WO1992001939A1

WO1992001939A1 - Testing for metal ions

Info

Publication number: WO1992001939A1
Application number: PCT/GB1991/001167
Authority: WO
Inventors: Martin Francis Finlan; Anthony Robert Rees; David Staunton
Original assignee: Amersham International Plc
Priority date: 1990-07-17
Filing date: 1991-07-15
Publication date: 1992-02-06
Also published as: EP0539482A1; GB9015683D0

Abstract

A method of testing for metal ions uses an antibody which binds both an antigen and the metal ions, the antibody affinity being reduced in the presence of the metal ions. Detection of antigen binding is preferably by means of surface plasmon resonance spectroscopy. Preparation of a suitable antibody, by modifying the complementarity determining regions of the anti-lysozyme antibody HyHEL-5, is described.

Description

TESTING FOR METAL IONS

There is a substantial need for rapid sensitive tests for metal ions in fluid samples. One area of importance is the environmental monitoring of heavy and toxic metals. This invention provides a method of testing for metal ions in a fluid sample, by the use of an antibody which binds both an antigen and the metal ions, the binding affinity of the antibody for the antigen being reduced in the presence of the metal ions, which method comprises contacting the antibody with, together or sequentially in either order, the antigen and the fluid sample, and monitoring binding or release of the antigen by the antibody as an indication of the presence of the metal ions in the sample. The key to this method is the provision of an antibody which binds both an antigen and the metal ions. Techniques for designing/producing/synthesising such antibodies are discussed below. It is probably necessary that the antigen and the metal ions compete for binding to the antibody, i.e. that the antigen binding site and the metal ion binding site of the antibody be overlapping or adjacent. It is probably necessary that the antibody have a binding affinity for the metal ions greater than for the antigen. At all events, it is necessary that the binding affinity of the antibody for the antigen be reduced in the presence of the metal ions.

The antibody is designed or chosen to bind the specific metal ions to be tested. The nature of the metal ions is not critical to the invention. The metal ions are present in a fluid sample which may be gaseous but is generally liquid. The method is suitable for testing any aqueous or non-aqueous liquid in which the immune properties of the antibody are capable of functioning. The antigen is any substance which is specifically bound by the antibody. This may be the same or different from the substance to which the antibody was originally raised. The antigen may be chosen to have other properties arranged for easy detection; for example it may comprise an enzyme or may be a substance of high refractive index or high volume as described in more detail below.

A method of the invention involves contacting the antibody with the antigen and the fluid sample. Several reaction schemes are envisaged:- i) The antibody is contacted first with the antigen and binds the antigen. Subsequently, when the antibody-antigen complex is contacted with the fluid sample, metal ions in the sample displace antigen from the antibody, and this displacement is monitored. This is the preferred arrangement. The antibody-antigen complex is formed beforehand. The test merely involves bringing the fluid sample into contact with this pre¬ formed complex. No other reagents are necessary, except possibly for the purpose of detecting displaced antigen. ii) The antibody is contacted simultaneously with the antigen and the fluid sample.. Binding of antigen to the antibody is monitored and is inversely proportional to the metal ion concentration of the fluid sample. iii) The antibody is contacted first with the fluid sample, and subsequently with the antigen. Binding of the antigen by the antibody is monitored and varies inversely as the metal ion content of the fluid sample. The antibody is preferably immobilised, either on small particles suspended in a fluid medium, or more preferably on a solid surface. Techniques for immobilising antibodies without compromising their immune properties are well known in the field and form no part of this invention.

Release of the antigen, or alternatively binding of the antigen, by the antibody can be monitored in various ways which are well known in the field. For example, if the antigen is covalently linked to an enzyme, the enzyme can be caused to catalyse a reaction which generates easily detected light or colour. Such reaction of course requires the addition of an enzyme substrate and perhaps other reactants.

