THEREOF

The present invention relates to a protein microarray and in particular to a method of obtaining proteins for use in a microarray, and to the use of a protein microarray. In particular, the invention relates to the use of a protein microarray to determine the general immunoglobulin profile of an individual to proteins they have been exposed to, and in particular, for the in vitro diagnosis of an allergy in an individual.

In the general population a large percentage of people are affected by allergic reactions to a variety of environmental immunogens, making atopic disorders one of the most prevalent disorders in the industrialised world. An allergic reaction is a reaction produced by the body's immune response to a substance that would normally be considered to be harmless. Allergic reactions can be caused by a multitude of environmental factors, including certain food products, pollen, animal dander, drugs, other chemicals, insect venom, parasites and commensal bacteria.

Allergic reactions can cause a variety of symptoms, including, irritation of the nose, sneezing, itching and redness of the eyes, narrowing of the airways (bronchoconstriction) , increased production of mucus in the lungs, shortness of breath (dyspnea), coughing, wheezing, abdominal pain, bloating, vomiting, diarrhoea, itchy skin, swelling of the skin, cutaneous reactions, oedema, hypotension, anaphylaxis, coma, and even death. The severity of the allergic response varies depending on the allergen and the individual.

In the light of the prevalence of allergic reactions there is a growing need to accurately determine what allergen is causing an allergic reaction in a particular individual, and/or to understand the immune response. Existing immunoglobulin (Ig) tests used to diagnose one or more allergens which are causing a particular allergic reaction in an individual only give a limited picture of the immunological response in an individual to a particular allergen, in particular with reference to food antigens. Furthermore, existing tests require large volumes of sample, cover only a limited number of allergens, and in particular only a limited number of foodstuffs, are not amenable to high sample through-put systems and the results are limited to normally just one immunoglobulin class. There is therefore a need for an improved method to identify allergens which cause an immune response in an individual, and/or to understand in more detail the nature of the immune response.

Sera of immune competent individuals contain antibodies of the IgG, IgA, IgM, IgD and IgE immunoglobulin classes. Despite their very short half -life, average of 2.5 days for IgE and about 3 weeks for the other classes, the overall concentration of antibodies in serum is remarkably stable during late childhood- adulthood, with only small variations at old age. Regarding their origin, "natural" antibodies of the IgM, IgA and IgG3 subclasses are usually derived from Bl antibody-secreting cells. In general they possess a low affinity to multiple antigens, a basic requirement for their regulatory role in innate immunity, and are maintained throughout life. Similarly after vaccination or infection, persistent levels of specific protective antibodies, which are the products of adaptive immunity and B2 plasma cells, are detectable in human serum for decades. The maintenance of this humoral memory, acquired and passively transferred in early development via milk and placenta in some species, depends on the activation of short and long-lived secreting Bl and B2 plasma cells whose lifetime and terminal differentiation is still poorly characterised.

The presence and prevalence of antibodies to microbial antigens and infectious agents are reasonably well characterised. Similarly a great number of reports have addressed the presence of food specific antibodies (Hvatum M et al, Gut

1992; 33: 632-638; Quinti I et al, Allergy 1989;44: 59-64; Barnes RMR, Clinical and Experimental Allergy 1995;25: 7-9.) however, the implications and roles of their undeniable presence has only recently started to be addressed

The gut immune system is bombarded daily by a plethora of protein antigens present in our normal diet. The expected outcomes from these exposures are mainly (i) induction of tolerance; (ii) systemic priming; or (iii) induction of local immunoglobulins. Breakdown or failure attributed to these responses result in yet not well understood immunological adverse reactions to foods, such as food allergy, food intolerance and inflammatory bowel diseases. Whereas food allergy, in particular the specific IgE response, has been investigated in some detail, many adverse reactions to food cannot be explained by IgE-mediated (type I) hypersensitivity alone (Hvatum M et al, Journal of Immunological Methods 1992; 148: 77-85) and little attention has been paid to the presence of other antibody classes to foods, or indeed other environmental allergens

According to a first aspect, the present invention provides a method for determining the specific immunoglobulin profile of an atopic or a non-atopic individual, comprising

i) providing a sample obtained from the individual;

ii) contacting the sample with a protein microarray containing proteins or protein extracts from one or more potential allergens;

iii) detecting one or more immunoglobulins bound to the microarray.

Preferably the sample is contacted with the microarray under conditions which will allow antibodies in the sample which are specific for proteins on the microarray to bind thereto.

Preferably the method can be used to detect more than one type of immunoglobulin. The method of the invention may be used to detect the binding of one or more of IgG, IgA, IgM, IgE or IgD immunoglobulins or subclasses thereof. The subclasses may be the subclasses of IgG, for example, IgGl, IgG2, IgG3 or IgG4 or subclasses of IgA, for example IgAl and IgA2. Preferably the method of the invention allows at least two, preferably at least three, preferably at least four different immunoglobulin classes or subclasses to be detected.

Preferably more than one type of immunoglobulin can be detected simultaneously or separately.

The terms "antibody" and "immunoglobulin" are used interchangeably herein.

The method of the invention may allow multiple potential allergens to be assayed at once, preferably the microarray includes at least 50, 100, 200, 300, 400 or more different proteins or protein samples. Accordingly, the method may be used in high throughput screening.

The proteins in the protein microarray may be proteins extracted from a foodstuff, parasite, aeroallergen, commensal bacteria or another source of potential environmental allergen. Alternatively the proteins may be synthetically produced or recombinantly produced. The microarray may comprise a mixture of two or more of proteins wherein one or more of the proteins is extracted from an allergen is a synthetic protein or is a recombinant protein, or a mixture thereof.

The microarray may contain protein extracts representative of most of the food products/ingredients consumed in the UK. This may represent about 16 different groups of food such as spices, seeds, meat, fish and dairy amongst others. Such an array may be used to determine simultaneously the presence in the sample of immunoglobulins from two or more different immunoglobulin classes, or subclasses, for example, at least two of IgA, IgM, IgG and IgE, directed against one or more potential allergens. Reference herein to "allergens" refers to a composition comprising a mixture of compounds including proteins which has the capacity to induce an immune response in an allergic individual. Preferably the allergens are protein. The immune response may cause the production of allergen specific antibodies, the antibodies may be one or more of IgG, IgA, IgD, IgM and IgE immunoglobulins or subclasses thereof.

Preferably the proteins in the microarray are immobilised on a nitrocellulose base slide (Fast slide) .

Preferably the sample is a sample of bodily fluid such as milk, saliva, blood or serum, preferably blood or serum.

