US 20040002078 A1
Protein arrays and their use to assay, in a parallel fashion, the protein products of highly homologous or related DNA coding sequences and described. By highly homologous or related it is meant those DNA coding sequences which share a common sequence and which differ only by one or more naturally occurring mutations such as single nucleotide polymorphisms, deletions or insertions, or those sequences which are considered to be haplotypes. Such highly homologous or related DNA coding sequences are generally naturally occurring variants of the same gene. Arrays according to the invention have two or more individual proteins deposited in a spatially defined pattern on a surface in a form whereby a property such as an activity or function of the proteins can be investigated or assayed in parallel by interrogation of the array.
1. A protein array comprising a surface upon which are deposited at spatially defined locations at least two protein moieties characterised in that said protein moieties are those of naturally occurring variants of a DNA sequence of interest.
2. A protein array as claimed in
3. A protein array as claimed in
4. A protein array as claimed in
5. A protein array as claimed in
6. A protein array as claimed in any of the preceding claims wherein the protein moieties comprise proteins associated with a disease state, drug metabolism or those which are uncharacterised.
7. A protein array as claimed in any of the preceding claims wherein the protein moieties encode wild type p53 and allelic variants thereof.
8. A protein array as claimed in any of the
9. A protein array as claimed in
10. A method of making a protein array comprising the steps of
a) providing DNA coding sequences which are those of two or more naturally occurring variants of a DNA sequence of interest
b) expressing said coding sequences to provide one or more individual protein moieties
c) purifying said protein moieties
d) depositing said protein moieties at spatially defined locations on a surface to give an array.
11. The method as claimed in
12. The method as claimed in
13. The method as claimed in
14. The method as claimed in
15. The method as claimed in
16. The method as claimed in
17. Use of an array as claimed in any of
18. A method of screening a set of protein moieties for molecules which interact with one or more proteins comprising the steps of
a) bringing one or more test molecules into contact with an array as claimed in any one of
b) detecting an interaction between one or more test molecules and one or more proteins on the array.
19. A method of simultaneously determining the relative properties of members of a set of protein moieties, comprising the steps of:
a) bringing an array as claimed in any one of
b) observing the interaction of said test substances with the set members on the array.
20. The method of
 This application claims the benefit of U.S. provisional patent application No. 60/335,806, filed Dec. 5, 2001, and of U.S. provisional patent application No. 60/410,815, filed Sep. 16, 2002, the complete disclosures of each of which are herein incorporated by reference.
 Single nucleotide polymorphisms (SNPs) are single base differences between the DNA of organisms. They underlie much of the genetic component of phenotypic variation between individuals with the exception of identical siblings and clones. Since this variation includes characteristics such as predisposition to disease, age of onset, severity of disease and response to treatment, the identification and cataloguing of SNPs will lead to ‘genetic medicine’ [Chakravarti, A. Nature 409 822-823 (2001)]. Disciplines such as pharmacogenomics are aiming to establish correlations between SNPs and response to drug treatment in order to tailor therapeutic programmes to the individual person. More broadly, the role of particular SNPs in conditions such as sickle cell anaemia and Alzheimer's disease, and issues such as HIV resistance and transplant rejection, are well appreciated. However, correlations between SNPs and their phenotypes are usually derived from statistical analyses of population data and little attempt is made to elucidate the molecular mechanism of the observed phenotypic variation. Until the advent of high-throughput sequencing projects aimed at determining the complete sequence of the human genome [The International Human Genome Mapping Consortium Nature 409 860-921 (2001); Venter, J. C. Science 291 1304-1351 (2001)], only a few thousand SNPs had been identified. More recently 1.42 million SNPs were catalogued by a consortium of researchers in a paper accompanying the human sequence [The International SNP Map Working Group Nature 409 928-933 (2001)] of which 60,000 were present within genes (‘coding’ SNPs). Coding SNPs can be further classified according to whether or not they alter the amino acid sequence of the protein and where changes do occur, protein function may be affected resulting in phenotypic variation. Thus there is an unmet need for apparatus and methodology capable of rapidly determining the phenotypes of this large volume of variant sequences.
 The Inventors herein describe protein arrays and their use to assay, in a parallel fashion, the protein products of highly homologous or related DNA coding sequences.
 By highly homologous or related it is meant those DNA coding sequences which share a common sequence and which differ only by one or more naturally occurring mutations such as single nucleotide polymorphisms, deletions or insertions, or those sequences which are considered to be haplotypes (a haplotype being a combination of variations or mutations on a chromosome, usually within the context of a particular gene). Such highly homologous or related DNA coding sequences are generally naturally occurring variants of the same gene.
 Arrays according to the invention have multiple (for example, two or more), individual proteins deposited in a spatially defined pattern on a surface in a form whereby the properties, (for example, the activity or function of the proteins) can be investigated or assayed in parallel by interrogation of the array.
 Protein arrays according to the invention and their use to assay the phenotypic changes in protein function resulting from mutations (for example, coding SNPs—i.e. those SNP mutations that still give rise to an expressed protein) differ completely from, and have advantages over, existing DNA based technologies for SNP and other mutational analyses [reviewed in Shi, M. M Clin Chem 47 164-72 (2001)]. These latter technologies include high-throughput sequencing and electrophoretic methods for identifying new SNPs, or diagnostic technologies such as high density oligonucleotide arrays [e.g. Lindblad-Toh, K. Nat Genet 24 381-6 (2000)] or high-throughput, short-read sequencing techniques which permit profiling of an individual's gene of interest against known SNPs [e.g. Buetow, K. H. Proc Natl Acad Sci USA 98 581-4 (2001)]. Importantly, and in contrast to the invention described herein, the phenotypic effects of a polymorphism remain unknown when only analysed at the DNA level.
 Indeed, the effects of coding SNPs on the proteins they encode are, with relatively few exceptions, uncharacterised. Examples of proteins with many catalogued SNPs but little functional data on the effect of these SNPs include p53, p10 (both cancer related) and the cytochrome P450s (drug metabolism). There are currently few if any methods capable of investigating the functionalities of SNP-encoded proteins with sufficiently high throughput required to handle the large volume of SNP data being generated. Bioinformatics or computer modelling is possible, especially if a crystal structure is available, but the hypotheses generated still need to be verified experimentally (i.e. through biochemical assay). Frequently though, the role of the mutation remains unclear after bioinformatic or computer-based analysis. Therefore, protein arrays as provided by the invention offer the most powerful route to functional analysis of SNPs.
 It would be possible to individually assay proteins derived from related DNA molecules, for example differing by one or more single nucleotide polymorphisms, in a test tube format, however the serial nature of this work and the large sample volumes involved make this approach cumbersome and unattractive. By arraying out the related proteins in a microtiter plate or on a microscope slide, many different proteins (hundreds or thousands) can be assayed simultaneously using only small sample volumes (few microlitres only in the case of microarrays) thus making functional analysis of, for example, SNPs economically feasible. All proteins can be assayed together in the same experiment which reduces sources of error due to differential handling of materials. Additionally, tethering the proteins directly to a solid support facilitates binding assays which require unbound ligands to be washed away prior to measuring bound concentrations, a feature not available in solution based or single phase liquid assays.
 Specific advantages over apparatus and methods currently known in the art provided by the arrays of the present invention are:
 massively parallel analysis of closely related proteins, for example those derived from coding SNPs, for encoded function
 sensitivity of analysis at least comparable to existing methods, if not better
 enables quantitative, comparative functional analysis in a manner not previously possible
 compatible with protein: protein, protein: nucleic acid, protein: ligand, or protein: small molecule interactions and post-translational modifications in situ “on-chip”
 parallel protein arrays according to the invention are spotting density independent
 microarray format enables analysis to be carried out using small volumes of potentially expensive ligands
 information provided by parallel protein arrays according to the invention will be extremely valuable for drug discovery, pharmacogenomics and diagnostics fields
 other useful parallel protein arrays may include proteins derived from non-natural (synthetic) mutations of a DNA sequence of interest. Such arrays can be used to investigate interactions between the variant protein thus produced and other proteins, nucleic acid molecules and other molecules, for example ligands or candidate/test small molecules. Suitable methods of carrying out such mutagenesis are described in Current Protocols in Molecular Biology, Volume 1, Chapter 8, Edited by Ausubel, F M, Brent, R, Kingston, R E, Moore, D D, Siedman, J G, Smith, J A, and Struhl, K.
 Thus in one aspect, the invention provides a protein array comprising a surface upon which are deposited at spatially defined locations at least two protein moieties characterised in that said protein moieties are those of naturally occurring variants of a DNA sequence of interest.
 A protein array as defined herein is a spatially defined arrangement of protein moieties in a pattern on a surface. Preferably the protein moieties are attached to the surface either directly or indirectly. The attachment can be non-specific (e.g. by physical absorption onto the surface or by formation of a non-specific covalent interaction). In a preferred embodiment the protein moieties are attached to the surface through a common marker moiety appended to each protein moiety. In another preferred embodiment, the protein moieties can be incorporated into a vesicle or liposome which is tethered to the surface.
