- Email: [email protected]
The interaction between different domains of staphylococcal protein a and human polyclonal IgG, IgA, IgM and F(ab')2: Separation of affinity from specificity
0161-5890/93$6.00+ 0.00 Pergamon Press Ltd
Immunology, Vol 30, No. 14, pp. 1279-1285, 1993 Printed in Great Britain.
Molecular
BETWEEN DIFFERENT THE INTERACTION STAPHYLOCOCCAL PROTEIN A AND HUMAN IgA, IgM and F(ab’),: SEPARATION OF AFFINITY
DOMAINS OF POLYCLONAL IgG, FROM
SPECIFICITY
ULLA K. LJUNGBERG,*§ BIRGER JANSSON,~ ULF NISS,* RUNE NILSSON,* BENGT E. B. SANDBERG* and BJ~RN NILSSON*~~ *Department of Immunology, Excorim AB, Box 10101, S-220 10 Lund, Sweden; and tDepartment of Biochemistry, Royal Institute of Technology, S-100 44 Stockholm, Sweden (First received I1 January 1993; accepted in revised form 14 April 1993)
Abstract-Binding properties of staphylococcal protein A (SPA) to different human immunoglobulins have been investigated. In this analysis, intact SpA as well as SPA-derived fragments containing one to five IgG-binding domains of different compositions, were used. The affinity binding constants of the different proteins to human polyclonal IgG, IgA, IgM and F(ab’),-fragments as well as their binding capacity to the immunoglobulin molecules were determined. The results show that although all the proteins bound to IgG, regardless of size or composition, the binding strength differed significantly. Proteins containing five domains have a stronger affinity for IgG than those containing one or two. There were no marked differences in binding strength between different domains. However, the binding ability to IgA and IgM showed a marked difference between the various SPA-derived proteins of different compositions. This discrepancy was correlated to differences in their relative binding properties to isolated F(ab’),-fragments of IgG. Hence, we conclude that the binding affinity is mainly affected by the number of domains, whereas the binding specificity is to a large extent determined by which domains are selected.
The gene encoding SpA has been cloned (LGfdahl et al., 1983) and sequenced (UhlCn et al., 1984). The analysis of the gene revealed that SpA contains five putative domains, designated E, D, A, B and C, respectively. Each domain was shown to be capable of binding IgG (Moks et al., 1986). A high degree of amino acid homology (average z 80%) exists between the A, B, C and D domains (Uhlen et al., 1984), whereas the E domain is more diverged. By recombinant methods, SPA-derived proteins containing different number and composition of domains have been produced (Moks et al., 1986; Nilsson et al., 1987). In this study, we have analysed the IgG, IgA, IgM and F(ab’), binding properties of intact SpA and some SPA-derived protein fragments. The results suggest that the binding is affected both by the number of immunoglobulin binding domains and by the composition of the domains.
INTRODUCTION
Protein A (SPA) from Staphylococcus aureus plays an important role in immunological methods due to its specific interaction with immunoglobulins. SpA is widely used for that purpose in a variety of immunochemical techniques (Langone, 1982a), and as a fusion partner to facilitate immobilization and purification of recombinant proteins (Nilsson and Abrahms&, 1990; Nilsson et al., 1991). It has also been successfully used clinically in extracorporeal immunotherapy (Bygren et al., 1985; Palmer et al., 1989). SpA interacts with immunoglobulins in a non-immune manner (Forsgren and Sjiiquist, 1966; Kronvall and Williams, 1969). The most extensive studies have been performed on the binding of SpA to the Fc-portion of immunoglobulin G (IgG) (for a review, see Langone, 1982b). The three-dimensional structure of domain B of SpA in a complex with Fc has been solved by X-ray crystallography (Deisenhofer, 1981). In addition to its IgG-binding, SpA has been shown to bind to IgA and IgM (Harboe and FGlling, 1974). It has been suggested that this interaction involves, not the Fc-part of the immunoglobulin, but rather the Fab-part (Ingantis, 1981). Although it has been shown that the SPA-binding site to Fab is different from the antigen binding site (ZikHn, 1980), the molecular structure of the SPA-Fab interaction is unknown.
MATERIALS Proteins
fPresent address: Department of Structural Biology, Kabi Pharmacia Bioscience Center, S-l 12 87 Stockholm, Sweden. GCorresponding author.
AND METHODS
Purified staphylococcal protein A (SPA) was obtained from Pharmacia LKB Biotechnology, Sweden (cat. No. 17-0770-01). Protein Vl, which is a truncated version of SpA (Guss et al., 1985), was kindly provided by Dr J. Sjaquist at the Biomedical Center, Uppsala, Sweden. The recombinant proteins Z-V, ZZ, Z and EB were produced in Escherichia coli by recombinant methods by growing strains transformed with the plasmids pEZV, pEZZ (Nilsson et al., 1987), pEZT (Nygren et al., 1988) and pASEB (Moks et al., 1986), respectively.
1279
U. K. LJUNGBERG et al.
1280
Recombinant proteins were purified by IgG affinity chromatography (Nilsson and Abrahmsen, 1990) using IgG Sepharose FF (Pharmacia LKB Biotechnology, Sweden). Polyclonal human IgG was obtained from Sandoz, Switzerland or Kabi, Sweden, polyclonal human serum IgA from Jackson, U.S.A. (cat No. 009000-01 l), and polyclonal human IgM from Calbiochem, U.S.A. (cat No. 401107). Zodination of proteins
Proteins were iodinated as described by Bolton and Hunter (1973). Protein (5 pg) was added to 0.2 mCi”‘Ilabelled Bolton-Hunter reagent (Amersham, U.K., cat. No. IM.5861) and stirred for 60 min, following the procedure described by Langone (1980). ‘251-proteins were separated from free or inactivated reagents by gel filtration on a Sephadex G-25 M column (Pharmacia LKB Biotechnology, cat. No. 17-0851-01). Specific activities of all different ‘2SI-labelled proteins were about 300 Ci/mmol. Preparation
and ptkjication
teins. Individual wells were counted for radioactivity in an LKB gamma counter (1 mimsample). The competitive method was used to determine the capacity of unlabelled proteins to inhibit the binding of ‘2SI-labelled proteins to immobilized immunoglobulins. A fixed amount of labelled proteins was mixed with serial dilutions of unlabelled proteins. The mixture of labelled and unlabelled proteins was diluted in PBS-Tween and 50 ~1 was added to each well. The wells were incubated, washed and counted for radioactivity as described for the saturation experiment. Data analysis
The data obtained from the binding assays were analysed with the Biosoft computer program package (Munson and Rodbard, 1980; McPherson, 1985). Using the included EBDA and LIGAND programs, the affinity constants and number of binding sites for the different proteins to the immunoglobulins were determined.