The bound immunoglobulins may be detected by any suitable method.

Preferably, the bound immunoglobulins are detected using labelled secondary antibodies or antibody fragments, specific to a particular immunoglobulin class.

For example, FITC-labelled anti-IgA antibodies will allow the detection of bound IgA antibodies. By using a different label, on different secondary antibodies, different classes of immunoglobulin bound to the microarray may be identified.

The labels may be detected by spectrophotoscopic, photochemical, biochemical, immunochemical, electrical, optical or chemical means. The label may be selected from the group comprising biotin, a fluorescent dye such as fluorescein, rhodamine or the green fluorescent protein, a radiolabel, and an enzyme such as horseradish peroxidise.

Preferably secondary antibodies directed to different immunoglobulin classes have different labels, for example, different fluorescent labels. An anti-IgG antibody may have a rhodamine label, an anti-IgE antibody may have a GFP label and an anti-IgA antibody may have a fluorescein label. Preferably, if the secondary antibodies are fluorescently labelled they can be detected using a spectrophotometer which is capable of measuring at more than wavelength at the same time, for example has two or more lasers. This will allow rapid throughput and screening.

In one embodiment the present invention provides a protein microarray comprising at least one extract of protein sample from at least 100 different foodstuffs, preferably at least 200, preferably at least 300, preferably at least 400 different foodstuffs. The one or more protein extracts for each foodstuff are preferably crude extracts of the foodstuffs. In addition, or alternatively, to foodstuffs the microarray may also comprise protein extracts obtained from human parasites, or bacteria, such as pathogenic and/or commensal bacteria. The microarray may further comprise proteins obtained from pollen, animal dander and insect venom, preferably for each foodstuff, bacteria or other potential allergen included in the array there is more than one protein extract.

By correlating the class of antibody bound and the protein sample to which it is bound the immunoglobulin profile of an individual can be prepared. Preferably an immunoglobulin profile contains information related to the classes/subclasses of immunoglobulin's in a sample to which are proteins or protein extracts in the microarray. This profile may then be used to diagnose whether an individual is allergic to a specific potential allergen or a family of allergens or their degree of susceptibility to certain diseases, such as inflammatory bowel disease, Crohn's disease or other gut related immunodeficiency.

Preferably the immunoglobulin profile of an individual provides information on the presence of one or more immunoglobulin specific to one or more protein sample. Preferably the immunoglobulin profile contains information about two or more, three or more, four or more, five or more, six or more, seven or more different immunoglobulin classes or subclasses. Preferably the immunoglobulin classes or subclasses are selected from the group comprising IgA, IgG, IgM, IgE, IgGl, IgG2, IgG3, IgG4, IgAl and IgA2.

Preferably by performing the method of this embodiment on a sample obtained from a human it is possible to determine whether the individual has had an allergic reaction to a particular substance.

The method of the invention also has the advantage that small sample volumes may be used. Much smaller than the sample volumes needed in most commercially available tests. The method of the invention may be performed on samples, for example of blood or serum, as small as lOOμl, 50μl, 25μl or less.

The protein microarray system has the advantage that the results are highly reproducible.

The method of the invention may be used to diagnose an allergy in an individual or to obtain the immunoglobulin profile of an individual.

According to a further aspect the invention provides a method of obtaining protein extracts from a sample, said method comprising the steps of a) obtaining a sample; b) homogenising the sample; c) adding a first buffer comprising a detergent to the sample; d) fractionating the product of c) to recover the extracted proteins; e) adding a second buffer comprising a chaotropic agent to the sample remaining after step d) . f) fractionating the product of e) to recover the extracted proteins.

The sample is preferably a sample of a potential allergen. The sample may be of a food product, or human parasites, or it may be bacterial sample, or a pollen sample, or an insect venom sample, or a sample of animal dander, or a sample of any other potential allergen.

The sample may be freeze dried before use, and may be stored frozen before use. The sample may be subjected to hexane extraction prior to step b), this may be required if the food sample contains 25% or more lipid.

In step b) the sample may be homogenised by mechanical grinding, this may be automatic or manual, for example, with a pestle-and-mortar.

The detergent in the first buffer may be a non-ionic, an anionic, or a zwitterionic detergent. Preferably the buffer comprises about 0.05 to 3% w/v of detergent. Preferably the detergent is Triton X, preferably the Triton X is used at a concentration of 0.5% w/v.

The second buffer preferably comprises a chaotropic agent at a concentration of about 4 to 8M. Preferably the chaotropic agent is urea, preferably the urea is used at about 6M.

The first and second buffer may also comprise one or more of the following: PBS; dithiothreitol; glycerol and phenylmethylsulfonyl fluoride (PMSF) or any other protease inhibitor.

Preferably if the first and/or second buffer contain PBS, the PBS is used at a concentration of 0.5X or IX PBS (wherein 0.5X PBS is 69 mM NaCl, 5mM Phosphate, 1.35 mM KCl, pH 7.4; and IX PBS is 137 mM NaCl, 10 mM Phosphate, 2.7 mM KCl, pH 7.4). Preferably IX PBS is used if the food sample contains meat.

Preferably if the first and/or second buffer contain dithiothreitol, it is used at a concentration of ImM dithiothreitol. Preferably if glycerol is used, it is used at a concentration of 20% w/v glycerol. Preferably if PMSF is used, it is used at a concentration of ImM PMSF.

Preferably in steps c) and e) at least some of the protein is solubilised.

Preferably in step c) and/or e) the sample and buffer are mixed to solubilise the proteins to allow them to be extracted. Preferably the sample and buffer are mixed by applying a mechanical shear force. This may be achieved by using ball bearings, or a pestle, or by mechanical grinders. Preferably the mechanical mixing is performed at a temperature of at least about 5 ⁰ C or less than room temperature. The mechanical mixing step may be carried out only once, or it be repeated several times. Preferably, in the method of the invention in step c) the sample and buffer are subjected to mechanical mixing twice.

Preferably in step e) the sample comprises the pellet remaining after at least some of the protein solubilised in step c) has been recovered by the fractionation in step d) .

Preferably in the fractionated step of step d) and f) the solubilised protein is extracted by recovering the supernatant.

Preferably the fractionated sample containing the protein extract is further clarified by centrifugation. This may be achieved by centrifugation at about 13000g for about 20 minutes.

The fractionated sample containing the protein extract may be further purified by passing it through a filter. Preferably a 0.45μm filter is used.