 A surface as defined herein is a flat or contoured area that may or may not be coated/derivatised by chemical treatment. For example, the area can be:
 a glass slide,
 one or more beads, for example a magnetised, derivatised and/or labelled bead as known in the art,
 a polypropylene or polystyrene slide,
 a polypropylene or polystyrene multi-well plate,
 a gold, silica or metal object,
 a membrane made of nitrocellulose, PVDF, nylon or phosphocellulose
 Where a bead is used, individual proteins, pairs of proteins or pools of variant proteins (e.g., for “shotgun screening”—to initially identify groups of proteins in which a protein of interest may exist; such groups are then separated and further investigated (analogous to pooling methods known in the art of combinatorial chemistry)) may be attached to an individual bead to provide the spatial definition or separation of the array. The beads may then be assayed separately, but in parallel, in a compartmentalised way, for example in the wells of a microtitre plate or in separate test tubes.
 Thus a protein array comprising a surface according to the invention may subsist as series of separate solid phase surfaces, such as beads carrying different proteins, the array being formed by the spatially defined pattern or arrangement of the separate surfaces in the experiment.
 Preferably the surface coating is capable of resisting non-specific protein absorption. The surface coating can be porous or non-porous in nature. In addition, in a preferred embodiment the surface coating provides a specific interaction with the marker moiety on each protein moiety either directly or indirectly (e.g. through a protein or peptide or nucleic acid bound to the surface). An embodiment of the invention described in the examples below uses SAM2™ membrane (Promega, Madison, Wis., USA) as the capture surface, although a variety of other surfaces can be used, as well as surfaces in microarray or microwell formats as known in the art.
 A protein moiety is a protein or a polypeptide encoded by a DNA sequence which is generally a gene or a naturally occurring variant of the gene. The protein moiety may take the form of the encoded protein, or may comprise additional amino acids (not originally encoded by the DNA sequence from which it is derived) to facilitate attachment to the array or analysis in an assay. In the case of the protein having only the amino acid sequence encoded by the naturally occurring gene, without additional sequence, such proteins may be attached to the array by way of a common feature between the variants. For example, a set of variant proteins may be attached to the array via a binding protein or an antibody which is capable of binding an invariant or common part of the individual proteins in the set. Preferably, protein moieties according to the invention are proteins tagged (via the combination of the protein encoding DNA sequence with a tag encoding DNA sequence) at either the N- or C-terminus with a marker moiety to facilitate attachment to the array.
 Each position in the pattern of an array can contain, for example, either:
 a sample of a single protein type (in the form of a monomer, dimer, trimer, tetramer or higher multimer) or
 a sample of a single protein type bound to an interacting molecule (for example, nucleic acid molecule, antibody, other protein or small molecule. The interacting molecule may itself interact with further molecules. For example, one subunit of an heteromeric protein may be attached to the array and a second subunit or complex of subunits may be tethered to the array via interaction with the attached protein subunit. In turn the second subunit or complex of subunits may then interact with a further molecule, e.g. a candidate drug or an antibody) or
 a sample of a single protein type bound to a synthetic molecule (e.g. peptide, chemical compound) or
 a sample of two different variant proteins or “haplotype proteins”, for example each possessing a different complement of mutations or polymorphisms, e.g. “protein 1” is derived from a DNA sequence carrying SNP “A” and a 3 base pair deletion “X” whilst “protein 2” is derived from a DNA sequence carrying SNP “A”, SNP “B” and a 3 base pair insertion “Y”. Such an arrangement is capable of mimicking the heterozygous presence of two different protein variants in an individual.
 Preferably the protein moiety at each position is substantially pure but in certain circumstances mixtures of between 2 and 100 different protein moieties can be present at each position in the pattern of an array of which at least one is tagged. Thus the proteins derived from the expression of more than one variant DNA sequence may be attached a single position for example, for the purposes of initial bulk screening of a set of variants to determine those sets containing variants of interest.
 An embodiment of the invention described in the examples below uses a biotin tag to purify the proteins on the surface, however, the functionality of the array is independent of tag used.
 “Naturally occurring variants of a DNA sequence of interest” are defined herein as being protein-encoding DNA sequences which share a common sequence and which differ only by one or more naturally occurring (i.e. present in a population and not introduced artificially) single nucleotide polymorphisms, deletions or insertions or those sequences which are considered to be haplotypes (a haplotype being a combination of variant features on a chromosome, usually within the context of a particular gene). Generally such DNA sequences are derived from the same gene in that they map to a common chromosomal locus and encode similar proteins, which may possess different phenotypes. In other words, such variants are generally naturally occurring versions of the same gene comprising one or more mutations, or their synthetic equivalents, which whilst having different codons, encode the same “wild-type” or variant proteins as those know to occur in a population.
 Usefully, DNA molecules having all known mutations in a population are used to produce a set of protein moieties which are attached to the arrays of the invention. Optionally, the array may comprise a subset of variant proteins derived from DNA molecules possessing a subset of mutations, for example all known germ-line, or inheritable mutations or a subset of clinically relevant or clinically important mutations. Related DNA molecules as defined herein are related by more than just a common tag sequence introduced for the purposes or marking the resulting expressed protein. It is the sequence additional to such tags which is relevant to the relatedness of the DNA molecules. The related sequences are generally the natural coding sequence of a gene and variant forms caused by mutation. In practice the arrays of the invention carry protein moieties which are derived from DNA molecules which differ, i.e. are mutated at 1 to 10, 1 to 7, 1 to 5, 1 to 4, 1 to 3, 1 to 2 or 1 discrete locations in the sequence of one DNA molecule relative to another, or more often relative to the wild-type coding sequence (or most common variant in a population). The difference or mutation at each discrete sequence location (for example a discrete location such as “base-pair 342” (the location can be a single base) or “base-pair 502 to base-pair 525” (the location can be a region of bases)) may be a point mutation such as a base change, for example the substitution of “A” for “G”. This may lead to a “mis-sense” mutation, where one amino acid in the wild type sequence is replaced by different amino acid. A “single nucleotide polymorphism” is a mutation of a single nucleotide. Alternatively the mutation may be a deletion or insertion of 1 to 200, 1 to 100, 1 to 50, 1 to 20 or 1 to 10 bases. To give an example, insertional mutations are found in “triplet repeat” disorders such as Huntington's Disease-protein variants corresponding to such insertional mutations can be derived from various mutant forms of the gene and attached to the array to permit investigation of their phenotypes.
 Thus, it is envisaged that proteins derived from related DNA molecules can be quite different in structure. For example a related DNA molecule which has undergone a mutation which truncates it, introduces a frame-shift or introduces a stop codon part-way through the wild-type coding sequence may produce a smaller or shorter protein product. Likewise mutation may cause the variant protein to have additional structure, for example a repeated domain or a number of additional amino acids either at the termini of the protein or within the sequence of the protein. Such proteins, being derived from related DNA sequences, are included within the scope of the invention.
 As stated above, also included within the scope of the invention are arrays carrying protein moieties encoded by synthetic equivalents of a wild type gene (or a naturally occurring variant thereof) of a DNA sequence of interest.
 Also included within the scope of the invention are arrays carrying protein moieties derived from related DNA molecules which, having variant i.e. mutated sequences, give rise to products which undergo differential pre-translational processing (e.g., alternatively spliced transcripts) or differential post-translational processing (e.g. glycosylation occurs at a particular amino acid in one expressed protein, but does not occur in another expressed protein due a codon change in the underlying DNA sequence causing the glycosylated amino acid to be absent).
 Generally, related DNA molecules according to the invention are derived from genes which map to the same chromosomal locus, i.e. the related DNA molecules are different versions of the same protein coding sequence derived from a single copy of a gene, which differ as a result of natural mutation.
 The wild-type (or the protein encoded by the most common variant DNA sequence in a population) of the protein is preferably included as one of the protein moieties on the array to act as a reference by which the relative activities of the proteins derived from related DNA molecules can be compared. The output of the assay indicates whether the related DNA molecule comprising a mutated gene encodes:
 (1) a protein with comparable function to the wild-type protein
 (2) a protein with lower or higher levels of function than the wild-type
 (3) a protein with no detectable function
 (4) a protein with altered post-translational modification patterns
 (5) a protein with an activity that can be modified by addition of an extra component (e.g. peptide, antibody or small molecule drug candidate).
 (6) a protein with an activity that can be modified by post-translational modification for example in situ on the chip, for example phosphorylation.
 (7) a protein with an altered function under different environmental conditions in the assay, for example ionic strength, temperature or pH.
 The protein moieties of the arrays of the present invention can comprise proteins associated with a disease state, drug metabolism, or may be uncharacterised. In one preferred embodiment the protein moieties encode wild type p53 and allelic variants thereof. In another preferred embodiment the arrays comprises protein moieties which encode a drug metabolising enzyme, preferably wild type p450 and allelic variants thereof.