RESULTS
of F(ab’),-fragment
Human polyclonal IgG (100 mg) was dissolved in 10ml of 20mM sodium acetate buffer, pH 3.0 and preincubated for 1 hr at room temperature. Thereafter, 5 ml of prewashed (4 x 10 ml 20 mM sodium acetate, pH 3.0) immobilized pepsin (Pierce, U.S.A., cat. No. 20343) was added and the mixture was stirred for 3 hr at 37°C. The proteolytic digestion of IgG was terminated by separating the gel using a sintered glass filter. The pH of the filtrate was adjusted to 6.5. F(ab’),-fragments were separated from intact IgG, Fc and low molecular weight fragments by using a TSK-G3000 SW HPLC column (Pharmacia LKB Biotechnology, cat. No. 2135-360). A final yield of 25 mg F(ab’)2 was collected. The purity was confirmed by SDS-polyacrylamide gel electrophoresis followed by silver staining. Contamination of undigested IgG was shown to be less than 1% and no traces of Fc-fragments could be detected.
In this study, the binding capacities and affinities of six different variants of SpA to different human immunoglobulins were determined. Linear representations of the SpA variants are shown in Fig. 1 and the proteins are described in Table 1. In contrast to intact SpA and Vl, which are natural products from specific strains of Staphylococcus aureus, Z-V, ZZ, EB and Z are products obtained by recombinant methods. EB consists of two IgG-binding domains which were fused at the DNA level (Moks et al., 1986). Z is derived from domain B by substituting the Asn-Gly dipeptide sequence with Asn-Ala (Nilsson et al., 1987). This substitution is expected not to have any effect on Fc binding (Nilsson et al., 1987; Deisenhofer, 1981). Z-V, ZZ and Z are multiplicities of domain Z (Fig. 1). Intact SpA and Z-V S~A)EIDIAIBICI
Binding assays
Removable flat-bottomed wells (Dynatech Laboratories, U.S.A., cat. No. 01 l-010-6302) were coated overnight at 4°C with 50 ~1 of immunoglobulins suspended in PBS (10 mM potassium phosphate buffer pH 7.4 and 150 mM NaCl), containing 5 pg/ml IgG, lOpg/ml IgA, lOpg/ml IgM or lOpg/ml F(ab’),, respectively. Before addition of ‘2sI-labelled proteins, the wells were washed repeatedly with PBS-Tween (PBS containing 0.05% Tween 20). Two methods were used to study binding: saturation experiments and competitive inhibition. For the saturation study, 50 ~1 of increasing amounts of ‘251-labelled protein, suspended in PBS-Tween, was added to each well. As a control of non-specific protein binding, some wells received labelled protein mixed with 1000 times excess of unlabelled protein. All wells were incubated for 3 hr at room temperature under agitation, then repeatedly washed with PBS-Tween to remove unbound pro-
x I
I
Protein Engineering
z-v)zlzlzl;Iz( Fig. 1. Schematic presentation of the protein A-derived proteins SpA. Vl, EB, Z, ZZ and Z-V. E, D, A, B and C refer to the different IgG-binding domains of SpA. X is the cell wall-binding domain of SpA. Z shows the slightly modified domain based on domain B.
Interaction
of protein A with immunoglobulins
1281
Table 1. Origin and characteristics of proteins used in the study Protein SPA
Vl
z-v,
EB
zz, z
Origin and characteristics
Reference(s)
Originates from Stuphylucoccu~ UhlCn ef al. (1984) aureus strain U511; contains five IgG-binding domains: ED-A-B-C and a region X which is in the cell wall; IV, z 50,000. Originates from a u.v.-induced Movitz et al. mutant of strain Cowan I; con(1979); tains two intact domains (E, D) Guss ef al. (1985) and one truncated (domain A); i%f,z 14,000. Protein A fragments obtained Nilsson et al. by protein engineering; contain (1987) 5, 2 or 1 identical IgG-binding domains based on the B-domain of protein A; M, (ZV)= 35,800; M,(ZZ)e 14,500; M,(Z)= 7000. Protein A fragments obtained Moks et al. (1986) by recombinant methods; contains 2 IgG-binding domains based on the E- and B-domains of protein A; &f, z 13,200.
contain five IgG-binding domains with different compositions. Protein Vi contains two intact IgG-binding domains (E and D) and one truncated (domain A) (Guss et al., 1985). The truncated portion of domain A spans the two cr-helices involved in binding and a similar peptide fragment based on domain B was recently shown to be sufficient for IgG-binding (Jansson et al., 1990). However, the binding strength may be lowered in the incomplete A-domain of Vl compared to the intact domain since it was shown that the peptide fragment of domain B was si~ificantly destabilized. The binding of SpA, VI, Z-V, ZZ, EB and Z to IgG, IgA, IgM and F(ab’), was first investigated in a saturation assay. In this study, ‘2sI-labelled proteins and immobilized immunoglobulins were allowed to interact for 3 hr under agitation at room temperature. These incubation conditions were necessary to reach equilibrium (data not shown). Figure 2 shows the obtained saturation curves for intact SpA and the different SpAderivates binding to the various immunoglobulins. Nonspecific binding was less than 0.5% and thus negligible. The most obvious conclusion from the experiment is that whereas the saturation level of all proteins binding to IgG is within the same range, there is a discrepancy in the binding to IgA, IgM, and F(ab’)* between intact SpA and VI on one side and Z-V, ZZ, EB and Z on the other. SpA and Vl show a high level of IgA, IgM and F(ab’), binding, the other proteins hardly bind to these immunoglobulins at all. These results are further presented in Table 2. Here, the data obtained from the saturation experiment were analysed using the LIGAND computer program (Munson and Rodbard, 1980; McPherson, 1985) and the number of accessible binding sites for the proteins on the immunoglobulins were determined. In order to simplify the comparison, the number of sites for
each protein on one specific immunoglobulin was normalized to the number of sites for intact SpA, defined as an arbitrary number of 100. Comparing the data of Table 2, the binding specificity difference is clearly seen. Whereas Z-V, ZZ, EB and Z show similar number of binding sites on IgG compared to SpA, significantly fewer sites are displayed on IgA, IgM and F(ab’),. Using the LIGAND computer program, the affinity constants for the interactions presented in Fig. 2 were determined (Table 3). Intact SpA has a strong binding to all immunoglobulins tested. The five-domain protein Z-V shows a high binding constant to IgG. However, its binding strength to IgA, IgM and F(ab’), is much lower. The smaller proteins, VI, ZZ, EB and Z, have five to ten times weaker affinity to IgG than SpA and Z-V. Under certain conditions two types of interactions were seen between the proteins and immunoglobulins, one high and one low (the low affinity was in the lO’/M range), demonstrating some kind of heterogeneity. The significance of this finding is unknown and hence, only the dominant affinity constants are presented. The binding properties of SPA, Vl, Z-V, ZZ, EB and Z were further investigated in a competitive inhibition assay. All six unlabelled proteins were allowed to compete with ‘251-labelled SpA for the sites on immobilized IgG, IgA, IgM or F(ab’)z (Fig. 3). A fixed amount of labelled SpA (4.0 x lo-l4 moles) was mixed with serial dilutions (‘log) of unlabelled SpA, Vl, Z-V, ZZ, EB and Z, starting with 500-fold excess. At 100% inhibition no ‘251-labelled SpA was bound to the immunoglobulins. The competition analysis shows that for IgG (Fig. 3A), SpA and Z-V are better inhibitors of ‘2SI-SpA binding than Vl, ZZ, EB and Z. Even at 500-fold excess, these smaller proteins could only inhibit binding to 50-75%. The binding discrepancy in the saturation curves for
1282
U.K. LJUNGBERG
A
ez al.