The fractionated sample may be further enriched in its protein content by using anion exchange chromatography, for example, by using a Q-sepharose column. Preferably, after the protein has been solubilised by buffer A or B, the solubilised protein is recovered by fractionating the sample and retaining the supernatant containing the soluble protein extract, at the end of the process the insoluble pellet is preferably discarded.

Preferably extracts of protein are recovered after some or all steps of the method. Protein recovered after some or all steps of the method may be used to make the microarray of the invention.

Preferably the protein extracts recovered in the method of the invention are used to produce a microarray. The total volume of the extract used may be in the range of nanolitres.

According to a further aspect, the invention provides a protein microarray comprising extracts of protein obtained from at least 10, 20, 30, 40, 50, 60, 100, 150, 200, 250, 300, 350, 400 or more potential allergens. More than one extract from each potential allergen may be included in the array. The potential allergens may be food stuffs, human parasites, bacteria, dander, venom or any other protein containing potential allergen.

The proteins in the protein microarray may be extracted from the potential allergen using the aforementioned method of the invention, and then used to produce a protein microarray according to the invention.

According to a yet further aspect, the invention provides a protein microarray comprising at least one protein extracted according to the aforementioned method of the invention. Preferably, the protein microarray comprises at least 10, 20, 30, 40, 50, 60, 100, 150, 200, 250, 300, 350, 400 or more protein extracts obtained using aforementioned method of the invention. According to another aspect, the invention provides a method for diagnosis of an allergy or for immunoglobulin profile comprising combining basophil degranulation with a protein microarray according to the invention.

Preferably the basophile degranulation is performed as described in Lin et al in Clinical and Experimental Allergy (2007) 37, 1854-1862, modified to use the microarray of the invention. Essentially, the method of this aspect of the invention comprises incobating basophils isolated from a patient, or basophils which have been stripped of immunoglobulins and resensitised with the serum of a patient, with a protein microarray and then determining where on the array the basophils bind, and using this data to produce on immunoglobulin profile and/or to determine the allergic status of the patient.

According to a yet another aspect, the invention provides a kit which comprises a microarray according to the invention and instructions to use the kit. Preferably the kit also includes at least one immunoglobulin detecting reagent. The immunoglobulin detecting reagent is preferably a secondary antibody (an antiimmunoglobulin antibody), specific to a particular class and/or subclass of immunoglobulin .

Preferably the kit also provides a control known antibody at a known concentration which is specific for a control protein on the microarray, preferably this can be used as a positive control when using the kit. Similarly, the kit may also include a second known antibody, again at a known concentration, which does not bind to any of the proteins in the array, this may be used as a negative control. Preferably the kit includes a reagent for detecting the first known antibody, and/or a reagent for detecting the second known antibody.

The skilled man will appreciate that preferred features discussed with reference to only some aspects of the invention can be applied to all. Embodiments of the present invention are described by way of example with reference to the following figures.

Figure 1 - shows the SDS-PAGE profile of a crude extract containing either triton/urea before sample loading, compared to elution fractions from MonoD ^{^} and MonoQ -Sepharose columns, using cod extracts as an example. MonoQ-Sepharose was found to have superior performance for all food extracts in terms of binding and elution yields.

Figure 2 - shows immunoglobulin-specific IgA (A), IgM (B), IgG (C),

IgE (D) calibration curves and a mixed reference serum curve (E) after correction for the spill over effect.

Figure 3 - illustrates the relative immunoglobulin cross reactivity of a panel of human sera to food products. The patients are designated: (1)

SOTON 004, (2) SOTON 029, (3) SOTON030, (4) SOTON040, (5)

SOTONOOl, (6) SOTON034, (7) SOTON036, (8) SOTON037, (9)

SOTON039, (10) SOTON051, (11) PEIOOl, (12) PEI007, (13) PEIOl 6,

(14) PEI016, (15) PEI023. In some panels the scale of concentrations has been raised in order to facilitate the visualization.

Figure 4 - illustrates the correlations between immunoglobulin classes within the human serum panel. The correlation values have been mapped to colours from blue ~0 to red < 0.8 (see colourbar on the right). The vertical axis denotes for which patient and the horizontal axis for which array the correlation has been calculated.

Figure 5 - shows the PCA scores for the four classes of immunoglobulins (n per serum sample/class = 4800) . Figure 6 - shows a comparison of total responses for IgG, IgE, IgA and IgM, grouped based on food type, as a comparison between non- atopic (black) and atopic (white) patients. Significance for each immunoglobulin is shown within each section by an asterisk.

Figure 7 - shows the results of competition of a food-specific immunoglobulin-A (non-atopic mother) micro-ELISA to food extracts of macadamia and sour cream. Plates were coated with approximately lOμg/ml of food extract. Food extract along with negative control proteins (PBS, BSA, rSFA8) were added to various diluted samples from the non-atopic mother to allow competition between binding immunoglobulins with supplemented proteins.

Figure 8 - shows the partial least square (PLS) results for prediction of UniCAP grade by IgE. a) Measured x predicted UniCAP grade showing the mathematical fitting of the model, b) VIP scores for the different food products. The higher the VIP scores, the more important the variable is for prediction.

Figure 9 - shows cluster analysis of the four classes of immunoglobulins and the serum samples (15). The main classes of food products are indicated. In order to facilitate the visualization, the food product data were log transformed, centered, normalized and then clustered by the K- means algorithm using k= 15 (Eisen et al, Proceedings of the National Academy of Sciences of the United States of America vol 95, pg 14863,

1998).

Materials and Methods Other proteins and Reagents Human albumin, chicken albumin, and bovine serum albumin were obtained from Sigma (Sigma, UK). Recombinant Brazil nut 2S albumin (Ber el) and recombinant sunflower 2S albumin (SFA 8) were produced in P. pastoris as previously described (Alcocer et al, Journal of Molecular Biology 2002; 324: 165-175). Grass pollen extract was kindly donated by Dr Helmut Haas (Research Centre Borstel, Germany) and natural Ara h 2 was provided by the University of Southampton. Control antibodies (IgG, IgM and IgA) were purchased from Calbiochem (Merck Biosciences, UK); IgE was purchased from Abd (Serotec, UK) and control serum from the ELISA IgA/IgG quantitation kits (Bethyl, Texas, USA). All other food samples were produced by the method outlined below. Nitrocellulose coated glass slides (FAST slides) were supplied by Whatman Schleicher & Schuell (Dassel, Germany). If not otherwise stated, all chemical reagents are from Sigma UK. The antibodies were purchased from: Cy3-conjugated Goat anti-human IgA (Jackson Immuno Reseach, Baltimore, USA), Alexa 488-conjugated Goat anti-human IgM (Molecular Probes Europe BV, Holland), biotinylated Goat anti-human IgE (Vector Laboratories, Burlingame, USA), Goat anti-human IgG (Zymed, USA) was labelled in-house using Alexa 594 Protein Labelling kit (Molecular Probes) .

