BACKGROUND There are many situations in which it is desirable to be able to identify potentially pathogenic organisms such as bacteria and viruses rapidly. Current laboratory methods typically involve culturing organisms and the use of immunodiagnostic tests, or preparation of histological specimens and use of specialised staining and/or immunohistochemical techniques. Such techniques demand, at best, hours and, where culturing of organisms is required, days. DNA-based identification, for instance PCR, although more sensitive, still requires hours, as well demanding considerable laboratory facilities and expertise.

The identification of specific micro-organism by means of specific polyclonal antisera or of monoclonal antibodies is standard practice. Immunohistochemistry allows identification of organisms present in tissues, and techniques such as enzyme-linked immunosorbent assays (ELISA) are used to detect pathogens, or antigens derived from them, in body fluids. However, unless being used only to confirm the presence of a particular pathogen suggested by other diagnostic criteria (the usual situation), these diagnostic techniques require the use of a battery of different specific antibodies and are poorly suited to the identification of a particular pathogen from a large number of potential targets. Immunoaffinity purification of cellular components is also well-known in the art, but is not generally a useful technique for identification purposes, since it usually requires knowledge of the organism concerned.

The use of cross-reacting sera to identify micro-organisms has been previously reported (Bonenberger et al, 2001). In this case, polyclonal antiserum raised against the BCG strain of Mycobacterium bovis was used to stain a wide variety of micro-organisms in biopsy specimens. This technique did not allow identification of specific individual organisms, but generically stained a range of bacterial, fungal and protozoal pathogens.

Mass spectrometry (MS) has been used increasingly for biological applications in recent years. New developments have allowed large biological molecules to be 10 analysed (reviewed in Bakhtier and Tse, 2000). In particular, matrix-assisted desorption ionisation (MALDI) and electrospray MS with their relatively gentle ionisation methods are particularly well-suited to protein applications (reviewed in Rowley et al, 2000). More recently, the introduction of ion trap MS has reduced the time required to analyse mixtures of biological molecules, particularly when small amounts are available (Henderson et al, 1999).

WO 00/29987 (Demirev/University of Maryland) discloses a method of measuring the molecular masses of various components of micro-organisms and using database searching to attempt to identify them. However, such an 20 approach has the disadvantages of having to process large amounts of information and of having to distinguish between a great number of components of similar molecular mass. There is no attempt to simplify the mass spectrometry by any preselection of informative biomarkers.

WO 96/37777 (Nelson et a/) discloses a method for analysing antibody/antigen analytes using mass spectrometry. However, the object of the application is to determine the presence or absence of specific antibodies and/or antigens, and, if so, to measure the amounts present. There is no suggestion of a method that might be used to unambiguously identify an unknown organism.

30 In combination with the use of specific monoclonal antibodies and immunoaffinity purification, MS has allowed detailed structural mapping of many molecules (reviewed in Downard, 2000). In particular, the analysis of molecules of interest by immunoaffinity chromatography, followed by MS analysis of the isolated molecules has been performed on many proteins, for example calnexin (Yamashita et al, 1999). In some cases, proteins purified by immunoaffinity are then subjected to enzymatic digestion to generate a set of defined peptides, which are then analysed by MS, for example the Ty1 Gag protein of Saccharomyces cerevisiae (Yu et al, 1998). Lacey et al (2001) report the analysis of isoforms of transferrin by means of immunoaffinity purification followed by MS analysis in order to establish the structure of the carbohydrate 10 modifications responsible for the heterogeneity of transferrin. However, this involved the use of specific anti-transferrin polyclonal antibodies binding to molecules of the same amino acid sequence. The differences between molecules were within carbohydrate part of the glycoproteins and the antibodies were not cross-reactive.

Affinity purification of molecules carrying a common structural feature may be performed with other ligands than immobilised antibodies and is a technique very well-known in the art. Bundy and Fenselau (2001) report the use both of lectins to capture a variety of complex carbohydrates from a variety of micro-organisms, 20 and of defined carbohydrates to capture bacteria expressing lectin molecules.

The captured molecules, or peptides derived from them by acid hydrolysis, were then analysed by MS. Although these could be described as generic ligands used to capture a variety of molecules for subsequent MS analysis, the method is not used for the identification of unknown organisms. There is no selection of suitable biomarkers for such an application. There is no teaching as to how such biomarkers might be identified. There is no suggestion that the ligands suggested would have sufficient specificity to allow purification of the specific biomarker required or that the biomarkers used would be consistently present across the necessary variety of organisms required for the present invention.

