EP2140266A1

EP2140266A1 - Quantification of analyte molecules using multiple reference molecules and correlation functions

Info

Publication number: EP2140266A1
Application number: EP08715628A
Authority: EP
Inventors: Anpan Han; Cornelia Steinhauer; Hanne Kaas; Kristine Garde
Original assignee: Toxispot AS
Current assignee: Toxispot AS
Priority date: 2007-03-26
Filing date: 2008-03-26
Publication date: 2010-01-06
Also published as: WO2008116471A1

Abstract

Disclosed is a method for quantifying analyte molecules in a sample comprising providing a substrate containing two or more reference re- gions containing different immobilised reference first binding partners and one or more analyte regions containing immobilised analyte first binding partner(s); applying a mixture containing reference second bind- ing partners and a sample suspected of containing analyte first or second binding partner(s), in which mixture the concentrations of the reference second binding partners are known; measuring the binding of the bind- ing partner(s) in the mixture to the corresponding immobilised analyte partner(s); and quantifying the concentration of analyte binding part- ner(s) in the sample, where the quantification is performed by correlat- ing the measurement of the binding of the analyte partners to the meas- urement of the binding of the reference binding partners. The method is useful for quantifying microcystin LR.

Description

QUANTIFICATION OF ANALYTE MOLECULES USING MULTIPLE REFERENCE MOLECULES AND CORRELATION FUNCTIONS

Introduction

The present invention relates to an assay method for quantifying analyte molecules in a sample. The present invention also relates to the use of two or more reference binding partners in an assay for quantifying analyte molecules in order to minimise the risk of false positive results. The method is useful for quantifying microcystin LR.

Background art

Biochemical assays are methods to analyse chemical substances through the application of biological molecules, e.g. proteins or DNA. An immunoassay is a biochemical assay involving antibodies which are proteins with the property of being able to bind a substance, an antigen, with great specificity. The antigen may be a protein, or a peptide, a carbohydrate, a small organic molecule etc. This property is utilised in an immunoassay, where antibodies are used to bind corresponding antigens that are present in a sample. For an introduction please refer to The Immu- noassay Handbook edited by Wild (2005). Thus, immunoassays are used to quantify analytes such as biomarkers, toxins, drugs, pesticides, viruses etc.

This invention relates to so-called heterogeneous assays which are also referred to as solid-phase assays. In heterogeneous immunoas- says the antibody or antigen is immobilised onto a surface. There are two general kinds of heterogeneous immunoassays; competitive and non-competitive, the latter of which includes direct and sandwich assays. There are many variations within the types of immunoassays. A typical, traditional immunoassay requires at least an antibody-antigen pair and commonly also a secondary antibody capable of binding either of the binding partners of the antibody-antigen pair and which secondary antibody is equipped with a tag allowing direct or an indirect detection. Of these the antigen is normally the analyte of interest. When the binding partners of the antibody-antigen pair interact this interaction can be detected by letting the secondary antibody bind to its corresponding binding partner and detecting this binding via the tag. Depending on the assay format either the antigen (e.g. competitive assay) or the antibody (e.g. sandwich or direct assay) may be immobilised on the substrate material. The detection system may consist of an enzyme linked to the secondary antibody (an indirect detection) or a label allowing direct observation, such as radioactive elements or fluo- rescent probes. Typical enzymes used as labels are horseradish peroxidase (HRP) or alkaline phosphatase (AP). In this case binding of the secondary antibody to its corresponding antigen may be detected by adding a chromogenic substrate of the enzyme and measuring the extent of the formation of the coloured product of the enzymatic reaction. In most clinical settings it is necessary to determine the absolute concentration of the analyte (antigen), i.e. a quantitative measurement of the antigen concentration in the sample. This is also true in environmental monitoring and many other domains of immunoassay applications. To determine the absolute concentration it is necessary to gen- erate a standard curve or a dose response curve, which is a function that relates the signal produced by the label to the absolute concentration of the antigen. To generate such a function, standards of the antigen with well-defined concentrations are typically assayed in parallel with the antigen in the sample. The results of the parallel assays with standards will enable the user to determine a function that relates the assay signal to the absolute concentration of the antigen. Using this function, it is then possible to relate the signal generated by the sample with an unknown amount of antigen.

In strong analogy to DNA microarrays the classical immunoas- say has recently been miniaturised to the analytical protein microarray format (Zhu and Snyder 2003; Wingren and Borrebaeck 2004; Dufva and Christensen 2005). Modern arraying technology enables the spotting and application of pL-amounts of liquid in a multiplex format. Thus, an unlimited number of different samples can be analysed in one assay us- ing μL-amounts of reaction-partners. Thereby the microarray format not only saves sample volume but also enables a more sensitive detection of analyte and reduces the number of handling steps significantly.

To be able to determine the absolute concentration of an anti- gen, as described above, a standard curve is required. In the microarray format, this is usually carried out by incubating several arrays with known concentrations of analyte target molecules. Examples of competitive immunoassays in microarray formats are the works of Han et al. (2003) and Belleville et al. (2004) who determined the absolute concen- trations of pesticides. To measure one sample up to 8 arrays were needed for the standard curve and control experiments. For assays using labelled primary antibodies, 8 different solutions with spiked antigen were prepared and the mixture was pipetted onto the arrays. Du et al. (2005) reported competitive immunoassays that detected drugs from se- rum samples. Also here a number of arrays were used to generate the standard curves.

There are several examples of sandwich microarray immunoassays (Pawlak et al., 2002; Urbanowska et al., 2003; Weissenstein et al., 2006; Wiese et al., 2001). Weissenstein et al. (2006) reported a quanti- tative immunoassay of cancer biomarkers in a tissue extract based on the ZeptoMARK (Zeptosens) technology platform. Twelve arrays were used to generate the dose response curve and an additional two were used as negative controls.

