Search International and National Patent Collections
Some content of this application is unavailable at the moment.
If this situation persists, please contact us atFeedback&Contact
1. (WO1992009633) IMMUNOGLOBULIN-BINDING PROTEINS AND RECOMBINANT DNA MOLECULES CODING THEREFOR
Note: Text based on automatic Optical Character Recognition processes. Please use the PDF version for legal matters

IMMUNOGLOBULIN-BINDING PROTEINS AND RECOMBINANT DNA MOLECULES
CODING THEREFOR
This invention relates to immunoglobulin-binding proteins and recombinant DNA molecules coding therefor.
Protein-A (SpA) is a cell wall component of Staphylococcus auxeus which binds to the Fc region of inununoglobulins from a variety of sources (Langone, 1982). For example, it can bind to human IgG sub-classes 1, 2 and 4 but not in general to IgG3,. It can also efficiently bind IgG from rabbit and pig, but it binds horse and cow IgG with lower affinity, and binds rat IgG only very weakly (Boyle and Reis, 1987).
This specific interaction with IgG molecules makes SpA a very useful immunological tool. It has been used in immunologlobulin purification when immobilised into chromatography columns, and as an antibody probe in enzyme-linked immunosorbent assay (ELISA) systems; together these uses have been exploited in the screening and
purification of monoclonal antibodies. Recently SpA has also found use in chemotherapy to remove immune complexes from serum (see Palmer et al., 1989 and references therein), and in biotechnology where it has been incorporated into cloning vectors in which the cloned gene can be expressed as a fusion with SpA (Nilsson et al, 1985).
The isolation of SpA from S. aureus cells is not
straightforward and also no entirely satisfactory techniques are available for anchoring SpA to solid supports. The gene for SpA has been cloned and sequenced (Lofdahl et al, 1983, Uhlen et al, 1984, Shuttleworth et al, 1987) and encodes a 42 kDa protein consisting of 5 homologous IgG binding domains termed E.D.A.B and C, and a C-terminal cell wall spanning and membrane anchoring region, region X. In addition an N-terminal signal sequence is thought to target the protein out of the cell. The crystal structure of a single IgG binding domain-fragment B (SpAB) bound to human Fc has been resolved at the 2.8 A level (Deisenhofer, 1981) and recent NMR studies show that SpAB contains 2 α-helices, the residues of which forms most of the contact points of Fc. The Fc binding α-helices of successive SpA domains are apparently separated by flexible polypeptide spacer regions.
Site directed mutagenesis is a powerful tool which could be used to probe the SpA-Fc interaction. However, mismatch primer mutagenesis is very difficult since the repeated nature of the gene means that the primer could anneal to multiple sites. Other workers (Nilsson et al, 1987: Saito et al, 1989) have reported the production of IgG binding proteins, based upon the B domain of SpA, from
synthetic genes. Their studies highlighted the difficulties often encountered when expressing small foreign proteins in E. coli eg.
proteolysis of the product by host enzymes or difficulties in the purification of the expressed protein. Accordingly it has hitherto not been practical to prepare mutated proteins, derived from SpA, which have properties adapted to specific purposes. Specifically, the production of modified forms of SpA which avoid the above- mentioned difficulties has consequently proved to be problematical.

The present invention has solved these problems by designing a synthetic Fc-binding domain which is highly amenable to site directed mutagenesis. Expression of polypeptides comprising this synthetic Fc-binding domain has enabled the production of immuno-globulin-binding proteins having distinct advantages compared to SpA, rendering them particularly useful in preparative and diagnostic techniques and in therapy.
According to one aspect thereof, the present invention provides a polypeptide capable of forming a complex with an
immunoglobulin, said polypeptide being characterised by having at least 2, but not more than 4 binding domains, each capable of binding to the Fc region of an immunoglobulin of the IgG class.
Preferably the polypeptide is characterised by having 2, but not more than 2 of said binding domains.
In one embodiment of the invention, the binding domains possess a high degree of sequence homology with the binding domains of Staphylococcus aureus Protein-A (SpA). Thus preferably each of said binding domains has at least 75% sequence homology, preferably at least 90% sequence homology, with at least one of the binding domains designated A, B, C, D and E of Staphylococcus aureus Protein-A.
Most preferably each of said binding domains has at least 75% sequence homology, preferably at least 90% sequence homology, with the binding domain designated B of Staphylococcus aureus Protein-A.
It is not necessary for the binding domains of the
polypeptide to match precisely the size of the binding domains of SpA, but preferably each of said binding domains consists of from 40 to 55 amino acid residues.

The following are especially preferred sequences for the binding domains of polypeptides according to the invention:
(1) the sequence
Ala Pro Lys Ala Asp Asn Lys Phe Asn Lys
Glu Glu Glu Asn Ala Phe Tyr Glu Ile Leu
His Leu Pro Asn Leu Asn Glu Glu Glu Arg
Asn Ala Phe Ile Glu Ser Leu Lys Asp Asp
Pro Ser Glu Ser Ala Asn Leu Leu Ala Glu
Ala
(2) sequences consisting of at least 40 amino acid residues and derived from sequence (1) by
(a) deleting up to 11, preferably not more than 8 and
most preferably not more than 3 amino acid residues
of sequence (1) and/or
(b) substituting up to 11, preferably not more than 8
and most preferably not more than 3 amino acid
residues of sequence (1) by other amino acid
residues and/or
(c) inserting up to 11, preferably not more than 8
and most preferably not more than 3 amino acid
residues into sequence (1) .