30 There is no suggestion that the use of one, or a small set, of biomarkers might allow rapid identification of a wide range of micro-organisms.

Although more rapid identification of pathogens in a laboratory setting would be advantageous in itself, there is a further need for portable field-based systems that could be employed in both peacetime epidemics of human or animal disease, and in a military situation as part of biological weapons counter- measures. In this context, pathogens, such as plague or anthrax bacteria, may be used by an aggressor, typically delivered as an aerosol. Laser measurements may be used to detect the presence of an aerosol, but this may simply be a mist of, say, water, delivered as a dummy weapon (Willeke and Baron, 1993). There is a need to be able rapidly to make an accurate 10 identification, in the field, of matter, however delivered, which is suspected to be a biological weapon. Ion trap MS-based approaches to identification of bacteria have been reported previously (Krishnamurthy et al, 1999). In this case, following separation by reverse-phase microcapillary chromatography, whole bacteria were directly analysed, and identified purely on the basis of the spectrum produced. Although the authors comment on the potential for miniaturisation of the equipment for field use, there is no suggestion of the use of any form of immunoaffinity selection to simplify the MS analysis required and to make such analysis practical for rapid identification of a large range of organisms. In addition, there is no reference to the identification of particular 20 proteins for use as biomarkers, and no consideration of the reproducibility the MS spectrum obtained in different environmental conditions. Without characterisation of the biomarkers used, this technique is also vulnerable to inconsistency in the behaviour of the biomarkers in, for instance, ionisation characteristics, which further reduces its reliability.

Direct MS analysis of viral proteins has also been reported (WO 99/58727), but again, no affinity purification or use of common biomarkers is suggested. MS analysis of bacterial cell lysates has also been reported (Chong et al, 1997).

Bacterial samples were solubilised with guanidinium hydrochloride and Triton X- 30 100 before analysis by MALDI-TOF MS. There was no use of any form of affinity purification and the aim was to profile induction and repression of protein synthesis in Escherichia coli, rather than to identify unknown organisms.

STATEMENT OF INVENTION The invention provides a method of identifying a micro-organism comprising determining the molecular mass of at least one protein extracted from the plurality of proteins which constitute the micro-organism.

The invention follows from the discovery that, of all the thousands of proteins 10 which typically constitute a micro-organism, an identification can be made by assessing a relatively very small selection of proteins, even as few as one.

It is well-known in the art that a number of proteins, often those that perform some ubiquitous and vital metabolic function within the cell, are highly structurally conserved across a broad range of species. The fact that they perform very similar functions in different species sharing common metabolic pathways results in evolutionary pressure to conserve structural features on which functional properties depend. Such highly conserved proteins include enzymes concerned with basic cellular processes like glycolysis (eg triose phosphate isomerase) and nucleotide metabolism (eg adenylate kinase), DNA polymerases and heat shock 20 proteins.

Regions of such proteins that are conserved show a high degree of homology in amino acid sequence. As a result, they bear common immunological epitopes to which cross-reacting antibodies may bind so that a single monoclonal antibody may used to identify, or to isolate, of any of a family of such conserved proteins from a variety of species. In some cases a single antibody may bind to such a very widely conserved epitope and so be useful in isolating proteins from many species. In many cases, however, a number of such antibodies, binding to different epitopes on the same, or other proteins, may be used in combination, in 30 order to maximise the number of species identifiable and minimise the chance of a micro-organism that is present remaining undetected. Surprisingly, despite their highly conserved structure, the current invention demonstrates that the small differences between such proteins allow rapid and consistent identification of the species from which they are derived by accurate determination of their mass. The resolution obtained from mass spectrometry is easily capable of identifying single amino acid differences between proteins or peptides derived from them. Thus, the combination of affinity purification of highly conserved proteins bearing common epitopes, and subsequent mass spectroscopic analysis of such proteins, or peptides derived from them, may form the basis of a rapid and reliable method of identifying the micro-organism from which they are obtained. Proteins, or other biological molecules, used in this way are known as 10 biomarkers.

This method depends on the availability of a database of biomarkers, relating accurate molecular masses of known biomarkers to the species from which they are derived. In some cases, it may be necessary to use more than one biomarker for unambiguous identification of a species, sub-species or strain.

Such databases are generated by growing the relevant micro-organisms under a range of conditions, mapping the proteomes by 2D-gel electrophoresis and with western blot using antibodies raised against the whole micro-organism cell lysate. Markers of interest, selected according to the criteria below, can rapidly 20 be identified, their masses accurately determined by mass spectroscopy and the mass entered into the database.

