General enquiries on this form should be made to:Defra, Science Directorate, Management Support and Finance Team,Telephone No. 020 7238 1612E-mail: research.competitions@defra.gsi.gov.uk

SID 5 Research Project Final Report

SID 5 (Rev. 3/06) Page 1 of 18

NoteIn line with the Freedom of Information Act 2000, Defra aims to place the results of its completed research projects in the public domain wherever possible. The SID 5 (Research Project Final Report) is designed to capture the information on the results and outputs of Defra-funded research in a format that is easily publishable through the Defra website. A SID 5 must be completed for all projects.

This form is in Word format and the boxes may be expanded or reduced, as appropriate.

ACCESS TO INFORMATIONThe information collected on this form will be stored electronically and may be sent to any part of Defra, or to individual researchers or organisations outside Defra for the purposes of reviewing the project. Defra may also disclose the information to any outside organisation acting as an agent authorised by Defra to process final research reports on its behalf. Defra intends to publish this form on its website, unless there are strong reasons not to, which fully comply with exemptions under the Environmental Information Regulations or the Freedom of Information Act 2000.Defra may be required to release information, including personal data and commercial information, on request under the Environmental Information Regulations or the Freedom of Information Act 2000. However, Defra will not permit any unwarranted breach of confidentiality or act in contravention of its obligations under the Data Protection Act 1998. Defra or its appointed agents may use the name, address or other details on your form to contact you in connection with occasional customer research aimed at improving the processes through which Defra works with its contractors.

Project identification

1. Defra Project code SE0308

2. Project title

Molecular pathogenesis of Brucellosis - unlocking the secrets of host specificity

3. Contractororganisation(s)

Veterinary Laboratories Agency                         

54. Total Defra project costs £ 544730(agreed fixed price)

5. Project: start date................ 01 April 2003

end date................. 31 March 2006

SID 5 (Rev. 3/06) Page 2 of 18

6. It is Defra’s intention to publish this form. Please confirm your agreement to do so...................................................................................YES NO (a) When preparing SID 5s contractors should bear in mind that Defra intends that they be made public. They

should be written in a clear and concise manner and represent a full account of the research project which someone not closely associated with the project can follow.Defra recognises that in a small minority of cases there may be information, such as intellectual property or commercially confidential data, used in or generated by the research project, which should not be disclosed. In these cases, such information should be detailed in a separate annex (not to be published) so that the SID 5 can be placed in the public domain. Where it is impossible to complete the Final Report without including references to any sensitive or confidential data, the information should be included and section (b) completed. NB: only in exceptional circumstances will Defra expect contractors to give a "No" answer.In all cases, reasons for withholding information must be fully in line with exemptions under the Environmental Information Regulations or the Freedom of Information Act 2000.

(b) If you have answered NO, please explain why the Final report should not be released into public domain

Executive Summary7. The executive summary must not exceed 2 sides in total of A4 and should be understandable to the

intelligent non-scientist. It should cover the main objectives, methods and findings of the research, together with any other significant events and options for new work.

The aim of this project was to utilise current genomic and post-genomic technologies to investigate species-specific differences between Brucella isolates with special emphasis on their possible relevance to recognised host preferences. The objectives were as follows:1. In silico post-genomics analysis2. Genomic comparisons via microarray studies3. Genomic comparisons via subtractive hybridisation studies. 4. Proteomic comparisons via two-dimensional gel electrophoresis (2-DE) and proteome identification via Practical considerations reflecting the complexity of the techniques undertaken in this project meant that work was confined to only three isolates that represented recent clinical isolates of the main pathogenic species. The isolates selected were:B. abortus biovar 1 UK 18/03-211 isolated from bovine lymph nodes (Scotland 2003)B. melitensis biovar 2 F8/01-155 isolated from bovine milk and swabs (Kosovo 2001)B. suis biovar 1 01-5744 a recent field isolate from a pig (France 2002).

Genomic differences between species were revealed through in silico analysis of published genome data, and supported by subtractive hybridisation (SH) studies using the three field isolates. Microarray development was not completed during the lifetime of the project and thus objective 2 was not completed (the customer was informed of this during the course of the project). Both the in silico analysis and SH approaches revealed few differences between the three main Brucella species highlighting the genetic conservation within the Brucella genus. In addition, no differences were observed between the field isolates used for the SH study and the genome sequence indicating that there are likely to be few large strain-specific differences within a species. It is possible that the few differences observed between species could affect host specificity although much more extensive experimental work would be required to investigate this.

