Eur. J. Biochem. 186, 557-561 (1989)
`‘C FEBS 1989
`
`Thiol-directed immobilization of recombinant Ig G-binding receptors
`
`Charlotta LJUNGQUIST. Birgcr JANSSON. Tomd\ MOK‘3 nnd Mnthias UHLEN
`Department of Biochemistry and Biotechnology, Royal Institute of Tcchnology, Stockholm, Sweden
`
`(Received July 28,’Seplember 18, 1989) - EJB 89 0942
`
`A genetic engineering approach is described for directed immobilization of IgG-binding receptors to a thiol-
`containing matrix using a single cysteine residue. The cysteine residue is introduced into the C-terminal part of
`receptors based on staphylococcal protein A. Receptors containing one, two or five IgG-binding domains were
`produced in Escherichiu coli and subscquently immobilized on thiopropyl-Sepharose. A high amount, 5 pmol/ml
`gel (45 mg/ml>, of recombinant receptor could be rapidly immobilized to the solid support and both the gel and
`the immobilizcd receptor could be regenerated by reduction and oxidation rcactions. The gel was used for affinity
`purification of hunian IgG and analysis of IgG-binding capacity at different amounts of immobilized recombinant
`protein show the same maximal IgG-binding capacity (20 - 25 mgiml) for all three immobilized receptors. How-
`ever, at low substitution grade of receptors, the immobilized receptor molecules were shown to bind one (Z-Cys)
`and two (ZZ-Cys) TgG molecules, respectively. These results demonstrate that the immobilized protein molecules
`are in a functionally active form and that a two-domain receptor can bind two molecules of IgG without steric
`hindrance. Interestingly, the five-domain receptor (ZV-Cys), with a structure similar to native protein A, can only
`bind approximately two IgG molecules, despite the five-domain structure of the molecule.
`
`Staphylococcal protein A plays an important role in immu-
`nology and biochemistry owing to its specific interaction with
`the F, part of immunoglobulin G (IgG) froin many mammals
`[I]. A variety of biological responses lo protein A has been
`reported. including activation of the complement system,
`mitogenic stimulation of lymphocytes and cell-mediated
`cytotoxicity [l]. F, receptors occur also on the cell surface
`of other Gram-positive bacteria such as a large variety of
`streptococcal strains [2 - 51. This suggests a common biologi-
`cal role for these receptors, although the significance for viru-
`lence has not yet been clarified.
`Structural, genetic and biochemical studies [h, 71 have
`revealed that the protein A molecule consists of five homolo-
`gous IgG-binding domains (E, D, A, B and C), while the IgG-
`binding receptor from two independent clinical isolates of
`streptococci has either two (81 or three [9] homologous IgG-
`binding domains. The difference in receptor valency raises the
`question whether different strain might have adapted their F,
`receptors to various external environments.
`Many immunological methods have been developed and
`refined using protein A as a reagent, including immuno-
`precipitation techniques and immunoassays [l]. Protein A
`immobilized on solid support has found extensive use, in
`particular for affinity purification of polyclonal and mono-
`clonal antibodies [l]. Recently, columns with immobilized pro-
`tein A have been used to treat patients with autoiminunc
`diseases using an extracorporal purification scheme [ 101.
`These affinity gels have been obtained by coupling proteins
`via primary amino groups (or similar nucleophilic groups) to
`
`a chemically activated matrix. The methods used have the
`disadvantage that a heterogencous immobilization is obtained
`and therefore some of the attachment sites interfere with func-
`tional regions of the protein. In addition, the matrix and the
`ligand cannot be regenerated, which might be desired for some
`applications.
`An alternative method for immobilizing proteins is to use
`a genetic engineering approach, in which a specific attachment
`site is selectively introduced into a non-essential part of the
`recombinant protein. Here, we show that a single cysteine
`residue introduced into the C-terminal part of different re-
`combinant protein A molecules can be used for immobiliza-
`tion on a thiol-containing solid matrix. A covalent bond is
`achieved with all molecules exposing the IgG-binding part to
`the surrounding medium. The immobilization is directed and
`easily reversible by reduction and oxidation.