An important factor in the selection of biomarkers is that that their masses should be constant irrespective of variable factors such as cell cycle, or of growth conditions such as temperature or availability of nutrients. This is particularly relevant to the identification of micro-organisms in the environment, such as biological weapons, where conditions may be far from optimal for the organism concerned, and in response to which stress it may change its pattern of gene expression or of post-translational modification. It is therefore preferable that 30 they are not transiently modified by phosphorylation, lipidation or ribosylation, although if such modifications were known and consistent, this would not preclude the use of such molecules for identification. Biomarkers should also be consistently expressed, in all conditions, at levels high enough to make extraction and isolation in quantities large enough to make identification practical.

In view of such considerations, the heat shock protein (Hsp) families of molecules are particularly suitable biomarkers. Molecules such Hsp60 are highly conserved across species, are not post-translationally modified, and are consistently and ubiquitously expressed. In fact, their expression, since it is related to cellular stress, is increased when organisms are in sub-optimal 10 environmental conditions. Hsp60 (GroEL, chaperonin), together with its co- chaperonin Hsp10 (GroES) is involved in the ATP-dependent, post-translational folding of nascent polypeptides into their correct tertiary structures, as well as refolding of non-native proteins back into their correct native conformation (Sigler et al, 1998).

In addition to the use of cross-reacting antibodies to isolate appropriate biomarkers from a range of micro-organisms, other ligands may be used for affinity capture of such biomarkers. Lectins may be used to capture glycoproteins of glycolipids carrying a specific structural feature in their 20 carbohydrate modifications. Immobilised nucleic acids may be used to capture DNA-binding proteins. These may be generic DNA-binding proteins (such as polymerases) or may be sequence-specific binding proteins (such as transcription factors or restriction endonucleases), depending on the ligand used.

Immobilised RNA aptamers and ribozymes may also be used to bind specific target structures (reviewed by Hoffman et al, 2001). Artificial dye ligands are capable of binding diverse molecules sharing a common structural feature such as a cofactor binding pocket (see Affinity Chromatography: principles and methods, Pharmacia LKB Biotechnology, 1988). As an example, Cibacron Blue F3G-A binds a variety of NAD or NADP-requiring enzymes, and enzymes that 30 have specificity for adenylyl substrates such as adenylate kinase, which is a useful conserved biomarker for the present invention.

The combination of immunoaffinity purification of one or more highly conserved biomarkers from cell lysates using a cross-reacting antibody or some other generic binding ligand, followed by mass spectroscopic analysis of the one or more biomarkers used, preferably by ion trap mass spectroscopy, by reference to a database, provides a rapid, reliable and reproducible method of identifying micro-organisms for a variety of applications.

In an alternative embodiment, the captured biomarker may be enzymatically digested to produce a predictable set of peptides consistent with the enzyme 10 used and the known amino acid sequence of the candidate molecules from the range of species recorded. The spectrum of masses produced is a fingerprint characteristic of the biomarker from which they originated and can be cross- referenced to a database for identification of the organism involved. The use of immobilised enzymes is a convenient way of simplifying the process for automation and also reducing the complication of enzyme molecules being present in the peptide mixture to be analysed.

For field biological warfare counter-measures applications, it is envisaged that the entire process from sampling the environment, through concentrating and 20 lysing the cells, affinity purifying the biomarker (s) of interest, eluting said biomarkers, delivering them to the mass spectrometer, recording the mass spectrum obtained, matching the spectrum to the best fit in a database and finally cross-referencing this information to obtain an identification of the organism detected, will be automated within a portable unit. Extra steps, such as enzymatic cleavage of captured biomarkers, would be invoked automatically if a definitive identification does not result from the first analysis. A proportion of the original eluate from the affinity purification step will be retained for this purpose.

Depending on the precise application, the ultimate read-out may be a precise identification of an organism or strain thereof. For battlefield applications, a 30 simple"safe"or"not safe"read-out might be appropriate.

Further automatic units may be designed for other applications. For instance, a bench-top unit for the analysis of blood, or tissue samples for hospital and laboratory use.

Accordingly the current invention provides a biomarker characterised in that species homologues of said biomarker derived from the majority of species in at least two genera of micro-organisms are substantially structurally similar, such that said structural similarity allows isolation of said biomarkers from different species of micro-organism and that each biomarker derived from each species of 10 micro-organism in said genera has a unique molecular mass.

Preferably, said biomarker is characterised in that it is a protein and in that said structural similarity consists of substantial similarity of amino acid sequence. It is also preferred that said micro-organisms are bacteria.