Proteomics techniques (2- Dimensional gel electrophoresis [2-DE] and Phoretix gel comparison software, or Liquid chromatography with on line mass spectroscopy [LC-MS]) were used to search for protein expression differences between in vitro cultivated whole cell preparations of the three isolates. Neither of the techniques used revealed consistent and reproducible differences between the proteomes of the three isolates. In line with the genomic analysis this finding suggests a high degree of similarity between the proteomes of the three main Brucella species. Moreover, species-specific expression of proteins that could be linked to the recognised host specificities was not apparent from this data. The underlying

SID 5 (Rev. 3/06) Page 3 of 18

hypothesis for this study was that the different species of Brucella (which have recognised host specificities or preferences) differentially express certain proteins, and that these proteins are responsible for, or indicative of, the observed host specificity. Our investigations of the proteome of in vitro cultivated Brucella isolates failed to reveal any conclusive expression differences between strains. However, it is possible that if host specificity determinants are expressed by Brucella strains, an interaction with the host is necessary to trigger the expression of these determinants. Thus to take into account the influence of the host system on the Brucella proteome further LC-MS work was undertaken to determine the proteome of B. abortus UK 18/03-211 infected bovine lymph node samples isolated from clinical cases of brucellosis. The aim of this experiment was to determine the differences observed between the in vitro cultivated proteome of B. abortus and that of B. abortus in association with the host. Notably the sensitivity of the LC-MS technique was not sufficient to allow detection of B. abortus specific proteins in the lymph node lysate, but was able to reveal distinct differences in the proteomes of infected compared with uninfected bovine lymph nodes. Although significant further analysis would be required to confirm these findings the preliminary data suggests a proteome indicative of immunologically active cells from the infected samples in comparison to a resting immune proteome profile from the uninfected samples. It is possible that with further investigation these differences could be confirmed as biomarkers specific to Brucella infection and eventually they could be exploited to produce diagnostic tests. Thus, whilst this approach did not “unlock the secrets of host specificity”, potentially important insights into the host response to infection were gained from this work.

Gottingen™ mini-pigs and Balb/c mice were used in experimental infection studies. Animals were infected with B. suis biovar 1 (01-5744), or Y. enterocolitica O:9 (YE 234/02) and the infection was monitored for a three month period. Control ‘uninfected’ animals were also used in these studies. Standard immunological responses of pigs (Brucella specific antibody production and recall IFN production) were measured at weekly intervals post-infection. These assays were used to demonstrate current problems with immunodiagnosis of porcine brucellosis: antibody detection methods have poor specificity and Yersinia infection triggers false positive serological reactions (FPSR), IFN assay has relatively poor sensitivity being useful for diagnosis at the herd level but unreliable for diagnosis of individual animals. An attempt to discover alternative specific biomarkers of infection that could be exploited for diagnosis was made through analysis of serum and plasma collected over the course of the study. Perbio™ (Endogen) SearchLight™ Cytokine array technology was used to measure the generation of nine different cytokines (including IFN) in the plasma from Brucellergene™ stimulated whole blood. This data revealed Brucella infection specific elevation of IL-6 and IL-8 as well as IFN. These cytokines have not previously been investigated for diagnostic potential in brucellosis and thus may represent new specific biomarkers of infection. Further work to verify the diagnostic relevance of specific IL-6 and IL-8 induction in a larger sample number of pigs, small ruminants and cattle is currently under investigation in the follow-up project SE0310.

Ciphergen™ SELDI protein-chip™ technology was used to assess the protein profile of serum samples collected from pigs or mice. Comparison of the SELDI trace data obtained from Brucella versus Yersinia, or Brucella versus uninfected animals did not reveal any striking differences in protein profiles. Furthermore, longitudinal analysis of the data in the mouse study did not reveal temporal evolution of biomarkers. Most notably the presence of known biomarkers (infection group specific antibody (an IgG marker at around 180 kDa)) could not be demonstrated by this technique, suggesting some refinement of the technique is necessary. However, despite this apparent lack of sensitivity in the high molecular mass ranges (15 – 200 kDa) a number of potentially useful diagnostic biomarkers (differential expression / concentration of a peak in one sample group compared to another) were indicated in the low mass range (0.1 – 15 kDa). These markers were detected through statistical analyses (non-parametric T tests) of the profiles and further investigation would be necessary to confirm these as specific biomarkers of brucellosis. Most importantly a larger sample number would be required to verify the detection of these markers in field samples. Furthermore, refinement of the techniques to include fractionation of sera and the assessment of different binding and washing conditions may be able to improve the sensitivity of the technique and reveal further markers.