International patent application WO2005/090546 describes an- tibody microarrays for identifying, quantifying and qualifying analyte molecules with a special focus on expression analysis. The principles of WO2005/090546 are based on traditional immunoassays. In order to improve normalisation of data the assays are further supplied with spots containing internal control molecules corresponding to molecules nor- mally unexpressed in the target samples.

WO2006/020126 also relates to microarrays for the determination of protein concentrations. For the achievement of this a target protein (of unknown concentration) and control proteins (of known concentration) are labelled with the same tag and immobilised in spots in an ar- ray. The tag may then be quantified by binding a binding pair member of the tag to the tag and then quantifying these interactions. Thus, in reality the quantification of the target protein is only indirect as it is the tag which is quantified and this quantification is then used for calculating the concentration of the target protein. A preferred tag in WO2006/020126 is the protein glutathione-S-transferase.

US-application US2005/0250109 relates to methods for the detection, identification and quantification of multiple target molecules using microarray technology. The methods of US2005/0250109 employ immobilised sets of capture molecules capable of binding two different sets of target molecules, so that the binding of a set of target molecules may be quantified relative to the binding of a second set of molecules with the aid of a "correction factor". This correction factor is "calculated through the use of at least one additional identical target molecule (able to bind first set of capture molecules and second set through an adaptor molecule) simultaneously quantified on both capture molecules". The methods are mainly intended for use with nucleotides, though despite mention of how the method may be employed also with proteins and peptides, the methods appear ill-suited for this purpose. International patent application WO2004/104230 also takes advantage of the microarray format to quantify analyte molecules in a sample using ELISA principles. In this case an array is formed containing spots capable of binding the analyte as well as a number of spots containing a "reference capture molecule". By using a reference molecule- reference-capture molecule pair of which neither are (expected to be) present in a sample of interest it is possible to use this to minimise internal variation. Since the reference capture spots in the array are designed to be identical, the signals recorded from interaction between the reference molecule and the corresponding capture molecule allow an es- timate of the internal variation of the array to be made. This estimate of internal variation may then be employed to improve the reading of the signal from the analyte molecule with respect to variation. Also the positioning of the reference spots may function as a map of the array for automated reading of the array. Microarrays for the detection and quantification of antibodies (of a given specificity) are described in US-application US2006/0154299. The arrays contain antigens for the analyte antibodies of interest and a series of deposited antibodies from the same species as the analyte anti- bodies. The deposited antibodies are used for constructing standard curves for the interaction between the immobilised antibodies and a fluo- rescently labelled, secondary antibody from another species (specific for antibodies from the target organism). These secondary antibodies will also bind to any antibodies from the sample with specificity to the im- mobilised antigens, therefore also serving to quantify the sample. Thus, despite not being fully identical to the analyte molecules of interest the origin and isotype of the immobilised antibodies make these functionally equivalent to the analyte molecules with respect to quantification so that the set-up is merely an internal standard curve. Considering the complexity of samples commonly encountered in clinical or environmental settings, i.e. such samples normally have a large number of different constituents, the use of immunoassays pose a risk that constituents other than the analyte of interest will interact and bind with antibodies thought only specific for the analyte molecule. This may lead to incorrect results from an immunoassay which in the case of competitive assays may be especially detrimental, as such errors will appear as highly undesired false positive results.

We have now found that by including reference molecules unrelated to the analyte molecules in an immunoassay the risk of obtaining false positive results may be minimised if two or more of such references are assayed in parallel with a sample containing analyte molecules, and the measurement of the sample molecule is correlated with the measurement of the two or more reference molecules.

Disclosure of the invention

The present invention relates to an assay method for quantifying analyte molecules in a sample comprising a) providing a substrate containing two or more reference regions containing different immobilised reference first binding partners and one or more analyte regions containing immobilised analyte first binding partner(s), where the reference first binding partners and the analyte first binding partner(s) have affinities for binding different second binding partners; b) applying a mixture containing reference second binding partners and a sample suspected of containing one or more analyte first or second binding partner(s), in which mixture the concentrations of the reference second binding partners are known; c) allowing the mixture to interact with the substrate; d) measuring the binding of the analyte binding partner(s) in the mixture to the corresponding immobilised analyte binding part- ner(s) and the binding of the reference second binding partners to the immobilised reference first binding partners; and e) quantifying the concentration of analyte binding partner(s) in the sample, where the quantification is performed by correlating the measurement of the binding of the analyte partners to the mea- surement of the binding of the reference binding partners.

This assay method is particularly suited for use in an immunoassay format where the binding partners comprise antibodies and their corresponding antigens. Depending on the design of an assay the analyte molecule may be either the antigen or the antibody part of the ana- lyte binding partner pair. However, as will be obvious to those skilled in the art the assay method is not limited to antigen-antibody binding partners but may well also include binding partners comprising other pairs of proteins capable of interacting specifically, proteins interacting specifically with non-proteinaceous molecules; or pairs of non-proteinaceous molecules.

Although the benefits of the present invention may be realised by using two different immobilised reference first binding partners, the use of three or more different immobilised reference first binding partners is preferred. In a preferred embodiment the binding partners com- prise microcystin LR (MC-LR), human chorionic gonadotropin (hCG), insulin and (Tyr[SO₃H]²⁷) cholecystokinin fragment 26-33 (CCK) and antibodies raised against MC, hCG, insulin and CCK.

In a preferred embodiment of the present invention a competi- tive assay format is used in which the mixture containing the reference second binding partners and the sample comprises analyte first and second binding partners. Thus, for the pair of analyte binding partners the mixture containing the sample contains both the binding partner of the immobilised binding partner as well as a solubilised form of the binding partner immobilised on the substrate.