C
l
e
3
3m
g ?
200
ii!
100
B
Added @mole)
Added WnOlel l
-0
0 5
0
5
Adciid
15 fpmole)
Fig. 2. Saturation curves for different 1z51-lab~ll~dproteins binding to immobilized IgG (A), IgA (B), IgM (C) and F{ab’), (D). Increasing amounts of SpA (O), VI (a), Z-V (r]), ZZ (m), EB (a) and Z (A) were added. Under the conditions used, non-specific binding was less than 0.5% of specific binding (data not shown).
IgA, IgM and F(ab’), (Fig. 2B-D) was also clearly shown in the competition curves (Fig. 3B-D). Binding of ‘251-labelled SpA to these immunoglobulins was only inhibited by SpA itself and to a lesser degree by Vl. The other proteins competed very poorly. The affinity constants were determined using the LIGAND computer program. Since difference in binding valencies might obscure the analysis of different proteins when competing with each other, only data obtained from competitive inhibition where unlabelled SpA competed
‘2SI-labelled SpA, unlabelled Vl competed with ‘2SI-labelled Vl, etc. (data not shown), were used in the binding constant calculations. The result is presented in Table 3 and compared with the affinity constants obtained from the saturation curve experiments. There is good agreement between the two experiments. The low binding occupancy of ZZ, EB and Z to IgA, IgM and F(ab’& made the calculation of their affinity constants from the competitive inhibition data impossible. with
DISCUSSION
Table 2. Relative number of binding sites for SpA, Vl, Z-V, ZZ, EB and Z to IgG, IgA, IgM and F(ab), Relative number of binding sites (%) Protein
IgG
IgA
IgM
F(abX
SPA Vl z-v zz EB Z
100 111 89 123 128 131
loo 109 6 6 4 6
100 133 11 3 5 1
100 110 17 20 22 23
The maximal number of binding sites was determined by LIGAND computer analysis of the saturation curves of Fig. 2. Only sites correlated with a high affinity constant were regarded. The relative number was obtained after normalizing all values to the number of binding sites for SPA. Data shown are the outcome of one specific experiment, though reflecting a general obtained result.
AND CONCLUSIONS
We have analysed the non-immune interactions between SpA and human polyclonal IgG, IgA, IgM and F(ab’),. The aim of this study was to compare intact SpA and SPA-derived proteins containing different number of immunoglobulins-binding domains and different composition of domains. Determination of the aflinity constants indicates that the number of IgG-binding domains greatly affects the binding strength; the fivedomain proteins SpA and Z-V have a relatively high binding constant when compared to the shorter proteins. In the analysis, saturation and competitive inhibition curves gave very similar results (Table 3). The number of accessible binding sites on the immunoglobulins for the various proteins have been determined. The result from the saturation study (Fig. 2, Table 2) shows a pronounced discrepancy in specificity between SpA and Vl on one side and Z-V, ZZ, EB and 2 on the other. Whereas SpA and VI bind to all immunoglobulins tested, Z-V, ZZ, EB and Z bind poorly to IgA, IgM and F(ab’),. An additional reflection of
Interaction
1283
of protein A with immunoglobulins
Table 3. Affinity constants of SpA, Vl, Z-V, ZZ, EB and Z bound to IgG, IgA, IgM and F(ab’), determined by saturation and competitive inhibition studies Affinity constants (x IO-* M-‘)
Protein
Sat.
SPA Vl z-v zz EB Z
26 5.7 30 4.8 3.9 2.2
1gG Comp. 25 6.1 20 4.3 2.2 2.8
Sat.
IgA Comp.
31 5.9 12 1.3 0.3 0.3
21 4.3 5.4 nd nd nd
Sat. 31 6.1 12 3.6 2.8 1.9
IgM Comp. 32 4.9 7.1 nd nd nd
F(ab’), Sat. Comp. 10 2.2 6.3 0.7 0.5 0.1
8.8 2.3 6.8 nd nd nd
“nd = not determined. The saturation curves of Fig. 2 were transformed to Scatchard plots using the LIGAND computer program and the affinity constants were determined. Data from competitive inhibition of 12SI-labelled SpA, VI, Z-V, ZZ, EB and Z with the respective unlabelled protein were likewise analysed. The mean values of three to five experiments are presented.
Fig. 2 is that the maximal binding of intact SpA to IgG, IgA and IgM is quite different, hence the different scales on the Y-axes. The maximal binding level to IgA is rather low compared to that of IgG and IgM. This could be due to partial inactivation of immobilized IgA or, as suggested by Langone (19826), that a sub-population of serum IgA does not bind to SpA. What is the molecular reason for the discrepancy in binding specificity between SpA and Vl versus the other
proteins? The most characterized non-immune interaction between protein and immunoglobulin is the binding of SpA to the Fc-portion of the IgG molecule (Forsgren and Sjbquist, 1966; Kronvall and Williams, 1969; Deisenhofer, 1981). An additional binding site for SpA on IgG has been demonstrated in the Fab part of the immunoglobulin (Inganas et al., 1980) and it has been suggested that the interaction between SpA and IgA and IgM is mediated via a Fab-binding (Inganls,
Added (Slog)
Fig. 3. Competitive inhibition of binding of ‘*‘I-labelled SpA and different SPA-derived immobilized IgG (A), IgA (B), IgM (C) and F(ab’), (D). The labelled SpA was allowed with unlabelled SpA (O), Vl (a), Z-V (a), ZZ (m), EB (A) or Z (A). The amount SpA was kept at 4.0 x 10-‘4moles. Competing protein was diluted in steps of 7og, 2.0 x lo-” moles.