Overall, this project has confirmed that there are limited differences in the genome and proteome of members of the Brucella genus. The few genomic differences identified by the SH work are completely supported by microarray data and in silico analyses of other researchers. Proteome analysis revealed significant differences between the proteomes of infected versus uninfected bovine lymph nodes but was unable to reliably elucidate differences between the three Brucella isolates. Consequently, with relatively few differences between the isolates we were unable to identify Brucella specific proteins that may contribute to host specificity. Similarly, biomarker profiling experiments were unable to conclusively identify markers which clearly differentiate brucellosis and yersiniosis infections in pigs, or markers that are indicative of different stages of brucellosis infection in pigs or mice. Differences in the protein profiles of mice and pigs were observed but these were unrelated to infection status and thus did not provide any insight into host specificities. Several potential avenues for further research were highlighted by these

SID 5 (Rev. 3/06) Page 4 of 18

investigations. Those with immediate practical value (for instance IL-6 and IL-8 based diagnosis) and are being pursued in further Defra funded projects.

Project Report to Defra8. As a guide this report should be no longer than 20 sides of A4. This report is to provide Defra with

details of the outputs of the research project for internal purposes; to meet the terms of the contract; and to allow Defra to publish details of the outputs to meet Environmental Information Regulation or Freedom of Information obligations. This short report to Defra does not preclude contractors from also seeking to publish a full, formal scientific report/paper in an appropriate scientific or other journal/publication. Indeed, Defra actively encourages such publications as part of the contract terms. The report to Defra should include: the scientific objectives as set out in the contract; the extent to which the objectives set out in the contract have been met; details of methods used and the results obtained, including statistical analysis (if appropriate); a discussion of the results and their reliability; the main implications of the findings; possible future work; and any action resulting from the research (e.g. IP, Knowledge Transfer).

The objectives have been reproduced from the original proposal (in blue font) with a report on work carried out towards each below. Figures and tables are included in a seperate document.

Objective 01:To elucidate the difference between species/biovars of Brucella using molecular techniques (in silico post-genomic analysis; microarray analysis; suppressive subtractive hybridisation and proteomics). A panel of Brucella strains will be assembled to include type strains, vaccine strains, those from diverse geographical locations and representatives of species and biovars in our possession. Representatives of closely related bacteria (Ochrobactrum, Rhizobium and Agrobacterium) will be included. These will be used to evaluate the above techniques as appropriate.

Practical considerations reflecting the complexity of the techniques undertaken in this project meant that work was confined to only three isolates that represented recent clinical isolates of the main pathogenic species. The isolates selected were:B. abortus biovar 1 UK 18/03-211 isolated from bovine lymph nodes (Scotland 2003)B. melitensis biovar 2 F8/01-155 isolated from bovine milk and swabs (Kosovo 2001)B. suis biovar 1 01-5744 a recent field isolate from a pig (France 2002).

The in-silico comparative genomics will give a detailed comparison of the genomes of three Brucella species, namely, B. melitensis, B. abortus and B. suis. Although this will not assist with biotype variance, it will however give detailed species differences between these isolates.

The summary below is extracted from the SE0308 Annual Report 2004-5.

The in-silico comparative genomics of the three main Brucella species, namely, B. melitensis, B. abortus and B. suis, has revealed the striking similarity between these species. Identity amongst most ORF’s was greater than 99%. Some polymorphism was identified among putative outer membrane proteins or associated with insertion/deletion events. A specific 2,750 bp sequence has been identified within an autotransporter encoding region on Chr II in B. abortus. In B. suis, a 7,738 fragment associated with metabolic capabilities appears species specific. B. melitensis unique ORF’s were found associated with a 20kb pathogenicity island, however, many of these were shared with B. abortus highlighting the greater similarity between these two species compared with B. suis. Whether these differences alone will account for the observed differential host preferences among the brucellae remains unresolved (Halling et al. 2005). Our results support the location of these genomic insertions and deletions for B. suis and B. melitensis (the genome has only just been released for B. abortus Halling et al. 2005).

Further work looking at B. abortus was not carried out at the VLA but extensive comparative genomics data is now publicly available (Halling et al., 2005; Chain et al., 2005). Findings of the subtractive hybridisation studies reported below are related to these comparative genomics studies.

SID 5 (Rev. 3/06) Page 5 of 18

Microarray slides will be produced using the cloned open reading frames produced through our ORFeome project, (further information available from : http://www.invitrogen.com/Content/World/gway_flyer.pdf). This will be used to compare the genomes of panel members with that of B. melitensis 16M. Identification of differentially possessed genes will be attempted by comparison to annotated genomes.

The report below is reproduced from a supplemental report to the 2004/5 SE0308 Annual Report. As predicted the microarray has only just been completed and is still undergoing validation at the Spanish laboratory.