Polypeptides according to the invention having at least one binding domain as specified in alternative (2) above may be produced by site-directed mutagenesis, using as a starting point recombinant DNA molecules containing DNA sequences coding for sequence (1) above.
It is particularly preferred according to the invention for the derived sequences 2(a), 2(b) and 2(c) to confer on the
polypeptides according to the invention a binding capacity which has a different pH dependence compared to that of protein A itself. This may be achieved for example by replacing a non-ionisable amino acid residue in sequence (1) by an ionisable amino acid residue.
Alternatively an ionisable residue may be replaced by a non-ionisable residue. As a further alternative, an ionisable residue may be replaced by another ionisable residue having a different pKa or pKb.

Examples of ionisable amino acid residues include

His, Arg, Lys, Glu, Asp, Cys and Tyr.

Of these residues, Glu, Asp, Tyr and Cys ionise when the pH is raised; whereas His, Arg and Lys ionise as the pH is lowered.

Thus according to a preferred aspect of the invention Tyr18 (the Tyr residue which occurs in the partial sequence Ala Phe Tyr Glu) may be replaced by a residue selected from:

His, Arg, lys, Glu, Asp and Cys.

Thus, for example, specific examples of derived sequences (b) include sequences wherein the or at least one of the partial sequences
Ala Phe Tyr Glu
is replaced by one of the following sequences:
Ala Phe Glu Glu
Ala Phe Phe Glu
Ala Tyr Tyr Glu
Ala Phe His Glu
Ala Phe Lys Glu
Ala Phe Cys Glu

Further examples of derived sequences (b) for the binding domains of polypeptides according to the invention are sequences having the following general sequence:
Ala Pro Lys Ala Asp Asn Lys Phe Asn Lys
Glu Glu Glu Asn Ala Phe X Glu Ile Leu
His Leu Pro Asn Leu Asn Glu Glu Glu Arg
Pro Ser Glu Ser Ala Asn Leu Leu Ala Glu
Ala
wherein X can be phenylalanine, glutamic acid, histidine, cysteine or lysine.
Sequences for the binding domains of polypeptides according to the invention such as those above are preferably mutated by cassette mutagenesis.
Preferably each of the binding domains of the polypeptide of the invention has the same amino acid sequence.

Thus for example it is preferred that where a polypeptide according to the invention comprises two or more derived sequences as defined in paragraph (2) above, each of said derived sequences is identical, i.e. the derived sequences contain the same amino acid substitutions) at the same position(s). As indicated, preferred polypeptides according to the invention having derived binding domain sequences as described above may exhibit modified, pH-dependent binding affinities. Particularly preferred polypeptides according to the invention are provided at their C-terminal ends with an amino acid residue having a functional group allowing the polypeptide to be bound covalently to a solid support. Thus preferably the polypeptides according to the invention are provided with a cysteine residue at the C-terminal end.
One such preferred polypeptide has the following C-terminal sequence
Ala Pro Lys Ala Asp Asn Lys Phe Asn Lys
Glu Glu Glu Asn Ala Phe Tyr Glu Ile Leu
His Leu Pro Asn Leu Asn Glu Glu Glu Arg
Asn Ala Phe Ile Gln Ser Leu Lys Asp Asp
Pro Ser Glu Ser Ala Asn Leu Leu Ala Glu
Ala Lys Lys Leu Asn Glu Ser Glu Ala Pro
Lys Ala Asp Asn Lys Phe Asn Lys Glu Gln
Glu Asn Ala Phe Tyr Glu Ile Leu His Leu
Pro Asn Leu Asn Glu Glu Glu Arg Asn Ala
Phe Ile Gln Ser Leu Lys Asp Asp Pro Ser
Glu Ser Ala Asn Leu Leu Ala Glu Ala Cys This polypeptide has two binding domains of formula (1) separated by a 7 amino acid linker (Lys Lys Leu Asn Glu Ser Gln) based upon the sequence linking adjacent IgG binding domains in native SpA. The above polypeptide is further provided with a Cysteine residue at the C-terminus.
Preferred polypeptides according to the invention are produced in the form of fusion proteins, especially fusion proteins having a molecular weight in the range 18 - 30 kDa. It is further preferred that the fusion proteins of the invention comprise a polypeptide according to the invention, fused to an amino acid sequence capable of acting as a nucleus for protein folding events. An example of such a sequence is the first 50 to 85 amino acids of DNase1.
In one preferred embodiment of the invention, said fusion protein is one having an N-terminal amino acid sequence comprising the sequence of the first 50 to 85 amino acids of DNase1, preferably the sequence of the first 8l amino acids of DNase1. Said fusion proteins in accordance with the invention may further be in the form of inclusion bodies.
According to a further aspect of the invention there is provided a recombinant DNA molecule having insert coding for the amino acid sequence
Ala Pro Lys Ala Asp Asn Lys Phe Asn Lys
Glu Glu Glu Asn Ala Phe Tyr Glu Ile Leu
His Leu Pro Asn Leu Asn Glu Glu Glu Arg
Asn Ala Phe Ile Glu Ser Leu Lys Asp Asp
Pro Ser Glu Ser Ala Asn Leu Leu Ala Glu
Ala and characterised by the presence of at least one unique restriction site, preferably at least two unique restriction sites. Preferred recombinant DNA molecules according to the invention are characterised by the presence of at least three, preferably four unique restriction sites, particularly restriction sites selected from DdeI MluI, BglII and MaeIII.
One such preferred DNA insert has the sequence
gcg cct aag gct gat aac aaa ttc aac aaa
gaa cag cag aac gcg ttc tac gag atc tta
cat ctg ccg aac ctg aac gaa gaa cag cgt
aac get ttc att cag tct ctg aaa gac gac
ccg agc cag tct gct aac ctg ctg gct gaa
gct
and a second has the sequence
atg gcg cct aag gct gat aac aaa ttc aac
aaa gaa cag cag aac gcg ttc tac gag atc
tta cat ctg ccg aac ctg aac gaa gaa cag
cgt aac gct ttc att cag tct ctg aaa gac
gac ccg agc cag tct get aac ctg ctg gct
gaa gct tgc
In the above sequences, one or more codon may be replaced by a degenerative codon (i.e. one coding for the same amino acid).
Thus the initial codon get in the first sequence (and the corresponding second codon in the second sequence) may be replaced by gcg, which is the codon present in the corresponding position in the sequence coding for the natural binding domain of SpAB. The
substitution gcg <╌> gct has no effect on the restriction map of the overall sequence.