Food Samples

A comprehensive list containing most food products/ingredients available in the UK (ca. 400) was compiled from websites of supermarkets

(www.sainsburystoyou.com www.tesco.com, www.waitrosedeliver.com - Oct

2005), the UK Food Standard Agency (McCance and Widdowson, 2002), dietary control (www.caloriedatabase.com, www.weightlossforgood.co.uk- Oct

2005), herb and spice (www.sulfolkherbs.com - Sep 2005) (www.simplyspice.co.uk, www.exotic-fruits.co.uk, www.chinaexporter.com -

Oct 2006) and allergy concern organisations (www.allergyuk.org - Oct 2005),

(www.food-database.co.uk- Oct 2005). Small samples (l-5g) or whole foods were then purchased or donated from shops and restaurants within the

Leicestershire-Nottinghamshire (UK) area. Samples were collected either in paper (manila-free) envelopes or plastic (10ml) tubes (Sterilin, UK) and immediately frozen (-20 ⁰ C) until further processing. Table 1 lists the main food products utilized in this array.

Vegetables Fruits Herbs and Spices

Alfalfa, Artichoke, Asparagus, Apple, Apricot, Banana, Arrowroot, Basil, Bay Leaves,

Aubergine, Avocado, Baby Blackberry, Blackcurrant, Black Pepper, Camomile,

Chard Leaf, Bamboo Shoot, Blueberry, Cherry, Cardamom, Cayenne Pepper,

Beansprout, Beetroot, Bhindi, Clementine, Coconut Chilli Powder, Cinnamon, Cloves,

Borlotti Bean, Broad Bean, (Desiccated), Coconut Milk, Coriander (Ground), Cumin,

Broccoli, Brussels Sprout, Currants, Dates, Fig, Curry Leaves, Curry Powder,

Butternut Squash, Cabbage, Gooseberry, Grape, Custard Powder, Dili, Earl Grey,

Caper, Carrot, Cauliflower, Grapefruit (Red), Guava, Fennel Seed, Fragrance spice,

Ceieriac, Celery, Chervil (Fresh), Italian Kiwi, Italian Kiwi Garam Masala, Ginger (Ground),

Chickpea (Cooked), Chilli (Hayward), Kiwi, Kiwi Juniper Berry, Lemongrass, Lime

(Green), Chilli (Red), Chinese (Protein), Lemon, Lemon Leaves, Liquorice Root,

Cabbage, Chives (Fresh), Rind, Lime, Lychee, Mandarin Marjoram, Mint, Nutmeg,

Coriander Leaf, Courgette, Peel, Mango, Melon, Oregano, Paprika, Parsley,

Cress, Cucumber, DaI Chini, Nectarine, Orange, Orange Peppermint, Pimiento Pepper,

Fennel, Ganna (Spain), Garlic, Marmalade, Papaya, Passion Rosemary, Saffron, Sage,

Gherkin, Leek, Lettuce, Lettuce Fruit, Paw Paw, Peach, Pear, Tarragon, Tea Leaves, Thyme,

(Lambs), Maize Babycorn, Physalis, Pineapple, Plum, Turmeric.

Mange Tout, Marrow, Pomegranate, Prune,

Mushroom, Mushroom (Oyster), Pumpkin, Raisins,

Mushroom (Shitake), Okra, Olive Rambutant, Raspberry,

(Black), Olive (Green), Onion Rhubarb, Sharon Fruit,

(Red), Onion (White), Pak Choi, Strawberry, Sultanas,

Parsnip, Peas, Pepper (Red), Tamarind (Dried), Tamarind

Potato, Radish, Rocket, Runner (Fresh), Tangerine,

Bean, Seaweed, Shallot, Watermelon. Spinach, Spring Onion, Swede, Sweetcorn, Texturised Vegetable protein, Tomato, Turnip, Water Chestnut, Watercress.

Beans Nuts Cereals

Baked Bean, Black Bean, Almond, Almond, Almond Basmati Rice, Bread (White), Black Canna (Chickpea), (Blanched), Barley Seed, Corn, Corn Starch, Cornflour, Black eyed Bean, Black Brazil Nut, Brazil Nut Cous Cous, Egg Noodle, Kidney Bean, Butter Bean, (Protein), Cashew, Flour (Chapatti), Lentil (Red),

Cannalini Bean (Cooked), Chestnut, Hazelnut, Oats, Pasta, Pearl Barley,

Chicory, Coffee Bean, Hazelnut (Organic), Lectin Rice (Arborio), Rice Flageolet Bean, Green (Peanut), Macadamia, (Basmati), Rice (Red), Rye, Mung Bean, Haricot Bean, Paleskin Nut, Peanut, Sago, Tapioca, Wheat, Lectin (Soya), Lentil Yellow Peanut (Blanched), Wheat Flour, Wheat Germ,

(Moog dal), Pinto Bean, Red Pecan, Pine Kernel, Pine Wheat LTP. Bean, Red Kidney Bean, Nut Kernel, Pistachio, Rye Soya (Total Protein), Soya Seed, Walnut, Walnut Bean, Soya Bean (UK Soy), (Organic). Soya Flour, Soya Milk, Soya Milk (Extract), Soya Sprout, Split Chick pea

(ChannaDal), Tofu (Extract).

Seeds Dairy Meat

Aniseed, Aniseed Star, Brie, Buttermilk, Cheddar, Bacon, Beef, Beef Stock, Caraway Seeds, Celery Cottage Cheese, Cow's Chicken, Chicken Liver, Seed, Cocoa, Coriander Milk, Cow's Milk (Whey), Duck, Gelatine, Ham, Kidney Seed, Fenugreek, Green Cream, Edam, Feta, (Pig), Lamb Liver, Pepperoni, Pepper Seeds, Hemp Seed, Fromage frais, Goat's Pig Liver, Pork, Sausage Linseed, Mustard (Black), Milk, Greek Yogurt, Casing, Smoked Bacon, Mustard (Yellow), Mustard Mozzarella, Parmesan, Suet, Turkey, Venison. Seeds (Black), Pomegranate Sour Cream, Stilton, Tofu Seed, Poppy Seed, Rape (Paneer), Yogurt (Plain). Seed, Sesame, Sunflower Seed, White Pepper Seed.