 The number of protein variants attached to the arrays of the invention will be determined by the number of variant coding sequences that occur naturally or that are of sufficient experimental, commercial or clinical interest to generate artificially. An array carrying a wild type protein and a single variant would be of use to the investigator. However in practice and in order to take advantage of the suitability of such arrays for high throughput assays, it is envisaged that 1 to 10000, 1 to 1000, 1 to 500, 1 to 400, 1 to 300, 1 to 200, 1 to 100, 1 to 75, 1 to 50, 1 to 25, 1 to 10 or 1 to 5 related DNA molecules are represented by their encoded proteins on an array. For example, in the case of the gene for p53 (the subject of one of the Examples described herein) there are currently about 50 known germ-line or inheritable mutations and more than 1000 known somatic mutations. An individual may of course inherit two different germ-line mutations. Thus a p53 variant protein array might carry proteins derived from the 50 germ-line mutations each isolated at a different location, proteins from a clinically relevant subset of 800 somatic coding mutations (where a protein can be expressed) each isolated at a different location (or in groups of 10 at each location) and all possible pair-wise combinations of the 50 germ-line mutations each located at a different location. It can therefore be seen that an array of the invention can usefully represent individual DNA molecules containing more than 1000 different naturally occurring mutations and can accordingly carry many more, for example 10000 or more, separate discrete samples or “spots” of the protein variants derived therefrom either located alone or in combination with other variants.
 In a second aspect, the invention provides a method of making a protein array comprising the steps of
 a) providing DNA coding sequences which are derived from two or more naturally occurring variants of a DNA sequence of interest
 b) expressing said coding sequences to provide one or more individual proteins
 c) purifying said proteins
 d) depositing said proteins at spatially defined locations on a surface to give an array.
 Steps c) and d) are preferably combined in a single step. This can be done by means of “surface capture” by which is meant the simultaneous purification and isolation of the protein moiety on the array via the incorporated tag as described in the examples below. Furthermore, step c) may be optional as it is not necessary for the protein preparation to be pure at the location of the isolated tagged protein—the tagged protein need not be separated from the crude lysate of the host production cell if purity is not demanded by the assay in which the array takes part.
 The DNA molecules which are expressed to produce the protein moieties of the array can be generated using techniques known in the art (for example see Current Protocols in Molecular Biology, Volume 1, Chapter 8, Edited by Ausubel, F M, Brent, R, Kingston, R E, Moore, D D, Siedman, J G, Smith, J A, and Struhl, K). The ease of in vitro manipulation of cloned DNA enables mutations, for example SNPs, to be generated by standard molecular biological techniques such as PCR mutagenesis using the wild-type gene as a template. Therefore, only knowledge of the identity of the mutation, for example SNP (often available in electronic databases), and not the actual mutation containing DNA molecule, is required for protein array fabrication. The wild-type gene, encoding the protein of interest, is first cloned into a DNA vector for expression in a suitable host. It will be understood by those skilled in the art that the expression host need not be limited to E. coli—yeast, insect or mammalian cells can be used. Use of a eukaryotic host may be desirable where the protein under investigation is known to undergo post-translational modification such as glycosylation. Following confirmation of expression and protein activity, the wild-type gene is mutated to introduce the desired SNPs. The presence of the SNP is confirmed by sequencing following re-cloning.
 To make the array, clones can be grown in microtiter plate format (but not exclusively) allowing parallel processing of samples in a format that is convenient for arraying onto slides or plate formats and which provides a high-throughput format. Protein expression is induced and clones are subsequently processed for arraying. This can involve purification of the proteins by affinity chromatography, or preparation of lysates ready for arraying onto a surface which is selective for the recombinant protein (‘surface capture’). Thus, the DNA molecules may be expressed as fusion proteins to give protein moieties tagged at either the N- or C-terminus with a marker moiety. As described herein, such tags may be used to purify or attach the proteins to the surface or the array. Conveniently and preferably, the protein moieties are simultaneously purified from the expression host lysate and attached to the array by means of the marker moiety. The resulting array of proteins can then be used to assay the functions of all proteins in a parallel, and therefore high-throughput manner.
 In a third aspect, the invention provides a method of simultaneously determining the relative properties of members of a set of protein moieties derived from related DNA molecules, comprising the steps of: providing an array as herein described, bringing said array into contact with a test substance, and observing the interaction of the test substance with each set member on the array.
 In one embodiment, the invention provides a method of screening a set of protein moieties derived from related DNA molecules for compounds (for example, a small organic molecule) which restore or disrupt function of a protein, which may reveal compounds with therapeutic advantages or disadvantages for a subset of the population carrying a particular SNP or other mutation. In other embodiments the test substance may be:
 a protein for determining relative protein:protein interactions within a set of protein moieties derived from related DNA molecules
 a nucleic acid molecule for determining relative protein:DNA or protein:RNA interactions
 a ligand for determining relative protein:ligand interactions
 Results obtained from the interrogation of arrays of the invention can be quantitative (e.g. measuring binding or catalytic constants KD & KM), semi-quantitative (e.g. normalising amount bound against protein quantity) or qualitative (e.g. functional vs. non-functional). By quantifying the signals for replicate arrays where the ligand is added at several (for example, two or more) concentrations, both the binding affinities and the active concentrations of protein in the spot can be determined. This allows comparison of SNPs with each other and the wild-type. This level of information has not been obtained previously from arrays. Exactly the same methodology could be used to measure binding of drugs to arrayed proteins.
 For example, quantitative results, KD and Bmax, which describe the affinity of the interaction between ligand and protein and the number of binding sites for that ligand respectively, can be derived from protein array data. Briefly, either quantified or relative amounts of ligand bound to each individual protein spot can be measured at different concentrations of ligand in the assay solution. Assuming a linear relationship between the amount of protein and bound ligand, the (relative) amount of ligand bound to each spot over a range of ligand concentrations used in the assay can be fitted to equation 1, rearrangements or derivations.
Bound ligand=B max/((K D /[L])+1) (Equation 1)
 [L]=concentration of ligand used in the assay
 Preferred features of each aspect of the invention are as defined for each other aspect, mutatis mutandis.
 Further features and details of the invention will be apparent from the following description of specific embodiments of a protein array, a p53 protein SNP array and a p450 array, and its use in accordance with the invention which is given by way of example with reference to the accompanying drawings, in which:
FIG. 1 shows p53 mutant panel expression. E. coli cells containing plasmids encoding human wild type p53 or the indicated mutants were induced for 4 h at 30 C. Cells were lysed by the addition of lysozyme and Triton X100 and cleared lysates were analysed by Western blot. A band corresponding to full length his-tagged, biotinylated p53 runs at around 70 kDa.
FIG. 2 shows a gel shift assay to demonstrate DNA binding function of E.coli expressed p53. 1 ul of cleared E.coli lysate containing wild type p53 (wt) or the indicated mutant was combined with 250 nM DIG-labelled DNA and 0.05 mg/ml polydI/dC competitor DNA. The -ve control contained only DNA. Bound and free DNA was separated through a 6% gel (NOVEX), transferred to positively charged membrane (Roche) and DIG-labelled DNA detected using an anti-DIG HRP conjugated antibody (Roche). The DNA:p53 complex is indicated by an arrow.
FIG. 3 shows microarray data for the p53 DNA binding assay. Lysates were arrayed in a 4×4 pattern onto streptavidin capture membrane as detailed in A) and probed with B) Cy3-labelled anti-histidine antibody or C) Cy3-labelled GADD45 DNA, prior to scanning in an Affymetrix 428 array scanner.
FIG. 4 shows CKII phosphorylation of p53. 2 ul of E.coli lysate containing p53 wild type (wt) or the indicated mutant protein were incubated with or without casein kinase II in a buffer containing ATP for 30 min at 30 C. Reactions were Western blotted and phosphorylation at serine 392 detected using a phosphorylation specific antibody.
FIG. 5 shows microarray data for the CKII phosphorylation assay. The p53 array was incubated with CKII and ATP for 1 h at 30 C. and analysed for phosphorylation at serine 392. Phosphorylation was detected for all proteins on the array except for the truncation mutants Q136X, R196X, R209X, R213X, R306X and for the amino acid mutants L344P and S392A.
FIG. 6 shows a solution phase MDM2 interaction assay. 10 ul of p53 containing lysate was incubated with 10 ul of MDM2 containing lysate and 20 ul anti-FLAG agarose in a total volume of 500 ul. After incubation for 1 h at room temperature the anti-FLAG agarose was collected by centrifugation, washed extensively and bound proteins analysed by Western blotting. P53 proteins were detected by Strep/HRP conjugate.
FIG. 7 shows microarray data for MDM2 interaction. The p53 array was incubated with purified Cy3-labelled MDM2 protein for 1 h at room temperature and bound MDM2 protein detected using a DNA array scanner (Affymetrix). MDM2 protein bound to all members of the array apart from the W23A and W23G mutants.
FIG. 8a shows replicate p53 microarrays incubated in the presence of 33P labelled duplex DNA, corresponding to the sequence of the GADD45 promoter element, at varying concentrations and imaged using a phosphorimager so individual spots could be quantified.
FIG. 8B shows DNA binding to wild-type p53 (high affinity), R273H (low affinity) and L344P (non-binder) predicting a wild-type affinity of 7 nM.
FIG. 9A shows a plasmid map of pBJW102.2 for expression of C-terminal BCCP hexa-histidine constructs.
FIG. 9B shows the DNA sequence of pBJW102.2.
FIG. 9C shows the cloning site of pBJW102.2 from start codon. Human P450s, NADPH-cytochrome P450 reductase, and cytochrome b5 ORFs, and truncations thereof, were ligated to a DraIII/Smal digested vector of pBJW102.2.
FIG. 10A shows a vector map of pJW45.
FIG. 10B shows the sequence of the vector pJW45.
FIG. 11A shows the DNA sequence of Human P450 3A4 open reading frame.