proteins to to compete of labelled starting at
U. K. LJUNGBERG et al.
1284
1981). As has been demonstrated in this paper (Fig. 2 and Table 2), the relative binding properties of SpA-derived proteins to isolated F(ab’),-fragments are similar to the binding properties to IgA and IgM. This implies that intact SpA and Vl are able to react with the immunoglobulins via the Fab-portion in addition to Fc. However, the other SPA-derived proteins bind to Fc, but the occupancy to IgA, IgM and F(ab’), is significantly lower (Table 2). Since SpA and Z-V both contain five IgG-binding regions, the discrepancy can not be explained by any differences in number of IgG-binding domains. Hence, the difference in binding specificity must depend on the type of domain included in the protein. As shown in Fig. 1, SpA contains five different domains called E, D, A, B and C (from N- to Cterminus). It has been shown that regions A, B and C are highly homologous. Region D contains an insertion of three amino acids and region E is the most diverged domain (Uhlen et al., 1984). Studies on separate domains have shown that all of them can interact with human polyclonal IgG (Moks et al., 1986). In contrast to SpA, protein Z-V contains five identical domains based on region B of intact SpA (Nilsson et al., 1987). Since Z-V, as well as ZZ and Z, show low binding occupancy to the F(ab’),-fragment of IgG (Table 2) this suggests that the A, D and/or E regions contain a structural motif, which is not present in Z, and which is necessary for Fab-binding. This suggestion is supported by the fact that Vl, containing the D, E and a truncated A region (Guss et al., 1985), shows full binding to F(ab’), in our analysis (Fig. 2 and Table 2). Since EB binds only poorly to IgA, IgM or F(ab’), this would imply that the addition of domain E to domain B is not sufficient for full Fab binding. Therefore, we propose that domain D possesses the ability of SpA to bind to Fab, since it possesses an interesting amino acid insertion compared to the other protein A domains, and this significant difference between this domain and the others could also be responsible for interacting with Fab. More studies using site specific mutagenesis and production of isolated domain D would be valuable to analyse the significance of this hypothesis. Acknowledgements-The
authors are grateful to Drs Mathias
Uhlen, Margareta Eliasson discussions. This project has Swedish National Board for corim AB. Dr John Sjiiquist of protein Vl.
and Tomas Moks for valuable been financially supported by the Technical Development and Exis acknowledged for the kind gift
REFERENCES Bolton A. E. and Hunter W. M. (1973) The labelling of proteins to high specific radioactivities by conjugation to a ‘*%containing acylating agent. Biochem. J. 133, 529-539. Bygren P., Freiburghaus C., Lindholm T., Simonsen O., Thysell H. and Wieslander J. (1985) Goodpasture’s syndrome treated with staphylococcal protein A immunoadsorption. Lancet ii, 1295-1296. Deisenhofer J. (1981) Crystallographic refinement and atomic models of a human Fc fragment and its complex with
fragment B of protein A from Staphylococcus aureus at 2.9and 2.8-A resolution. Biochemistry 20, 2361-2370. Forsgren A. and Sjijquist J. (1966) ‘Protein A’ from S. aureus. I. Pseudo-immune reaction with human y-globulin. J. Immunol. 97, 822-827. Guss B., Leander K., Hellman U., Uhlen M., Sjijquist J. and Lindberg M. (1985) Analysis of protein A encoded by a mutated gene of Staphylococcus aureus Cowan I. Eur. J. Biochem.
153, 579-585.
Harboe M. and Felling I. (1974) Recognition of two distinct groups of human IgM and IgA based on different binding to Staphylococci. Stand. J. Immunol. 3, 471482. Inganas M. (1981) Comparison of mechanisms of interaction between protein A from Staphylococcus aureus and human monoclonal IgG, IgA and IgM in relation to the classical Fey and the alternative F(ab’),c protein A interactions. Scand.
J. Immunol.
13, 343-352.
InganHs M., Johansson S. G. 0. and Bennich H. H. (1980) Interaction of human polyclonal IgE and IgG from different species with protein A from Staphylococcus uureus: Demonstration of protein-A-reactive sites located in the F(ab’), fragment of human IgG. Stand. J. Immunol. 12, 23-31. Jansson B., Palmcrantz C., Uhlen M. and Nilsson B. (1990) A dual affinity gene fusion system to express small recombinant proteins in a soluble form: Expression and characterization of protein A deletion mutants. Protein Engng 2, 555-561.
Kronvall G. and Williams R. C. (1969) Differences in anti-protein A activity among IgG subgroups. J. Immunol. 103, 828-833.
Langone J. J. (1980) ‘251-labelled protein A: Reactivity with IgG and use as a tracer in radioimmunoassay. Methods Enzymol.
70, 356375.
Langone J. J. (1982~) Use of labelled protein A in quantitative immunochemical analysis of antigens and antibodies. J. Immunol. Meth. 51, 3-22. Langone J. J. (19826) Protein A of Staphylococcus aureus and related immunoglobulin receptors produced by Streptococci and Pneumonococci. Adc. Immunol. 32, 157-252. Liifdahl S., Guss B., Uhlen M., Philipson L. and Lindberg M. (1983) Gene for staphylococcal protein A. Proc. natn. Acad. Sci.
U.S.A.
80, 697-701.
McPherson G. A. (1985) Analysis of radioligand binding experiments. J. Pharmacol. Meth. 14, 213-228. Moks T., Abraham& L., Nilsson B., Hellman U., Sjiiquist J. and Uhltn M. (1986) Staphylococcal protein A consists of five IgG-binding domains. Eur. J. Biochem. 156, 637-643. Movitz J., Masuda S. and Sjiiquist J. (1979) Physico- and immunochemical properties of staphylococcal protein A extracellularly produced by a set of mutants from Stuphylococcus aureus Cowan I. Microbial. Immunol. 23, 5160. Munson P. J. and Rodbard D. (1980) LIGAND: A versatile computerized approach for characterization of ligand-binding systems. Anal. Biochem. 107, 220-239. Nilsson B. and Abrahmsen L. (1990) Fusions to staphylococcal protein A. Methods Enzymol. 185, 144-161. Nilsson B., Forsberg G. and Hartmanis M. (1991) Expression and purification of recombinant insulin-like growth factors from Escherichia coli. Methods Enzymol. 198, 3-16. Nilsson B., Moks T., Jansson B., Abrahmsen L., Elmblad A., Holmgren E., Henrichson C., Jones T. A. and Uhlen M. (1987) A synthetic IgG-binding domain based on staphylococcal protein A. Protein Engng 1, 107-113. Nygren P.-A., Eliasson M., Palmcrantz E., Abrahmsen L. and Uhlen M. (1988) Analysis and use of the serum albumin
Interaction
of protein
binding region of streptococcal protein G. J. Mol. Recognition 1, 69-74. Palmer A., Welsh K., Gjorstrup P., Taube D., Bewick M. and Thick M. (1989) Removal of anti-HLA antibodies by extracorporeal immunoadsorption to enable renal transplantation. Lancer i, 10-12.
A with immunoglobulins
1285
UhlCn M., Guss B., Nilsson B., Gatenbeck S., Philipson L. and Lindberg M. (1984) Complete sequence of the staphylococcal gene encoding protein A. J. Biol. Chem. 259, 1695-1702. Zikan J. (1980) Interactions of pig Faby fragments with protein A from Staphylococcus aureus. Folia Microbial. 25, 246253.