Preferably antibody-antigen binding is monitored by spectroscopic means which do not require the further addition of chemical reagents. The spectroscopic methods are:- i) Evanescent wave spectroscopy ii) Fluorescence detection (non-evanescent wave) . iii) Absorption spectroscopy (non-evanescent wave) . Evanescent wave spectroscopy is defined in the present context as embracing three related methods: (i) attenuated total reflection (ATR) spectroscopy (ii) total internal reflectance fluorescence (TIRF) spectroscopy, and (iii) surface plasmon resonance (SPR) spectroscopy. Each of these three spectroscopic techniques examines an optical property of a solution bordering a surface where total internal reflection of a light beam has occurred. In each case the incident and reflected beams are on the side of the surface distal to (i.e. remote from) the solution under study, whereas the evanescent wave is established on the solution side of the surface but extends into that solution for a very short distance, typically less than the wavelength of the incident/reflected beam. Spectroscopy by ATR or TIRF requires only a transparent material such as glass or quartz to create the interfacial surface with the solution, whereas SPR spectroscopy requires that the glass or quartz surface be coated with a thin (e.g. 50nm) metal layer of, for example, silver. All three methods can detect the exchange of solute molecules between the bulk phase of the solution and the interfacial surface, albeit by different means. ATR spectroscopy detects the absorption of evanescent wave light by molecules, with appropriate absorption spectra, that lie within the evanescent wave region. If the absorbed light is re- emitted as fluorescence then the emission can be measured with a suitable detector, such as a photo- multiplier tube, leading to TIRF spectroscopy. Thus both ATR and TIRF spectroscopy measure the absorption of light by molecules at or close to the interfacial surface, the difference being that ATR measures the absorption directly whereas TIRF measures it indirectly, as re-emitted fluorescence. By contrast, SPR spectroscopy measures changes of refractive index that may occur in the SPR evanescent wave region, but, just as with ATP and TIRF, those changes arise due to redistribution of molecules between the bulk phase of the solution and the evanescent wave region. For a review of these methods, see Sutherland and Dahne, 1987, "Biosensors - Fundamentals and Applications" pp 655-678, Oxford University Press, Oxford.

Preferably, binding or release of the antigen by the antibody is monitored by means of SPRS. Particularly if the antigen is a macromolecule, its binding to antibodies immobilised on an SPRS surface is readily detected. To increase sensitivity, the antigen may be chosen to be or comprise a substance having a high refractive index. This may be a molecule or particle with a high refractive index or a large size, to confer a higher refractive index on the antigen reagent as a whole thus giving rise to a larger SPRS signal than would be the case with an unmodified antigen. Possible substances include heavy species (e.g. halogens or metal ions other than those being tested for), highly electronically delocalised species (e.g. polycyclic aromatics or dyes), metal or metal oxide particles such as titania particles or high refractive index organic species such as ferritin. Alternatively, the substance may be one having a low refractive index i.e. a refractive index lower than that of the environment close to the solid surface. This aspect is described in international application WO 90/11525 published 4 October 1990.

The antibody may be immobilised on the solid surface used for SPRS. Alternatively, the antigen may comprise an enzyme which is caused to catalyse a reaction resulting in the production of a reaction product which is deposited on the solid surface. This aspect is described in international application WO 90/11510 published 4 October 1990.

The method of the invention may be qualitative, i.e. simply to detect the presence or absence of the metal ions in the fluid sample, or quantitative. For quantitative assays, measurements may be made of the rate of change of refractivity, and/or of the absolute refractivity at a given time. Contact between the antigen and the fluid sample on the one hand and the solid surface carrying the antibodies on the other hand may be static, but is more preferably dynamic e.g. by the fluid medium being caused to flow across the solid surface. The following description concerns the design/preparation/synthesis of antibodies which bind to antigens and also to metal ions.

The wide range of specificities exhibited by antibodies is a function of the sequence and length variability of six hypervariable loops or complementarity-determining regions (CDRs) which form the antigen combining site. These six CDRs supported on a highly conserved β-sheet framework region constitute the variable region of the antigen binding fragment (Fab) .