FIG. 11B shows the amino acid sequence of full length human P450 3A4.
FIG. 12A shows the DNA sequence of human P450 2C9 open reading frame.
FIG. 12B shows the amino acid sequence of full length human P450 2C9
FIG. 13A shows the DNA sequence of human P450 2D6 open reading frame.
FIG. 13B shows the amino acid sequence of full length human P450 2D6.
FIG. 14 shows a western blot and coomassie-stained gel of purification of cytochrome P450 3A4 from E. coli. Samples from the purification of cytochrome P450 3A4 were run on SDS-PAGE, stained for protein using coomassie or Western blotted onto nitrocellulose membrane, probed with streptavidin-HRP conjugate and visualised using DAB stain:
 Lanes 1: Whole cells
 Lanes 2: Lysate
 Lanes 3: Lysed E. coli cells
 Lanes 4: Supernatant from E. coli cell wash
 Lanes 5: Pellet from E. coli cell wash
 Lanes 6: Supernatant after membrane solublisation
 Lanes 7: pellet after membrane solublisation
 Lanes 8: molecular weight markers: 175, 83, 62, 48, 32, 25, 16.5, 6.5 Kda.
FIG. 15 shows the Coomassie stained gel of Ni-NTA column purification of cytochrome P450 3A4. Samples from all stages of column purification were run on SDS-PAGE:
 Lane 1: Markers 175, 83, 62, 48, 32, 25, 16.5, 6.5 KDa
 Lane 2: Supernatant from membrane solublisation
 Lane 3: Column Flow-Through
 Lane 4: Wash in buffer C
 Lane 5: Wash in buffer D
 Lanes 6&7: Washes in buffer D+50 mM Imidazole
 Lanes 8-12: Elution in buffer D+200 mM Imidazole.
FIG. 16 shows the assay of activity for cytochrome P450 2D6 in a reconstitution assay using the substrate AMMC. Recombinant, tagged CYP2D6 was compared with a commercially available CYP2D6 in terms of ability to turnover AMMC after reconstitution in liposomes with NADPH-cytochrome P450 reductase.
FIG. 17 shows the rates of resorufin formation from BzRes by cumene hydrogen peroxide activated cytochrome P450 3A4. Cytochrome P450 3A4 was assayed in solution with cumene hydrogen peroxide activation in the presence of increasing concentrations of BzRes up to 160 μM.
FIG. 18 shows the equilibrium binding of [3H]ketoconazole to immobilised CYP3A4 and CYP2C9. In the case of CYP3A4 the data points are the means±standard deviation, of 4 experiments. Non-specific binding was determined in the presence of 100 μM ketoconazole (data not shown).
FIG. 19 shows the chemical activation of tagged, immobilised P450 involving conversion of DBF to fluorescein by CHP activated P450 3A4 immobilised on a streptavidin surface.
FIG. 20 shows the stability of agarose encapsulated microsomes. Microsomes containing cytochrome P450 2D6 plus NADPH-cytochrome P450 reductase and cytochrome b5 were diluted in agarose and allowed to set in 96 well plates. AMMC turnover was measured immediately and after two and seven days at 4° C.
FIG. 21 shows the turnover of BzRes by cytochrome P450 3A4 isoforms. Cytochrome P450 3A4 isoforms WT, *1, *2, *3, *4, *5 & *15, (approximately 1 μg) were incubated in the presence of BzRes (0-160 μM) and cumene hydrogen peroxide (200 μM) at room temperature in 200 mM KPO4 buffer pH 7.4. Formation of resorufin was measured over time and rates were calculated from progress curves. Curves describing conventional Michaelis-Menton kinetics were fitted to the data.
FIG. 22 shows the inhibition of cytochrome P450 3A4 isoforms by ketoconazole. Cytochrome P450 3A4 isoforms WT, *1, *2, *3, *4, *5 & *15, (approximately 1 μg) were incubated in the presence of BzRes (50 μM), Cumene hydrogen peroxide (200 μM) and ketoconazole (0, 0.008, 0.04, 0.2, 1, 5 μM) at room temperature in 200 mM KPO4 buffer pH 7.4. Formation of resorufin was measured over time and rates were calculated from progress curves. IC50 inhibition curves were fitted to the data.
 Use of a Protein Array for Functional Analysis of Proteins Encoded by SNP-Containing Genes—the p53 Protein SNP Array
 Mutations in the tumour suppresser protein p53 have been associated with around 50% of cancers, and more than a thousand SNPs of this gene have been observed. Mutations of the p53 gene in tumour cells (somatic mutation), or in the genome of families with a predisposition to cancer (germline mutation), provide an association between a condition and genotype, but no molecular mechanism. To demonstrate the utility of protein arrays for functional characterisation of coding SNPs, the Inventors have arrayed wild type human p53 together with 46 germline mutations (SNPs). The biochemical activity of these proteins can then be compared rapidly and in parallel using small sample volumes of reagent or ligand. The arrayed proteins are shown to be functional for DNA binding, phosphorylated post-translationally “on-chip” by a known p53 kinase, and can interact with a known p53-interacting protein, MDM2. For many of these SNPs, this is the first functional characterisation of the effect of the mutation on p53 function, and illustrates the usefulness of protein microarrays in analysing biochemical activities in a massively parallel fashion.
 Materials and Methods for Construction of p53 SNP Array.
 Wild type p53 cDNA was amplified by PCR from a HeLa cell cDNA library using primers P53F (5′ atg gag gag ccg cag tca gat cct ag 3′) and P53R (5′ gat cgc ggc cgc tca gtc agg ccc ttc tg 3′) and ligated into an E.coli expression vector downstream of sequence coding for a poly Histidine-tag and the BCCP domain from the E. coli AccB gene. The ligation mix was transformed into chemically competent XL1Blue cells (Stratagene) according to the manufacturer's instructions. The p53 cDNA sequence was checked by sequencing and found to correspond to wild type p53 protein sequence as contained in the SWISS-PROT entry for p53 [Accession No. P04637].
 Construction of p53 Mutant Panel
 Mutants of p53 were made by using the plasmid containing the wild type p53 sequence as template in an inverse PCR reaction. Primers were designed such that the forward primer was 5′ phosphorylated and started with the single nucleotide polymorphism (SNP) at the 5′ end, followed by 20-24 nucleotides of p53 sequence. The reverse primer was designed to be complementary to the 20-24 nucleotides before the SNP. PCR was performed using Pwo polymerase which generated blunt ended products corresponding to the entire p53-containing vector. PCR products were gel purified, ligated to form circular plasmids and parental template DNA was digested with restriction endonuclease DpnI (New England Biolabs) to increase cloning efficiency. Ligated products were transformed into XL1Blue cells, and mutant p53 genes were verified by sequencing for the presence of the desired mutation and the absence of any secondary mutation introduced by PCR.
 Expression of p53 in E. coli
 Colonies of XLIBlue cells containing p53 plasmids were inoculated into 2 ml of LB medium containing ampicillin (70 micrograms/ml) in 48 well blocks (QIAGEN) and grown overnight at 37° C. in a shaking incubator. 40 μl of overnight culture was used to inoculate another 2 ml of LB/ampicillin in 48 well blocks and grown at 37° C. until an optical density (600 nm) of ˜0.4 was reached. IPTG was then added to 50 μM and induction continued at 30° C. for 4 hours. Cells were then harvested by centrifugation and cell pellets stored at −80° C. For preparation of protein, cell pellets were thawed at room temperature and 40 μl of p53 buffer (25 mM HEPES pH 7.6, 50 mM KCl, 10% glycerol, 1 mM DTT, 1 mg/ml bovine serum albumin, 0.1% Triton X100) and 10 μl of 4 mg/ml lysozyme were added and vortexed to resuspend the cell pellet. Lysis was aided by incubation on a rocker at room temperature for 30 min before cell debris was collected by centrifugation at 13000 rpm for 10 min at 4° C. The cleared supernatant of soluble protein was removed and used immediately or stored at −20° C.
 Western Blotting
 Soluble protein samples were boiled in SDS containing buffer for 5 min prior to loading on 4-20% Tris-Glycine gels (NOVEX) and run at 200 V for 45 min. Protein was transferred onto PVDF membrane (Hybond-P, Amersham) and probed for the presence of various epitopes using standard techniques. For detection of the histidine-tag, membranes were blocked in 5% Marvel/PBST and anti-RGSHis antibody (QIAGEN) was used as the primary antibody at 1/1000 dilution. For detection of the biotin tag, membranes were blocked in Superblock/TBS (Pierce) and probed with Streptavidin-HRP conjugate (Amersham) at 1/2000 dilution in Superblock/TBS/0.1% Tween20. The secondary antibody for the RGSHis antibody was anti-mouse IgG (Fc specific) HRP conjugate (Sigma) used at 1/2000 dilution in Marvel/PBST. After extensive washing, bound HRP conjugates were detected using either ECLPlus (Amersham) and Hyperfilm ECL (Amersham) or by DAB staining (Pierce).