The expression, purification and activity of novel
Fc-binding proteins according to the invention designated 81-SpAB*-2 and 53-SpAB*-2 consisting of two such synthetic domains fused to part of the bovine DNAase1 gene will now be described by way of example, with particular reference to the following drawings of which:
Figure 1 shows the complete nucleotide sequence and the
encoded amino acid sequence of the synthetic
SpAB* gene.
Figure 2 shows the formation of a gene encoding two SpAB*
domains.
Figure 3a shows the construction of gene fusion plasmid
p81-SpAB*-2.
Figure 3b shows the construction of gene fusion plasmids
p81-SpAB*-2 and p53-SpAB*-2
Figure 4 shows the complete DNA and amino acid sequence of
fusion protein 81-SpAB*-2.
Figure 5 shows an SDS-PAGE gel illustrating the time course
of induction of fusion protein 81-SpAB*-2.
Figure 6 shows an SDS-PAGE gel illustrating inclusion body
purification.
Figure 7 shows an enzyme linked immunosorbant assay for IgG
binding.
Figure 8 shows the formation of 81-SpAB*-1 or
81-SpAB*-2-IgG complexes by light scattering.
Figure 9 shows helical wheel representation of residues
13-21 and 29-40 from SpAB*.

Figure 10 shows the affinities of 81-SpAB-2 and SpA for IgG
at different pHs.
Figure 11 shows the relationship between binding of IgG by
mutated 81-SpAB*-2 proteins and pH.
In the following Example the production of IgG-binding proteins (81-SpAB*-2 and 53-SpAB*-2) by total gene synthesis is described. Unique restriction sites have been placed along the genes to facilitate the production of variant proteins. 81-SpAB*-2 is the product of the fusion of part of the gene for bovine DNAasel and a gene coding for the two B domains (SpAB) of Protein A from
Staphylococcus aureus.
The fusion product is expressed in high yields in
Escherichia coli JM103 as an inclusion body which can be purified by centrifugation and washing with aqueous denaturants such as Triton and urea. The protein may be extracted into 2.5M urea and IgG-binding activity is restored on removal of the urea by dialysis.
The protein has a single cysteine residue placed at the carboxyl terminal of the protein which facilitates either
immobilisation of the protein to an insoluble matrix or the labelling of the protein by radioactive or fluorescent reagents and has the same affinity and specificity for IgG from various sources as Protein A.
The protein can be precipitated from solution by adjusting the pH to 6.0 and is very heat stable and loses no activity by heating at 85°C for 30 min.