The method of the present invention may advantageously be employed in a microarray format, although other formats, such as mi- crotiter plates of e.g. 96, 384 or other numbers of wells are also suitable. In such cases each well of the microtiter plate may comprise one region, or each well may comprise a microarray of regions. In a microarray the regions for immobilising analyte or reference binding partners are preferably discrete, although by careful choice of detection methods several regions may well be superimposed without jeopardising the measurement of bound binding partners. For example, by employing a combination of principally different detection methods, e.g. fluorescent and radioactive labelling simultaneous reading of superimposed signals may be possible. In cases where the regions are discrete in the substrate it may be advantageous to position a plurality of identical regions in close vicinity so that the identical regions can be said to be arranged in sets.

In the present invention, a preferred method of detecting interactions between respective binding partners involves using one or more additional binding partners which additional binding partners are capable of binding to the analyte second binding partners and/or to the reference second binding partners, and which additional binding partners are not capable of binding to the immobilised first binding partners, and where the additional binding partners facilitate measurement. In a more preferred embodiment, these additional binding partners comprise secondary antibodies labelled with a tag allowing detection. This tag may be an enzyme capable of converting a colourless substrate molecule into a coloured or fluorescent form for indirect detection, examples of such enzymes being horseradish peroxidase or alkaline phosphatase, or the tag may be a fluorescent moiety, such as green fluorescent protein, Cy3 or Cy5, or a radioactive atom, such as ¹²⁵I, allowing direct detection. Other tags facilitating detection will be well known to those skilled in the art.

The present invention also relates to the use of two or more first reference binding partners in an assay for quantifying analyte molecules in a sample in order to minimise the risk of false positive results which assay comprises a) measuring the binding between analyte first and second binding partners and the binding between the two or more reference first binding partners and corresponding reference second binding partners; and b) quantifying the concentration of analyte binding partner(s) in a sample, where the quantification is performed by correlating the measurement of the binding of the analyte partners to the measurement of the binding of the reference binding partners. In a preferred embodiment this assay is a competitive assay.

Brief description of the figures

Fig. 1 shows an exemplary schematic illustration of a substrate with mi- croarrays.

Fig. 2 depicts the steps in an example of a competitive ELISA assay ac- cording to the method of the present invention.

Fig. 3 shows images of microarrays after binding and detection of first and secondary binding partners

Fig. 4 shows correlation curves for MC

Fig. 5 shows a standard curve for MC Detailed description of the invention

The present invention is related to the field of quantification of biochemical analyte molecules. In a preferred embodiment it is related to the measurement of microcystin LR in a competitive immunoassay in a mi- croarray format.

Binding partners As used herein the term "binding partner" refers to either of two parts of a pair of molecules capable of binding specifically to each other, and it may therefore refer to either the antigen or the antibody of an antigen- antibody pair, to either the carbohydrate or the lectin of a carbohydrate- lectin pair, to either strand of DNA of a double stranded molecule of complementary sequence etc. In general, suitable examples of pairs of binding partners include those formed by antigen-antibody interactions, compound-aptamer interactions, antibody-antibody interactions, protein- small molecule interactions, enzyme-substrate interactions, enzyme- substrate analogue interactions, lectin-carbohydrate interactions, drug- receptor interactions, streptavidin-biotin interactions, nucleic acid- nucleic acid interactions, nucleic acid-protein interactions etc. The term "antibody" refers to antibodies of any suitable isotype as well as antibody fragments containing the antigen binding specificity of a complete antibody. Antibodies may be of monoclonal or polyclonal origin, or they may be produced recombinantly.

The "first binding partner" refers to the part of a pair of binding partners immobilised on a substrate, whereas the "second binding partner" is not immobilised; the fact that "first binding partner" refers to the immobilised form does not exclude that the same type of molecule is used in a non-immobilised form. The term "analyte binding partner" is either part of a pair of binding partners comprising an analyte molecule, whereas the "reference binding partner" refers to either part of a pair of binding partners employed to construct a correlation curve between the binding of analyte binding partners and reference binding partners. Basically, the sample to be analysed by the present assay method can be of any origin. In a certain aspect of the invention, the sample is a water sample and the analyte binding partner it is suspected of containing is selected from the group consisting of bacterial toxins, fungal toxins, algal toxins, shellfish toxins, fish toxins, enzymes occur- ring in the synthesis pathway of toxins, markers for the occurrence of pathogenic microorganisms, geosmins, plant protection products and pharmaceutical products, and degradation products of the above mentioned group of compounds. A certain group of binding partners of interest in the present invention is algal toxins, such as cyanotoxins, especially hepatoxins and neurotoxins. Specific examples of hepatoxins include microcystin (MC) like microcystin-LR (MC-LR) and microcystin-RR (MC-RR), nodularin, and cylindrospermopsin. Specific examples of neurotoxins include ana- toxins and saxitoxins. Other algal toxins of interest include paralytic shellfish toxins/poisoning (PST, PSP, saxitoxins and gonyautoxins), diar- rheic shellfish toxins/poisoning (DST, DSP, ocadeic acid, yessotoxin, etc.), amnesic shellfish toxins/poisoning (AST, ASP, domoic acid), ciguatera fish toxins/poisoning (CFT, CFP, ciguatera etc.), neurotoxins shell- fish toxins/poisoning (NST, NSP, brevetoxin, etc.), and endotoxins (lipopolysaccharides, LPS).

In general, the reference binding partners are selected from molecules not expected to be present in a sample to be analysed. The reference binding partners may be of the same chemical nature as that of the analyte binding partners, e.g. proteins when proteins are to be analysed or nucleic acids when nucleic acids are to be analysed, or they may be of a different chemical nature. Preferably, the chemical nature of the reference binding partners is the same as that of the analyte binding partners. Suitable pairs of reference binding partners include proteins known to give an immune response upon immunisation of animals, such as mice, rabbits or goats, and the corresponding antibodies raised against the proteins upon the immunisation.