 DNA Gel Shift Assay
 DNA binding function of expressed p53 was assayed using a conventional gel shift assay. Oligos DIGGADD45A (5′DIG-gta cag aac atg tct aag cat gct ggg gac-3′) and GADD45B (gtc ccc agc atg ctt aga cat gtt ctg tac 3′) were annealed together to give a final concentration of 25 μM dsDNA. Binding reactions were assembled containing 1 μl of cleared lysate, 0.2 μl of annealed DIG-labelled GADD45 oligos and 1 μl of polydI/dC competitor DNA (Sigma) in 20 μl of p53 buffer. Reactions were incubated at room temperature for 30 min, chilled on ice and 5 μl loaded onto a pre-run 6% polyacrylamide/TBE gel (NOVEX). Gels were run at 100 V at 4° C. for 90 min before being transferred onto positively charged nitrocellulose (Roche). Membranes were blocked in 0.4% Blocking Reagent (Roche) in Buffer I (100 mM maleic acid, 150 mM NaCl, pH 7.0) for 30 min and probed for presence of DIG-labelled DNA with anti-DIG Fab fragments conjugated to HRP (Roche). Bound HRP conjugates were detected using ECLPlus and Hyperfilm ECL (Amersham). p53 Phosphorylation Assay
 Phosphorylation of p53 was performed using purified casein kinase II (CKII, Sigma). This kinase has previously been shown to phosphorylate wild type p53 at serine 392. Phosphorylation reactions contained 2 μl of p53 lysate, 10 mM MgCl2, 100 μM ATP and 0.1U of CKII in 20 μl of p53 buffer. Reactions were incubated at 30° C. for 30 min, reaction products separated through 4-20% NOVEX gels and transferred onto PVDF membrane. Phosphorylation of p53 was detected using an antibody specific for phosphorylation of p53 at serine 392 (Cell Signalling Technology), used at 1/1000 dilution in Marvel/TBST. Secondary antibody was an anti-rabbit HRP conjugate (Cell Signalling Technology), used at 1/2000 dilution.
 MDM2 Interaction Assay
 The cDNA for the N-terminal portion of MDM2 (amino acids 17-127) was amplified from a cDNA library and cloned downstream of sequences coding for a His-tag and a FLAG-tag in an E. coli expression vector. Plasmids were checked by sequencing for correct MDM2 sequence and induction of E. coli cultures showed expression of a His and FLAG tagged soluble protein of the expected size. To test for interaction between MDM2 and the p53 mutant panel, binding reactions were assembled containing 10 μl p53 containing lysate, 10 μl MDM2 containing lysate, 20 μl anti-FLAG agarose in 500 μl phosphate buffered saline containing 300 mM NaCl, 0.1% Tween20 and 1% (w/v) bovine serum albumin. Reactions were incubated on a rocker at room temperature for 1 hour and FLAG bound complexes harvested by centrifugation at 5000 rpm for 2 min. After extensive washing in PBST, FLAG bound complexes were denatured in SDS sample buffer and Western blotted. Presence of biotinylated p53 was detected by Streptavidin/HRP conjugate. p53 Microarray Fabrication and Assays
 Cleared lysates of the p53 mutant panel were loaded onto a 384 well plate and printed onto SAM2™ membrane (Promega, Madison, Wis., USA) using a custom built robot (K-Biosystems, UK) with a 16 pin microarraying head. Each lysate was spotted 4 times onto each array, and each spot was printed onto 3 times. After printing, arrays were wet in p53 buffer and blocked in 5% Marvel/p53 buffer for 30 min. After washing 3×5 min in p53 buffer, arrays were ready for assay.
 For DNA binding assay, 5 μl of annealed Cy3-labelled GADD45 oligo was added to 500 μl p53 buffer. The probe solution was washed over the array at room temperature for 30 min, and washed for 3×5 min in p53 buffer. Arrays were then dried and mounted onto glass slides for scanning in an Affymetrix 428 array scanner. Quantification of Cy3 scanned images was accomplished using ImaGene software.
 For the phosphorylation assay, 10 μl CKII was incubated with the arrays in 320 μl p53 buffer and 80 μl Mg/ATP mix at 30 ° C. for 30min. Arrays were then washed for 3×5 min in TBST and anti-phosphoserine 392 antibody added at 1/1000 dilution in Marvel/TBST for 1 h. After washing for 3×5 min in TBST, anti-rabbit secondary antibody was added at 1/2000 dilution for 1 h. Bound antibody was detected by ECLPlus and Hyperfilm.
 For the MDM2 interaction assay, 1 μl of purified Cy3 labelled MDM2 protein was incubated with the arrays in 500 μl PBS/300 mM NaCl/0.1% Tween20/1% BSA for 1 h at room temperature. After washing for 3×5 min in the same buffer, arrays were dried, mounted onto glass slides and analysed for Cy3 fluorescence as for the DNA binding assay.
 Expression of p53 in E.coli and Construction of Mutant Panel
 The full length p53 open reading frame was amplified from a Hela cell cDNA library by PCR and cloned downstream of the tac promoter in vector pQE80L into which the BCCP domain from the E.coli gene ACCB had already been cloned. The resultant p53 would then be His and biotin tagged at its N-terminus, and FIG. 1 shows Western blot analysis of soluble protein from induced E.coli cultures. There is a clear signal for His-tagged, biotinylated protein at around 66 kDa, and a band of the same size is detected by the p53 specific antibody pAb1801 (data not shown). The plasmid encoding this protein was fully sequenced and shown to be wild type p53 cDNA sequence. This plasmid was used as the template to construct the mutant panel, and FIG. 1 also shows analysis of the expression of a selection of those mutants, showing full length protein as expected for the single nucleotide polymorphisms, and truncated proteins where the mutation codes for a STOP codon. The mutants were also sequenced to confirm presence of the desired mutation and absence of any secondary mutations.
 Although the Inventors have used His and biotin tags in this example of a SNP array, other affinity tags (eg FLAG, myc, VSV) can be used to enable purification of the cloned proteins. Also an expression host other than E. coli can be used (eg. yeast, insect cells, mammalian cells) if required.
 Also, although this array was focussed on the naturally occurring germline SNPs of p53, other embodiments are not necessarily restricted to naturally occurring SNPs (“synthetic” mutants) or versions of the wild type protein which contain more than one SNP. Other embodiments can contain versions of the protein which are deleted from either or both ends (a nested-set). Such arrays would be useful in mapping protein:ligand interactions and delineating functional domains of unknown proteins.
E. coli Expressed p53 is Functional for DNA Binding
 To demonstrate functionality of our p53, the Inventors performed electrophoretic mobility shift assays using a DNA oligo previously shown to be bound by p53. FIG. 2 shows an example result from these gel shift assays, showing DNA binding by wild type p53 as well as mutants R72P, P82L and R181C. The first 2 mutants would still be expected to bind DNA as these mutations are outside of the DNA binding domain of p53. Having demonstrated DNA binding using a conventional gel based assay, the Inventors then wanted to show the same function for p53 arrayed on a surface. FIG. 3C shows the result of binding Cy3-labelled DNA to the p53 mutant panel arrayed onto SAM2™ membrane (Promega, Madison, Wis., USA). Although the Inventors have used SAM2™ membrane in this example of a SNP array, other surfaces which can be used for arraying proteins onto include but are not restricted to glass, polypropylene, polystyrene, gold or silica slides, polypropylene or polystyrene multi-well plates, or other porous surfaces such as nitrocellulose, PVDF and nylon membranes. The SAM2™ membrane specifically captures biotinylated molecules and so purifies the biotinylated p53 proteins from the mutant panel cell lysates. After washing unbound DNA from the array, bound DNA was visualised using an Affymetrix DNA array scanner. As can be seen from FIG. 3, the same mutants which bound DNA in the gel shift assay also bound the most DNA when arrayed on a surface. Indeed, for a DNA binding assay the microarray assay appeared to be more sensitive than the conventional gel shift assay. This is probably because in a gel shift assay the DNA:protein complex has to remain bound during gel electrophoresis, and weak complexes may dissociate during this step. Also the 3-dimensional matrix of the SAM2™ membrane used may have a caging effect. The amount of p53 protein is equivalent on each spot, as shown by an identical microarray probed for His-tagged protein (FIG. 3B).
 Use of the p53 Array for Phosphorylation Studies
 To exemplify the study of the effect of SNPs on post-translational modifications, the Inventors chose to look at phosphorylation of the p53 array by casein kinase II. This enzyme has previously been shown to phosphorylate p53 at serine 392, and the Inventors made use of a commercially available anti-p53 phosphoserine 392 specific antibody to study this event. FIG. 4 shows Western blot analysis of kinase reactions on soluble protein preparations from p53 wild type and S392A clones. Lane 1 shows phosphorylation of wild type p53 by CKII, with a background signal when CKII is omitted from the reaction (lane 2). Lanes 3 and 4 show the corresponding results for S392A, which as expected only shows background signal for phosphorylation by CKII. This assay was then applied in a microarray format, which as can be seen from FIG. 5 shows phosphorylation for all of the mutant panel except the S392A mutant and those mutants which are truncated before residue 392.
 Use of the p53 Array to Study a Protein:Protein Interaction
 To exemplify the study of a protein:protein interaction on a SNP protein array, the interaction of MDM2 with the p53 protein array was investigated. FIG. 6 shows that FLAG-tagged MDM2 pulls down wild type p53 when bound to anti-FLAG agarose. However the W23A mutant is not pulled down by FLAG agarose bound MDM2, which would be expected as this residue has previously been shown to be critical for the p53/MDM2 interaction (Bottger, A., Bottger, V., Garcia-Echeverria, C., et al, J. Mol. Biol. (1997) 269: 744-756). This assay was then carried out in a microarray format, and FIG. 7 shows the result of this assay, with Cy3-labelled protein being detected at all spots apart from the W23A and W23G mutant spots.