Variations of 81-SpAB*-2 have been produced by amino acid substitutions, and some of these mutated proteins show changes in IgG binding activity.
In this Example, a gene encoding a single synthetic IgG binding domain was contructed by automated DNA synthesis. This synthetic domain, termed SpAB* was based upon one of the five IgG binding domains of Protein A; domain B (SpAB) which has an amino acid sequence closest to the consensus sequence of the five domains. Further it is strongest binding of all isolated single domains. The amino acid numbering system used to refer to residues in the synthetic binding domains throughout this description is based upon that devised by Uhlen et al (1984) and is shown in Figure 1 for ease of reference. Bacterial strains, cloning vectors and cell growth
E. coli JM103 (Messing et al, 198l) was used as a bacterial host. Plasmid and phage vectors used were pUC19 (Yanisch-Perron et al, 1985) pkk223-3 (Brosius and Holy, 1984) and phage M13mp19
(Yanisch-Perron et al, 1985). Bacteria were routinely grown in
L-broth (1% bactotryptone, 0-5% yeast extract, 0.5% NaCl) supplemented where appropriate with 50 μg/ml ampicillin (Sigma).
DNA techniques
Restriction enzymes (purchased from Boehringer Mannheim, Northumbria Biologicals Ltd) were used according to the supplier's recommendations, as were the enzymes T4 DNA ligase, T4 polynucleotide kinase and calf intestinal alkaline phosphatase (Boehringer Mannheim). DNA sequencing was performed using 'Sequenase', modified phage T7 DNA polymerase (Tabor and Richardson, 1987; 'Sequenase' kit purchased from United States Biochemical Corporation). All sequencing protocols including template preparation, were performed according to the supplier's recommendations. Oligonucleotides were synthesized on a fully automated Applied Biosynthesis 380A DNA synthesiser which employs the phosphoramidite method of solid phase synthesis (Atkinson and Smith, 1984). De-protected oligonucleotides were purified by electrophoresis on a 7M Urea 12% polyacrylamide gel from which the band corresponding to the full length DNA sequence was excised and eluted (Maniatis et al, 1983).
DNA constructions
A synthetic gene, termed SpAB* was constructed, based on the B domain of SpA.
The DNA sequence was modified to maximise where possible the codon usage for translation in E. coli (Guoy and Gautier, 1982;
Grosjean and Fiers, 1982). Oligonucleotide cassette based site directed mutagenesis is facilitated by the introduction of a series of unique restriction sites at intervals in the DNA sequence.
Specifically, SpAB* was constructed as a series of six oligonucleotides of length 58-66 bp (see Fig 1); adjacent
oligonucleotide pairs had a 7 bp cohesive overlap with the
neighbouring pair. The internal 5' ends were phosphorylated
separately, then complementary oligonucleotide pairs were annealed together by heating separately, to 85°C followed by slow cooling to room temperature. The three pairs were ligated together and cloned into BamHI/Pstl cut M13mp19. DNA sequencing was performed to check the construction, then the resultant BamHI/Pstl SpAB* insert was subcloned into pUC19 to create plasmid pSpAB*. Two α-helices within the encoded domain are predicted to be largely involved in IgG binding and these are represented in Figure 1 by boxes over the amino acid sequence. Residues which are predicted to make close contacts with the Fc molecule have been underlined. The amino acid sequence of SpAB* remains identical to that of SpAB except for the
substitution of an alanine residue for glycine-29, and the
introduction of a C-terminal cysteine.
The gly --> ala replacement occurs at a non-essential position in the 2nd α-helix (i.e. away from the face that interacts with Fc) and is not believed to affect Fc binding (Nilssen et al, 1987). This substitution was done to remove the single Asn-Gly peptide bond, making the domain resistant to hydroxylamine treatment. This will permit the inclusion of such a bond at the junction between the DNAasel and SpAB moieties of the fusion protein so that the two may be split by hydroxylamine and separately purified.
The additional Cys is introduced at the C-terminus which is away from the Fc-binding region and provides a reactive site for possible fluorescent labelling or immobilisation onto Sepharose to give an IgG purification column.
A gene encoding two Fc-binding domains (SpAB*-SpAB*) was constructed by linking 2 of the SpAB* genes together by the
methodology shown in Figure 2. The synthetic linker DNA encodes those amino acids which separate adjacent Fc-binding domains in native Protein A. This technique also ensures that the cysteine residue and stop codons are removed from domain 1, giving an in-frame protein with a single C-terminal cysteine residue. This construction was also cloned into pUC19 to give plasmid pSpAB*-2.
The technique used for linking genes for 2 SpAB* domains is shown in Figure 2. pSpAB* was digested with BamHI/HindlII and the 185 bp fragment was purified away from vector DNA to give the
'upstream' domain 1. In a parallel procedure pSpAB* was digested with Ddel/Pstl and the 166 bp fragment corresponding to the
'downstream' domain 2 was purified. These 2 molecules were then ligated together using a short synthetic linker sequence (see Table 1) containing the appropriate HindIII/Ddel restriction site cohesive ends. The resultant SpAB* - SpAB* gene was cloned into M13mp19, sequenced, then sub-cloned into pUC19 to create pSpAB*-2.
Although SpAB* was designed with its own Shine-Dalgarno ribosome binding site (Figure 1), high level expression was not achieved following sub-cloning into expression vector pkk223-3, so a gene fusion approach was used to increase expression.
The SpAB*-SpAB* sequence was linked to synthetically constructed genes encoding mutated and inactive bovine DNAase1 protein (Worrall and Connolly, 1990) to create two genes, the first encoding the first 81 amino acids of DNAase1 followed by a 12 amino acid spacer and then the 111 amino acids of SpAB*-SpAB* and the second
encoding the first 53 amino acids of DNase1 followed by a 12 amino acid spacer and then the 111 amino acids of SpAB*-SpAB*. The resulting plasmids containing the fused constructs cloned into the polylinker of pkk223-3 were termed p81-SpAB*-2 and p53-SpAB*-2 and their construction is shown in Figures 3 and 3a. Figure 4 shows the complete DNA and amino acid sequence of the fusion protein
p81-SpAB*-2 which consists of 204 amino acids and has a calculated molecular weight of 27.0 kDa. On induction of E. coli JM103
(p81-SpAB*-2 ) with 2mM IPTG, fusion protein 81-SpAB*-2
accumulates within the cell (Figure 5) such that it becomes the major cell protein, equivalent to at least 15% of total cell protein as estimated by gel scanning of a Coomassie Blue stained SDS-PAGE gel whereas the fusion protein 53_SpAB*-2 was expressed at a lower level. This is almost certainly an underestimate since only the protein moiety derived from the DNAase 1 gene is stained well by this reagent.
Construction of gene fusion plasmids
Gene fusion plasmids were constructed by purifying fragments of the SpAB*-2 gene from pSpAB*-2 and inserting these into
appropriately restricted pAW2 (Worrall and Connolly, 1991). Figure 3b summarises the construction of the plasmids encoding two IgG binding domains and 81 or 53 residues from the N-terminus of the DNase1.
p81-SpAB*-2 was created by ligation of the Kpn I-Pst I fragment of pSpAB*-2 into Kpn I-Pst I cut pAW2. p53-SpAB*-2 was constructed in the same way, using the XbaI-PstI restriction sites. The plasmids encoding fusion proteins with single SpAB* domains were constructed by digesting each respective plasmid with Bgl II, removing the released fragment and religating the shortened, linearised plasmid. Restriction analysis and DNA sequencing were performed to confirm the generation of recombinant DNA molecules. The complete
nucleotide/amino acid sequence of the encoded fusion protein 81-SpAB*-2 is shown in Figure 4 and its amino acid composition is shown in Table 1.
Table 1. Amino acid composition
parts.
DNase I 'INTER' SpAB*-SpAB* TOTAL
Ala 5 2 14 21
Cys - - 1 1
Asp 5 - 6 11
Glu 4 2 11 17
Phe 2 - 6 8
Gly 3 1 - 4
His 2 - 2 4
Ile 6 1 4 11
Lys 4 - 10 14
Leu 8 1 13 22
Met 2 1 1 4
Asn 5 - 15 20
Pro 2 1 6 9
Gln 2 - 11 13
Arg 7 2 2 11
Ser 6 - 7 13
Thr 3 1 - 4
Val 8 - - 8
Trp - - - - Tyr 7 - 2 9