The approaches taken to modelling the antibody combining site, so far, fall into two groups; knowledge based and ai initio. An algorithm combining these approaches has already been described in some detail (A R C Martin e_fc al. Proc. Natl. Acad. Sci. U.S.A., Vol 86, pp 9268- 9272, December 1989, Biophysics). It has been used to model the combining sites of the anti-lysozy e antibodies HyHEL-5 and GLOOP 2 according to the following procedure. First, the framework region was taken from the Fab x-ray structure. Second, CDR conformations were derived by application of the scheme shown in Figure 1 thereof, where the modelling for each CDR is carried out in the presence of the crystal structure conformations of the other five. Thus, the scheme is essentially a single-CDR method and is useful for modelling loop replacements, insertions, deletions, mutations etc. or for providing a starting model for interpretation of electron density maps.

More recently the inventors have devised a scheme using which, it is now feasible to model the combining site of an antibody by reference only to the amino acid sequences of the variable regions and applications of the following protocol:

1 ) Select framework region of each variable domain based on sequence homology;

2) Carry out framework sequence substitutions using an automated procedure;

3) Orient variable domains based on known packing factors;

4) Select order in which CDRs are to be constructed;

5) Model combining site according to Figure

1 of the aforesaid Martin publication. From the data of several antibodies recently completed and in progress, it can be predicted that, within the near future the modelling of antibody combining sites by a procedure in which the user is not required to make subjective structural judgments will become routine.

The introduction of a metal binding site within the antibody combining site requires detailed knowledge of the interaction between the antibody and antigen, and antibody cDNA for site-directed mutagenesis. The complex between the raurine antibody HyHEL-5 and its antigen, hen egg lysozyme (HEL) , is known to high resolution (Sheriff S. s± al (1988) PNAS Si, 8075) and the antibody sequence determined by cDNA sequencing. However, no cDNA clones are available for HyHEL-5 (due to multiple chain problems), but the complimentary determining regions (CDRs) of HyHEL-5 can be grafted onto an available frame-work structure to create an antibody with the same specificity (Verhoyen M. e_fc al (1988) Science 223., 1534). The full length antibody cDNA clones of Gloop

2 were used to provide the framework and the CDRS replaced by oligonucleotide directed mutagenesis. The oligonucleotides were designed using sequence data from both antibodies. In the case of CDR H3, mutations were not confined to the variable region but extended into the framework region. For example residues 93 and 94 at the beginning of H3 CDR encode the unusual araino acids Leu and His respectively.

For the light chain all the CDRs were grafted onto the framework at the same time, but for the heavy chain a full length clone was constructed from two separate cDNA clones due to spurious mutations created in the constant region of the individual clones.

The HyHEL-5/Gloop 2 antibody was expressed in Xenophus oocytes (Roberts S. si ai (1987) Nature 223., 731) and found to bind HEL specifically with high affinity in the order 10~⁸ to 1θ^"9M. As with HyHEL-5 but unlike Gloop 2, the hybrid antibody did not bind the loop region of HEL in a radioimmunoassay.

To introduce a metal binding site into the antibody combining site (ACS) while maintaining antigen binding, only residues not involved in antigen binding or the structural integrity of the ACS could be mutated. In HyHEL-5 all 6 CDRs contribute at least one residue interacting with HEL. For L1 the interacting residues are Asn-31 and Tyr-32, while Tyr-34 interacts with other residues from the ACS. The remainder of L1, residues 25 to 30, project into the surrounding solvent at the edge of the ACS, and this portion of the CDR loop is a suitable candidate for mutagenesis. Two approaches can be taken to design a metal binding site within L1 : i) Search structural databases for appropriate metal binding sites with homology to the L1 CDR, and crea^'te a "chimeric" loop from both structures that retains both antigen and metal binding groups ;in their appropriate conformations. ii) Replace side chain groups of the appropriate conformation in L1 with liganding groups, e.g. Cys, Glu, His, to create a metal binding site while retaining the overall structure of L1. Example 1

The following experimental work was performed with the object of creating an antibody that would bind simultaneously to both its antigen and metal ions. So far as i) is concerned, the attempt was not successful; it is possible however that the attempt would have been successful if the object had been to create an antibody which bound to antigen or metal ion but not both at the same time. So far as ii) was concerned, the obtained antibody showed substantially reduced affinity for antigen in the presence of metal ion; and was therefore suitable for use in the method of the present invention.