Fish Drink and Sauce Others

Anchovy, Cod, Coley Fish, Ale, Beer, Black Bean Baking Powder, Chocolate, Herring, Kipper, Mackerel, Sauce, Black Pepper Chocolate (Plain), Cream of Pilchards, Plaice, Salmon, Sauce, Cranberry Sauce, Tartar, Glutamine (MSG), Sardine, Smoked Haddock, Horseradish Sauce, Malt Pichia, Rock Salt, Sugar Sprat, Trout, Tuna. Vinegar, Marmite, Satay (Brown), Sugar (Muscavado),

Sauce, Sherry, Soya Yeast (Dry).

Sauce, Vegetable Stock,

Worcestershire Sauce.

Shellfish Eggs Inhaled Allergens

Cockles, Crab, Crab Duck Egg, Duck Egg Birch Pollen, Grass Pollen, (Cooked), Mussels, Oyster, (White), Duck Egg (Yolk), Grass Pollen (Velvet), House Oyster Sauce, Prawn, Duck egg white, Egg dust mite, Scallop, Squid. Albumin, Egg Ovalbumin,

Hen Egg, Hen Egg

(White), Hen Egg (Yolk).

Table 1 - List of all food ingredients (about 400) included in the "all-diet" protein microarray.

Sample Processing Food samples were freeze-dried and then immediately returned to a freezer in sealed airtight containers in the presence of silica gel to avoid rehydration. Batches of samples were then homogenised based on food type. Some food samples were amenable to mechanical grinding either by hand, or in a pestle-and- mortar whilst others, whose lipid levels were more than 25% (meat, fish, nuts and some seeds), required a hexane extraction prior to grinding. The proteins were extracted using two buffers (A & B) containing 0.5x Phosphate Buffered Saline, (0.5xPBS- 69 mM NaCl, 5mM Phosphate, 1.35 mM KCl, pH 7.4); for a meat sample Ix PBS was used (137 mM NaCl, 10 mM Phosphate, 2.7 mM KCl, pH 7.4) . Both buffers also contained ImM dithiothreitol, 20% (w/v) glycerol (both from Sigma, UK) and ImM phenylmethylsulfonyl fluoride (PMSF - Perbio Science, UK). PMSF was added just prior to use due to its relatively short half- life. Buffer A contained 0.5% w/v Triton-100 (Sigma, UK) and Buffer B 6M urea (Fisher, UK). Protein were extracted mechanically using a Mixer/Miller 8000 (Glen Creston Ltd, UK) and 1.2ml 96-well format 8-strip microtubes (Starlab, UK), each containing three 3mm stainless steel ball bearings (Midland Bearings Ltd, UK) and either (1.2ml) of Buffer A or B. The conditions were optimised to avoid generation of excessive heat, i.e. a short time (10 min) and temperatures of 5°C below room temperature prior to and after milling. After two rounds of extractions and centrifugations (MSE Mistral 300Oi, Sanyo, UK) 250Og - 10 minutes with Buffer A, supernatants were collected and the pellet re- extracted twice with Buffer B using the same procedure.

Sample Clarification, Purification and Normalisation The food extracts were further clarified by high speed centrifugation in eppendorf tubes (Hawk 15/05 refrigerated, Sanyo, UK) at 13,00Og for 20 min. The supernatant was then filtered under a vacuum using a 96-well format silica based filtration unit (Pall-96, 0.45μm pore-Qiagen USA). The filtrates were further purified by anion exchange chromatography using Q-Sepharose 96-well micro- spin columns according to the manufacturers' instructions (VIVAPure, UK). The columns were equilibrated with extraction buffer (200 μl) and centrifuged at lOOOg for 5 minutes. Two (300 μL) volumes of sample were loaded to the column and the centrifugation repeated at the same conditions. The column was washed with two volumes (2 x 200 μL) of extraction buffer and then eluted with IM NaCl. The eluates from all steps were collected, sealed, labelled and stored frozen at -20°C. Protein concentration of each extract was determined by the bicinchonic acid (BCA) method according to the manufacturer's instructions (Pierce, UK) using a robotic arm (Biomek® 3000 Laboratory Automation Workstation, Beckman, UK). The food extracts were then normalised using dedicated software supplied by the mechanical arm manufacturer (Beckman) to lmg/ml of protein using the appropriate buffer. Eluate samples were diluted to 0.5mg/ml. Samples that had an original protein concentration below these levels were left undiluted. All normalised samples were then transferred to 96 well plates (Sterilin, UK) and finally to 384 well microarray plates (Genetix, UK), sealed and stored at -20°C until further use.

Protein Array Printing

Normalised samples were printed onto FAST slides (Whatman Schleicher &

Schuell, Dassal, Germany) using a QArraylite arrayer (Genetix, UK) at the Centre for Biomolecular Sciences, University of Nottingham to a final spot density of 4800 spots/slide. The printing of each sample with spot size of approximately 200μm diameter and the same size spacing was replicated three times across the slide into 48 individual blocks. Spots of PBS and diluted Strepavidin-Cy3/Cy5 were included for quality and alignment purposes. Slides were left to dry overnight and then blocked in 3% BSA (w/v) with shaking in PBS at 37 ⁰ C. The slides were then washed three times for 2 minutes in PBS containing 0.05% (w/v) Tween-20, followed by five times 60 seconds washes with purified water, and finally dried by centrifugation (MSE Mistral 3000i, Sanyo, UK) at 30Og for ten minutes at room temperature.

Sera/Maternal Milk

Sera from patients previously characterised and reported as monospecific and polyspecific to Brazil Nut (Murtagh et al, 2003. Clinical and Experimental Allergy 33:1147-1152) and Soya Bean proteins (Lin et al., International Archives or Allergy and Immunology 2006; 141:91-207) was used. SPT/RAST/UniCap data were either provided with the samples or were conducted by an accredited commercial lab, Pathology Lab (TDL Pathology, London, UK). Total IgE values were confirmed in-house by total-IgE ELISA to ensure correlation between all sample sets. Further new samples of matching sera and maternal milk were obtained from 4 non-atopic and 6 atopic mothers (University of Southampton, Local Research Ethics Committee approval 04/Q1702/57). The mothers were skin tested to a panel of allergens: cow's milk, hen's egg, house dust mite, cat dander, grass pollen mix, tree pollen mix, dog dander, Alternaria, Aspergillus on the velar surface of the forearm using a 1 mm lancet (Soluprick, ALK, Denmark). Saline was used as a negative control and histamine (10 mg/ml) as positive control. Skin prick tests were considered positive if larger than 3mm.