81 12 111 204

Protein induction and inclusion body isolation
The expression of recombinant fusion proteins was induced by the addition of ImM IPTG (Northumbria Biologicals Ltd) to cultures of E. coli JM103 containing the appropriate plasmid which had reached an optical density of A600 0.7 - 0.9. Cell growth was at 37°C in a 251 batch fermentation vessel. Cells were harvested 4 hours post induction and were stored frozen at -20°C. Aliquots equivalent to 10g wet cell paste were thawed and resuspended in 30 ml 10 mM Tris-HCL pH.8.5, then treated with lysozyme at 0.1 mg/ml, stirring at 4°C for 10 minutes. Cells were disrupted by sonieation (4 × 20 second bursts at medium amplitude, MSE Soniprep 150) and the sonicate was treated with 10 μg/ml DNAase1 for 30 minutes on ice.
Further induction analysis (Figure 6) indicated that
81-SpAB*-2 was produced as an inclusion body within the cell, a theory which was confirmed by microscopy (data not shown). When whole cells of an induced culture are disrupted by sonieation and then subjected to low speed centrifugation, 81-SpAB*-2 is found
exclusively in the pelleted insoluble fraction (as shown in Figure 6, lane 3). This enabled a purification protocol to be developed involving repeated sonieation and washes in detergent to solubilise as much contaminating protein as possible away from the inclusion bodies. Urea is then used to solubilise 81-SpAB*-2, which is found to regain IgG binding activity on removal of urea by dialysis (see Materials and Methods Section for details). It was found that a similar pattern of results was obtained for the fusion proteins 81-SpAB*-2,
53-SpAB*-1 which were synthenised with one SpAB* domain and
53-SpAB*-2 and the inclusion bodies containing these proteins could be recovered by the procedure described above. Inclusion bodies were pelleted by centrifugation at 7.500 g for 15 minutes and were resuspended in 10 mM Tris-HCL pH 8.5, 1% v/v Triton X-100 detergent. The sonieation process was repeated to ensure that all cells had been disrupted, the sonicate was stirred at 4°C for 15 minutes then the inclusion bodies were pelleted by centrifugation as before. Two washes were performed in this way, then the pellet was washed x 2 in 10 mM Tris-HCL pH 8.5 containing 1 M urea. Solubilisation of the fusion protein was achieved by extraction in 10 mM Tris-HCL pH 8.5 containing 2.5 M urea at 4°C for 1 hour. The solution was spun at 27,000 g for 20 minutes and the supernatant was retained. A further extraction in 10 mM Tris-HCL pH 8.5 containing 4 M urea was performed on the pellets and again the supernatant was retained. The urea was dialysed away against 2 × 4 litres 20 mM KP buffer pH 8.0, and aliquots of the resulting protein solution were freeze dried.
The observation that the level of expression of the fusion proteins is dependent upon the number of residues from the DNase 1 included in the fusion protein is significant and possibly explained by close examination of the 3 dimentional structure of DNase1 (Suck et al, 1984). The amino terminal 80-85 amino acid residues appear to exist as a domain distinct from the remainder of the protein. This domain includes several secondary structural features. Two a-helices (I and II, see Suck et al, 1984) consisting of residues 18-29 and 42-54 respectively are present, and six B-strands (A,B,C,D,E and F) four of which (A,C,E and F) form a B-pleated sheet and the other two form a parallel B-pleated sheet with each other. Thus it is possible that the N-terminal 81 residues of 81-SpAB*-2 or 2 may fold into the same. stable tertiary structure as in DNase 1. The first 53 residues of DNase 1 (used in 53-SpAB*-1 or 2) also contain helices I and II but contain only four B-strands (A,B,C and D). This truncated sequence may still fold into its 'native' structure but will be less stable having lost possible hydrogen bonds between strands C and F.