 The Inventors have used a novel protein chip technology to characterise the effect of 46 germline mutations on human p53 protein function. The arrayed proteins can be detected by both a His-tagged antibody and also a p53 specific antibody. This array can be used to screen for mutation specific antibodies which could have implications for p53 status diagnosis.
 The Inventors were able to demonstrate functionality of the wild type protein by conventional gel based assays, and have achieved similar results performing the assays in a microarray format. Indeed, for a DNA binding assay the microarray assay appeared to be more sensitive than the conventional gel shift assay. These arrays can be stored at −20 C. in 50% glycerol and have been shown to still be functional for DNA binding after 1 month (data not shown).
 The CKII phosphorylation assay results are as expected, with phosphorylation being detected for all proteins which contained the serine at residue 392. This analysis can obviously be extended to a screen for kinases that phosphorylate p53, or for instance for kinases that differentially phosphorylate some mutants and not others, which could themselves represent potential targets in cancer.
 The MDM2 interaction assay again shows the validity of the protein array format, with results for wild type and the p53 mutants mirroring those obtained using a more conventional pull down assay. These results also show that our protein arrays can be used to detect protein:protein interactions. Potentially these arrays can be used to obtain quantitative binding data (ie KD values) for protein:protein interactions in a high-throughput manner not possible using current methodology. The fact that the MDM2 protein was pulled out of a crude E. coli lysate onto the array bodes well for envisioned protein profiling experiments, where for instance cell extracts are prepared from different patients, labelled with different fluorophores and both hybridised to the same array to look for differences in amounts of protein interacting species.
 Indeed, in Example 2 below the applicant has gone on to demonstrate that these arrays can be used to obtain quantative data.
 Quantitative DNA Binding on the p53 Protein Microarray
 DNA-Binding Assays.
 Oligonucleotides with the GADD45 promoter element sequence (5′-gta cag aac atg tct aag cat gct ggg gac-3′ and 5′-gtc ccc agc atg ctt aga cat gtt ctg tac-3′) were radiolabelled with gamma 33P-ATP (Amersham Biosciences, Buckinghamshire, UK) and T4 kinase (Invitrogen, Carlsbad, Calif.), annealed in p53 buffer and then purified using a Nucleotide Extraction column (Qiagen, Valencia, Calif.). The duplex oligos were quantified by UV spectrophotometry and a 2.5 fold dilution series made in p53 buffer. 500 μl of each dilution were incubated with microarrays at room temperature for 30 min, then washed three times for 5 min in p53 buffer to remove unbound DNA. Microarrays were then exposed to a phosphorimager plate (Fuji, Japan) overnight prior to scanning. ImaGene software (BioDiscovery, Marina del Rey, Calif.) was used to quantify the scanned images. Replicate values for all mutants at each DNA concentration were fitted to simple hyperbolic concentration-response curves R=Bmax/((Kd/L)+1), where R is the response in relative counts and L is the DNA concentration in nM.
 Binding of p53 to GADD45 Promoter Element DNA.
 Replicate p53 microarrays were incubated in the presence of 33P labelled duplex DNA, corresponding to the sequence of the GADD45 promoter element, at varying concentrations (FIG. 8A). The microarrays were imaged using a phosphorimager and individual spots quantified. The data were normalised against a calibration curve to compensate for the non-linearity of this method of detection and backgrounds were subtracted. Replicate values for all mutants were plotted and analysed by non-linear regression analysis allowing calculation of both Kd and Bmax values (Table 1).
FIG. 8B shows DNA binding to wild-type p53 (high affinity), R273H (low affinity) and L344P (non-binder) predicting a wild-type affinity of 7 nM.
 DNA Binding.
 Quantitative analysis of the DNA binding data obtained from the microarrays yielded both affinities (Kd) and relative maximum binding values (Bmax) for wild-type and mutant p53. Protein function microarrays have not previously been used in this way and this data therefore demonstrate their usefulness in obtaining this quality and amount of data in a parallel fashion. The approach of normalising binding data for the amount of affinity-tagged protein in the spot provides a rapid means of analysing large data sets [Zhu, H. et al. Global analysis of protein activities using proteome chips. Science 293, 2101-2105 (2001).], however it takes into account neither the varying specific activity of the microarrayed protein nor whether the signal is recorded under saturating or sub-saturating conditions. The quantitative analysis carried out here allowed the functional classification of mutants into groups according to GADD45 DNA binding: those showing near wild-type affinity; those exhibiting reduced stability (low Bmax); those showing reduced affinity (higher Kd); and those showing complete loss of activity (Table 1).
 Proteins with near wild-type affinity for DNA generally had mutations located outside of the DNA-binding domain and include R72P, P82L, R306P and G325V. R337C is known to affect the oligomerisation state of p53 but at the assay temperature used here it is thought to be largely tetrameric [Davison, T. S., Yin, P., Nie, E., Kay, C. & Arrowsmith, C. H. Characterisation of the oligomerisation defects of two p53 mutants found in families with Li-Fraumeni and Li-Fraumeni like syndrome. Oncogene 17, 651-656 (1998).], consistent with the affinity measured here. By contrast, total loss of binding was observed for mutations introducing premature stop codons (Q136X, R196X, R209X and R213X) and mutations that monomerise the protein (L344P [Lomax, M. E., Barnes, D. M., Hupp, T. R., Picksley, S. M. & Camplejohn, R. S. Characterisation of p53 oligomerisation domain mutations isolated from Li-Fraumeni and Li-Fraumeni like family members. Oncogene 17, 643-649 (1998).] and the tetramerisation domain deficient R306X) as expected.
 Within the DNA-binding domain, the applicant found that mutations generally reduced or abolished DNA binding with the notable exceptions of R181C/H, S227T and H233N/D; these are all solvent exposed positions, distant from the protein-DNA interface and exhibit wild-type binding. Mutations R248Q/W, R273C/H and R280K, present at the protein-DNA interface, exhibited low affinities with Kd values 2-7 times higher than wild-type (Table 1) consistent with either loss of specific protein-DNA interactions or steric hindrance through sub-optimal packing of the mutated residue.
 Many of the remaining mutants fall into a group displaying considerably reduced specific activities, apparent from very low Bmax values, even when normalised according to the amount of protein present in the relevant spot. For some mutants, DNA binding was compromised to such a level that although binding was observed, it was not accurately quantifiable due to low signal to background ratios e.g. P151S and G245C. For others such as L252P, low signal intensities yielded measurable Kd values, but with wide confidence limits.
 To further demonstrate the applicability of the invention to protein arrays comprising at least two protein moieties derived from naturally occurring variants of a DNA sequence of interest such as, for example, those encoding proteins from phase 1 or phase 2 drug metabolising enzymes (DME's) the invention is further exemplified with reference to a p450 array. Phase 1 DME's include the Cytochrome p450's and the Flavin mono oxygenases (FMO's) and the Phase 2 DME's, UDP-glycosyltransferase (UGTs), glutathione S transferases (GSTs), sulfotransferases (SULTs), N-acetyltransferases (NATs), drug binding nuclear receptors and drug transporter proteins.
 Preferably, the full complement, or a significant proportion of human DMEs are present on the arrays of the invention. Such an array can include (numbers in parenthesis currently described in the Swiss Prot database): all the human P450s (119), FMOs (5), UDP-glycosyltransferase (UGTs) (18), GSTs (20), sulfotransferases (SULTs) (6), N-acetyltransferases (NATs) (2), drug binding nuclear receptors (33) and drug transporter proteins (6). This protein list does not include those yet to be characterised from the human genome sequencing project, splice variants known to occur for the P450s that can switch substrate specificity or polymorphisms known to affect the function and substrate specificity of both the P450s and the phase 2 DMEs.
 For example it is known that there are large differences in the frequency of occurrence of various alleles in P450s 2C9, 2D6 and 3A4 between different ethnic groups (see Tables 2, 3 and 4). These alleles have the potential to affect enzyme kinetics, substrate specificity, regio-selectivity and, where multiple products are produced, product profiles. Arrays of proteins described in this disclosure allow a more detailed examination of these differences for a particular drug and will be useful in predicting potential problems and also in effectively planning the population used for clinical trials.
 1. Johansson, I., Oscarson, M., Yue, Q. Y., Bertilsson, L., Sjoqvist, F. & Ingelman-Sundberg, M. (1994) Mol Pharmacol 46, 452-9.
 2. Sachse, C., Brockmoller, J., Bauer, S. & Roots, I. (1997) Am J Hum Genet 60, 284-95.
 3. Griese, E. U., Zanger, U. M., Brudermanns, U., Gaedigk, A., Mikus, G., Morike, K., Stuven, T. & Eichelbaum, M. (1998) Pharmacogenetics 8, 15-26.
 4. Marez, D., Legrand, M., Sabbagh, N., Guidice, J. M., Spire, C., Lafitte, J. J., Meyer, U. A. & Broly, F. (1997) Pharmacogenetics 7, 193-202.