These milestones all related to the use of microarrays to ascertain uniquely possessed/expressed genes present in B. abortus; B. melitensis and B. suis through comparison with B. melitensis 16M used to construct the array. To deliver this within the allocated budget, it was necessary for us to work as part of a European consortium with the array currently in construction in Spain. Although progressing, we do not have direct control over this and delivery will not be within the remaining life of this project. Furthermore, another group based in USA has now published comparative whole genome hybridisations of all six brucellae with B. melitensis 16M (Rajashekara et al. 2004). This data revealed 217 ORF’s that differed among these species, characteristically clustered into genomic islands. Deleted ORF’s when compared with B. melitensis 16M are given in the table below:

Analysis of multiple strains of these species demonstrated that these changes were usually conserved within species (see table below).

SID 5 (Rev. 3/06) Page 6 of 18

Several deletions were detected among other Brucella species when compared with B. melitensis 16M. These were tested in multiple isolates of each species. If felt necessary, we could further develop this, looking at representative biovars and further isolates not tested in this publication. Any work we subsequently do with regard to this milestone will consequently be a duplication of this study and unlikely to produce further tangible outputs, or if on different isolates, further add to the table above.

Once again the findings of the microarray array work by Rajashekara et al. (2004) will be correlated to the output of our subtractive hybridisation work discussed below.

Subtractive hybridisation will be used to focus on differentially possessed genes representative of our panel members. We will initially have to select appropriate restriction endonucleases to cut test genomes into appropriately sized fragments for inclusion with the vectors used (Clontech).

Subtractive hybridization is a method for isolation of differentially distributed nucleic acids. The substrates for subtraction are nucleic acids (tester) containing specifically possessed sequences to be extracted (target) and nucleic acids used for comparison of target molecules (driver). During subtractive hybridization, the tester sample is mixed with the driver to form driver-tester hybrid molecules that are non-target molecules. Tester-tester hybrid molecules and single-stranded tester molecules, also formed during subtractive hybridization, represent the fractions enriched in target molecules. After hybridization, non-target molecules and fractions enriched in target molecules are separated ultimately allowing identification of genomic differences between strains.

The aim of the work in this project was to identify genomic differences between strains that could serve as potentially useful diagnostic markers. The three recent field isolates discussed above representing the three most common Brucella species were used for comparison. Each of these strains was subtracted from the two other strains in both directions resulting in a total of six subtractions. Thus, for example, the B. suis and B. abortus strains were compared using both B. suis as the driver (to identify sequences specific to the B. abortus strain) and B.abortus as the driver (to identify sequences specific to the B.suis strain). Briefly, genomic DNA was prepared from these isolates and digested with HaeIII and subtraction was carried out as described in the PCR-Select Bacterial Genome Subtraction Kit User Manual (http://www.clontech.com/techinfo/manuals/PDF/PT3170-1.pdf). The two subtracted DNA samples were then cloned into plasmid pAL9 to generate a library and transformed into E. coli. Ninety-six clones obtained from each of the specific libraries were picked and used to amplify a PCR product corresponding to the insert. These were spotted onto membranes and subject to differential screening with both of the subtracted DNAs that had been radioactively labelled used as probes. Comparison of the resulting blots allowed selection of clones that appeared to be specific for one of the strains.

Results of this screening suggested that a number of clones containing strain specific inserts had been generated. For the B. melitensis v B. abortus comparison 16 B. abortus specific clones were obtained while 43 B. melitensis specific clones were obtained. For the B. melitensis v B. suis comparison 36 B. suis specific clones were obtained while 41 B. melitensis specific clones were obtained. Finally, for the B. suis v B. abortus comparison 35 B. abortus specific clones were obtained while 39 B. suis specific clones were obtained. Six of the clones for each subtraction were selected for Southern Blot analysis to confirm differential screening results. HaeIII digested genomic DNA representing tester and driver samples was resolved on an agarose gel and transferred to nylon membranes. These were hybridised to radioactive probes prepared from six randomly selected differential clones. An example of such an experiment is shown below where six clones obtained as

SID 5 (Rev. 3/06) Page 7 of 18

specific to B. suis by using B. abortus as the driver are used to probe B. abortus and B. suis genomic DNA. The results confirm that 5 of the clones appear specific for B. suis while the other appears to represent a restriction polymorphism (C7). Given the relatively small number of clones obtained and the fact that Southern blotting of the samples of six showed that in most cased the majority of clones were strain specific it was decided to go ahead and sequence the inserts in all 210 clones obtained rather than screen all samples by Southern blotting.

Sequences of all 210 inserts were obtained by sequencing from both ends of the insert using primers complementary to the vector sequence. By the time this work was completed genome sequences of representatives of all three strains examined here were publicly available and thus the sequences of all inserts were compared against these genomes to confirm whether the differences determined here were also present in the sequenced strains.