The existence of the complete and thus presumably more stable N-terminal domain of DNase 1 in 81-SpAB*-1 or 2 may act as a nucleus for protein folding events and hence maylead to a compact fusion protein, better protected from proteolysis.
Protein Analysis
Protein concentration was estimated by the absorption of light at 280nm, (E280=10,800). An alternative method for protein concentration estimation is the bicinchoninic acid protein assay of Smith et al (1985) (Sigma).
Determination of IgG-binding activity
The interaction of 81-SpAB*-2 and 81-SpAB*-1 with Swine anti-sheep IgG coupled to Horse radish peroxidase was determined by ELISA experiments. In the standard method 81-SpAB*-2 was shown to have the same affinity for this IgG species as whole Protein A from Stapkylococcus aureus (figure 7). It was found, however, that protein 81-SpAB*-1 interacts approximately 100 fold more weakly than either SpA or 81-SpAB*-2. Both IgG binding domains of 81-SpAB*-2 are functional since immunoprecipitates are formed when 81-SpAB*-2 is mixed with IgG. These precipitates can be detected either by observation of an increase in light scattering or by using Ouchterlony plates. Light scattering experiments used to monitor the interaction between IgG binding proteins and IgG were performed using a Perkin Elmer 650S Spectrofluorimeter with an incident and emission wavelength of 320nm.

IgG binding activity was quantified using an Enzyme Linked Immunoabsorbent Assay (ELISA) technique modified from Hudson and Hay (1980). Serial dilutions of protein in 50mM sodium carbonate buffer, about pH 9.0 (coating buffer) were used to coat the wells of a microtitre plate at 37°C for at least two hours. Wells were then washed three times in 0.1% Tween 20 in phosphate buffered saline, pH 8.2, (PBS-Tween)) before incubation at room temperature with 100μl Swine anti-sheep-horse radish peroxidase conjugate (Serotec, UK) diluted to 1μg per ml in the same buffer. After 30 min the wells were washed three times again with PBS-Tween, then 200μl of substrate was added to each well (0.35 mg/ml O-phenylene diamine, and 0.1% v/v H2O2 in 0.1M sodium citrate-phosphate buffer pH 5.0). The
reaction was stopped after 15 min by the addition of 50μl per well of 12.5% H2SO4, and the absorbance read at 495nm.
Competitive ELISA Experiments
IgG from other species was also shown to be bound by 81-SpAB*-2 by using competitive ELISA techniques. For competitive ELISA all wells were coated with 200ng of 81-SpAB*-2 in sodium carbonate buffer pH 9.6 as above. Serial dilutions of the test antibody in PBS-Tween were then made and allowed to bind to the 81-SpAB*-2 coating the wells for about 10min. 50μl of Swine
anti-sheep-horse radish peroxidase conjugate was then added (2μg per ml in PBS-Tween) and the rest of the ELISA technique carried out as described above.

Experiment 1
The results of experiment 1, summarised in Table 2 demonstrate that 81-SpAB*-2 binds IgG with a similar species
specificity as whole Protein A. The affinity for the IgG decreases in the following order:
Guinea Pig/Human/Pig/Mouse/Rabbit/Cow/Horse/Rat/Chicken/and Goat.
Table 2. Inhibition of 81-SpAB*-2 binding to peroxidase
conjugated swine-anti-sheep IgG from different species.
The figures given represent the amount (ng) of
competing antibody required to inhibit binding of the
test antibody by 50%, according to the conditions of
the ELISA described (see Materials and Methods).
Amount of competing
antibody required to
Source of IgG give 50% inhibition
ng
Guinea pig < 20
Human 40
Pig 50
Mouse 50
Rabbit 60
Cow 320
Rat 3200
Horse 4000
Chicken > 5000
Goat > 5000
A further comparison was carried out as follows.
Experiment 2
Experiment 2 used an identical procedure to Experiment 1, except that the serial dilutions of the test antibody in PBS-Tween were allowed to bind to the 81-SpAB*-2 coating the wells of the microtitre plate for 15 minutes, instead of about 10 minutes. The results of this experiment are shown in Table 3 .

Table 3
ng of immunoglobulin giving 50% inhibition

Source of IgG SpA SpAB*-2

Goat 2,500 6,000
Rat 2,250 5,600
Cow 900 400
Mouse 150 70
Pig 20 65
Guinea Pig 40 50
Human 10 25
Rabbit 10 22

The affinity for the IgG decreases in the following order from this experiment as follows:
Rabbit/Human/Guineapig/Pig/Mouse/Cow/Rat/Goat.
Horse and Chicken IgG were not tested in this experiment.
This experiment also demonstrates that 81-SpAB*-2 binds IgG with a similar species specificity as whole Protein A.