`This approach has been used to investigate the interaction
`between the affinity gel and IgG by varying the number of
`IgG-binding domains of the immobilized receptors. A syn-
`thetic IgG-binding domain (Z) based on protein A [Ill was
`polymerized to yield recombinant receptors with one, two or
`five binding domains. A single cysteine residue was genetically
`introduced into the C-terminal part of each receptor to enable
`a thiol-directed immobilization. This concept ensures repro-
`ducible immobilization, with all IgG-binding domains func-
`tional, and has allowed studies of the binding between
`immunoglobulins and immobilized protein A with different
`IgG-binding capacity.
`
`Corre,Tpondenc.c 10 M. Uhltn, Department or Biochemistry and
`Biotechnology. Royal Institute of Tcchnology, S-100 44 Stockholm,
`Sweden
`Abbreviations. Z-Cys, ZZ-Cys and ZV-Cys, synthetic proteins
`containing one. two or five IgG-binding domains based on
`staphylococcal protein A, containing a C-terminal cysteine residue.
`
`MATERIALS AND METHODS
`D N A constructions
`Restriction enzymes and T4 DNA Ligase (Pharmacia AB,
`Sweden, New England Biolabs, USA and Boehringer
`
`Page 1 of 5
`
`JSR Exhibit 1032
`
`

`

`558
`
`Mannheim, FRG) were used according to the suppliers rec-
`ommendations. DNA manipulations were done essentially
`as described by Maniatis et al. [12]. Oligonucleotides were
`synthesized by KabiGen AB, Sweden, using solid-phase chem-
`istry.
`A linker including the codon for cysteine and a stop codon
`was introduced between the EcoRI and the BamHl restriction
`sites in the multilinker of the plasmid pEZZl8 [13] to give the
`plasmid pEZZCYS. This plasmid was digested with NotI and
`EcoRI and the small fragment comprising the part of the gene
`fusion derived from protein A was exchanged for analogous
`NotI - EcoRI fragments from the plasmids pEZ and pEZV
`[ I l l resulting in the plasmids pEZCYS and pEZVCYS. The
`DNA fragments encoding one, two or five IgG-binding do-
`mains are thus followed by the cysteine-encoding linker and
`a stop codon.
`
`the proteins
`Expression and purification of
`Transformation and growth of Escherichia coli was
`performed essentially as described by Maniatis et al. [12]. E.
`coli HBlOl [14] was used as bacterial host and fermentations
`were performed in 8 1 in a Braun Biostat E fermentor. E. ci~li
`HBlOl cells containing pEZCYS, pEZZCYS and pEZVCYS
`were grown in complex media supplemented with 70 mg/l
`ampicillin.
`The culture media were recovered by cross-flow micro-
`filtration and subjected to affinity chromatography on IgG-
`Sepharose FF (Pharmacia AB, Sweden) [l 51. The eluted pro-
`teins were lyophilized and analyzed by SDS/PAGE (Fig. 1)
`using the Phast system (Pharmacia AB, Sweden). The proteins
`were reduced with 0.5 M 2-mercaptoethanol in 50 mM Trisi
`HCI, pH 8.5. for 1 h at 50'C, purified on Sephadex G-75
`(Pharmacia AB, Sweden) and the content of free thiol groups
`were determined spectrophotometrically 1161.
`
`the protein A derivatives
`Immobilization of
`The immobilization of recombinant protein A derivatives
`to thiopropyl-Sepharose 6B (Pharmacia AB, Sweden) was
`performed batchwise in coupling buffer (0.1 M TrisIHC1,
`pH 7.5, 0.1 M NaCl, 1 mM EDTA) at room temperature for
`15 min.
`the release of 2-
`The coupling was monitored by
`thiopyridone from the gel measured at 343 nm 1171. After
`binding the gel was washed with coupling buffer and washing
`buffer A (25 mM Pipes, pH 7.0, 0.1 M NaCI), before human
`polyclonal IgG (Kabi AB, Sweden) diluted in washing buffer
`A was applied to the gel for 15 min at room temperature.