Preferably, said biomarker is characterised in that least three species homologues share at least one common epitope allowing isolation by immunoaffinity chromatography. More preferably, at least one common epitope is shared by at least five species. Even more preferably, it is a heat shock 20 protein and, most preferably, it is Hsp60.

Alternatively, said biomarker may be adenylate kinase.

Also provided is a method of identifying micro-organisms comprising: a. identifying a biomarker characterised in that species homologues of said biomarker derived from the majority of species in at least two genera of micro-organisms are substantially structurally similar, such that said structural similarity allows isolation of said biomarkers from different species of micro- organism and that each biomarker derived from each species of micro- 30 organism in said genera has a unique molecular mass; b. isolating said biomarkers by affinity chromatography chromatography directed towards regions of structural similarity; c. measuring the mass of said biomarkers by mass spectrometry, and; d. analysing the combination of molecular mass data obtained with reference to a database and thereby deducing the species of micro-organism present.

Preferably, said biomarkers are isolated from a cell lysate.

More preferably said biomarkers are isolated by means of immunoaffinity chromatography and, most preferably, by immobilised antibodies that bind specifically to cross-reacting epitopes present on marker molecules derived from 10 a variety of micro-organism species.

Alternatively, the method may include the additional step of cleaving the isolated biomarkers into defined fragments before determining their molecular mass by means of mass spectroscopy. Preferably, said cleavage of biomarkers is achieved by means of enzymatic digestion.

Preferably, the measurement of molecular mass of biomarkers or fragments thereof is by means of ion trap mass spectrometry.

20 Also provided is a method of identifying macromolecular toxins comprising: a. Isolating one or more toxins by affinity chromatography; b. measuring the molecular mass of said toxin (s) by means of mass spectrometry; and c. analysing the combination of molecular mass data obtained with reference to a database and thereby deducing the identity of the toxin (s) present.

Another embodiment of the invention comprises an apparatus for the automatic performance of any of the above comprising: a. a means for isolating said biomarkers or toxins; 30 b. a unit comprising a mass spectrometer capable of determining the molecular masses of said biomarkers or toxins c. a data processing device capable of matching the data obtained with a database of known molecular masses and thereby deducing the identity of the micro-organism or toxin detected.

Alternatively, said apparatus further comprises a unit comprising one or immobilised proteolytic enzymes capable of cleaving said biomarkers.

Definitions 10 As used herein"biomarker"means an environmental biochemical parameter, detection or quantification of which may be used as a means of identifying a potential biological hazard. In this case, it specifically refers to structurally conserved biological macromolecules, including proteins, that may be isolated from a wide range of micro-organisms, and used to identify said micro- organisms.

As used herein"affinity chromatography"means"a type of adsorption 20 chromatography in which the molecule to be purified is specifically and reversibly adsorbed by a complementary binding substance (ligand) immobilised on an insoluble support (matrix)" (see Affinity Chromatography: principles and methods, Pharmacia LKB Biotechnology, 1988).

As used herein"immunoaffinity chromatography"means a form of affinity chromatography in which the immobilised ligand is an antibody or epitope- binding derivative thereof.

As used herein"species homologue"means an equivalent gene or gene product 30 from another species. Such homologues perform equivalent functions and share a degree of sequence similarity at the amino acid level. As used herein, no assumptions are made as to the evolutionary relationship between the organisms involved.

DETAILED DESCRIPTION OF THE INVENTION The invention will now be described by way of example, with reference to the figure of the drawings in which: 10 Figure 1 is a schematic layout of the functional elements in a system utilising the method according to the invention.

Figure 2 shows a graphical comparison the masses of Hsp60 protein from a variety of potentially pathogenic bacteria Figure 3 shows an indirect ELISA measure of the binding affinities of monoclonal IgG1 A57-E4 to recombinant Hsp60 proteins from Francisella tularensis and Burkholderia pseudomallei.

20 Figure 4 is a graphical comparison of the peptide fingerprints resulting from Arg- C digests of HSp60 proteins of Brucella abortus and Staphylococcus epidermidis.

Figure 5 is a graphical comparison of the molecular masses of adenylate kinase from a range of potentially pathogenic micro-organisms compared with human Hsp60.