Correlation functions

In the present invention the use of dose response curves and control experiments for measuring and quantifying and analyte molecule is replaced by another set of functions that relate a measured signal to the concentration of an analyte molecule. In the case a microarray format is employed, and this set of functions is stored in e.g. a computer, three arrays at the most, preferably a single array is needed to determine the absolute concentration of multiple analytes when analysing a sample. In case e.g. each of the regions are formed in separate wells of a microtiter plate in place of the microarray format the term "microtiter plate" will generally substitute the term "microarray" in the discussions below.

The method of the present invention involves the generation of a set of correlation functions correlating the signals obtained from the binding of analyte binding partners to that of the signals of the binding between the two or more respective reference binding partners.

When the methods of the present invention are employed with a substrate containing microarrays with immobilised first binding partners, the microarray is spotted with (please see also Fig. 1): - two or more (sets of) regions of reference first binding partners; the regions are denoted r_mir where the subscripted / is the index denoting the region, r denotes a reference binding partner, and the subscripted m is and index identifying the reference binding partner. - one or more (sets of) regions of analyte first binding partners; the regions are denoted a_n], where the subscripted j is the index denoting the region, a denotes an analyte binding partner, and the subscripted n is and index identifying the analyte binding partner. - optionally (a set of) control regions; these optional regions are denoted Ck, where the subscripted k is the index denoting the region, and c indicates a control. Such control regions may contain only the buffers otherwise used for immobilising the first binding partners, molecules, such as proteins, not expected to bind any of the binding partners or they may contain a substance facilitating detection. Thus, these may serve as negative controls and may further act as landmarks in the microarray landscape for the software employed for analysis or for normalisation pur- poses.

Fig. 1 schematically illustrates a substrate with two microarrays (numbered 1 and 2) according to the present invention. Also shown is an enlarged section with a microarray depicting the positions of the analyte and reference binding partners (denoted a and r, respectively as discussed above) as well as of potential control regions (denoted c).

To such a microarray is applied a mixture (of standards) containing reference second binding partners and/or analyte second binding partners in known and well-defined concentrations; these second binding partners are denoted ~r_m and ~a_π, respectively, with the ^v~' indicating a second binding partner. The concenctrations of the second binding partners are indicated as [~r_m], and [~a_π]/, respectively, where / denotes the index of different concentrations. Preferably 10 such standards are used. After incubation of the standards of each concentration, /, with at least one microarray, M/ (preferably in duplicates or triplicates), the microarray signals are measured: the average signal intensities of the regions caused by the complexes formed between the reference binding partners are denoted Sr_m. - the average signal intensities of the regions caused by the complexes formed between the analyte binding partners are denoted Sa_n. the optional signal intensities from the optional control regions are denoted Sc_k. For each measured array the values R(Sa_n, Sr_m)ι are calculated; in case control regions are used, R(Sa_n, Sc_k)ι is also calculated. As an example this value could be the ratio between Sa_n and Sr_m, i.e. Sa_n/Sr_m. R(Sa_n, Sr_m)ι is now plotted as a function of [~a_π]/, and functions will be generated by fitting data. Hence for each analyte there will be a set of functions, F_nm, that correlate the analyte concentrations with the signals from the reference and the analyte regions, R(Sa_n, Sr_m)ι = F_πm([~a_π]/). The number of functions is equal to the number of reference binding partners. For example if there are 3 analyte binding partner and 3 reference binding partner, there will be 9 functions in total. With the correlation functions stored in a database the following procedure is needed to carry out a multiplex quantitative assay using a single microarray.

1. Mix sample with the reference standard 2. Incubate the mixture with a single microarray

3. Optionally include a washing procedure

4. Incubate with an optional mixture containing additional binding partners facilitating detection

5. Optionally a further washing procedure 6. Read the microarray

7. If secondary detection molecules are needed, incubate the microarray with secondary detection molecules and apply necessary signal generation steps before reading the array.

Determine R(Sa_n,Sr_m). Determine the analyte concentrations using the obtained ratios and the correlation functions stored in the database. For each analyte a set of concentrations will be obtained [~a_π]. The average of this set of concentrations <[~a_π]> will give an accurate determination of the target concentration in the original sample.

Sandwich assays

In the case of sandwich assays the analyte and reference first binding partners could be capture antibodies, and the corresponding second binding partners would then be antigens with multiple epitopes, so that the analyte molecule(s) would in this case be antigens. Additional bind- ing partners facilitating detection would then be necessary to measure the binding. These additional binding partners could be different antibodies each capable of binding to a different epitope on the antigens (i.e. the second binding partners).

In a direct approach each of these additional binding partners would be coupled to e.g. HRP, AP, fluorescent dyes or radioactive labels allowing detection of the interactions between the first and second binding partners via the corresponding additional binding partners. However, when designing an assay involving many analyte and possibly also many reference binding partner pairs this approach would be rather cumber- some as all the different additional binding partners would individually have to be chemically linked to a tag. In a more elegant approach the additional binding partners would be antibodies from the same species (if necessary also of the same isotype) raised against the second binding partners. This would allow all additional binding partners to be detected using only a single appropriately labelled antibody raised against antibodies of the origin (and possibly isotype) of the additional binding partners. Such labelled antibodies against antibodies of various origins are readily available commercially in contrast to labelled antibodies raised against a certain antigen which would have to be custom made. Advantageously, no chemical modifications to the constituents of the sample are needed in these cases. However, both approaches described above for sandwich assays do fall within the scope of the present invention.

Direct binding assays

In the case of direct binding assays the procedures to both generate the correlation functions and the assay itself are simplified in terms of complexity of liquid handling. If the detection methods require a label the reference second binding partners must be labelled. During measure- ment of a signal from a sample containing unknown amounts of ana- lytes, the sample must be labelled. However, in case of label-free detection methods no labelling (and thereby no chemical modifications) of the sample constituents would be required.