 5. Aklillu, E., Persson, I., Bertilsson, L., Johansson, I., Rodrigues, F. & Ingelman-Sundberg, M. (1996) J Pharmacol Exp Ther 278, 441-6.
 6. Dandara, C., Masimirembwa, C. M., Magimba, A., Sayi, J., Kaaya, S., Sommers, D. K., Snyman, J. R. & Hasler, J. A. (2001) Eur J Clin Pharmacol 57, 11-7.
 7. Aithal, G. P., Day, C. P., Kesteven, P. J. & Daly, A. K. (1999) Lancet 353, 717-9.
 8. Imai, J., Ieiri, I., Mamiya, K., Miyahara, S., Furuumi, H., Nanba, E., Yamane, M., Fukumaki, Y., Ninomiya, H., Tashiro, N., Otsubo, K. & Higuchi, S. (2000) Pharmacogenetics 10, 85-9.
 9. Dickmann, L. J., Rettie, A. E., Kneller, M. B., Kim, R. B., Wood, A. J., Stein, C. M., Wilkinson, G. R. & Schwarz, U. I. (2001) Mol Pharmacol 60, 382-7.
 10. Sata, F., Sapone, A., Elizondo, G., Stocker, P., Miller, V. P., Zheng, W., Raunio, H., Crespi, C. L. & Gonzalez, F. J. (2000) Clin Pharmacol Ther 67, 48-56.
 11. Eiselt, R., Domanski, T. L., Zibat, A., Mueller, R., Presecan-Siedel, E., Hustert, E., Zanger, U. M., Brockmoller, J., Klenk, H. P., Meyer, U. A., Khan, K. K., He, Y. A., Halpert, J. R. & Wojnowski, L. (2001) Pharmacogenetics 11, 447-58.
 12. Dai, D., Tang, J., Rose, R., Hodgson, E., Bienstock, R. J., Mohrenweiser, H. W. & Goldstein, J. A. (2001) J Pharmacol Exp Ther 299, 825-31.
 13. Hsieh,K. P., Lin, Y. Y., Cheng, C. L., Lai, M. L., Lin, M. S., Siest, J. P. & Huang, J. D. (2001) Drug Metab Dispos 29, 268-73.
 Cloning of Wild-Type H. sapiens Cytochrome P450 Enzymes CYP2C9, CYP2D6 and CYP3A4
 The human cytochrome p450s have a conserved region at the N-terminus, this includes a hydrophobic region which faciliates lipid association, an acidic or ‘stop transfer’ region, which stops the protein being fed further into the membrane, and a partially conserved proline repeat. Three versions of the p450s were produced with deletions up to these domains, the N-terminal deletions are shown below.
 The human CYP2D6 was amplified by PCR from a pool of brain, heart and liver cDNA libraries (Clontech) using specific forward and reverse primers (T017F and T017R). The PCR products were cloned into the pMD004 expression vector, in frame with the N-terminal His-BCCP tag and using the Not1 restriction site present in the reverse primer. To convert the CYP2D6 for expression in the C-terminal tag vector pBJW102.2 (FIG. 9A&B), primers were used which incorporated an Sfi1 cloning site at the 5′ end and removed the stop codon at the 3′ to allow in frame fusion with the C-terminal tag. The primers T017CR together with either T017CF1, T017CF2, or T017CF3 allowed the deletion of 29, 18 and 0 amino acids from the N-terminus of CYP2D6 respectively.
 Primer sequences are as follows:
 PCR was performed in a 50 μl volume containing 0.5 μM of each primer, 125-250 μM dNTPs, 5 ng of template DNA, 1x reaction buffer, 1-5 units of polymerase (Pfu, Pwo, or ‘Expand long template’ polymerase mix), PCR cycle=95° C. 5 minutes, 95° C. 30 seconds, 50-70° C. 30 seconds, 72° C. 4 minutes×35 cycles, 72° C. 10 minutes, or in the case of Expand 68° C. was used for the extension step. PCR products were resolved by agarose gel electrophoresis, those products of the correct size were excised from the gel and subsequently purified using a gel extraction kit. Purified PCR products were then digested with either Sfi1 or Not1 and ligated into the prepared vector backbone (FIG. 9C). Correct recombinant clones were determined by PCR screening of bacterial cultures, Western blotting and by DNA sequence analysis.
 CYP3A4 and CYP2C9 were cloned from cDNA libraries by a methodology similar to that of CYP2D6. Primer sequences to amplify CYP3A4 and CYP2C9 for cloning into the N-terminal vectors are as follows:
 Primers to convert the N-terminal clones for expression in the C-terminal tagging vector are as follows:
 The full length or Hydrophobic peptide (C3) version of 2C9 was produced by inverse PCR using the 2C9-stop transfer clone (C1) as the template and the following primers:
 Cloning of NADPH-Cytochrome P450 Reductase
 NADPH-cytochrome P450 reductase was amplified from fetal liver cDNA (Clontech), the PCR primers [NADPH reductase F1 5′-GGATCGACATATGGGAGACTCCCACGTGGACAC-3′; NADPH reductase R1 5′-CCGATAAGCTTATCAGCTCCACACGTCCAGGGAG-3′] incorporated a Nde I site at 5′ and a Hind III site at the 3′ of the gene to allow cloning. The PCR product was cloned into the pJW45 expression vector (FIG. 10A&B)), two stop codons were included on the reverse primer to ensure that the His-tag was not translated. Correct recombinant clones were determined by PCR screening of bacterial cultures, and by sequencing.
 Cloning of Polymorphic Variants of H. sapiens Cytochrome P450s CYP2C9, CYP2D6 and CYP3A4
 Once the correct wild-type CYP450s (FIGS. 11, 12, & 13) were cloned and verified by sequence analysis the naturally occurring polymorphisms of 2C9, 2D6 and 3A4 shown in Table 5 were created by an inverse PCR approach (except for CYP2D6* 10 which was amplified and cloned as a linear PCR product in the same way as the initial cloning of CYP2D6 described in Example 3). In each case, the forward inverse PCR primer contained a 1 bp mismatch at the 5′ position to substitute the wild type nucleotide for the polymorphic nucleotide as observed in the different ethnic populations.
 The following PCR primers were used.
 Expression and Purification of P450 3A4
E. coli XL-10 gold (Stratagene) was used as a host for expression cultures of P450 3A4. Starter cultures were grown overnight in LB media supplemented with 100 mg per litre ampicillin. 0.5 litre Terrific Broth media plus 100 mg per litre ampicillin and 1 mM thiamine and trace elements were inoculated with 1/100 dilution of the overnight starter cultures. The flasks were shaken at 37° C. until cell density OD600 was 0.4 then δ-Aminolevulinic acid (ALA) was added to the cells at 0.5 mM for 20 min at 30° C. The cells were supplemented with 50 μM biotin then induced with optimum concentration of IPTG (30-100 μM) then shaken overnight at 30° C.
 The E. coli cells from 0.5 litre cultures were divided into 50 ml aliquots, cells pelleted by centrifugation and cell pellets stored at −20° C. Cells from each pellet were lysed by resuspending in 5 ml buffer A (100 mM Tris buffer pH 8.0 containing 100 mM EDTA, 10 mM β-mercaptoethanol, 10× stock of Protease inhibitor cocktail-Roche 1836170, 0.2 mg/ml Lysozyme). After 15 minutes incubation on ice 40 ml of ice-cold deionised water was added to each resuspended cell pellet and mixed. 20 mM Magnesium Chloride and 5 μg/ml DNaseI were added. The cells were incubated for 30 min on ice with gentle shaking after which the lysed E.Coli cells were pelletted by centrifugation for 30 min at 4000 rpm. The cell pellets were washed by resuspending in 10 ml buffer B (100 mM Tris buffer pH 8.0 containing 10 mM β-mercaptoethanol and a 10× stock of Protease inhibitor cocktail-Roche 1836170) followed by centrifugation at 4000 rpm. Membrane associated protein was then solubilised by the addition of 2 ml buffer C (50 mM potassium phosphate pH 7.4, 10× stock of Protease inhibitor cocktail-Roche 1836170, 10 mM β-mercaptoethanol, 0.5 M NaCl and 0.3% (v/v) Igepal CA-630) and incubating on ice with gentle agitation for 30 minutes before centrifugation at 10,000 g for 15 min at 4° C. and the supernatant (FIG. 14) was then applied to Talon resin (Clontech).
 A 0.5 ml column of Ni-NTA agarose (Qiagen) was poured in disposable gravity columns and equilibrated with 5 column volumes of buffer C. Supernatant was applied to the column after which the column was successively washed with 4 column volumes of buffer C, 4 column volumes of buffer D (50 mM potassium phosphate pH 7.4, 10× stock of Protease inhibitor cocktail-Roche 1836170, 10 mM β-mercaptoethanol, 0.5 M NaCl and 20% (v/v) Glycerol) and 4 column volumes of buffer D+50 mM Imidazole before elution in 4 column volumes of buffer D+200 mM Imidazole (FIG. 15). 0.5 ml fractions were collected and protein containing fractions were pooled aliquoted and stored at −80° C.