A summary of these findings are shown in Table 1 in the Appendix. For each experiment the table shows the number of clones that mapped to a particular location, the identity of the genes shown to be differentially present and whether the clone sequences were confirmed to be present or absent in the B. abortus (A), B. melitensis (M) and B. suis (S) genome sequences. Most clones that were confirmed to be differentially present in the genomes were found to correspond to the same genomic difference. Thus for example considering the target B. suis/driver B. melitensis subtraction 27 of the clones corresponded to one genomic region while 3 clones corresponded to another region. The remaining clones were found to be present in all three genomes and all represented individual hits on different parts of the genome. These hits most likely represent ‘leakage’ in the technique (i.e. sequences that are not differentially possessed) although the possibility that they may be strain specific differences such that they represent sequences deleted in the strain used in our work relative to the genome sequence strain of the same species cannot be discounted. However the fact that all such clones were isolated hits across the genome in contrast to clusters of hits seen for clones confirmed to be missing from the genomes suggests the vast majority represent ‘leakage’.

Overall relatively few regions of difference were apparent with only 16 differentially possessed regions apparent from all six of the subtraction experiments. In addition most of these were seen in multiple experiments (as would be expected for true differences). Thus for example the first region of difference (shown in blue), a region apparently absent in B. suis but present in B. melitensis and B. abortus was seen in both the target B. abortus/driver B. suis and the target B. melitensis/driver B. suis subtractions. Seven regions were identified in two extractions and these are shown in the corresponding colours in Table 1. Thus the overall number of genomic differences seen amounted to nine. Comparison with data in the literature confirmed that all of these differences had noted previously either in microarray experiments (Rajashekara et al, 2004) or in comparative genomic studies (Chain et al, 2005; Halling et al, 2005).

The blue region of difference (RD) corresponds to a 20.8 Kb deletion in B. suis previously noted by comparison with a B. melitensis microarray as containing mostly hypothetical genes (Rajashekara et al, 2004) and noted in genomic comparions (Halling et al, 2005). The dark yellow RD that is specific to S. suis corresponds to an 18.3 Kb region noted in genome comparisons (Chain et al, 2005; Halling et al, 2005) and extensively characterised recently by Lavigne et al. (2005). This region appears to have been acquired recently by phage integration and encodes a number of hypothetical proteins, a transcriptional regulator and a type IV conjugal transfer cluster. The orange RD represents a smaller B. suis specific fragment of 2.7 Kb noted in genomic comparisons (Chain et al, 2005; Halling et al, 2005). It was noted that this deletion affects an amino-acid ABC type transporter and a glutathione S-transferase and concluded that the loss of this transporter (likely responsible for transport of arginine, ornithine and lysine) may account for the observed inability of B. melitensis and B. abortus to oxidise these amino acids. The grey RD represents a 25.1 Kb region absent in only B. abortus. As noted by Halling et al. (2005) this region had previously been identified by Vicainzo et al. (2001) and encodes a number of ORFs likely to be involved in polysaccharide biosynthesis. The light yellow RD that appears to be deleted from B. abortus represents a small region of 2.7 Kb also noted by Halling et al. (2005) as encoding a probable surface protein and two partial ORFs corresponding to insertion sequences. The green RD corresponds to a B. abortus specific region BruAb2–0168 identified by Halling et al. (2005) within a complex region of diversity. Of the two RDs not

SID 5 (Rev. 3/06) Page 8 of 18

seen in two reciprocal subtractions one represents a region absent from B. abortus. It corresponds to a 1.2 Kb deletion noted by comparison with a B. melitensis microarray and in genomic comparisons (Chain et al, 2005; Halling et al, 2005) encoding a prokaryotic signalling diguanylatecyclase/phosphodiesterase domain (GGDEF) protein. The other white RD corresponds to a 969 bp deletion in B. melitensis noted by Halling et al. (2005) as encoding a NAD dependent epimerase and a hypothetical protein.

The SH thus fulfilled its objectives by identifying differences between the three genomes examined. The work has somewhat been overtaken by time as illustrated by the fact that all of the differences observed had been reported in the last two years based on comparisons of genome sequences and/or microarray comparisons. However the congruence of data presented here with these studies does confirm the validity of the approach. The small number of differences observed overall again highlights the genetic conservation within the Brucella group. In addition, although different strains were used here, no differences were observed between these and the genome sequences (i.e there were no strain-specific differences within a species). It is possible that some of the differences observed could affect host specificity although much more extensive experimental work would be required to investigate this.

Two-dimensional gel electrophoresis will be used to compare the spot profiles of panel members. Where differentially expressed proteins are seen, the spots will be excised and identified using MALDI-TOF technology (equipment based in collaborating institutes). This should yield the amino acid sequence of excised spots, facilitating protein identification.