Observation of the formation of SpAB* - IgG complexes
Precipitation of SpAB -IgG complexes can only occur if both participants have two or more functional sites for the other. Nephelometry measurements were made in order to determine whether both IgG binding domains of our construct 81-SpAB*-2 were able to bind IgG. Figure 8 illustrates the changes in the light scattering pattern from a cell containing 81-SpAB*-2 on addition of isolated Fc
fragments or whole IgG (human or chicken). A large precipitation indicated by an increase in light (320nm) scattering at 90° occurs if 81-SpAB*-2 is mixed with 5 fold excess of Fc or human IgG wheareas only slight precipitation occurs when human IgG is mixed with
81-SpAB*-1 No precipitation occurs if chicken IgG is added to
81-SpAB*-2 human fc is added to 81-SpAB*-1 or any immunoglobulin or Fc is added to a solution of DNAasel. Thus it appears that both IgG binding sites in 81-SpAB*-2 are functional in the presence of human Fc or IgG but that there is either no affinity for IgG from chicken or that the complexes formed remain soluble. Evidence from competitive ELISA experiments described above suggest that the former is the case. The lack of formation of any precipitate with DNase 1 and IgG removes the possibility that the DNase 1 part of the fusion is involved in any binding phenomenon to either Fab or Fc. The slight increase in light scattering observed when 81-SpAB*-1 is mixed with human IgG must arise from possible interactions involving the Fab (but not antigen binding site) of the IgG since no such event occurred when Fc or chicken IgG was used. The possibility of a recognition site for SpA on the Faba section of IgG has been noted before (Langone, 1982, Lindmark, 1983). Our studies suggest that such recognition is to part of the SpAB domain.
SDS-PAGE
SDS-PAGE was performed according to the method of Laemmli (1970).

Electrophoretic Transfer of Protein from SDS-PAGE Gel to PVDF membrane

The electrophoretic transfer of proteins from SDS-PAGE gels to membranes was performed essentially as described by Towbin et al (1979). However, a PVDF membrane was used in preference to
nitrocellulose, and CAPS buffer was used as the transfer buffer
(Matsudaira, 1987). Electroelution was carried out for 2 hours at a constant 45 volts. For total protein staining the membrane was immersed in Coomassie Blue stain for 2-5 min, then destained as for SDS-PAGE gels. For the qualitative detection of IgG-binding activity of protein bands electro-blotted onto PVDF membrane, the membrane was blocked with 3% gelatine in PBS for 5-10 min, then the IgG-HRP conjugate was added (diluted 1:5,000 in 1% gelatine). The solution was agitated at intervals for 20-30 min, then the solution was discarded and the filter was washed 5 times in 0.1% v/v Tween 20 (in phosphate buffered saline (PBS) -Tween) and once in PBS.
Chloronaphthol solution was added to cover the filter along with H2O2 (to 0.1% v/v) and the solution was agitated. A purple/blue colour is indicative of IgG-binding activity.
Heat denaturation studies on 81-SpAB*-2 have shown that the protein is extremely resilient, and any denaturation caused by heating up to 85°C for periods up to 30 min was found to be completely reversible with no loss of activity being detected in by ELISA carried out at room temperature or 37°C after recooling the heat treated 81-SpAB*-2. However, the protein is extremely susceptible to ienaturation and precipitation from solution if subjected to very mild acidic conditions. The protein precipitates from solution when the pH is 6.2 or lower. This provides a simple method for removing the protein from solution under extremely mild conditions. The
precipitated 81-SpAB*-2 can be refolded to produce active protein by dissolving the precipitate in 4M urea and removing the urea by dialysis.
PRODUCTION OF MUTANTS