Competitive assays

In the case of competitive assays the procedures to both generate the correlation functions and the assay itself are simplified in terms of complexity of liquid handling.

Fig. 2 illustrates a competitive ELISA according to a preferred embodiment of the present invention. In step 1, the substrate is func- tionalised with an analyte first binding partner (a^), three different reference first binding partners (r_lr r_2r r₃) and a control region (c) not contan- ing any immobilised binding partners. In step 2, analyte first and second binding partners (a^ and ^a₁, respectively) and reference second binding partners (^r_2-5) are added and allowed to interact with the immobilised first binding partners. In step 3, additional binding partners are added to facilitate the measurement of the binding between respective binding partners.

Theoretical considerations

Considering the design of heterogeneous assays, such as microarray immunoassays, most are performed under non-equilibrium conditions. Therefore, assay time will normally be a parameter to consider. For most solid-phase immunoassays the initial reactions (binding of soluble binding partners to immobilised ones) are diffusion limited, so that the amount of complex formed between a soluble second binding partner and an immobilised first binding partner is generally proportional to the distance between the binding partners, the concentration of the soluble binding partner and time, as well as a molecule specific diffusion coefficient. The correlation functions described above employ the ratio between signals from the interactions between the analyte binding partners to that of the reference binding partners. As is the case in the method of the present invention, when a liquid containing both the analyte second binding partner as well as the reference second binding partner in a well- mixed form (i.e. the concentrations of both these binding partners are homogeneous throughout the volume of the liquid) is applied to an array, the time dependencies of the diffusion of the reference and analyte binding partners cancel each other. This is a very important effect of the method of the present invention, since the method becomes independent of assay time.

When two, or more preferably three or more, reference binding partners are employed to generate the correlation functions further advantages are provided. These reference partners will be selected from molecules not expected to bind to molecules present in a sample to be analysed. However, there is always a risk that cross-binding may occur, but by using two, three or more reference binding partners, the risk of incorrect results caused by cross-binding can be nearly eliminated. This is particularly relevant in the case of competitive assays, where any un- wanted cross-binding may results in false positive measurements which are highly undesirable. Additionally by correlating the measurement of the binding of analyte binding partners with that of the binding of reference binding partners in a complex way where all obtained information is integrated and considered in the correlation functions, as is done in the methods of the present invention, much greater confidence will be obtained in the analysis of a sample with an unknown composition.

Empirical correlation functions In the method of the present invention the correlation functions may be empirical, though as will be appreciated by those of skill in the art, more complex correlations based on theoretical consideration may also be used. An example of such an empirical function is the four parameter logistic function presented below. R(Sa_n, SrJ = Sa_n /Sr_m = Fn([~ a_n]₀) =

Where A is the maximal signal (upper asymptote), E is the minimal signal (lower asymptote), C or IC₅O is the analyte concentration at the midpoint between A and E ((A + E)II) and B(A - f)/4 is the slope of the four-parameter function at IC₅O-

Preparation of mϊcroarrays

The substrate material used in the method of the present invention may be any conventional material and may have any convenient design. To be able to fit into a standard scanning apparatus the dimension of the substrate, when a microarray format is employed, usually resembles that of a microscope slide (e.g. 25 mm x 76 mm). The wells of a micro- titer plate may also serve as a substrate for preparation of microarrays and furthermore, the microarrays to be used in the method of the present invention may also be comprised in such devices commonly known as microfluidic devices. The surface of the material may be hydrophobic, hydrophilic, thiophilic or it may contain a contain mixture of areas with these characteristics. Suitable materials for the substrate include any rigid transparent or non-transparent material, however materials, such as glass, silicon, gold or polymers, such as polystyrene, polymethylmethacrylate), polypropylene, cyclic olefin copolymers (COC), polyethylene terephthalate, polycarbonate etc. are preferred. In addition to rigid materials soft materials, such as polydimethylsiloxane, agarose or poly- acrylamide gels may also be used. An example of a suitable material is MaxiSorp™ produced by Nunc A/S (Denmark).

The surface of the substrate material may be employed in an unmodified or modified form, or it may be activated to carry groups such as amines, thiols, aldehydes, epoxies, amino- or other silanes, polylysine etc. to allow the immobilised first binding partners to couple covalently to the surface. However, in general, the first binding partners are immobilised on the surface of the substrate material by physical adsorption, though in certain cases it may be necessary to induce covalent links be- tween the first binding partners and appropriate chemical moieties on the surface. Methods and conditions for immobilisation via chemical coupling of binding partners to surfaces are well-known to those skilled in the art.

Immobilisation of the first binding partners to the surface of the substrate material is generally performed by applying a liquid comprising the binding partner to the surface and allowing the binding partner to interact with the surface; the liquid may be in the form of a solution or suspension. The liquid of the first binding partners to be used for immobilisation is preferably an aqueous solution, and it may contain salts, polymers, and/or further additives. An organic solvent may be selected in special circumstances, e.g. in the event the binding partner to be immobilised is not sufficiently soluble in water. The content of the organic solvent(s) may be from 0-100%. As an example, the immobilisation solution may contain a certain amount of formaldehyde, betaine, or DMSO for reducing the evaporating time or to avoid or inhibit degradation of the first binding partner.

For the preparation of microarrays it is generally desired to add to the immobilisation solution such additives which will provide for a sufficiently high viscosity that reduces mass transport by capillary actions and provide for sufficiently low evaporation for allowing the binding partner to interact with the surface. An adjusted evaporation rate is also important for spot homogeneity and morphology. Various commercial spotting buffers are suitable. An example of such buffers is Genetix' amine spotting buffer.

The immobilisation liquid is generally applied in a quantity in the sub-μl range. In a preferred embodiment, the solution of the first binding partner is spotted in an amount of 10 nl or less. In a more preferred embodiment, each spot is applied in an amount of about 1 nl or less. A deposited amount of solution of 1 nl corresponds to a spot diameter of around 100 μm.