 Determination of Heme Incorporation into P450s
 Purified P450s were diluted to a concentration of 0.2 mg/ml in 20 mM potassium phosphate (pH 7.4) in the presence and absence of 10 mM KCN and an absorbance scan measured from 600-260 nm. The percentage bound heme was calculated based on an extinction coefficient ε420 of 100 mM−1 cm−1.
 Reconstitution and Assay of Cytochrome P450 Enzymes into Liposomes with NADPH-Cytochrome P450 Reductase
 Liposomes are prepared by dissolving a 1:1:1 mixture of 1,2-dilauroyl-sn-glycero-3-phosphocholine, 1,2-dileoyl-sn-glycero-3-phosphocholine, 1,2-dilauroyl-sn-glycero-3-phosphocholine in chloroform, evaporating to dryness and subsequently resuspending in 20 mM potassium phosphate pH 7.4 at 10 mg/ml. 4 μg of liposomes are added to a mixture of purified P450 2D6 (20 pmol), NADPH P450 reductase (40 pmol), cytochrome b5 (20 pmol) in a total volume of 10 μl and preincubated for 10 minutes at 37° C.
 After reconstitution of cytochrome P450 enzymes into liposomes, the liposomes are diluted to 100 μl in assay buffer in a black 96 well plate, containing HEPES/KOH (pH 7.4, 50 mM), NADP+(2.6 mM), glucose-6-phosphate (6.6 mM), MgCl2 (6.6 mM) and glucose-6-phosphate dehyrogenase (0.4 units/ml). Assay buffer also contains an appropriate fluorogenic substrate for the cytochrome P450 isoform to be assayed: for P450 2D6 AMMC, for P450 3A4 dibenzyl fluorescein (DBF) or resorufin benzyl ether (BzRes) can be used and for 2C9 dibenzyl fluorescein (DBF). The reactions are stopped by the addition of ‘stopping solution’ (80% acetonitrile buffered with Tris) and products are read using the appropriate wavelength filter sets in a fluorescent plate reader (FIG. 16).
 P450s can also be activated chemically by, for example, the addition of 200 μM cumene hydroperoxide in place of the both the co-enzymes and regeneration solution (FIG. 17).
 In addition fluorescently measured rates of turnover can be measured in the presence of inhibitors.
 Detection of Drug Binding to Immobilised P450s CYP3A4
 Purified CYP3A4 (10 μg/ml in 50 mM HEPES/0.01% CHAPS, pH 7.4) was placed in streptavidin immobiliser plates (Exiqon) (100 μl per well) and shaken on ice for 1 hour. The wells were aspirated and washed twice with 50 mM HEPES/0.01% CHAPS. [3H]-ketoconazole binding to immobilised protein was determined directly by scintillation counting. Saturation experiments were performed using [3H]ketoconazole (5 Ci/mmol, American Radiochemicals Inc., St. Louis) in 50 mM HEPES pH 7.4, 0.01% CHAPS and 10% Superblock (Pierce) (FIG. 18). Six concentrations of ligand were used in the binding assay (25-1000 nM) in a final assay volume of 100 μl. Specific binding was defined as that displaced by 100 μM ketoconazole. Each measurement was made in duplicate. After incubation for 1 hour at room temperature, the contents of the wells were aspirated and the wells washed three times with 150 μl ice cold assay buffer. 100 μl MicroScint 20 (Packard) was added to each well and the plates counted in a Packard TopCount microplate scintillation counter (FIG. 18).
 Chemical Activation of Tagged, Immobilised CYP3A4
 CYP3A4 was immobilised in streptavidin immobiliser plates as described in Example 9 and was then incubated with dibenzyl fluorescein and varying concentrations (0-300 μM) of cumene hydrogen peroxide. End point assays demonstrated that the tagged, immobilised CYP3A4 was functional in a turn-over assay with chemical activation (FIG. 19).
 Immobilisation of P450s through Gel Encapsulation of Liposomes or Microsomes
 After reconstitution of cytochrome P450 enzymes together with NADPH-cytochrome P450 reductase in liposomes or microsomes, these can then be immobilised on to a surface by encapsulation within a gel matrix such as agarose, polyurethane or polyacrylamide.
 For example, low melting temperature (LMT) (1% w/v) agarose was dissolved in 200 mM potassium phosphate pH 7.4. This was then cooled to 37° C. on a heating block. Microsomes containing cytochrome P450 3A4, cytochrome b5 and NADPH-cytochrome P450 reductase were then diluted into the LMT agarose such that 50 μl of agarose contained 20, 40 and 20 pmol of P450 3A4, NADPH-cytochrome P450 reductase and cytochrome b5 respectively. 50 μl of agarose-microsomes was then added to each well of a black 96 well microtitre plate and allowed to solidify at room temperature.
 To each well, 100 μl of assay buffer was added and the assay was conducted as described previously (for example, Example 8) for conventional reconstitution assay. From the data generated a comparison of the fundamental kinetics of BzRes oxidation and ketoconazole inhibition was made (Table 6) which showed that the activity of the CYP3A4 was retained after gel-encapsulation.
 For estimation of KM and Vmax for BzRes assays were performed in the presence of varying concentrations of BzRes up to 320 μM. Ketoconazole inhibition was performed at 50 μM BzRes with 7 three-fold dilutions of ketoconazole from 5 μM. Values in parenthesis indicate standard errors derived from the curve fitting.
 The activity of the immobilised P450s was assessed over a period of 7 days (FIG. 20). Aliquots of the same protein preparation stored under identical conditions, except that they were not gel-encapsulated, were also assayed over the same period, which revealed that the gel encapsualtion confers significant stability to the P450 activity.
 Quantitative Determination of Affect of 3A4 Polymorphisms on Activity
 Purified cytochrome P450 3A4 isoforms * 1, *2, *3, *4, *5 & * 15 (approx 1 μg) were incubated in the presence of BzRes and cumene hydrogen peroxide (200 μM) in the absence and presence of ketoconazole at room temperature in 200 mM KPO4 buffer pH 7.4 in a total volume of 100 μl in a 96 well black microtitre plate. A minimum of duplicates were performed for each concentration of BzRes or ketoconazole. Resorufin formation of was measured over time by the increase in fluorescence (520 nm and 580 nm excitation and emission filters respectively) and initial rates were calculated from progress curves (FIG. 21).
 For estimation of KM app and Vmax app for BzRes, background rates were first subtracted from the initial rates and then were plotted against BzRes concentration and curves were fitted describing conventional Michaelis-Menton kinetics:
V=V max/(1+(K M /S))
 where V and S are initial rate and substrate concentration respectively. Vmax values were then normalised for cytochrome P450 concentration and scaled to the wild-type enzyme (Table 7).
 For estimation of IC50 for ketoconazole, background rates were first subtracted from the initial rates which were then converted to a % of the uninhibited rate and plotted against ketoconazole concentration (FIG. 22). IC50 inhibition curves were fitted using the equation:
 where V and I are initial rate and inhibitor concentration respectively. The data obtained are shown in Table 7:
 The parameters were obtained from the fits of Michaelis-Menton and IC50 inhibition curves to the data in FIGS. 21 & 22. Values in parenthesis are standard errors obtained from the curve fits.
 Array-Based Assay of Immobilised CYP3A4 Polymorphisms
 Cytochrome P450 polymorphisms can be assayed in parallel using an array format to identify subtle differences in activity with specific small molecules. For example, purified cytochrome P450 3A4 isoforms *1, *2, *3, *4, *5 & *15 can be individually reconstituted in to liposomes with NADPH-cytochrome P450 reductase as described in Example 11. The resultant liposomes preparation can then be diluted into LMP agarose and immobilised into individual wells of a black 96 well microtitre plate as described in Example 11. The immobilised proteins can then be assayed as described in Example 11 by adding 100 μl of assay buffer containing BzRes+/−ketoconazole to each well.
 Chemical activation (as described in Example 12) can also be used in an array format. For example, purified cytochrome P450 3A4 isoforms *1, *2, *3, *4, *5 & *15 can be individually reconstituted in to liposomes without NADPH-cytochrome P450 reductase and the resultant liposomes can be immobilised via encapsulation in agarose as described in Example 11. The cytochrome P450 activity in each well can then be measured as described in Example 12 by 100 μl of 200 mM KPO4 buffer pH 7.4 containing BzRes and cumene hydrogen peroxide (200 μM), +/−ketoconazole, to each well.
 In summary, the Inventors have developed a novel protein array technology for massively parallel, high-throughout screening of SNPs for the biochemical activity of the encoded proteins. Its applicability was demonstrated through the analysis of various functions of wild type p53 and 46 SNP versions of p53 as well as with allelic variants of p450. The same surface and assay detection methodologies can now be applied to other more diverse arrays currently being developed. Due to the small size of the collection of proteins being studied here, the spot density of our arrays was relatively small, and each protein was spotted in quadruplicate. Using current robotic spotting capabilities it is possible to increase spot density to include over 10,000 proteins per array.
 The entire disclosure of each of the aforementioned patent and scientific documents cited hereinabove is expressly incorporated by reference herein.
 The invention may be embodied in other specific forms without departing from the spirit or essential characteristics thereof. The foregoing embodiments are therefore to be considered in all respects illustrative rather than limiting on the invention described herein. The scope of the invention is thus indicated by the appended claims rather than by the foregoing description, and all changes that come within the meaning and range of equivalency of the claims are intended to be embraced by reference therein.