`The gel was subsequently washed with washing buffer A and
`washing buffer B (SO mM ammonium acetate, pH 5.5, 0.1 M
`NaCI). 'The bound IgG was then eluted with elution buffer
`(0.2 M acetic acid, pH 3.3, 0.1 M NaCI) and measured
`spectrophotometrically at 280 nm (ac. = 107000 M-'
`. cm-l, molecular mass 150 kDa). Immobilized protein A de-
`rivative was eluted with reducing buffer (0.1 M 2-mer-
`captoethanol, 0.1 M Tris/HCl, pH 8.5, 0.1 M NaCI, 1 mM
`EDTA) and desalted on a small column of Sephadex (3-25
`(NAP-10, Pharmacia AB, Sweden) before analysis of the pro-
`tein content at 280 nm. The molar absorption coefficients used
`were 2400 M-' . cm-', 3600 M-' . cm-', 7190 M-' . cm-'
`for Z-Cys, ZZ-Cys and ZV-Cys, respectively.
`The Z-Cys saturation experiment was performed with a
`mixed gel to facilitate handling of small quantities of gel. The
`inert
`thiopropyl-Sepharose 6B was diluted 1 : 10 with
`
`Table 1 The tnoleculor nzuss of the reroinbinant receptor7 ond their
`number of IgG-binding domains
`
`Gcnc product
`
`Molecular
`mass
`
`Number of IgG-
`binding domains
`
`z - c y s
`z z - c y s
`z v - c y s
`
`kDa
`9.3
`15.9
`35.8
`
`1
`2
`5
`
`1
`
`2
`
`3
`
`4
`
`5
`
`6
`
`7
`
`8
`
`
`
`Fig. 1. Anulysis of IgG - Sepharose-purified thiol-coniaining receptors
`on u 8-25% gradient SDS.IPAGE. Lanes 2-4, non-reducing con-
`ditions. Lane 2, Z-Cys; lanc 3, ZZ-Cys; lane 4, ZV-Cys. Lanes 5 - 7,
`reducing conditions. showing. from lert to right, Z-Cys. ZZ-Cys and
`ZV-Cys. Lanes 1 and 8 show different molccular mass markers with
`sizes in kDa indicated
`
`Sepharose CL6B, (Pharmacia AB, Sweden) containing no
`thiol groups, to give a theoretical coupling capacity of2 pmol/
`in1 gel, which for Z-Cys represents 18 mg/ml. The results were
`recalculated to values corresponding to 100% thiopropyl-
`Sepharose.
`
`RESULTS
`the recornbinant proteins
`Expression and purification of
`The synthesis of a gene encoding a synthetic IgG-binding
`domain Z similar to region B of protein A has been described
`previously [ll]. Based on this gene, a set of three expression
`plasmids were constructed, encoding one. two or five Z do-
`mains, respectively, each having a C-terminal 19-amino-acid
`extension containing a single cysteine residue. The sequence
`is Val-Asp-Ala-Asn-Ser-Leu-Glu-Met-Tyr-Cys-Ala-Pr~-Leu-
`Lys-Pro-Ala-Lys-Ser-Ala. The predicted molecular masses of
`the recombinant receptors are shown in Table 1.
`The plasmids were introduced into E. coli and the cells
`were grown in a fermentor as described earlier [15]. The yields
`of the gene products after affinity purification wcrc approxi-
`mately 35,20 and 15 mg/l culture medium for Z-Cys, ZZ-Cys
`and ZV-Cys, respectively. The proteins were analyzed by SDS/
`PAGE under both non-reducing and reducing conditions
`(Fig. 1). Highly purified material is obtained for all three
`proteins (Fig. 1, lanes 5,6 and 7), although some minor degra-
`dation can be observed for the gene product ZV-Cys (Fig. 1,
`
`Page 2 of 5
`
`

`

`A
`
`559
`
`ADDED Z-cys
`(pmol/ml gel)
`Fig. 2. Suturation of thiopropyl-Sepharose with Z-Cys. Z-Cys was
`added batchwise to thiopropyl-Sepharose until saturation of thc gel.