Immunology, Vol 30, No. 14, pp. 1279-1285, 1993 Printed in Great Britain.
Molecular
BETWEEN DIFFERENT THE INTERACTION STAPHYLOCOCCAL PROTEIN A AND HUMAN IgA, IgM and F(ab’),: SEPARATION OF AFFINITY
DOMAINS OF POLYCLONAL IgG, FROM
SPECIFICITY
ULLA K. LJUNGBERG,*§ BIRGER JANSSON,~ ULF NISS,* RUNE NILSSON,* BENGT E. B. SANDBERG* and BJ~RN NILSSON*~~ *Department of Immunology, Excorim AB, Box 10101, S-220 10 Lund, Sweden; and tDepartment of Biochemistry, Royal Institute of Technology, S-100 44 Stockholm, Sweden (First received I1 January 1993; accepted in revised form 14 April 1993)
Abstract-Binding properties of staphylococcal protein A (SPA) to different human immunoglobulins have been investigated. In this analysis, intact SpA as well as SPA-derived fragments containing one to five IgG-binding domains of different compositions, were used. The affinity binding constants of the different proteins to human polyclonal IgG, IgA, IgM and F(ab’),-fragments as well as their binding capacity to the immunoglobulin molecules were determined. The results show that although all the proteins bound to IgG, regardless of size or composition, the binding strength differed significantly. Proteins containing five domains have a stronger affinity for IgG than those containing one or two. There were no marked differences in binding strength between different domains. However, the binding ability to IgA and IgM showed a marked difference between the various SPA-derived proteins of different compositions. This discrepancy was correlated to differences in their relative binding properties to isolated F(ab’),-fragments of IgG. Hence, we conclude that the binding affinity is mainly affected by the number of domains, whereas the binding specificity is to a large extent determined by which domains are selected.
The gene encoding SpA has been cloned (LGfdahl et al., 1983) and sequenced (UhlCn et al., 1984). The analysis of the gene revealed that SpA contains five putative domains, designated E, D, A, B and C, respectively. Each domain was shown to be capable of binding IgG (Moks et al., 1986). A high degree of amino acid homology (average z 80%) exists between the A, B, C and D domains (Uhlen et al., 1984), whereas the E domain is more diverged. By recombinant methods, SPA-derived proteins containing different number and composition of domains have been produced (Moks et al., 1986; Nilsson et al., 1987). In this study, we have analysed the IgG, IgA, IgM and F(ab’), binding properties of intact SpA and some SPA-derived protein fragments. The results suggest that the binding is affected both by the number of immunoglobulin binding domains and by the composition of the domains.
INTRODUCTION
Protein A (SPA) from Staphylococcus aureus plays an important role in immunological methods due to its specific interaction with immunoglobulins. SpA is widely used for that purpose in a variety of immunochemical techniques (Langone, 1982a), and as a fusion partner to facilitate immobilization and purification of recombinant proteins (Nilsson and Abrahms&, 1990; Nilsson et al., 1991). It has also been successfully used clinically in extracorporeal immunotherapy (Bygren et al., 1985; Palmer et al., 1989). SpA interacts with immunoglobulins in a non-immune manner (Forsgren and Sjiiquist, 1966; Kronvall and Williams, 1969). The most extensive studies have been performed on the binding of SpA to the Fc-portion of immunoglobulin G (IgG) (for a review, see Langone, 1982b). The three-dimensional structure of domain B of SpA in a complex with Fc has been solved by X-ray crystallography (Deisenhofer, 1981). In addition to its IgG-binding, SpA has been shown to bind to IgA and IgM (Harboe and FGlling, 1974). It has been suggested that this interaction involves, not the Fc-part of the immunoglobulin, but rather the Fab-part (Ingantis, 1981). Although it has been shown that the SPA-binding site to Fab is different from the antigen binding site (ZikHn, 1980), the molecular structure of the SPA-Fab interaction is unknown.
MATERIALS Proteins
fPresent address: Department of Structural Biology, Kabi Pharmacia Bioscience Center, S-l 12 87 Stockholm, Sweden. GCorresponding author.
AND METHODS
Purified staphylococcal protein A (SPA) was obtained from Pharmacia LKB Biotechnology, Sweden (cat. No. 17-0770-01). Protein Vl, which is a truncated version of SpA (Guss et al., 1985), was kindly provided by Dr J. Sjaquist at the Biomedical Center, Uppsala, Sweden. The recombinant proteins Z-V, ZZ, Z and EB were produced in Escherichia coli by recombinant methods by growing strains transformed with the plasmids pEZV, pEZZ (Nilsson et al., 1987), pEZT (Nygren et al., 1988) and pASEB (Moks et al., 1986), respectively.
1279
U. K. LJUNGBERG et al.
1280
Recombinant proteins were purified by IgG affinity chromatography (Nilsson and Abrahmsen, 1990) using IgG Sepharose FF (Pharmacia LKB Biotechnology, Sweden). Polyclonal human IgG was obtained from Sandoz, Switzerland or Kabi, Sweden, polyclonal human serum IgA from Jackson, U.S.A. (cat No. 009000-01 l), and polyclonal human IgM from Calbiochem, U.S.A. (cat No. 401107). Zodination of proteins
Proteins were iodinated as described by Bolton and Hunter (1973). Protein (5 pg) was added to 0.2 mCi”‘Ilabelled Bolton-Hunter reagent (Amersham, U.K., cat. No. IM.5861) and stirred for 60 min, following the procedure described by Langone (1980). ‘251-proteins were separated from free or inactivated reagents by gel filtration on a Sephadex G-25 M column (Pharmacia LKB Biotechnology, cat. No. 17-0851-01). Specific activities of all different ‘2SI-labelled proteins were about 300 Ci/mmol. Preparation
and ptkjication
teins. Individual wells were counted for radioactivity in an LKB gamma counter (1 mimsample). The competitive method was used to determine the capacity of unlabelled proteins to inhibit the binding of ‘2SI-labelled proteins to immobilized immunoglobulins. A fixed amount of labelled proteins was mixed with serial dilutions of unlabelled proteins. The mixture of labelled and unlabelled proteins was diluted in PBS-Tween and 50 ~1 was added to each well. The wells were incubated, washed and counted for radioactivity as described for the saturation experiment. Data analysis
The data obtained from the binding assays were analysed with the Biosoft computer program package (Munson and Rodbard, 1980; McPherson, 1985). Using the included EBDA and LIGAND programs, the affinity constants and number of binding sites for the different proteins to the immunoglobulins were determined.