Example 1 An automatic sampling and identification system With reference to Figure 1, a vacuum device (not shown) is used to capture a 30 sample of an aerosol suspected to contain pathogenic bacteria. The aerosol mixed with a carrier liquid, and the suspension is fed into the system (1) via a sampler (2). From there, the suspension is delivered to an ultrasonicator (4) within which ultrasound is used to break down the cell walls of any bacteria within the suspension, thereby releasing bacterial constituent proteins into a lysate. Inevitably, the lysate will also contain debris, so downstream from the ultrasonicator (4) is a filter (5), which prevents the passage of unwanted matter.

In some cases, lysis may be improved by the use of a detergent, although this should not interfere with the immunoaffinity step downstream. Suitable mild non- ionic detergents are well-known in the art and include polyoxyethylene based detergents (such as Triton X-100 and X-114, Nonidet P40, and the Brij series) and n-octyl a-D-glucopyranoside.

10 Next comes an immunoaffinity module (6) in which one or more bacterial biomarkers, if present in the suspension, are isolated. Within the module (6) are one or more immobilised antibodies, specific for said biomarker (s). Biomarkers in the lysate passing through are thereby bound, whilst the remaining fluid passes through and is discarded. This step not only isolates the relevant biomarkers, but effectively concentrates them from what may be a very dilute lysate. After washing through the lysate, a small volume of elution buffer is admitted to the unit, to remove bound biomarkers.

20 The released biomarkers are delivered to a de-salter (8) whereupon they are de- salted before passing to an ion trap mass spectrometer (10) in which their individual molecular masses are determined. The combination of molecular masses obtained is then cross-referenced with a database of the molecular masses of the relevant biomarkers in a range of bacteria so as to identify any match. The output may be a specific identification, or the operator may simply be notified that the area is either"safe"or that it is"un-safe"and that appropriate protective measures are required.

In the event that biomarker proteins proteins are too large for individual analysis, 30 eluted proteins may be sent to the de-salter (8) via an enzymatic digester (12) in which the proteins are cleaved at predictable points in their amino acid sequence and the resultant peptides analysed. The pattern of peptide molecular weights

produced is diagnostic when compared to a database of such predicted peptides (see Example 3) Example 2 The use of Hsp60 as a biomarker to identify potentially pathogenic bacteria The average molecular mass of Hsp60 from a wide variety of organisms may be both predicted to a high degree of accuracy from the known amino acid sequence (corrected for mixture of isotopes present) and directly measured using the appropriate purified recombinant protein. Although Hsp60 is highly conserved across many species, not just bacteria, mass spectrometry allows highly accurate determination of mass and allows proteins molecules differing by as little as three mass units to be distinguished. Comparison of such measured values with a database of known values allows identification of the species involved, as shown in Table 1.

Table 1 Hsp60 Mr Bacterium 58015.3 Da Chlamydia trachomatis 57301.7 Da Francisella tularensis 57154.8 Da Salmonella typhimurium 56757.3 Da Burkholderia pseudomallei Figure 2 shows a graphical comparison the Hsp60 masses of a wider range of organisms illustrating that many species may be identified purely on the basis of their Hsp60 mass, as measured by mass spectrometry However, In order to reduce the background of other proteins, some of which might have confusingly similar masses, affinity purification of a relevant biomarker is preferred. In the case of Hsp60, it is possible to immunoaffinity

purify protein from cell lysates by means of cross-reacting antibodies. As an example, monoclonal antibody A57-E4 (Affinity Bioreagents Inc) binds to the linear epitope RGIDKA present in the Hsp60 of many potentially pathological organisms, including Bordetella pertussis, Burkholderia cepacia, Burkholderia pseudomallei, Chlamydia trachomatis, Chlamydophila pneumoniae, Chlamydophila psittaci, Coxiella burnetii, Haemophilus influenzae, Escherichia coli, Francisella tularensis, Klebsiella pneumoniae, Legionella pneumophila, Neisseria meningitidis, Pseudomonas aeruginosa, Salmonella typhi, Vibrio cholera, Yersinia enterocolitica.

The use of such an antibody would therefore allow purification of Hsp60 from, and identification of, a wide range of potential pathogens. The binding of this antibody to the Hsp60 proteins of Francisella tularensis and Burkholderia pseudomallei was confirmed and quantified by a standard colorimetric indirect ELISA as shown in Figure 3.