Two general approaches may be used for the application of immobilisation liquid to the substrate: contact and non-contact (printing) methods. Microarray fabrication using contact printing is based on high definition pins that, upon contact with the microarray substrate deposit a small amount of a probe solution. The pins are attached to a robotic arm that moves the pins between the different probe solutions, the microarray substrate and a washing station.

Non-contact printing is similar in terms of robotics but instead of pins, small dispensing systems are mounted on the robotic arm. The dispensing system can be based on inkjet, bubble-jet or piezo actuation technology and can usually dispense in the range of 100 pL to 2 ml_.

An example of a suitable apparatus for creating microarrays using contact printing is the QArray2 (Genetix Ltd., New Milton, L)K) which may be operated with a suitable printhead, such as the microarray pins of an SMP3 printhead (Arrayit, USA). Other types of contact printing equipment may be equally suited for creating microarrays for use in the methods of the present invention.

Following immobilisation of binding partners to the substrate surface for use in the method of the present invention, the substrate may further be inactivated by immobilisation of chemical entities considered inert to the assay concerned. Thus, in a preferred embodiment the substrate is blocked with bovine serum albumin (BSA) or ovalbumin (OA) after immobilisation of analyte and reference first binding partners. After the first binding partners have been immobilised and the substrate optionally inactivated, the substrate should be handled in a suitable fashion to avoid degeneration or degradation. Depending on the specific design of the microarray, the substrate may be used immedi- ately upon manufacture or it may be stored under dry or moist conditions at ambient or cold temperatures.

Measurement of signals

The interactions between binding partners may be detected and meas- ured using any appropriate method. In a preferred embodiment the assay method of the present invention involves the use of binding partners linked with enzymes, such as HRP or AP. When the method is performed in a microarray format, reaction products formed in the chromogenic, enzymatic reaction may be measured with an image scanner of sufficient resolution. Thus, for example a flat-bed scanner, such as the Cano- Scan 8400F (Canon) with a resolution of 1200 dpi (corresponding to a pixel size of around 20 μm), may be employed for measuring the signal. Relative quantification of the signals may then be performed using appropriate software, such as ScanAlyze2 (www.rana.lbl.gov/EisenSoftware.htm)

Examples

Example 1

Microarray fabrication A microarray was created for use in an assay method according to the present invention for the quantification of the algal hepatotoxin micro- cystin LR (MC).

Transparent MaxiSorp™ polymer microarray slides (of 76 x 25 mm² size), obtained from Nunc A/S (Denmark), were used as a substrate to immobilise analyte and reference first binding partners. The analyte first binding partner {a^} was microcystin LR (MC-LR) from DHI Water & Environment, Denmark. (Tyr[SOsH]²⁷) Cholecystokinin fragment 26-33 Amide (CCK) {o}, insulin {r₂} from Sigma-Aldrich and hu- man chorionic gonadotropin (hCG) {r₃} obtained from Fitzgerald, USA were employed as the reference first binding partners.

The first binding partners were each mixed with a spotting reagent (phosphate buffered saline, pH 7.2) to a final concentration of 0.5 g/L and spotted using the microarray pins of an SMP3 printhead (Ar- raylt, USA) operated by a microarray robot (Qarray 2, Genetix, UK) according to the manufacturers' guidelines. Following the spotting the slides were incubated for 1 hour at ambient temperature to allow the molecules to adsorb to the substrate. Each binding partner was spotted in a row of 8 replicates, and a row of spotting reagent was spotted between the rows of MC and hCG spots to function as a negative control. Thus, this resulted in an 8 x 5 array of spots, where the rows were: MC, spotting reagent {c}, hCG, insulin and CCK (see Fig. 1).

Eight arrays were spotted on each microarray slide. Six slides were fabricated in each batch. Circles were drawn around each array using a hydrophobic pen (Mini PAP PEN, Invitrogen, Denmark), which confined the liquid applied to the array during operation. The slides were then further treated and blocked with BSA in PBS buffer for 10 minutes at ambient temperature. After blocking, the slides were rinsed in water and dried.

Generation of correlation functions

The created microarray substrates were used for generating correlation functions for quantifying MC in a competitive ELISA format (Fig. 2). For this purpose the second binding partners were murine monoclonal antibodies to the respective first binding partners. Thus, the second binding partners were anti-MC (Alexis, Switzerland), anti-CCK (Statens Serum Institute, Denmark), anti-hCG (Fitzgerald) and anti-insulin (Sigma- Aldrich). For the detection of binding, a secondary anti-mouse antibody labeled with horseradish peroxidase (anti-mouse-HRP, Sigma-Aldrich) was employed with a peroxidase specific substrate (Sigma-Aldrich). This secondary antibody was capable of binding specifically to the second binding partners but could not bind to the first binding partners.

The eight arrays on a slide were each incubated for 1 hour at ambient temperature with the murine antibodies (second binding partners) in a buffer spiked with MC (analyte first binding partner) in different concentrations. The concentrations of MC were 0 μg/L, 0.1 μg/L, 0.3 μg/L, 0.6 μg/L, 1 μg/L, 2 μg/L, 3 μg/L and 10 μg/L, respectively. The concentrations of the four murine antibodies (second binding partners) in the mixture were 0.2 μg/mL for anti-MC and anti-CCK, 0.1 μg/mL for anti-insulin and anti-hCG. The buffer used was PBS with 0.05% BSA and 0.05% Tween. The volume applied to each array was 50 μL.