RESULTS
of F(ab’),-fragment
Human polyclonal IgG (100 mg) was dissolved in 10ml of 20mM sodium acetate buffer, pH 3.0 and preincubated for 1 hr at room temperature. Thereafter, 5 ml of prewashed (4 x 10 ml 20 mM sodium acetate, pH 3.0) immobilized pepsin (Pierce, U.S.A., cat. No. 20343) was added and the mixture was stirred for 3 hr at 37°C. The proteolytic digestion of IgG was terminated by separating the gel using a sintered glass filter. The pH of the filtrate was adjusted to 6.5. F(ab’),-fragments were separated from intact IgG, Fc and low molecular weight fragments by using a TSK-G3000 SW HPLC column (Pharmacia LKB Biotechnology, cat. No. 2135-360). A final yield of 25 mg F(ab’)2 was collected. The purity was confirmed by SDS-polyacrylamide gel electrophoresis followed by silver staining. Contamination of undigested IgG was shown to be less than 1% and no traces of Fc-fragments could be detected.
In this study, the binding capacities and affinities of six different variants of SpA to different human immunoglobulins were determined. Linear representations of the SpA variants are shown in Fig. 1 and the proteins are described in Table 1. In contrast to intact SpA and Vl, which are natural products from specific strains of Staphylococcus aureus, Z-V, ZZ, EB and Z are products obtained by recombinant methods. EB consists of two IgG-binding domains which were fused at the DNA level (Moks et al., 1986). Z is derived from domain B by substituting the Asn-Gly dipeptide sequence with Asn-Ala (Nilsson et al., 1987). This substitution is expected not to have any effect on Fc binding (Nilsson et al., 1987; Deisenhofer, 1981). Z-V, ZZ and Z are multiplicities of domain Z (Fig. 1). Intact SpA and Z-V S~A)EIDIAIBICI
Binding assays
Removable flat-bottomed wells (Dynatech Laboratories, U.S.A., cat. No. 01 l-010-6302) were coated overnight at 4°C with 50 ~1 of immunoglobulins suspended in PBS (10 mM potassium phosphate buffer pH 7.4 and 150 mM NaCl), containing 5 pg/ml IgG, lOpg/ml IgA, lOpg/ml IgM or lOpg/ml F(ab’),, respectively. Before addition of ‘2sI-labelled proteins, the wells were washed repeatedly with PBS-Tween (PBS containing 0.05% Tween 20). Two methods were used to study binding: saturation experiments and competitive inhibition. For the saturation study, 50 ~1 of increasing amounts of ‘251-labelled protein, suspended in PBS-Tween, was added to each well. As a control of non-specific protein binding, some wells received labelled protein mixed with 1000 times excess of unlabelled protein. All wells were incubated for 3 hr at room temperature under agitation, then repeatedly washed with PBS-Tween to remove unbound pro-
x I
I
Protein Engineering
z-v)zlzlzl;Iz( Fig. 1. Schematic presentation of the protein A-derived proteins SpA. Vl, EB, Z, ZZ and Z-V. E, D, A, B and C refer to the different IgG-binding domains of SpA. X is the cell wall-binding domain of SpA. Z shows the slightly modified domain based on domain B.
Interaction
of protein A with immunoglobulins
1281
Table 1. Origin and characteristics of proteins used in the study Protein SPA
Vl
z-v,
EB
zz, z
Origin and characteristics
Reference(s)
Originates from Stuphylucoccu~ UhlCn ef al. (1984) aureus strain U511; contains five IgG-binding domains: ED-A-B-C and a region X which is in the cell wall; IV, z 50,000. Originates from a u.v.-induced Movitz et al. mutant of strain Cowan I; con(1979); tains two intact domains (E, D) Guss ef al. (1985) and one truncated (domain A); i%f,z 14,000. Protein A fragments obtained Nilsson et al. by protein engineering; contain (1987) 5, 2 or 1 identical IgG-binding domains based on the B-domain of protein A; M, (ZV)= 35,800; M,(ZZ)e 14,500; M,(Z)= 7000. Protein A fragments obtained Moks et al. (1986) by recombinant methods; contains 2 IgG-binding domains based on the E- and B-domains of protein A; &f, z 13,200.
contain five IgG-binding domains with different compositions. Protein Vi contains two intact IgG-binding domains (E and D) and one truncated (domain A) (Guss et al., 1985). The truncated portion of domain A spans the two cr-helices involved in binding and a similar peptide fragment based on domain B was recently shown to be sufficient for IgG-binding (Jansson et al., 1990). However, the binding strength may be lowered in the incomplete A-domain of Vl compared to the intact domain since it was shown that the peptide fragment of domain B was si~ificantly destabilized. The binding of SpA, VI, Z-V, ZZ, EB and Z to IgG, IgA, IgM and F(ab’), was first investigated in a saturation assay. In this study, ‘2sI-labelled proteins and immobilized immunoglobulins were allowed to interact for 3 hr under agitation at room temperature. These incubation conditions were necessary to reach equilibrium (data not shown). Figure 2 shows the obtained saturation curves for intact SpA and the different SpAderivates binding to the various immunoglobulins. Nonspecific binding was less than 0.5% and thus negligible. The most obvious conclusion from the experiment is that whereas the saturation level of all proteins binding to IgG is within the same range, there is a discrepancy in the binding to IgA, IgM, and F(ab’)* between intact SpA and VI on one side and Z-V, ZZ, EB and Z on the other. SpA and Vl show a high level of IgA, IgM and F(ab’), binding, the other proteins hardly bind to these immunoglobulins at all. These results are further presented in Table 2. Here, the data obtained from the saturation experiment were analysed using the LIGAND computer program (Munson and Rodbard, 1980; McPherson, 1985) and the number of accessible binding sites for the proteins on the immunoglobulins were determined. In order to simplify the comparison, the number of sites for
each protein on one specific immunoglobulin was normalized to the number of sites for intact SpA, defined as an arbitrary number of 100. Comparing the data of Table 2, the binding specificity difference is clearly seen. Whereas Z-V, ZZ, EB and Z show similar number of binding sites on IgG compared to SpA, significantly fewer sites are displayed on IgA, IgM and F(ab’),. Using the LIGAND computer program, the affinity constants for the interactions presented in Fig. 2 were determined (Table 3). Intact SpA has a strong binding to all immunoglobulins tested. The five-domain protein Z-V shows a high binding constant to IgG. However, its binding strength to IgA, IgM and F(ab’), is much lower. The smaller proteins, VI, ZZ, EB and Z, have five to ten times weaker affinity to IgG than SpA and Z-V. Under certain conditions two types of interactions were seen between the proteins and immunoglobulins, one high and one low (the low affinity was in the lO’/M range), demonstrating some kind of heterogeneity. The significance of this finding is unknown and hence, only the dominant affinity constants are presented. The binding properties of SPA, Vl, Z-V, ZZ, EB and Z were further investigated in a competitive inhibition assay. All six unlabelled proteins were allowed to compete with ‘251-labelled SpA for the sites on immobilized IgG, IgA, IgM or F(ab’)z (Fig. 3). A fixed amount of labelled SpA (4.0 x lo-l4 moles) was mixed with serial dilutions (‘log) of unlabelled SpA, Vl, Z-V, ZZ, EB and Z, starting with 500-fold excess. At 100% inhibition no ‘251-labelled SpA was bound to the immunoglobulins. The competition analysis shows that for IgG (Fig. 3A), SpA and Z-V are better inhibitors of ‘2SI-SpA binding than Vl, ZZ, EB and Z. Even at 500-fold excess, these smaller proteins could only inhibit binding to 50-75%. The binding discrepancy in the saturation curves for
1282
U.K. LJUNGBERG
A
ez al.