`Released 2-thiopyridone was measured as the absorbance at 343 nm
`(0) and immobilized Z-Cys was eluted with reducing huffer and
`detcrmined spectrophotometrically a t 280 nm ( 0 )
`
`lane 7). A large fraction of the purified Z-Cys and of ZZ-Cys
`have spontaneously formed dimers (Fig. 1, lanes 2 and 3),
`which become monomeric by reduction (Fig. 1, lanes 5 and
`6). The dimers have most likely been formed in vivu after
`synthesis and secretion, but it cannot be excluded that spon-
`taneous dimerization has occurred during protein recovery. It
`is interesting to note that Z-Cys yields a considerably higher
`amount of dimers than ZZ-Cys and that very little dimer is
`formed by ZV-Cys. To our knowledge, a similar size depen-
`dence of spontaneous dimer formation has not yet been de-
`scribed and the reason for this phenomenon is not clear.
`The accessibility of the cystcine residue to the solvent
`was determined according to Ellman [I 61 after reduction and
`desalting of the purified proteins. The results show that the
`cysteine is fully exposed to the solvent in all three cases (data
`not shown).
`
`Immobilization to thioprupyl-Sepharose
`The immobilization kinetics of the reduced forms of Z -
`Cys, ZZ-Cys and ZV-Cys to thiopropyl-Sepharose were
`analyzed. The recombinant proteins were separately mixed
`with the gel at room temperature and the amount of
`immobilized recombinant protein was determined as released
`2-thiopyridone. In each case, less than 10 min were required
`to achieve complete immobilization and further incubation
`for up to 24 h did not significantly increase the immobilization
`yield. The fast immobilization kinetics are in contrast to the
`times required for naturally cysteine-containing proteins such
`as parvalbumin, bovine serum mercaptalbumin and cerulo-
`plasmin that need 2, 6 and 15 h, respectively, for complete
`immobilization [ 181.
`Thiopropyl-Sepharose contains approximately 20 pmol2-
`pyridyl disulphide groups/ml gel, that potentially can be used
`for directed immobilization of cysteine containing proteins.
`To investigate the available capacity of Z-Cys for immobiliza-
`tion to the thiopropyl-Sepharose, increasing amounts of the
`recombinant protein were added to the gel. The amount of
`immobilized Z-Cys after 15 min at room temperature was
`determined using two independent methods. First, as de-
`
`I.-
`'0
`
`-
`' 500
`' 600
`' 300
`100
`200
`' 400
`BOUND Z DERIVATIVE (nrnol/ml gel)
`
`' 700
`
`B
`a t
`
`
`
`.
`
`-I--
`
`BOUND IgG BINDING DOMAINS (nmol/ml gel)
`Fig. 3. The ZgG-binding cupucity ufrhe immobilized recombinunf recep-
`tors. Increasing amounts of Z derivativc werc batchwise immobilized
`to thiopropyl-Sepharose; ( 0 ) Z-Cys; [ 0) ZZ-Cys; ( A ) ZV-Cys. An
`excess of IgC was allowed to bind to the immobilized 2 derivative.
`After washing, the bound IgG was eluted with acctic acid, pH 3.3,
`and measured at 280 nm. (A) IgG-binding capacity of the gcl relatcd
`to the amount immobilized receptor. (B) IgG-binding capacity of the
`recombinant receptors relaled to Lhe number of immobilized IgG-
`binding domains; (B) is gencrated from (A) by multiplying the amount
`bound to receptor by its number of IgG-binding domains
`
`scribed above, bound Z-Cys was monitored as i-cleased 2-
`thiopyridone, and second, the amount of eluted Z-Cys was
`determined upon subsequent reduction. The rcsults in Fig. 2
`show good correlation between the two methods. Note that
`gels with different substitution grades can be obtained simply
`by varying the amount of added reduced Z-Cys to the
`thiopropyl-Sepharose. The amount of immobilized Z-Cys re-
`aches a maximum at 5 pmol (45 mg)/ml gel. which corre-
`sponds to 25% of the theoretical sites occupied. This value is
`markedly higher than the value for commercial protein A gels
`obtained by chemical immobilization, which have approxi-
`mately 70 nmol (3 mg) protein Aim1 gel.