Protein Microarray Hybridisation

Microarray hybridizations were carried out using a Lucidea Slide Pro Hybridiser (Amersham, UK), optimised in-house for the protein array. Essentially the overall protocol involves multiple washes with PBS + T (0.2% w/v Tween-20) containing 1% BSA and incubations of diluted sera (1:15) in the same buffer for two hours at 26 ⁰ C. After another round of washings, the second antibody was then added and incubated for a further 1 hour at 26 ⁰ C. Final washes were carried out, three times with PBS + T 0.2% and five times with water for two and one minute duration respectively. The slides were then dried by centrifugation at 300g for 10 minutes.

Slide Scanning and Data Handling

The dried microarray slides were scanned in a four colour Axon 4200AL autoloader scanner (Axon Instruments, USA) with a PMT gain set to 100% and an adjustable PMT voltage and power (percentage variable) to vary detection sensitivity. The maximal laser excitation wavelengths used were 488, 532, 594 and 635nm. While emission filters with a narrow bandwidth between 15-20nm as specified by the manufacturer (Axon Instruments) ensured a reduction in spillover from other fluorophores. Slides were scanned at lOμm resolution with

'lines to average' set to 1.

Data Processing

Data from the scanner was processed using GenePix Pro software v6.0.1.27 (Axon Instruments). Triplicate spot readings were averaged for both the sample slide and the control slide (no sera/milk). Control data was subtracted from the sample slide to eliminate non-specific binding and inherent autofluorescence of some proteins using Visual Basic macros inbuilt into Microsoft Excel software

(Microsoft, USA). Data was converted to a relative concentration using internal standard curves for IgM, IgA (Calbiochem, USA), IgG (Sigma, UK), and IgE

(Serotec, UK) certified pure immunoglobulins. The spill-over effect was eliminated by subtraction of best-fit lines of the pure immunoglobulins measured by the four lasers and plotted along the linear range of curves. Sera/milk sample data were compiled for all immunoglobulin classes and transferred from Excel to Matlab (version 7.1 (R14SP3), The Mathworks Inc., USA) using Excel link toolbox (Mathworks) and the Dataset Object (Version 5.0, Eigenvector Research Inc., USA). In order to visualise the variance in the arrays, the multivariate method principal component analysis (PCA) was applied on the array datasets using the PLS toolbox (Version 4.1.0, Eigenvector Research). Additionally, Partial Least Squares Regression (PLS) was performed to establish the multivariate link between the array data sets and the highest UniCAP results. The centering, normalization and clustering analysis was carried out using the open source software Cluster 3.0 developed by Michael Eisen at Stanford University and made available by the Michiel de Hoon, Human Genome Center, University of Tokyo.

Micro ELISAs

Total human-IgA, IgG and IgE content were determined by ELISA using commercial kits (Bethyl Laboratories, USA) following the manufacturer's instructions and adapted to a 384 well scale. Competitive micro ELISAs were carried out essentially as previously described (Lin J et al, International Archives of Allergy and Immunology 2006; 141: 91-102) with some modifications. Micro- ELISA plates were coated with a range of food extracts (squid, sour cream, macadamia, egg white and others) diluted 1:100 in coating buffer, except for macadamia which was diluted 1:1000. Serum was preincubated with a 1:10 serial dilution series of the competitive proteins from 0.0053μg/ml to 500μg/ml for one hour at RT.

Results

Assembling the "all diet" protein microarray In order to determine the specificity of the serum immunoglobulins (IgA, M, G and E) towards food products, a comprehensive "all-diet" protein microarray was produced that used protein extracts, rather than pure proteins, to effectively detect specific immunoglobulin responses. The "all-diet" array contains extracts from most food products/ingredients (n = 400) available in the UK representing 16 groups of products such as spices, seeds, meat, fish, dairy and others. Each product was represented in the array by at least 12 spots representing different stages of protein purification and solubility in buffer solutions containing detergent and chaotropic agents. The food extracts produced using these procedures were of a complex nature and were aimed to be representative of the natural exposure of the digestive tract to food products.

In order to extract the largest number of proteins from different food groups many extraction buffers were tested with mixed results (data not shown). After exhaustive tests the 2 step buffer system using PBS/detergent and chaotropic agent was shown to be effective. To limit problems during loading in PBS subsequent Q-Sepharose chromatography steps, a lower initial concentration of PBS (0.5x) was used in most extracts, with the exception of meat and fish products. Addition of DTT, glycerol and PMSF to the extraction buffers had no apparent effect on the proteins extracts, this merely controlled protein stability after extraction. The time and temperature used during extractions were also optimized to improve the solubilisation and avoid protein degradation and denaturation. A chaotropic agent was used to solubilise proteins otherwise insoluble in the phosphate-detergent buffer. Four ion exchange chromatographic resins were evaluated and MonoQ-Sepharose in particular demonstrated superior performance and yields at the physiological pH employed as shown by SDS- PAGE analysis (Figure 1). After extraction, a filtration step (0.45 μm filter) was deemed preferable due to column fouling in subsequent chromatographic steps.

The food extracts were printed on the 3D-polymeric solid surface of FAST slides, previously shown (Lin J et al, Clin Exp Allergy 2007;37: 1854-1862; Pang S et al, Journal of Immunological Methods 2005;302: 1-12) to provide a stable matrix for protein arrays. Several techniques were tried for the printing of the array with the food extracts amongst them split pins, solid pins and inkjet printing. The results presented here were all the product of an automated protocol using array slides printed using solid pins only.

In general the spots of food extracts and purified fractions represented in the array were complex mixtures of proteins of discrete composition.

As discussed elsewhere (Kusnezow and Hoheisel 2003; 16: 165-176; Renault NK et al, Biotechnol Lett 2007;29: 333-339) one key factor when choosing the surface platform on which to immobilize the proteins is the spot morphology. Crude and protein purified extracts were printed to the 3-D surface of FAST slides with an average spot diameter of 200 μim. Some spots however had a diameter of up to 300 μm due to the increased viscosity caused by urea on foods such as wheat, corn, orange rind and olive. These tended to have a very complex carbohydrate-lipid structure resulting in more sample attached to the outside of the pins.

