IC₅₀ ^IM) of the polypeptide sequence of the invention for Fc fragments to the binding affinity (IC ,' ) of unmodified protein G for such fragments; is such that the value of one of (IC₅ 7lC₅₀") or (IC_s_{)"7lC ₀ ^,v) is less than 2 (preferably less than 1.5) and the other is greater than 3 (preferably greater than 5 and most preferably greater than 7).

4. A polypeptide sequence as claimed in claim 3 wherein (1C^ 7IC_5I1") is less than 2 (preferably less than 1.5) and 1C ₀'"/IC^₀' ) is greater than 3 (preferably greater than 5 and most preferably greater than 7) whereby the polypeptide sequence is selective for Fab.

5 A polypeptide sequence as claimed in claim 2 which is capable of selectively binding the Fab fragment

6. A polypeptide sequence as claimed in claim 4 oi 5 wheiein the mutation is at position 40 of Domain I. II oi III

7. A polypeptide sequence as claimed in claim 6 wherein the mutation is the N40A mutation.

8 A polypeptide sequence as claimed in claim 7 which is a Protein G fragment comprising said Domain II incorporating the N40A mutation

9. A polypeptide sequence as claimed in claim 3 oi 4 wherein the mutation is E32A, K33A, Q37A and/or E47A of any one Domains I. II or III.

10. A polypeptide sequence as claimed in claim 3 wherein (IC _I7lC ₍₎") is greater than 3 (preferably greater than 5 and most preferably greater than 7) and (IC^ "7lC^ ^lv) is less than 2 (preferably less than 1 5) whereby the polypeptide sequence is selective for Fc

1 1. A polypeptide sequence as claimed in claim 2 which is capable of selectively binding the Fc fragment.

12 A polypeptide sequence as claimed in claim 10 or 1 1 wherein the mutation is T16A, T21A, T22A, Y38F and N42A ol any one of Domains 1. II or III.

13 A method of effecting separation of a mixture of Fab and Fc fragments of an antibody from a mixture thereof, the method comprising ti eating said mixtuie with a polypeptide sequence in accordance w ith claim 1 which selectively binds to a predetermined one of the Fab and T. fragments to form a complex theiewith. and separating the unbound fragment iiom said complex

14 A method as claimed in claim 1 which is an affinity sepaiation in which the polypeptide sequence is immobilised on a support which is then treated with the mixture of Fab and Fc fiagments to be sepaiated

15. A method of producing a polypeptide sequence as claimed in claim 1 composing expressing, in a host, the polypeptide sequence Ii om a vector (the "mutation construct")

16 A method as claimed in claim 1 5 wheiein the mutation construct is piepared by the steps of.

(i) replicating a vector incorporating the wild type coding sequence (the "wild type construct") in the presence of deoxyui idine so as to pioduce a construct which misincorporates deoxyuridine (the "deoxyui idine constiuct"), and

(ii) subjecting the "deoxyuridine construct" to mismatch site directed mutagenesis, copying and selection to produce the "mutation construct^" in which the region coding for the selective polypeptide is a mutation of the wild type gene coding sequence, and there is no deoxyuridine misincorporation

17 A method as claimed in claim 16 wheiein the selection for the mutation construct is effected by transfection of the mutation constiuct and the deoxvuπdine construct into an organism that is capable of lepaiπng DNA with deoxvuπdine misincorporation, and gi owing the organism