1) Deslσn of metal binding site by ho ology

The Brookhaven Protein Structure Database was searched for loop conformations similar to that of L1 initially by using C α distances from the N and C termini. The loop structures selected were screened by "fitting" them onto the L1 framework take off positions (see Martin ^~± al, 1989). Any poor fits were eliminated. Among the structures selected was a cadmium binding site of rat liver metallothionein (MT 2), residues 36 to 45, derived by X-ray crystallography (Furey WF. eJt al (1986) Science 231. 704) .

From the crystal structure, cysteine residues 36, 37, and 41 all bound to the same cadmium ion and so the loop structure provided the majority of ligands. From a comparison of the antibody and metallothionein structures, a hybrid loop was constructed consisting of MT-2 residues Cys-36, Cys-37, Pro-38, Cly-40 and Cys- 41, and L1 residues Ser-27, Asn-30, Tyr-31, Met-32, and Tyr-33. The hybrid loop was subjected to conformational search using methods described in Martin et al. (1989) with and without a zinc ion present. Although the hybrid loop conformation was very close to the parent antibody structure, cysteine 24 could not be rotated into an appropriate liganding position without a high free energy penalty due to the tight packing at the base of the CDR and the restriction of rotational freedom by the presence of a disulphide bond between residue 23 and 87.

ϋ) Creation of a metal binding site ab initio

From the previous modelling strategy it was decided to utilise residues with the appropriate orientation that already existed in the CDR. This orientation would be towards a common point within the loop, and examination of the L1 CDR residues indicated that alanine 25 and valine 29 would be suitable for conversion to cysteines. For chelate stability, at least three binding groups are required. For the third ligand group no residue with the correct orientation ^was present and instead serine 27 was replaced with glutamic acid whose propyl component of the side chain could allow a carbonyl oxygen to donate electrons to the metal ion as in the case of carboxypeptidase A. Modelling of this loop structure indicated that the region 26 to 29 is very flexible allowing the liganding groups, 29, 27 and 25, to adopt appropriate conformations to create a metal binding site with ligand to metal distances of just over 2A. Examination of the nearest neighbours to this site revealed that isoleucine residue 2 of the light chain N terminus approached the metal binding site within a distance of 2A. Replacement of this residue with histidine allowed the creation of a distorted tetrahedral arrangement of liganding groups within the L1 loop. Creation gf the modelled proteins

Amino acid substitutions were created by oligonucleotide site-directed mutagenesis using the Eckstein methodology. The V domains of the mutants were sequenced to check for spurious mutations and transferred to SP6 transcription vectors. A mutant covering all four metal liganding groups (His 2, Cys 25, Glu 27 and Cys 29) was found to bind the antigen HEL with an affinity comparable to the parent antibody. However, at zinc concentrations of 0.02raM the affinity of the parent antibody for HEL decreased by 40% while the putative metal binding antibody retains its high affinity.

Example 2

The inventors have designed and constructed 2 putative raetalloantibodies based on the structure of the anti-hen egg lysozyme (HEL) antibody HYHel-5. Mutations were introduced into the L1 complementarity determining region (CDR) and the amino terminus of the light chain. Two mutant proteins were created, LM contained only the mutations within the L1 while LM-2N contained both L1 and light chain amino terminus mutations. Figure 1 of the accompanying drawings shows the structure of the original complementarity determining region (CDR) L 1 for HyHEL-5 (a) and models of two metal binding sites LM (b) and LM - 2N(c) derived from it with a zinc ion present. In LM, residues 25, 27 and 29 have been mutated to cysteine, glutamic acid and cysteine respectively. LM-2N contains all the mutations of LM but in addition contains a mutation of the light chain amino terminal isoleucine to a histidine, supplying a fourth ligand for the metal binding site. From previous experiments with expression of the mutated antibodies in Xenopus laevis oocytes, it was known that the engineered proteins retained affinity and specificity for the antigen, HEL. The proteins were then expressed as Fv fragments in E.coli. purified by affinity chromatography using HEL immobilised onto 4B sepharose, and analysed by SDS polyacrylamide gel electrophoresis (PAGE) and isoelectric focussing (IEF) . The proteins obtained were greater than 90% pure and exhibited an affinity and specificity similar to the original antibody.