`
`Page 3 of 5
`
`

`

`5 60
`
`IgG-bindiq capacity
`Various amounts of reduced recombinant receptors were
`mixed with the thiopropyl-Sepharose for 10 min at room tem-
`perature. Gels containing increasing amounts of the three
`different receptors were subsequently assayed for IgC-binding
`capacity. The gels were saturated with polyclonal human IgG
`and then the IgG was eluted by lowering the pH and deter-
`mined at 280 nm. The IgG-binding capacity is presented as a
`function of the amount immobilized recombinant receptor
`(Fig. 3A) or as a function of the amount of immobilized IgG-
`binding domain (Fig. 3 B).
`Significant differences in IgG-binding capacity can be ob-
`served depending on the number of binding domains of the
`receptor. For Z-Cys, an initial 1 : 1 binding ratio can be ob-
`served between immobilized receptor and affinity-bound IgG.
`Thus, each immobilized recombinant receptor molecule can
`bind one IgG molecule. Similarly, ZZ-Cys has an initial 1 :2
`ratio, which suggests that this two-domain recombinant recep-
`tor may indeed bind two IgG molccules. In contrast, ZV-Cys
`has a ratio close to 1 : 2> which suggests that this receptor can
`bind no more than two molecules of IgG despite its five-
`domain structure.
`For all three receptors, the maximum IgG-binding ca-
`pacity of the gels are approximately 150 nmol/ml, which corre-
`sponds to 22 mg IgG/ml. This capacity is similar to the ca-
`pacity for commercial protein-A - Sepharose, despite the fact
`that some of the gels contain more than 10 times the amount
`of immobilized Z-Cys and ZZ-Cys as compared to the com-
`mercial protein A gel (Fig. 3). Thus, the same IgG-binding
`capacity can be accomplished by less than 1/10 of the amount
`of receptor using orientated immobilization. This is probably
`due to steric effects depending on the matrix and already
`bound IgG molecules.
`
`Regenerution of the solid matrix and the recombinant ligand
`To investigate if the thiopropyl-Sepharose containing the
`recombinant receptors could be regenerated, gels containing
`each of the recombinant proteins were treated with 0.1 M
`2-mercaptoethanol as a reducing agent. Using this simple
`procedure, all three receptors were eluted from the gel (Fig. 2
`shows data for Z-Cys). The gcls were thereafter treated with
`the gel.
`1.5 mM 2,2’-dipyridyl disulphide to reactivate
`lmmobilization studies using Z-Cys, as described above, dem-
`onstrated no difference in capacity for the regenerated gel as
`compared to the original thiopropyl-Sepharose for covalent
`binding to the cysteine-containing proteins (data not shown).
`Finally, the proteins eluted from the regenerated gels were
`used for immobilization to new thiopropyl-Sepharose. No
`difference in immobilization efficiency could be observed for
`all the three recombinant receptors (data not shown). These
`results demonstrate that both the solid matrix and the re-
`combinant ligands can be regenerated by reduction and oxi-
`dation reactions.
`
`DISCUSSION
`Recently, approaches to facilitate purification of re-
`combinant proteins have been described involving the intro-
`duction of cysteines into a non-functional part of the gene
`product. Carter and Wells [19] showed that a recombinant
`subtilisin in which the serine residue at position 24 is changed
`to a cysteinc can be purified on activated thiol-Sepharose.
`Similarly, Persson et al. [20] showed that galactokinase with
`
`four cysteines introduced into the C-terminus was retarded on
`a column of thiopropyl-Sepharose, while native galactokinase
`was not retarded by the solid support under the conditions
`used.