C
l
e
3
3m
g ?
200
ii!
100
B
Added @mole)
Added WnOlel l
-0
0 5
0
5
Adciid
15 fpmole)
Fig. 2. Saturation curves for different 1z51-lab~ll~dproteins binding to immobilized IgG (A), IgA (B), IgM (C) and F{ab’), (D). Increasing amounts of SpA (O), VI (a), Z-V (r]), ZZ (m), EB (a) and Z (A) were added. Under the conditions used, non-specific binding was less than 0.5% of specific binding (data not shown).
IgA, IgM and F(ab’), (Fig. 2B-D) was also clearly shown in the competition curves (Fig. 3B-D). Binding of ‘251-labelled SpA to these immunoglobulins was only inhibited by SpA itself and to a lesser degree by Vl. The other proteins competed very poorly. The affinity constants were determined using the LIGAND computer program. Since difference in binding valencies might obscure the analysis of different proteins when competing with each other, only data obtained from competitive inhibition where unlabelled SpA competed
‘2SI-labelled SpA, unlabelled Vl competed with ‘2SI-labelled Vl, etc. (data not shown), were used in the binding constant calculations. The result is presented in Table 3 and compared with the affinity constants obtained from the saturation curve experiments. There is good agreement between the two experiments. The low binding occupancy of ZZ, EB and Z to IgA, IgM and F(ab’& made the calculation of their affinity constants from the competitive inhibition data impossible. with
DISCUSSION
Table 2. Relative number of binding sites for SpA, Vl, Z-V, ZZ, EB and Z to IgG, IgA, IgM and F(ab), Relative number of binding sites (%) Protein
IgG
IgA
IgM
F(abX
SPA Vl z-v zz EB Z
100 111 89 123 128 131
loo 109 6 6 4 6
100 133 11 3 5 1
100 110 17 20 22 23
The maximal number of binding sites was determined by LIGAND computer analysis of the saturation curves of Fig. 2. Only sites correlated with a high affinity constant were regarded. The relative number was obtained after normalizing all values to the number of binding sites for SPA. Data shown are the outcome of one specific experiment, though reflecting a general obtained result.
AND CONCLUSIONS
We have analysed the non-immune interactions between SpA and human polyclonal IgG, IgA, IgM and F(ab’),. The aim of this study was to compare intact SpA and SPA-derived proteins containing different number of immunoglobulins-binding domains and different composition of domains. Determination of the aflinity constants indicates that the number of IgG-binding domains greatly affects the binding strength; the fivedomain proteins SpA and Z-V have a relatively high binding constant when compared to the shorter proteins. In the analysis, saturation and competitive inhibition curves gave very similar results (Table 3). The number of accessible binding sites on the immunoglobulins for the various proteins have been determined. The result from the saturation study (Fig. 2, Table 2) shows a pronounced discrepancy in specificity between SpA and Vl on one side and Z-V, ZZ, EB and 2 on the other. Whereas SpA and VI bind to all immunoglobulins tested, Z-V, ZZ, EB and Z bind poorly to IgA, IgM and F(ab’),. An additional reflection of
Interaction
1283
of protein A with immunoglobulins
Table 3. Affinity constants of SpA, Vl, Z-V, ZZ, EB and Z bound to IgG, IgA, IgM and F(ab’), determined by saturation and competitive inhibition studies Affinity constants (x IO-* M-‘)
Protein
Sat.
SPA Vl z-v zz EB Z
26 5.7 30 4.8 3.9 2.2
1gG Comp. 25 6.1 20 4.3 2.2 2.8
Sat.
IgA Comp.
31 5.9 12 1.3 0.3 0.3
21 4.3 5.4 nd nd nd
Sat. 31 6.1 12 3.6 2.8 1.9
IgM Comp. 32 4.9 7.1 nd nd nd
F(ab’), Sat. Comp. 10 2.2 6.3 0.7 0.5 0.1
8.8 2.3 6.8 nd nd nd
“nd = not determined. The saturation curves of Fig. 2 were transformed to Scatchard plots using the LIGAND computer program and the affinity constants were determined. Data from competitive inhibition of 12SI-labelled SpA, VI, Z-V, ZZ, EB and Z with the respective unlabelled protein were likewise analysed. The mean values of three to five experiments are presented.
Fig. 2 is that the maximal binding of intact SpA to IgG, IgA and IgM is quite different, hence the different scales on the Y-axes. The maximal binding level to IgA is rather low compared to that of IgG and IgM. This could be due to partial inactivation of immobilized IgA or, as suggested by Langone (19826), that a sub-population of serum IgA does not bind to SpA. What is the molecular reason for the discrepancy in binding specificity between SpA and Vl versus the other
proteins? The most characterized non-immune interaction between protein and immunoglobulin is the binding of SpA to the Fc-portion of the IgG molecule (Forsgren and Sjbquist, 1966; Kronvall and Williams, 1969; Deisenhofer, 1981). An additional binding site for SpA on IgG has been demonstrated in the Fab part of the immunoglobulin (Inganas et al., 1980) and it has been suggested that the interaction between SpA and IgA and IgM is mediated via a Fab-binding (Inganls,
Added (Slog)
Fig. 3. Competitive inhibition of binding of ‘*‘I-labelled SpA and different SPA-derived immobilized IgG (A), IgA (B), IgM (C) and F(ab’), (D). The labelled SpA was allowed with unlabelled SpA (O), Vl (a), Z-V (a), ZZ (m), EB (A) or Z (A). The amount SpA was kept at 4.0 x 10-‘4moles. Competing protein was diluted in steps of 7og, 2.0 x lo-” moles.