After 1 hour incubation, the slides were washed and dried. Anti- mouse-HRP (0.5 μg/mL, 50 μL) was applied to all arrays and incubated for 1 h. The slides were then washed and incubated with the HRP substrate for 10 min. The (colourless) HRP substrate was converted to a blue stain by the HRP which bound to the microarray spots. The arrays were washed and dried. The slide was then scanned using a flatbed scanner (CanoScan 8400F, Canon) at 1200 dots per inch resolution. The signals were recorded (and expressed in arbitrary units) using microarray image analysis software (ScanAlyze2). The data was transferred to a computer for further data analysis using appropriate software (Windows Excel or Origin v. 7.0). Selected microarray images after the reaction between the HRP and the peroxidase substrate are presented in Fig. 3. The upper-most row of spots are signals for MC, the second row is the negative control (spotting reagent alone), the third row is for hCG, the fourth is for insulin, and the last row is for CCK. As is seen from Fig. 3, the spotting reagent (SR) itself did not give any signal which indicates that neither of the second binding partners were adsorped to the microarray substrate material after blocking this with BSA. The signal for CCK was very weak. We observed that the signal of the reference spots did not vary much with the MC concentra- tion, while we clearly observed that the MC signal decreased with increasing free MC in solution. Thus, the analyte binding partners in solution did not seem to interact with the immobilised first reference binding partners.

The assay signal for the analyte binding partners decreased with an increasing concentration of the analyte first binding partner in the sample solution. This is typical for competitive assays where the assay signal decreases with increasing amount of antigen due to the competition of binding between the second analyte binding partner in solution and the immobilised and dissolved first analyte binding partners. Hence, we show that we have a competitive assay for MC.

The mean spot intensities at different MC concentrations are given in Table 1. Using the mean intensities, we calculated the ratios between MC and the reference signals, and plotted the ratios as a function of MC concentration (Fig. 4). A spline interpolation was used in place of curve fitting for data treatment. From the correlation curves we observed that the dynamic range of the assay was between 1 μg/L to 3 μg/L.

Table 1 Mean spot intensities

MC cone. (μg/L)

0 0,1 0,3 0,6 10

MC 4690 4816 4470 3935 3771 2453 189 199 spotting 246 417 470 334 428 322 146 100 reagent hCG 4637 6631 5879 5640 5020 5148 4845 4631 insulin 4153 4234 4100 4443 4157 3735 4059 3731

CCK 1215 1288 1349 1126 1229 694 1250 868

Probing assays The next step is to probe the correlation functions and compare them with the traditional method of conducting immunoassays, i.e. creating a standard curve in parallel with the test of samples. We generated a standard curve using 4 arrays on a slide. The MC concentrations used to generate the standard curve were 0 μg/L (blank), 0.3 μg/L, 1 μg/L, and 3 μg/L to cover the dynamic range observed above. The other 4 arrays on the same slide were used to test and probe the correlation functions and the standard curve. The probe MC concentrations were 0 μg/L (blank), 0.6 μg/L, 2 μg/L, and 10 μg/L. The assay conditions and time were the same as for the assays used for generation of the correlation curve. The standard curve (a linear fit) is shown in Fig. 5. From the standard curve it seems that there was a difference between 0 and 1 μg/L, however, since the errors in our assay was about 15%, the dynamic range for the standard curve was practically also between 1 and 3 μg/L.

Using the signals and ratio results for the probe assays we de- termined the concentrations by reading from the standard and correlation curves, respectively. For example, for the blank probe we calculated an MC/hCG ratio of 1.3, and we then read from the MC/hCG correlation curve in Fig. 4 and obtained the corresponding MC concentration, which in our case was below 1 μg/L. The results are listed in Table 2.

Table 2. Experiment 1 - Dynamic range: 1 - 3 μg/L Probe MC (μg/L) O 0,6 2 10

^~ΪAC 4316 4105 2372 ¹^T

MC/HCG 1 ,3 1, 3 0 ,65 -0 ,01 MC/CCK 6 ,3 5, 6 2 ,4 -0 ,06 MC/insulin 1 ,3 1, 5 0 ,78 -0 ,02 MC cone, read from standard curve (μg/L) <1 <1 \J_ > 3

MC cone, read from correlation functions

MC/HCG (μg/L) < 1 < 1 1,3 > 3

MC/CCK (μg/L) < 1 < 1 2,3 > 3

MC/insulin (μg/L) < 1 < 1 1,4 > 3

Average (μg/L) <1 <1 1,7 > 3

From Table 2, it is evident that the correlation functions gave the same results as the standard curve.

The above experiments were repeated using different batches of slides with small assay adjustments in the range of MC concentrations and the dynamic range. The results are summarised in Table 3 and Table 4.

Table 3. Experiment 2 - Dynamic range: 1 - 3 μg/L Probe MC (μg/L) O 0,6 2 10

MC cone, read from standard curve (μg/L) \Λ_ <1 \β_ >_3

MC cone, read from correlation functions

MC/hCG (μg/L) < 1 < 1 1.9 > 3

MC/CCK (μg/L) < 1 < 1 1.7 > 3

MC/insulin (μg/L) 1,2 < 1 2.1 > 3

Average (μg/L) <1 <1 1.9 > 3

Table 4. Experiment 3 - Dynamic range: 0.3 - 3 μg/L Probe MC (μg/L) O 0.3 1 3

MC cone, read from standard curve (μg/L) 0.6 0.4 1.2 > 3

MC cone, read from correlation functions

MC/hCG (μg/L) 0.4 <0.3 0.8 > 3 MC/CCK (μg/L) <0.3 <0.3 0.7 > 3

MC/insulin (μg/L) <0.3 <0.3 1.6 > 3

Average (μg/L) <0.3 <0.3 1.0 > 3

For the blank sample in experiment 2 (Table 3), we see that the MC concentration determined using the MC/insulin ratio was 1.2 μg/L, i.e. a false positive. However, since the other two ratios indicate that no microcystin was present, we could rule out the presence of MC based on the average of all three reading. On the other hand, if we used only the standard curve (1.0 μg/L), we would be in a very hesitant situation, since the concentration read from the standard curve was just on the limit of our dynamic range. In experiment 3 (Table 4) we had a similar situation as in experiment 2. This time the MC/hCG gave 0.4 μg/L for the blank sample. Again using the other two ratios we could determine with confidence that the signal was beyond our dynamic range. However, if the standard curve was used alone, a false positive for the blank sample was ob- tained.