`Here, we show that protein immobilization can be
`performed using a similar genetic engineering approach. The
`advantages compared to conventional chemical techniques
`for covalent binding to solid matrices are several. First, the
`approach allows a directed immobilization, which gives homo-
`geneously bound molecules in a reproducible manner and
`ensures that only recombinant molecules with an intact C-
`terminus is immobilized. Thus, the concept is suitable for
`biochemical studies, in which the effect of a heterogeneous
`mixture of immobilized molecules is unpredictable and there-
`fore not desired. Second, the binding is reversible, which al-
`lows the solid support and the ligand to be regenerated. Third,
`the immobilization gives high yields of bound molecules,
`simply by mixing the reduced recombinant protein with the
`solid support without the use of hazardous chemicals. Finally,
`the amount of coupled protein can be easily quantified by
`several independent methods, which facilitate biochemical and
`biophysical studies. The disadvantage with the cysteine ap-
`proach is that native cysteine residues might interfere with the
`binding. Thc gcnetic cngineering approach is therefore most
`suitable for rccombinant proteins lacking cysteines, but may
`also be applicable for cysteinc containing proteins with non-
`exposed thiol groups. Another possible disadvantage is the
`the presence of reducing agents after
`need to avoid
`immobilization, which might not be suitable for enzymes re-
`quiring a reducing environment.
`With this directed immobilization method we have gener-
`ated an IgG-binding gel with a capacity of 20 - 25 mg IgG,/ml
`gel. Kato et al. [21] have described a non-directed method for
`immobilization of protein A on activated thiol-Sepharose 4B
`were protein A was mercaptosuccinylated then coupled to the
`gel. They report an IgG-binding capacity of 0.3 mg/ml gel
`which is 80-fold less than the directed immobilization de-
`scribed in this paper.
`Using our concept it has been possible to investigate the
`molecular interaction in affinity binding between IgG and
`recombinant derivates of protein A. The genetic engineering
`approach ensures little or no interference with functional IgG
`binding upon immobilization and has allowed us to study the
`ratio of IgG molecules able to bind to an affinity gel con-
`taining various amount of recombinant IgG-binding recep-
`tors. Several models have been proposed to explain how
`IgG . protein-A complex in solution are formed [22-241 all
`based on the reaction between IgG and protein A. However,
`Moks et al. (1986) [7] have recently demonstrated that native
`protein A consists of five IgG-binding domains and these
`models must therefore be re-evaluated. Interestingly, the tive-
`domain receptor (ZV), immobilized by the genetic approach,
`can only bind approximately two molecules of IgC (Fig. 3),
`which confirms the binding studies performcd in free solution
`[24]. The effect is most likely steric as both one- and two-
`domain receptors (Z and ZZ) can bind equimolar amounts of
`IgG molecules (one and two, respectively). The question arises
`why Staphylococcus aureus has evolved a five-domain IgG-
`binding receptor despite thc fact that only two IgG molceules
`can bc bound. It is noteworthy that the functionally similar
`but structurally different receptor streptococcal protein G
`only has two or three IgG-binding domains [8, 91.
`Although the concept has here been used for reversible
`immobilization to thiopropyl-Sepharose, it is also possible to
`use the same recombinant proteins for irreversible immobili-
`
`Page 4 of 5
`
`

`

`zation. Using tresyl-chloride- or tosyl-chloride-activated gels
`[25], a thiol ester instead of a disulphide bond can be formed,
`which gives the advantagc that the immobilized recombinant
`receptor is not sensitive to reducing agents.
`In conclusion, we have shown that an IgG affinity gel can
`be produced simply by directed immobilization of a cysteine-
`containing recombinant protein A molccule to thiopropyl-
`Sepharose. A comparison of the three receptors with different
`numbers of IgG-binding domains (Fig. 3) shows that the two-
`domain receptor (ZZ-Cys) gives high IgG-binding capacity
`even at low amounts of immobilized ligand. As little as
`100 nmol ZZ-Cys immobilizedim1 gel gives almost maximal
`IgG-binding capacity (1 50 nmol IgG/ml gel). The lack of ad-
`ditional binding for the five-domain receptor, which is similar
`in structure to native protein A, suggests that a two-domain
`protein A molecule might be better for IgC affinity chro-
`matography, espccially for applications where the ligand size
`is important.
`This investigation was supported by grants from the Swedish
`board for Technical Development and Nordisk Industri Fond. We
`thank Dr Bjiirn Nilsson, Dr Erik Holmgren and Lennart Pettersson
`for cooperation and stimulating discussions.