The protein array data are quantifiable and reproducible

In order to determine four classes of immunoglobulins simultaneously using the microarray, four different fluorochrome dyes were chosen and either covalently linked in house to secondary antibodies or were purchased already conjugated to secondary antibodies. The aim being that immunoglobulins of different classes would bind to different spots on the microarray, these could then be detected by using labeled second antibodies, for example, IgE could be detected using a labeled anti-IgE antibody. The automated hybridization and selection of the second antibody-fluorochromes were subsequently optimized. Great attention was paid to the cleanliness of the equipment and slides, as trace amounts of dyes, markers, dust and non degassed liquids affected the results. As shown in Figure 2, the use of the mixture of secondary antibody-fluorophore complexes resulted in an efficient hybridisation and labeling protocol. Within the secondary antibody mixture all but one contained the fluorophores covalently attached to the antibodies. In order to achieve the amplification required for a quantitative IgE detection, specific anti-IgE secondary antibody was biotinylated and detected by a further labeling step with strepavidin-Cy5 incubation. The major concern was specificity of the secondary antibody and measurement of the expected spillover effect, product of multiple fluorophores by using different excitation lasers and various emission filters. Careful calibration curves with known concentrations of pure immobilized immunoglobulins and mixtures (Figure 2) have shown that the system is specific for each class of immunoglobulin and, where emission spectra were close together, minimal spillover could be cancelled out. Within the system described, the fluorescence response shown in Figure 2 is relative to the concentration of standards i.e., is not directly proportional to the antibody concentration in solution but a function of the immobilized amount of protein after blocking and washing steps. These standard spots were essential internal controls and used mainly for measurement of batch-to-batch variation. As shown in Figure 2, after correction of spillover a linear dynamic range for the immobilized standards was obtained for 3 classes (IgA, IgG, and IgE). The blue filter for the pentameric IgM, although quantifiable, did saturate at lower concentration than the others (Figure 2b). The PMT gains of the scanner were also optimised for the individual lasers and kept constant throughout. Under these experimental conditions the profile of immunoglobulin of a standard human serum to food protein fractions was determined (Figure 2e) .

Due to the auto-fluorescence of some food extracts it became evident that a blank control array slide, run at the same conditions and able to establish the fluorescent background of each protein spot was preferred. Under these conditions the intra and inter slide reproducibility of 1600 protein fractions printed in triplicate blocks with the same human serum, after the spill over and background correction, were in the range of 98% and 94% (p < 0.001) respectively (data not shown). Although large variations were observed between different patients, stressing here the importance of their genetic background and previous exposure, the overall reproducibility of samples from the same patient withdrawn in different days (0, 1 and 60 days) were surprisingly stable with similar correlations within the range of 0.91 to 0.99 (P < 0.001) within different immunoglobulin classes and different days (data not shown). No significant statistic differences were found (P > 0.2) between days for the same subject. Overall correlations of antibody contents between milk and sera from non-atopic (NA) patients were in the range of 0.7-0.9 for all immunoglobulin classes. Atopic mothers on the whole showed a particular lower range between 0.5-0.7 (P < 0.05) for IgA being particularly low for some individuals in this class.

Different immunoglobulin classes have shown distinct specificity In order to evaluate the sensitivity and specificity of the "all-diet" food protein microarray a retrospective panel of well-characterized sera from atopic and non- atopic patients was employed (Table 2). Although reduced in numbers the panel of atopic sera possessed a large variation of total IgE contents and was representative of various UniCAP grades. With the exception of milk allergens, the panel covered important groups of food allergens commonly found in the UK. Furthermore, samples of young children and adults were represented. As shown in Figure 3, the profile of four classes of immunoglobulins to food components for this particular panel of human sera showed a widespread selectivity. Overall significant correlations were observed between subjects and their immunoglobulin classes (Figure 4). In general IgA has shown a better correlation to IgM than IgG, and IgG correlates better with IgA than IgM. No significant correlation was observed between IgE and any of the other class of immunoglobulin.

In order to facilitate the visualization of the putative differences in responses between individual immunoglobulin classes, the collective results were analyzed by a multivariate technique such as principal component analysis (PCA) (Figure 5) . PCA decomposes a data matrix in a set of loading and scores vector pairs (the principal components) which describe the most important data trends present in the dataset. In such a vector pair (or PC) the loading vector explains which linear combination of variables (in our case: food products) are responsible for a particular data trend. The scores vector contains the information of how important or strong this trend is for each sample.

Figure 5 includes the data for non-atopic volunteer control sera which is shown a triangles connected by solid lines. Figure 5 shows that the IgE class has the greatest discrimination power between atopic and non-atopic subjects. One interesting feature of the IgE cross reactivity concerned the pattern of response exemplified by patients PEI007, 016 and 023. These patients, clinically characterized as soy milk allergic subjects, were characterized in a previous study and did not cross-react with the ubiquitous major allergenic family of 2S albumins (Lin J et al, International Archives of Allergy and Immunology 2006; 141: 91-102). In the more comprehensive array used here, these three subjects clearly show a typical polyclonal response. This difference is even more accentuated if their IgE responses were compared to mono-specific subjects such as PEIOOl. Interestingly, IgG has also shown some degree of discrimination for some of the atopic sera.

Some groups of foods were more relevant for specific immunoglobulin classes In order to confirm the PCA results ANOVA analyses were used as a discriminatory tool. Due to the discontinuous nature of the data, the non- responders were removed from the dataset and a near normal distribution was achieved after transformation. With the exception of IgE, the fractions of atopic and control non-responders of the other classes have shown to be equivalent. Within this dataset no statistical differences were observed between the total IgM and IgA content of the non-atopic patients and atopic patients (p= 0.26 and 0.78 respectively). This is reflected in the PCA analysis in Figure 5 where controls and atopic groups are not well separated. Furthermore, although higher antibody levels could be observed for IgA in groups such as meat, fish and shellfish, the relationship is reversed in milk samples and practically non-existent for IgM (Figure 6).

As expected, large differences were observed between atopic and control subjects within the IgE class (p < 0.001). As a group response, these elevated IgE levels were elevated in atopic subjects within all food groups particularly in seeds, fish, beans, nuts, dairy and eggs (Figure 6). Unexpectedly, the observed the total IgG content of atopic subjects were reduced when compared to non-atopic ones

(p < 0.001). This reduction in IgG content was also observed at the group level (Figure 6) with all average IgG levels to food groups of atopic subjects being reduced when compared to the control group, with particular emphasis to meat, fish, shellfish and beans.

The array binding data can be confirmed by ELISA Competitive ELISAs were performed to establish whether the antibodies detected to food extracts were specific rather than cross-reactive. Extracts ranging from squid, macadamia, sour cream and hen egg yolk were immobilized and used to confirm the specificity of various immunoglobulins to diluted food extracts. As exemplified in Figure 7 the strong positive array results obtained with total extracts of macadamia and sour cream were specifically competed out in an ELISA technique.