Experimental approach

Evaluation and comparison of the total protein content of each Brucella isolate has the potential to reveal protein expression differences that cannot be predicted through genome and sequence based approaches such as SSH. Depending on the techniques used, proteome comparisons may reveal quantitative differences in protein content (up or down regulation relative to other isolates), or qualitative differences (presence or absence of particular proteins, or post-translational modifications) between species and/ or isolates. Our aim for this study was to define and compare the proteomes of the three Brucella isolates, in order to elucidate species-specific differences that may account for the reported differences in host-preference. The panel of 3 Brucella isolates investigated by SSH were also selected for proteomic analysis.

Initial cultivation of the isolates for proteome isolation was attempted in a minimal broth media designed to mimic both the acidity and the nutrient deprived conditions of the intracellular habitat or ‘Brucellosome’. Unfortunately, sufficient biomass for downstream processing was not obtained under these conditions and protocols had to be revised. Standard Serum Dextrose Agar (SDA) based culture was used for the growth of Brucella for protein isolation. The culture conditions were standardised for the three isolates of Brucella to reduce variability in experimental procedures. For each isolate clonal isolates of the strain were streaked onto SDA p lates and incubated at 37oC, 10% CO2, for 72 hours. Following incubation, growth was harvested from the plates by scraping into 0.1M PBS. Each culture was washed by centrifugation with 0.1M PBS, and resuspended to a final volume of 1 ml. After re-suspension methanol was added to each culture to inactivate any live Brucella. Sterility and purity checks were completed. For purity the morphology of the cultures on SDA was observed and a sample of the culture was assessed by 16S RNA sequencing. For each isolate three replicate culture preparations were prepared in order to reflect variability in cultivation, and protein isolation techniques.

Total protein preparations (proteomes) of Brucella isolates were prepared from the inactivated pellets according to the methods described by Wagner et al. (2002). Total protein concentration was determined by Bradford assay. A total of 100 g of protein was separated by isoelectric focussing (IEF) (First dimension), and then SDS-PAGE (Second dimension). Following electrophoresis all gels were stained with ImperialTM Protein Stain (Pierce) according to manufacturers instructions. Gels were preserved in 10% glycerol, 2% Sodium azide and covered with cellophane until analysis. For each preparation four replicate gels were produced in order to account for variation in the 2-DE procedure.

For analysis, each of the final 36 gels were assessed using specialised software (Phoretix version 6.01 (Nonlinear Dynamics, Newcastle)). The software defines spots from scanned images of the gel. Adjustments to spot defining protocols were made to account for the quality of the gel and the image. The best gel / image for each Brucella isolate was selected from the 12 replicates (4 X gels, from 3 X culture isolates), to represent the basic proteome of that isolate. These optimal gels were selected on the basis of having the maximum number of clearly defined spots for that particular isolate. These optimal gels were then compared against one another to elucidate differential expression of proteins or spots between the different Brucella isolates. For this comparison the three selected images / gels were re-analysed under the following conditions: sensitivity 9315, noise factor 5, operator size 39 and background 1. The procedure utilised statistical methods to compare spot position, intensity and size in order to determine presence, absence and up or down regulation. The intention was to select and excise differentially expressed spots for protein sequence identification following MALDI-ToF through collaboration at the University of East London (UEL).

SID 5 (Rev. 3/06) Page 9 of 18

An alternative approach to proteomics was also investigated. Prepared samples (methods as above for 2-DE) were further treated with by sonication in a lysis buffer using procedures adapted from Molloy et al., (1998). Protein concentration post-sonication was determined by Bradford assay, and 100 g of total protein was subjected to overnight tryptic digestion in ambic buffer. Samples were then analysed by LC-MS. First dimension separation of samples was performed by cation exchange chromatography, and the second dimension separation was completed by reverse phase HPLC with on line mass analysis. Proteins were identified and proteomes compiled with SEQUEST using the data from the B. abortus, B. melitensis and B. suis genomes. The proteomes were further compared based on the spectrum count as a measure of relative protein expression using Microsoft Access.

In addition to the in-vitro cultivated Brucella frozen lymph nodes isolated from cattle associated with this Brucella outbreak in 2003 were also used in the LC-MS approach. It was intended that a comparison of culture grown B. abortus and lymph node associated B. abortus may reveal Brucella specific proteome differences attributed to the different growth conditions or representative of host-pathogen interaction. Ultimately, analysis of clinical material was considered more likely to reveal protein components associated with host-specificity than culture derived material. For this work a selection of lymph nodes, which had been identified as culture positive, were selected for analysis along with some non-infected lymph nodes taken from Brucella free experimental animals housed on-site at VLA Weybridge. Each lymph node was macerated in PBS and total protein was isolated, quantified and subject to LC-MS procedures as previously described.

Results

Assessment of proteomic differences between B. abortus, B. suis and B. melitensis by 2-DE (IEF + SDS-PAGE).