`
`REFERENCES
`Langone, J. J. (1982) Adv. Immzmol. 32, 157-252.
`1.
`2.
`Rjiirck, L. & Kronvall, G. (1984) J . Zmmunol. 133, 969-9974,
`Yarnall, M. & Boyle, M. D. P. (1986) J . Zmmunol. 136, 2670-
`3.
`2673.
`Yarnvall, M. & Royle, M. D. P. (1986) M o l . Cell. Biochenz. 70,
`57 - 66.
`Reis, K. J., Ayoub, E. M. & Boyle, M. n. P. (1984) J . Immunol.
`132,3091 - 3097.
`Deisenhofer, J. (1981) Biochemistry 20, 2361 -2370.
`Moks, T., Abrahtnsen, L., Nilsson, B., Hellman, U., Sjoquist,
`J. & Uhlen, M. (1986) Eur. J. Biochern. 156, 637-643.
`
`6.
`7.
`
`4.
`
`5.
`
`561
`
`8. Fahnestock. S. R., Alexander, P., Nagle, J. & Filpula. D. (1986)
`J . Bacteriol. 167, 870 -880.
`9. (iuss, B., Eliasson, M., Olson, A,. Uhltn, M., Freij, A.-C., Flock,
`J.-I. & Lindberg, M. (1986) EMBO J. 5, 1567-1575.
`10. Pakmer, A,, Welsh, K., Gjornstrup, P., Taube, D.; Bewick, M. &
`Thick, M. (1989) Lancet 8628, 10-12.
`2 1 . Nilsson. B., Moks, T., Jansson. B., Abrahmsen, L., Elmblad, A,,
`Holmgren, E., Henrichson, C., Jones. T. A. & Uhlen, M. (1987)
`Protein Enng. 1, 107-113.
`12. Maniatis. T.. Fritsch, E. F. & Sambrook, J. (1982) Molecular
`cloning: a laboratory manual, Cold Spring Harbor Laboratory
`Press, New York.
`13. Lowenadler, B., Jansson, B., Paleus, S., Holmgren. E., Nilsson;
`B., Moks, T., Palm, G., Josephson, S. & Uhltn, M. (1987) Gene
`38, 87- 97.
`14. Boyer, H. W. & Roulland-Dussoix, 1). (1969) J . Mol. Biol. 41,
`459-472.
`15. Moks, T., Abrahmsen, L.. Osterliif, B., Josephson, S., Ostling,
`M.. Enfors, S.-O., Pcrsson, I., Nilsson. B. & Uhlen, M. (1987)
`Bio/Technologj> 5, 379 - 383.
`16. Ellman, G. L. (1959) Arch. Biochem. Biophys. 82, 70-77.
`17. Stuchbury, T., Shipton, M.. Norris. R.. Malthouse, J. P. G. &
`Brocklehurst, K. (1975) Biochvm. .I. 1.71: 41 7-432.
`18. Egorov, T. A.. Svensson, A., Rydkn, L. & Carlsson, .I. (1975)
`Yroc. Natl Acad. Sci. USA 72, 3029 - 3033.
`19. Carter, P. & Wells. J. A. (1987) Science 237, 394-399.
`20. Persson, M., G:son Bergstrand, M., Biilow, L. & Mosbach, K.
`(1988) Anal. Biochem. 172, 330 - 337.
`21. Kalo, K., Yamamoto, R., Umeda, Y. & Kosaka, A. (1981) J.
`Appi. Biochenz. 3, 75- 83.
`22. Langone, J. J., Boyle, M. D. P. & Borsos, T. (1978) J . Zmmunol.
`121. 333 - 338.
`23. Mota, G., Ghetie, V. & Sjoquist, J. (1978) Zmmunockemistry 15.
`643 - 65 1.
`24. Hansson, D. C., Phillips, M. L. & Schumaker, V. N. (1984) J .
`Zmmunol. 132. 1386- 1396.
`25. Nilsson, K., Mosbach, K. (1981) Biochem. Biiphvs. Res. Commun.
`102,449-457.
`
`Page 5 of 5
`
`