The array IgE results correlated with the UniCAP data

It was clear from the above results that the "all-diet" protein microarray may be used effectively in the detection of IgE. In particular, it is evident that the global read-out obtained from such arrays could be of benefit to clinical situations. With this in mind, clinically characterized and previously analyzed serum samples were sent for further UniCAP analysis at an independent and accredited clinical lab and the results are shown in Table 2. Table 2

κ>

00

Total IgE I 25 I 1 I N/A I N/A I 60 I 95 I 54 I 40 I 77 I 40 I 1789 I 262 I 1550 I 3855 I 2080 I KU/ml

I 312) {Existing Clinical Data acquired after microarray testing

Table 2:

Clinical Allergy Reactivity and UniCAP IgE Quantitation for Control non-atopic patients (SOTON 004, 029, 030, 040), atopic patients (SOTON 001, 034, 036, 037, 039, 051), Monospecific patient (PEIOOl), and Polyspecific patients (PEI 006, 007, 016, 023) 'SoybeanO' denotes positive oral challenge to soybean. SPT (Skin Prick Tests) determined at Southampton University; C = Cat, D = Dog, G = Grass Pollen, P = Peanut, E= Egg, HDM = House dust mite, T = Tree Pollen Mix. N/ A = Not Available. ND = Not Determined. UniCAP classifications are negative (0) 0-0.35 kUA/L; borderline (1) 0.35-0.69 kUA/L; positive (2) 0.70-3.49 kUA/L; strong positive (3) 3.5-17.49 kUA/L; highly positive (4) 17.5-49 kUA/L; highly positive (5) 50-99 kUA/L; highly positive (6) > 99 kUA/L.

The comparison of the UniCAP and array results has shown that subjects PEIO 16/023 for instance, although originally described as soy milk allergic individuals, had a typical polyclonal response that contrasted to the monoclonal response of PEIOOl. In order to confirm these results some of the samples with similar high total IgE content were re-analysed by the UniCAP technique, based this time on the array results (shaded boxes with bold outline - Table 2). The patient classed as monospecific to Brazil nut, PEIOOl in Table 2, showed on re- analysis a low reactivity to peanut and egg (yolk). Interestingly, polyspecific patient PEI 016 when tested for egg (yolk), peanut and hazelnut, resulted in UniCAP classes of 3, 3 and 6 respectively. Likewise PEI 023 also returned very high UniCAP results for egg (yolk) (class 4), grass pollen (class 6) and hazelnut (class 5) . Both polyspecific patients showed a whole range of high IgE reactions in the microarray analysis (Figure 3d, patients 14 and 15). Therefore, although only for few samples, the new UniCAP re-analysis did confirm the polyclonal nature of the PEIO 16/023 samples as suggested by the array. Some groups of food are recognized by all patients

In order to produce a graphic overview of the specific reactivity of the four classes of immunoglobulins, a clustering analysis was carried out as shown in

Figure 9. Different immunoglobulin classes showed cross-reactivity with different families of food products. Most of the serum IgGs cross-reacted with samples of milk, egg and shell fish products (Figure 9). Likewise IgM classes were largely biased towards sea food products. In agreement with Figure 4, IgA showed the same bias towards sea food products as detected for IgM. The clear distinction of the different patterns of IgE specificity and the few positive peaks of the control subjects are shown.

Discussion

The data produced from this small set of subjects using the "all-diet" protein microarray were considerable. An average of 20,000 data points per run of patient was obtained. Overall it is clear that the data were of a discontinuous nature, with serum samples either responding to the extracts of proteins in a poissonic distribution mode or not responding at all. The class of antibody dictated the specificity of the food group recognized. Dairy and egg products were recognized by IgGs from most subjects, atopic or controls. Others, such as fish and shellfish, were mostly recognized by IgA and M. interestingly, the overall results of multiple serum samples withdrawn from the same adult over a period of time were remarkably constant. The same results were observed with breast milk of several volunteers collected daily over a 3 week period.

The results presented here demonstrate that the method of the invention can be used to produce an individual's immunoglobulin profile. The serum samples analysed in this work represented a well characterised cohort with a variety in reactivity, gender and age for which previous results were available. The low UniCAP grade of some samples, although borderline for atopy, reflects the reality of many clinical practices. The microarray of the invention has been shown to be versatile and able to profile simultaneously the four major immunoglobulins classes found in human sera from samples as small as a drop of blood from a finger prick. It is clear from the results that one of the great advantages of the technique described is a very small volume of sera is required for the analysis ( < 25 μl) . Small blood samples (2-4 drops) from pricked fingers may now be used, a stark contrast to present day tests that require substantial volumes (milliliters), to screen for but a few allergens (Renault et al, Biotechnol Lett 2007;29: 333-339; Zar et al, American Journal of Gastroenterology 2005;100: 1550-1557).

The data presented shows that using the method of the invention it has been demonstrated that food specific IgA and IgM do correlate to each other and that IgE has a low correlation with any of the other immunoglobulin classes. Specific food groups such as dairy /eggs and fish were recognized by most subjects by IgG and IgM respectively. The array system has shown a great discrimination between atopic and non atopic individuals for IgE and surprisingly has shown a reduction in the overall IgG content for atopic patients. Poly and mono-specific IgE responders were easily identified. The results show that the "all-diet" protein microarray can compete with present standard IgE techniques. The array system possesses many advantages over traditional systems such as requirement of low sample volume, high sensitivity and a global view of the immune response.

The results demonstrate that by using a multivariate system of data analysis, the data obtained from a microarray of the invention can be used to produce mathematical models that can describe an event and be used for predictions. In areas such as microarray technology where the number of measurements can be high, an understanding of the whole, instead of a series of independent variables can be advantageous.

Finally, the "all-diet" protein microarray is a viable and useful tool for the global profiling of the four major human food-specific immunoglobulins and their subclasses. One advantage is the global view of the immune response, or "fingerprinting" of the immune response to the abundant food proteins which the microarray allows. Another advantage is the requirement for low sample volumes. The inclusion in the array of proteins from parasites and commensal bacteria, or other allergens, would further expand this "fingerprinting" concept of the immune response. Due to its high sensitivity, the technique is also successful on some otherwise intractable samples. Recently a sample of blood from a case of fatal anaphylaxis was analyzed using the "all-diet" microarray and the results showed quite compellingly that the unfortunate individual had extremely high levels of IgE to a range of food products. The serum sample, due to its age, haemolysis and nature was unable to be processed using the conventional systems. The method of the invention may be used as a substrate for basophil degranulation, for maternal milk profiling, for post-mortem assessment and for investigating aspects of food preference, as well as allergy diagnosis.