Table 2 Tryptic peptide map of Hsp60 from C trachomatis Average Mass* Position Peptide sequence 4029.607 503-543 SALESAASVAGLLLTTEALIAEIPEEKPAAAPAMPGAGMDY 3791.379 231-265 DFLPVLQQVAESGRPLLIIA EDIEGEALATLWNR 3183.553 197-224 GYLSSYFATNPETQECVLED ALVLIYDK 2717.020 142-167 EIAQVATISANNDAEIGNLI AEAMEK 2575.881 395-420 VDDAQHATIAAVEEGILPGG GTALIR 2462.838 421-442 CIPTLEAFLPMLTNEDEQIG AR 2333.760 286-307 AMLEDIAILTGGQLISEELG MK 2111.313 84-104 AGDGTTTATVLAEAIYTEGL R 1842.076 181-196 GFETVLDIVEGMNFNR 1723.935 484-499 DAYTDMLEAGILDPAK 1379.617 462-473 EGAIIFQQVMSR 1309.430 327-338 EDTTIVEGMGEK 1287.279 351-361 QIEDSSSDYDK 1231.409 105-116 NVTAGANPMDLK 1174.400 308-318 LENANLAMLGK 1145.336 65-74 HENMGAQMVK 1096.141 474-483 SANEGYDALR 1074.271 133-141 ISKPVQHHK 1049.227 380-389 VGAATEIEMK 1048.138 171-180 NGSITVEEAK 951.067 42-50 SFGSPQVTK 872.055 371-379 LSGGVAVIR 828.987 125-131 VWDQIR 803.887 58-64 EVELADK 781.800 7-12 YNEEAR 772.879 454-461 QIAANAGK 731.867 21-27 TLAEAVK 719.775 277-283 APGFGDR 710.851 36-41 HWIDK 699.868 447-453 ALSAPLK 689.786 51-57 DGVTVAK 688.758 339-344 EALEAR 614.762 28-33 VTLGPK 579.689 345-349 CESIK £.."S.. x A,,. r, '_dS-. v : za,. , Q 545,2 = V K. , t:. s I_ 45 6'f6' 364-367 w LQER 533.602 75-79 VCAVK 517.646 226-230 ISGIK Peptides of mass < 500 excluded.

* Average isotopically corrected mass (M+H) Table 3 Tryptic peptide map of Hsp60 from C pneumoniae Average Mass* Position Peptide sequence 3805.406 232-266 DFLPVLQQVAESGRPLLIIAEEIEGEALATLWNR 3213.580 198-225 GYLSSYFSTNPETQECVLED ALILIYDK 2722.019 143-168 EIAQVATISANNDSEIGNLI AEAMEK 2728.109 504-530 SALESAASIAGLLLTTEALI ADIPEEK 2561.854 396-421 VDDAQHATIAAVEEGILPGG GTALVR 2405.786 422-443 CIPTLEAFLPMLANEDEAIG TR 2319.733 287-308 AMLEDIAILTGGQLVSEELG MK 2097.286 85-105 AGDGTTTATVLAEAIYSEGL R 1957.147 328-345 EDTTIVEGLGNKPDIQAR 1828.049 182-197 GFETVLDWEGMNFNR 1739.934 485-500 DAYTDMIDAGILDPTK 1372.507 531. 544 SSSAPAMPSAGMDY ,-. :. : : 1 1 4 = ; _s :, r p. 30 5 6, ; 463=474 ; ; EGAIICQGULAiR 19PST 30.6w. =.,,, 352-362 QIEDSTSDYDK:.. 1231.409 106-117 NVTAGANPMDLK 1191.428 309-319 LENTTLAMLGK 1145.336 66-75 HENMGAQMVK 1096.141 475-484 SAN EGYDALR 1074.271 134-142 ISKPVQHHK 1049.227 381-390 VGAATEIEMK 951.067 43-51 NGSITVEEAK 875.951 59-65 EIELEDK 872.055 372-380 LSGGVAVIR 801.958 126-132 VWDELK 788.878 455-462 QIASNAGK 781.800 8-13 YNEEAR 731.867 22-28 TLAEAVK 719.775 278-284 APGFGDR 713.895 448-454 ALTAPLK 710.851 37-42 HWIDK 689.786 52-58 DGVTVAK 614.762 29-34 VTLGPK 592.687 346-350 CDNIK 559.726 323-327 VIVTK 545.616 365-368 LQER 533.602 76-80 EVASK 519.680 273-277 VCAVK 517.646 227-231 ISGIK Peptides of mass < 500 excluded.

* Average isotopically corrected mass (M+H) Example 3 Hsp60 peptide maps derived from Hsp60 by trypsin digestion As illustrated in Tables 2 and 3 above, enzymatic digestion of closely related Hsp60 proteins of similar overall molecular mass yields distinctive patterns of peptides that may be resolved by mass spectrometry. Trypsin cleaves peptides at the carboxy-peptide link of arginine and lysine residues (except where the next residue is a prolin). Allowing for a mass accuracy of 0.01 %, a few peptides are too similar to distinguish (boxed). In other cases, some very short peptides share identical composition and so have identical masses, and single free amino 10 acids result from the cleavages. Even allowing for this, each peptide set constitutes a unique fingerprint, diagnostic of a specific organism from which the protein is derived.