Thirty-six gels (twelve per isolate) were generated in this study. Representative gels from each of the Brucella isolates are shown in Figure 1. Using Phoretix software to define spots on these gels (under defined and standardised sensitivity and specificity parameters) 286 spots were detected on the B. abortus gel, 149 spots for B. melitensis and 257 spots from the B. suis gel. None of the spots identified were unique to one species: i.e.: each spot (protein) was represented in at least one other species. Differential spot size was indicated for a large proportion of the spots in the selected gels, but overall the variation observed for spot size between isolates was not significant when compared to the internal variation observed for a single spot on replicate gels. Thus, although the software can identify spot size differences between isolates, these are unlikely to be real and biologically relevant differences between the species.

Since inherent variability in spot definition did not permit reliable identification of differentially expressed proteins between the three isolates, appropriate ‘species-specific’ protein spots could not be selected for identification by MALDI-ToF procedures.

Assessment of proteomic differences between B. abortus, B. suis and B. melitensis by LC-MS

Three replicates of each isolate were investigated by LC-MS. This approach allows two dimensional separation of tryptic digested peptides according to mass and cationic charge. Peptides were identified by mass and charge and the results were cross-referenced against the respective genome database for each particular species of Brucella in order to identify the proteins. As with the 2-DE process a certain amount of variability was demonstrated between replicate tests of the same isolate, therefore for purposes of inter-isolate comparison an average proteome was determined for each isolate. An average total of 680, 593 and 659 proteins from B. abortus, B. suis, and B. melitensis respectively, were identified using this technique. Thus, the number of proteins observed for any isolate using this LC-MS technique was considerably greater than that observed from standard 2-DE methods.

To enable comparison of data between different species the protein identification data from the B. melitensis and B. suis isolates was also interpreted using the annotation of the B. abortus genome database. These data are summarised in Table 2. A list of the first 100 (most abundant) proteins identified for each isolate is provided in Appendix 1. Initial comparisons of these average proteomes suggested differences between the three species and the possibility of unique species-specific protein expression. There were 481 proteins identified in B. melitensis but not B. abortus, similarly 460 identified in B. suis but not in B. abortus. However, further data mining revealed that observed differences could be attributed to annotational differences in the genome databases and were not true differences in the protein component of the isolates.

Assessment of the proteome of bovine lymph nodes from uninfected and B. abortus infected (confirmed culture positive) cattle.

SID 5 (Rev. 3/06) Page 10 of 18

This work did not reveal the presence of Brucella specific proteins in the protein preparations from known infected animals, but did indicate a shift in the proteomes of infected versus uninfected bovine lymph nodes (bLN). Notably, an overall reduction in the number of proteins represented in the samples of infected bLN was determined in comparison to the protein isolation from uninfected bLN. The three samples from uninfected LN were found to contain an average (mean ± standard deviation) of 1186 ± 135 identifiable proteins, whereas the three samples from infected bLNs were assessed as containing an average of 844 ± 114 identifiable proteins. Data analysis revealed the consistent presence of 232 specific proteins in all three uninfected bLN replicate samples that were not represented in the infected group. Conversely, 91 of the consistently identified proteins detected the infected bLN samples were not represented in the uninfected sample group. This data is summarised in Table 3. A full list of proteins identified as differentially expressed between these two sample populations is provided in Appendix 2.

Discussion

This experiment was devised in order to determine basic differences in the proteomes of B. suis, B. abortus and B. melitensis which may relate to the reported host preferences exhibited by these pathogens. Unfortunately, using both the 2-DE and LC-MS approaches adopted in this study, a reliable and accurate assessment of proteomic profiles for the three isolates was not achieved.