Mutants were produced by site-directed mutagenesis in which the amino acid residues designated 17Phe and 18Tyr (see below) were replaced as follows:
17Phe —> 17Tyr
18Tyr —> 18Glu
18Tyr —> 18Phe
Tyr —> His
18Tyr —> 18Lys
18Tyr —> 18Trp
18Tyr —> 12Cys
Ala Pro Lys Ala Asp Asn Lys Phe Asn Lys
2
Glu Gln Gln Asn Ala Phe Tyr Glu Ile Leu
12 17 18
His Leu Pro Asn Leu Asn Glu Glu Gln Arg
22
Asn Ala Phe Ile Gln Ser Leu Lys Asp Asp
32
Pro Ser Gin Ser Ala Asn Leu Leu Ala Glu
42
Ala
52 Figure 9 is a helical wheel representation of the amino acid residues in the SpAB domain shown by X-ray crystallographic studies of Diesenhofer (1981) to fall into two helical secondary structural motifs. Residue 18 in the helix closest to the amino terminus of the protein has been implicated to be essential for binding to the Fc of IgG and possibly to participate in a hydrogen bond (ring hydroxyl group) with the carbonyl group of the peptide bond formed between residue 432 and 433 of the IgG. We have examined the requirement for this tyrosine residue and the effect of the disruption of the proposed hydrogen bond. Mutations described herein have therefore all been made at this position numbering 111 and 169 in the fusion protein with two IgG binding domains. Three mutations are described in
81-SpAB*-2, both domains having the same residue altered. Firstly, Tyr 111, 169 to Phe 111, 169 replacements, the most conservative of all, remove the possibility of the intermolecular hydrogen bonds to two IgG molecules. Secondly, the Tyr was replaced in each domain by Trp to examine the possibility of accomodating a bulkier side chain. Finally, the Tyr was replaced by Glu to assess the effect of charge at this position in each domain. These mutants were termed
81-SpAB*-2(Y111F, Y169F), 81-SpAB*-2(Y111W, Y169W) and
81-SpAB*-2(Y111E, Y169E) and the interactions between them and porcine IgG-HRP conjugate were determined by the modified ELISA technique described above.
Figure 7 shows the level of porcine IgG-HRP activity retained per well containing various amounts of each IgG binding protein. The data demonstrate clearly that the residue Tyr 111 (and 169) may be replaced by both Phe 111 (and 169) or Trp 111 (and 169) without severe disruption of the binding interactions. Analysis of such binding data shows that replacement of Y111, Y169 with Fill, F169 causes a three fold decrease in affinity for the IgG, this being equivalent to a loss of 0.6kcal per mol of binding energy due to the loss of the intermolecular hydrogen bond per domain. The replacement by the bulkier amino acid tryptophan decreases the affinity of the protein for IgG by a factor of two suggesting that the loss of the hydrogen bond is compensated by an increase in other favourable interactions, presumably hydrophobic. In contrast, the presence of a negative charge at this position, accomplished by the replacement of the Tyr residue in each domain by Glu, appears to almost completely destroy the interactions between the two proteins. This data suggests that the loss of binding of SpAB to IgG after nitration of the Tyr residue is not due solely to the loss of the hydrogen bond or to steric effects (Sjoholm et al, 1973) since the Phe and Trp mutants described above still possess reasonable binding activity. DNase 1 alone does not give any positive signal in ELISA tests.
In a further experiment, mutants with Tyr18 replaced by other amino acids were made by digestion of the plasmid 81-SpAB*-2 with MlaI and Bgl II to release a short fragment encoding residues

17-19 inclusive. This was replaced by short synthetic
oligonucleotides bearing single amino acid substitutions in position

18, coding for Phe, Glu, His or Lys (see Table 4) but having the same cohesive ends. The manipulations necessary to generate a gene encoding the same fusion protein but bearing identical mutations in the two IgG binding domains have been described elsewhere (Popplewell, 1991). In order to demonstrate that the IgG binding domains in the fusion protein behave as those in native Protein A both were compared for IgG binding ability by the modified ELISA protocol described above. The results in Fig. 10 demonstrate that the non-mutated protein 81-SpAB-2has a very similar affinity for IgG as Protein A in the pH range 6.0 to 8.0 although it has a three fold lower affinity at pH 5.0. This latter difference arises due to an instability of the fusion protein at this low pH.
The data displayed in Fig. 11 shows the amount of porcine IgG-HRP conjugate bound to various amounts of fusion proteins in the wells of a standard microtitre plate. The mutant Y111H,Y169K shows less binding, approximately 8% of that of the native protein and the mutant Y111E.Y169E shows virtually no IgG binding under these conditions. However, the interaction of all three proteins with Porcine IgG-HRP was found to be very sensitive to pH unlike the native protein. At pH values where the replaced residue would be expected to have a charge, the binding is less strong than under conditions where the equilibrium between charged and uncharged species lies towards the uncharged side. The mutant Y111E,Y169E therefore shows maximal binding at pH 4 and minimal binding at pH 6 or above i.e. at pH values where the carboxyl group of the side chain has a negative charge. In contrast, both the mutants Y111H,Y169H and Y111K,Y169K show the increased binding of IgG-HRP at higher pH where both side chains become less protonated. Significantly the apparent pK (6.5) of the binding curve shown by Y111H,Y169H is lower than that shown by
Y111K,Y169K (7.4), as would be expected. Table 4 gives the percentage of IgG binding shown by each mutant compared with the non-mutated construct at two pH values where the binding is minimal or maximum. It can be seen that the Glu mutant has a maximum binding of only 10% of the 'native' protein at pH 4.0 whereas much higher relative binding is obtained for the Lys (20%) or His (50%) mutants at pH 9.0.
SUMMARY

From the foregoing it can be seen that the present invention has successfully overcome the problems of the prior art. Recombinant DNA techniques are provided for the production of polypeptides having between 2 and 4 modified IgG binding domains which allow high levels of expression in E coli without incurring proteolysis by host enzymes or difficulties in purification. Said binding domains are highly amenable to site-directed mutagenesis and therefore enable the production of immunoglobulin-binding proteins having distinct advantages compared to Protein A. Examples of such mutated proteins are given and their properties investigated.

Table 4

Mutant

Construct: oH Relative Binding Native

Y111, 169 4.0 100%
(native)
9.0 100%

Y111H, Y169H 4.0 15%
9.0 50%

Y111K, Y169K 4.0 <5%
9.0 23%

Y111E, Y169E 10%
9.0 <0.5%