Conclusions

From the above experiments it is seen that the correlation curves gave correct results. The results were comparable to the results obtained us- ing conventional standard curves. By using an odd number of correlation spots, in this case three, potential false positives could be corrected for. One false positive was avoided compared to using the standard curve. The correlation functions proved to be more robust and confident in cases where the signal was on the limit of the dynamic range. Thus, using the assay method of the present invention decreases the risk of obtaining false positive results in the analysis of a sample.

References

Belleville, E., Dufva,M., Aamand,J., Bruun,L., Clausen, L., and Christen- sen, C. B. V. (2004). Quantitative microarray pesticide analysis. Journal of Immunological Methods 286, 219-229.

Du, H. W., Yang,W.P., Xing, W. L, Su,Y., and Cheng,J. (2005). Parallel detection and quantification using nine immunoassays in a protein microar- ray for drug from serum samples. Biomedical Microdevices 7, 143-146. Dufva,M., Christensen,C. (2005). Diagnostic and analytical applications of protein microarrays. Expert Rev. Proteomics 2(1), 41-48 Han,A., Dufva,M., Belleville,E., and Christensen,C.B.V. (2003). Detection of analyte binding to microarrays using gold nanoparticle labels and a desktop scanner. Lab Chip 3, 329-332.

Pawlak,M., Schick,E., Bopp,M.A., Schneider,MJ., Oroszlan,P., and Eh- rat,M. (2002). Zeptosens' protein microarrays: A novel high performance microarray platform for low abundance protein analysis. Proteomics 2, 383-393. Urbanowska,T., Mangialaio,S., Hartmann,C, and Legay,E. (2003). Development of protein microarray technology to monitor biomarkers of rheumatoid arthritis disease. Cell Biology and Toxicology 19, 189-202. Weissenstein,U., Schneider,MJ., Pawlak,M., Cicenas,J., Eppenberger- Castori,S., Oroszlan,P., Ehret,S., Geurts-Moespot,A., Sweep,F.C.GJ., and Eppenberger,U. (2006). Protein chip based miniaturized assay for the simultaneous quantitative monitoring of cancer biomarkers in tissue extracts. Proteomics 6, 1427-1436.

Wiese,R., Belosludtsev,Y., Powdrill,T., Thompson, P., and Hogan,M. (2001). Simultaneous multianalyte ELISA performed on a microarray platform. Clinical Chemistry 47, 1451-1457.

Wild, D. (2005). The Immunoassay Hand Book., D. Wild, ed. (Amsterdam: Elsevier).

Wingren,C, Borrebaeck,C. (2004). High-throughput proteomics using antibody microarrays. Expert Rev Proteomics 1(3), 355-64 Zhu,H., Snyder,M. (2003). Protein Chip Technology. Current Opinion in Chemical Biology 7(1), 55-63

Claims

P A T E N T C L A I M S

1. An assay method for quantifying analyte molecules in a sample comprising a) providing a substrate containing two or more reference regions containing different immobilised reference first binding partners and one or more analyte regions containing immobilised analyte first binding partner(s), where the reference first binding partners and the analyte first binding partner(s) have affinities for binding different second binding partners; b) applying a mixture containing reference second binding partners and a sample suspected of containing one or more analyte first or second binding partner(s), in which mixture the concentrations of the reference second binding partners are known; c) allowing the mixture to interact with the substrate; d) measuring the binding of the analyte binding partner(s) in the mixture to the corresponding immobilised analyte binding part- ner(s) and the binding of the reference second binding partners to the immobilised reference first binding partners; and e) quantifying the concentration of analyte binding partner(s) in the sample, where the quantification is performed by correlating the measurement of the binding of the analyte partners to the measurement of the binding of the reference binding partners.

2. Assay method according to claim 1, wherein the mixture containing the reference second binding partners and the sample comprises analyte first and second binding partners.

3. Assay method according to any of claims 1 or 2, wherein the two or more reference regions and the one or more analyte regions are arranged in sets.

4. Assay method according to any of claims 1 to 3, wherein the regions are discrete.

5. Assay method according to any of claims 1 to 4, wherein the substrate contains three or more reference regions.

6. Assay method according to any of claims 1 to 5, further comprising the step of adding one or more additional binding partners which additional binding partners are capable of binding to the analyte second binding partners and/or to the reference second binding partners, and which additional binding partners are not capable of binding to the immobilised first binding partners, and where the additional binding part- ners facilitate measurement.

7. Assay method according to any of claims 1 to 6, wherein the binding partners comprise proteins or peptides.

8. Assay method according to any of claims 1 to 7, wherein the binding partners comprise antibodies.

9. Assay method according to any of claims 1 to 8, wherein the binding partners comprise microcystin LR (MC-LR), human chorionic gonadotropin (hCG), insulin and (Tyr[SO₃H]²⁷) cholecystokinin fragment 26-33 (CCK) and antibodies raised against MC, hCG, insulin and CCK.

10. Use of two or more first reference binding partners in an as- say for quantifying analyte molecules in a sample in order to minimise the risk of false positive results which assay comprises a) measuring the binding between analyte first and second binding partners and the binding between the two or more reference first binding partners and corresponding reference second binding partners; and b) quantifying the concentration of analyte binding partner(s) in a sample, where the quantification is performed by correlating the measurement of the binding of the analyte partners to the measurement of the binding of the reference binding partners.

11. Use according to claim 10, wherein the assay is a competitive assay.