proteins to to compete of labelled starting at
U. K. LJUNGBERG et al.
1284
1981). As has been demonstrated in this paper (Fig. 2 and Table 2), the relative binding properties of SpA-derived proteins to isolated F(ab’),-fragments are similar to the binding properties to IgA and IgM. This implies that intact SpA and Vl are able to react with the immunoglobulins via the Fab-portion in addition to Fc. However, the other SPA-derived proteins bind to Fc, but the occupancy to IgA, IgM and F(ab’), is significantly lower (Table 2). Since SpA and Z-V both contain five IgG-binding regions, the discrepancy can not be explained by any differences in number of IgG-binding domains. Hence, the difference in binding specificity must depend on the type of domain included in the protein. As shown in Fig. 1, SpA contains five different domains called E, D, A, B and C (from N- to Cterminus). It has been shown that regions A, B and C are highly homologous. Region D contains an insertion of three amino acids and region E is the most diverged domain (Uhlen et al., 1984). Studies on separate domains have shown that all of them can interact with human polyclonal IgG (Moks et al., 1986). In contrast to SpA, protein Z-V contains five identical domains based on region B of intact SpA (Nilsson et al., 1987). Since Z-V, as well as ZZ and Z, show low binding occupancy to the F(ab’),-fragment of IgG (Table 2) this suggests that the A, D and/or E regions contain a structural motif, which is not present in Z, and which is necessary for Fab-binding. This suggestion is supported by the fact that Vl, containing the D, E and a truncated A region (Guss et al., 1985), shows full binding to F(ab’), in our analysis (Fig. 2 and Table 2). Since EB binds only poorly to IgA, IgM or F(ab’), this would imply that the addition of domain E to domain B is not sufficient for full Fab binding. Therefore, we propose that domain D possesses the ability of SpA to bind to Fab, since it possesses an interesting amino acid insertion compared to the other protein A domains, and this significant difference between this domain and the others could also be responsible for interacting with Fab. More studies using site specific mutagenesis and production of isolated domain D would be valuable to analyse the significance of this hypothesis. Acknowledgements-The
authors are grateful to Drs Mathias
Uhlen, Margareta Eliasson discussions. This project has Swedish National Board for corim AB. Dr John Sjiiquist of protein Vl.
and Tomas Moks for valuable been financially supported by the Technical Development and Exis acknowledged for the kind gift
REFERENCES Bolton A. E. and Hunter W. M. (1973) The labelling of proteins to high specific radioactivities by conjugation to a ‘*%containing acylating agent. Biochem. J. 133, 529-539. Bygren P., Freiburghaus C., Lindholm T., Simonsen O., Thysell H. and Wieslander J. (1985) Goodpasture’s syndrome treated with staphylococcal protein A immunoadsorption. Lancet ii, 1295-1296. Deisenhofer J. (1981) Crystallographic refinement and atomic models of a human Fc fragment and its complex with
fragment B of protein A from Staphylococcus aureus at 2.9and 2.8-A resolution. Biochemistry 20, 2361-2370. Forsgren A. and Sjijquist J. (1966) ‘Protein A’ from S. aureus. I. Pseudo-immune reaction with human y-globulin. J. Immunol. 97, 822-827. Guss B., Leander K., Hellman U., Uhlen M., Sjijquist J. and Lindberg M. (1985) Analysis of protein A encoded by a mutated gene of Staphylococcus aureus Cowan I. Eur. J. Biochem.
153, 579-585.
Harboe M. and Felling I. (1974) Recognition of two distinct groups of human IgM and IgA based on different binding to Staphylococci. Stand. J. Immunol. 3, 471482. Inganas M. (1981) Comparison of mechanisms of interaction between protein A from Staphylococcus aureus and human monoclonal IgG, IgA and IgM in relation to the classical Fey and the alternative F(ab’),c protein A interactions. Scand.
J. Immunol.
13, 343-352.
InganHs M., Johansson S. G. 0. and Bennich H. H. (1980) Interaction of human polyclonal IgE and IgG from different species with protein A from Staphylococcus uureus: Demonstration of protein-A-reactive sites located in the F(ab’), fragment of human IgG. Stand. J. Immunol. 12, 23-31. Jansson B., Palmcrantz C., Uhlen M. and Nilsson B. (1990) A dual affinity gene fusion system to express small recombinant proteins in a soluble form: Expression and characterization of protein A deletion mutants. Protein Engng 2, 555-561.
Kronvall G. and Williams R. C. (1969) Differences in anti-protein A activity among IgG subgroups. J. Immunol. 103, 828-833.
Langone J. J. (1980) ‘251-labelled protein A: Reactivity with IgG and use as a tracer in radioimmunoassay. Methods Enzymol.
70, 356375.
Langone J. J. (1982~) Use of labelled protein A in quantitative immunochemical analysis of antigens and antibodies. J. Immunol. Meth. 51, 3-22. Langone J. J. (19826) Protein A of Staphylococcus aureus and related immunoglobulin receptors produced by Streptococci and Pneumonococci. Adc. Immunol. 32, 157-252. Liifdahl S., Guss B., Uhlen M., Philipson L. and Lindberg M. (1983) Gene for staphylococcal protein A. Proc. natn. Acad. Sci.
U.S.A.
80, 697-701.
McPherson G. A. (1985) Analysis of radioligand binding experiments. J. Pharmacol. Meth. 14, 213-228. Moks T., Abraham& L., Nilsson B., Hellman U., Sjiiquist J. and Uhltn M. (1986) Staphylococcal protein A consists of five IgG-binding domains. Eur. J. Biochem. 156, 637-643. Movitz J., Masuda S. and Sjiiquist J. (1979) Physico- and immunochemical properties of staphylococcal protein A extracellularly produced by a set of mutants from Stuphylococcus aureus Cowan I. Microbial. Immunol. 23, 5160. Munson P. J. and Rodbard D. (1980) LIGAND: A versatile computerized approach for characterization of ligand-binding systems. Anal. Biochem. 107, 220-239. Nilsson B. and Abrahmsen L. (1990) Fusions to staphylococcal protein A. Methods Enzymol. 185, 144-161. Nilsson B., Forsberg G. and Hartmanis M. (1991) Expression and purification of recombinant insulin-like growth factors from Escherichia coli. Methods Enzymol. 198, 3-16. Nilsson B., Moks T., Jansson B., Abrahmsen L., Elmblad A., Holmgren E., Henrichson C., Jones T. A. and Uhlen M. (1987) A synthetic IgG-binding domain based on staphylococcal protein A. Protein Engng 1, 107-113. Nygren P.-A., Eliasson M., Palmcrantz E., Abrahmsen L. and Uhlen M. (1988) Analysis and use of the serum albumin
Interaction
of protein
binding region of streptococcal protein G. J. Mol. Recognition 1, 69-74. Palmer A., Welsh K., Gjorstrup P., Taube D., Bewick M. and Thick M. (1989) Removal of anti-HLA antibodies by extracorporeal immunoadsorption to enable renal transplantation. Lancer i, 10-12.
A with immunoglobulins
1285
UhlCn M., Guss B., Nilsson B., Gatenbeck S., Philipson L. and Lindberg M. (1984) Complete sequence of the staphylococcal gene encoding protein A. J. Biol. Chem. 259, 1695-1702. Zikan J. (1980) Interactions of pig Faby fragments with protein A from Staphylococcus aureus. Folia Microbial. 25, 246253.