Example 4 Comparison of Arg-C Hsp60 peptides from Brucella abortus and Staphylococcus epidermidis The endopeptidase Arg-C (clostripain), as its name suggests, cleaves the carboxy-peptide bonds of arginine. Figure 4 shows a graphical comparison of the peptide fingerprints obtained from Arg-C digestion of Hsp60 from B. abortus 20 and S. epidermidis. As shown in Figure 3, the masses of the whole Hsp60 proteins from these organisms are similar (57649 and 57529, respectively including N-terminal methionines). However, the peptide sets obtained are quite distinct and characteristic of the organisms involved.

Example 5 Use of adenylate kinase as a diagnostic biomarker Figure 5 shows a comparison of the masses of the highly conserved intracellular enzyme adenylate kinase from a variety of micro-organisms (bacteria and the protozoal parasite Shistosoma mansion/) as well as the human protein. Adenylate 30 kinase is a nucleoside monophosphate kinase that catalyses the reversible phosphotransferase reactions between adenosine monophosphate, diphosphate and triphosphate. This enzyme plays an important role in the synthesis of nucleotides that are required for a variety of cellular metabolic processes, as well as for RNA and DNA synthesis. Adenylate kinase fulfils the criteria of a useful biomarker for the disclosed invention, in that it is highly conserved across species and yet each species has a unique protein distinguishable by mass. It is also consistently expressed and essential for metabolism.

Example 6 Identification of E coli from molecular mass measurement of whole Hsp60 biomarker by electrospray mass spectrometry 10 To demonstrate the practicality of the invention, an anti-Hsp60 immunoaffinity column together with electrospray mass spectrometry were used to identify a bacterium, as follows.

Methods Preparation of antibody columns The ligand (monoclonal antibody A57-E4 (Affinity Bioreagents Inc) was dialysed into 0.2M NaHCO3, 0.5M NaCI, pH8.3 (coupling buffer) before binding to the column. The optimal volume was 1ml with an optimal concentration of between 1 and 10mg/ml.

20 A 1 ml NHS-activated Sepharose 4 in a Fast Flow Hi-Trap column (Pharmacia Biotech) was used. The column was washed with 3x2ml volumes of 1 mM HCI to remove the storage solution (isopropanol), keeping the flow rate to below a drop every two seconds to avoid compressing the matrix. The column was injected with ligand solution and incubated at room temperature for 30 minutes. The column was washed and deactivated by alternate washes with 0.5M ethanolamine, 0.5M NaCI, pH8.3 (buffer A) and 0.1 M acetate, 0.5M NaCI, pH4.0 (buffer B) (3x2ml of buffer A, 3x2ml of buffer B followed by 3x2m1 buffer A). The column was then equilibrated and stored in phosphate buffer containing 0.1% 30 (w/v) sodium azide.

Sample purification method All samples were run on NHS-activated columns on the AKTA prime system (Pharmacia Biotech). Samples were loaded in 20mM sodium phosphate, pH7.5 and eluted in 3M urea at either pH8.0 or pH2.0. A sample volume of 1 ml was loaded onto the column via the sample loop and impurities were washed away with 5ml of the phosphate buffer. 6ml of elution buffer was sent through the column, the first 1 ml of buffer was allowed to flow to waste while the following 2mls were collected for analysis by electrospray mass spectroscopy. The column was regenerated by washing with 4ml phosphate buffer, followed by 10ml 3M 10 urea and then 1 owl phosphate buffer in preparation for the next sample. The flow rate for all steps was 1 ml/min.

Mass Spectrometrv experimental procedure Eluent from the immunoaffinity column was collected as 2ml fractions and stored overnight at 4°C. Samples were injected into a 2 ml holding loop. The contents of the holding loop were then loaded on to a C8 cartridge (Hichrom) at a flow of 1 ml/min in 20% B (A=0. 1% TFA in water B= 0.1% TFA in acetonitrile/water 90/10 (v/v)). After washing to remove buffer salts the protein was eluted into the Quattro II tandem quadruple mass spectrometer (Micromass UK Ltd) using 20 90% B at a flow of 25 pLI/min. Acquisition was performed in continuum mode.