Example 3

The metal binding potential of the engineered proteins has been investigated using a metal ion blot derived from a protocol from Schiff et al. The protein is immobilised on nitrocellulose filters and its ability to bind metal radioisotopes compared against known metalloproteins (alcohol dehydrogenase, carbonic anhydrase, bovine serum albumin (low affinity) and negative controls (HEL, HyHEL-5 Fv) . After incubation with 10 mM EDTA to remove bound metals, 5 μg of protein was dotted onto nitrocellulose filters and the filters washed extensively with metal binding buffer (100 mM Tris pH 6.8 50 mM NaCl in Analar water). The filters were then incubated for 30 minutes with 50 μCi of the appropriate metal radioisotope (specific activity 10 to 100 raCi/mg) in metal binding buffer and 1mM DTT. After washing to remove unbound radioisotope the filters were autoradiographed for 30 mins to 12 hrs depending on the specific activity of the radioisotope. The intensity of signal from the putative metalloantibodies was then compared to the other proteins present on the filter.

The positive results were as follows:

65

Zn Carbonic anhydrase Fv LM-2N and Fv LM-2N with its antigen, HEL.

¹⁰⁹Cd Fv LM-2N and Fv LM-2N with HEL. Co Alcohol dehydrogenase, carbonic anhydrase,

BSA, FVLM-2N and Fv LM-2N with HEL. 63 Ni Fv LM-2N and a much reduced signal in the presence of its antigen, HEL. The metal blot assay allows rapid screening of engineered proteins for metal binding and indicates characteristics of the binding site as its preference

2+ 2+ for "hard" (e.g. Mn ) or "soft" (e.g. Cd ) cations.

In addition the present experiments with Ni 2+ suggest a possible alteration in the metal binding site conformation on binding antigen. The difference in metal binding between proteins LM and LM-2N suggests that the introduction of the histidine as a fourth ligand creates chelate stability and a high affinity binding site.

Claims

£ U£

1. A method of testing for metal ions in a fluid sample, by the use of an antibody which binds both an

5 antigen and the metal ions, the binding affinity of the antibody for the antigen being reduced in the presence of the metal ions, which method comprises contacting the antibody with, together or sequentially in either order, the antigen and the fluid sample, and monitoring 10 binding or release of the antigen by the antibody as an indication of the presence of the metal ions in the sample.

2. A method as claimed in Claim 1 , wherein the antibody is contacted first with the antigen and then

■ 15 with the fluid sample, and release of the antigen by the antibody is monitored.

3. A method as claimed in Claim 1 or Claim 2, wherein the antibody is immobilised.

4. A method as claimed in Claim 3, wherein 20 binding or release of antigen by the antibody is monitored by surface plasmon resonance spectroscopy.

5. A method as claimed in Claim 4, wherein the antigen is chosen to generate a large SPR signal.

6. A method as claimed in any one of Claims 1 to 25 5, wherein the antibody incorporates a metal binding site in one of the six hypervariable loops which form the antigen combining site.

7. A method as claimed in any one of Claims 1 to 6, wherein the antibody is based on the anti-lysozyme 0 antibody HyHEL-5.

8. A method as claimed in Claim 7, wherein the complementarity-determining region L1 of the antibody has been modified by mutating residues 25, 27 and 29 to cysteine, glutamic acid, and cysteine respectively, and

35 mutating the light chain amino terminal isoleucine to histidine.