Scan range m/z 700-2000 at 5s per scan. Capillary voltage was 3kV and cone voltage was ramped from 33V to 74V over the m/z range scanned. Source temperature was 80°C and both LM Res and HM Res were set to 15.5. The elution peak from the cartridge was approximately 1 min in duration. The instrument was calibrated using horse heart myoglobin.

Results Cross-reactivity of antibody to Hsp60 biomarkers 30 As shown in Figure 6, a standard binding assay demonstrates that Hsp60 from a number of bacterial species cross-react with a monoclonal antibody raised against Chlamydia trachomatis Hsp 60 and binding a conserved epitope (RGIDKA). The binding curves indicate that such an antibody is suitable for immunopurification of Hsp60 biomarkers from a range of bacterial species.

Identification of E coli Hsp60 K12 by molecular mass Figure 7 shows the mass spectrum detected from eluate from the anti-Hsp60 immunoaffinity column that had been loaded with bacterial protein. The indicated molecular mass peak at 57203.1 1.8 Da matches that of Hsp 60 from an E. coli K12 variant reported by Burland et a/ (1995). The standard E coli Hsp60 mass, as detailed in the Swiss-Prot database entry P06130, is given as 10 57137 Da. However, the variant reported by Burland et a/has two mutations, A261 L and 1266M, which together give an expected mass of 57197. This is within the known mass accuracy of the instrument ( 0.01 %) and allows an unambiguous identification of not just the organism, but an individual strain and/or mutant.

REFERENCES Bakhtier R and Tse FL. Biological mass spectrometry: a primer. Mutagenesis 20 15 : 415-430 (2000).

Bonenberger TE, Ihrke PJ, Naydan DK and Affolter VK. Rapid identification of tissue micro-organisms in skin biopsy specimens from domestic animals using polyclonal BCG antibody. Vet Dermatol 12 : 41-47 (2001) Bundy JL and Fenselau C. Lectin and carbohydrate affinity capture surfaces for mass spectroscopic analysis of microorganisms. Anal Chem 73: 751-757 (2001).

30 Burland VD, Plunkett G, Sofia HJ, Daniels DL and Blattner FR. Analysis of the Escherichia coli genome VI : DNA sequence of the region from 92.8 through 100 minutes. Nucleic Acids Res 23: 2105-2119 (1995) Chong BE, Wall DB, Lubman DM and Flynn SJ. Rapid profiling of E. coli proteins up to 500kDa from whole cell lysates using matrix-assisted laser desorption/ionization time-of-flight mass spectrometry. Rapid Commun Mass Spectrom 11 : 1900-1908 (1997).

Downard KM. Contributions of mass spectrometry to structural immunology. J Mass Spectrom 35: 493-503 (2000).

10 Henderson SC, Valentine SJ, Counterman AE and Clemmer DE. ESI/ion mobility/time-of-flight mass spectrometry for rapid and sensitive analysis of biomarker mixtures. Anal Chem 71 : 291-301 (1999).

Hoffman D, Hesselberth J and Ellington AS. Switching nucleic acids for antibodies. Nature Biotechnol 19 : 313-314 (2001).

Krishnamurthy T, David MT, Stahl DC and Lee TD. Liquid chromatography/ microspray mass spectrometry for bacterial investigations. Rapid Commun Mass Spectrom 13 : 39-49 (1999).

20 Lacey JM, Bergen HR, Magera MJ, Naylor S and O'Brien JF. Rapid determination of transferrin isoforms by immunoaffinity liquid chromatography and electrospray mass spectrometry. Clin Chem 47: 513-518 (2001).

Rowley A, Choudhary JS, Marzioch M, Ward MA, Weir M, Solari RCE and Blackstock WP. Applications of protein mass spectrometry in cell biology.

Methods 20: 383-397 (2000).

Sigler PB, Xu Z, Rye HS, Burston SG, Fenton WA and Horwich AL. Structure 30 and function in GroEL-mediated protein folding. Annu Rev Biochem 67: 581- 608 (1998).

Willeke K and Baron PA. Aerosol Measurement-Principles, Techniques and Applications. Van Nostrand Reinhold (1993).

Yamashita T, Kiyoki E, Tomita Y and Taira H. Immunoaffinity purification and identification of the molecular chaperone calnexin. Biosci Biotechnol Biochem 63 : 1491-1493 (1999).

Yu L, Gaskell SJ and Brookman JL. Epitope mapping of monoclonal antibodies by mass spectrometry: identification of protein antigens in complex biological systems. J Am Soc Mass Spectrom 9 : 208-215 (1998).