Using the 2-DE approach we encountered significant variability in results (number and positioning of spots) for replicate samples, suggesting poor reproducibility of technique and making identification of ‘real differences’ impossible to confirm. Reproducibility is a frequently cited encumbrance in proteomic research and is considered particularly problematic for manual 2-DE procedures as employed in this project (Fey and Larsen, 2001., Dowsey et al, 2003). To combat this problem we endeavoured to produce average gels based upon multiple protein isolations encompassing biological variation occurring during Brucella cultivation, and multiple replicate 2-DE analysis of these samples to reflect technology (IEF and SDS-PAGE) induced variability. The generation of 12 replicate samples (4 gel replicates of 3 cultivation preparations) for each isolate allowed us to define an average or composite gel for the isolate, which could then be compared against composite gels from the other isolates. Significantly, under the conditions defined for analysis, there were more differences observed between the 12 gels that were collated to form the composite for each particular isolate, than observed following comparison of the three isolate specific composites. Notably, on analysis of the selected ‘best gels’ for each species, none of the comparisons (B. abortus versus B. suis, B. melitensis versus B. suis, or B. abortus versus B. melitensis) revealed the presence or absence of spots unique to a single species of Brucella. Differences in spot size (area) and intensity (volume), which are often interpreted as differential expression of particular proteins were not reliably reproducible across replicate gel samples and therefore not considered to be true reflections of biological variation. Moreover, basic comparison of the number of spots revealed in this investigation against published data on the B. melitensis 16M proteome (Wagner et al., 2002, DelVecchio et al., 2002) indicated that our techniques lacked the sensitivity required to elucidate species specific differences. Using almost identical methodologies Delvecchio et al. (2002) observed 883 non-redundant spots, and proceeded to identify 557 proteins in the B. melitensis 16M proteome, whereas our efforts only revealed 164 spots for B. melitensis strain F8/01-155. Whilst some of these differences may represent possible real differences between the two strains it is extremely unlikely that this discrepancy is attributable solely to differential protein expression. In a basic comparison, it appears that the use of robotic technologies and more sensitive staining techniques (SYPRO ruby) by DelVecchios team resulted in a greater than five fold increase in the number of B. melitensis protein spots resolved by our 2-DE procedures. Consequently, we conclude that our techniques were able to measure only a fraction (< 20%) of the anticipated proteome for strain F8/01-155 and with such a restricted sample it is impossible to draw robust conclusions about isolate specific differences.

Despite extensive effort it was not possible for us to optimise our manual 2-DE techniques to match the levels of sensitivity and reproducibility achieved by other investigators in this field. Furthermore, formal identification of spots was not possible in-house at VLA, meaning that gels were analysed and interpreted at VLA but had to be transferred to a collaborating institute for analysis. In conclusion, the 2-DE approach for proteome analysis proved to be impractical.

The technology and expertise for 2-DE work was limited within VLA during the lifetime of this project, leading us to investigate an alternative proteomics approach (LC-MS) that did not rely upon the time consuming and expensive production of 2-DE gel images. The LC-MS technique has previously been employed for proteome investigations of Salmonella typhimurium (Coldham and Woodward, 2004), and expertise and technology is readily available on site at VLA. Using the LC-MS technique we were able to identify an average of 884 ± 31 proteins in the B. melitensis F8/01-155 replicates. 659 of these proteins were common to all three replicates. This represents a almost four fold increase (X 3.97) in the number of proteins of B. melitensis F8/01-155 that were resolved by LC-MS in comparison to the 2-DE technique. More modest, but none the less significant improvements, were also noted for B. suis (X 2.56 increase in protein detection), and B. abortus (X 2.37 increase in protein detection) isolates using this technique. However, despite apparent increases in protein detection

SID 5 (Rev. 3/06) Page 11 of 18

capability we were still unable to determine proteomic differences between the three Brucella isolates under investigation.

In addition to the increased resolving power, a further advantage to the LC-MS technique is that data defines protein identities rather than just mass and charge as seen in 2-DE (Protein identification has to be established following spot picking and MALDI-Tof for the 2-DE approach), allowing a more in depth critique of the data. The comparison of the consensus proteomes (ie: proteins seen in all three replicates for one isolate), initially suggests some differential protein content between each isolate. However, caution needs to be exercised when interpreting this data. Notably, proteins indicated as present in one isolate and not another include the products of house-keeping genes or known constitutively expressed proteins. The apparent absence of these proteins from any of the tested preparations is not logical, and reflects the limitations of this technology rather than true differences in the isolate proteomes. Since the LC-MS protein identification technique is influenced by the relative abundance of the particular peptide sequences in the preparation, it is more likely that these differences represent quantitative differences in protein expression rather than absolute measures of presence or absence. Moreover, the discrepancy in house-keeping gene products between isolates casts doubt on the validity of expression differences observed for other categories of proteins. In conclusion, although this approach was able to resolve and identify more proteins per isolate than the 2-DE approach, considerable further optimisation of protein preparation techniques or LC-MS techniques would be required to generate biologically meaningful results.

Brucella spp., are intracellular pathogens. Doubts were raised regarding the relevance of the agar culture procedure for accurately reflecting an encounter with the host and thus promoting discovery of host specificity determinants in Brucella. To address this issue attempts were made to cultivate Brucella in a minimal media more representative of the acidic and nutrient deprived intracellular compartment in which Brucella reside (Media adapted from that described by Rafie-Kolpin et al, 2003), but sufficient biomass for downstream processing could not be obtained by these methods. In an alternative strategy, we sought to more accurately reflect host-pathogen interaction through direct examination of clinical material (lymph nodes) from cattle associated with the outbreak from which B. abortus strain UK 18/03-211 was isolated. Lymph nodes from infected cattle (and appropriate uninfected controls) were processed to generate whole protein preparations for analysis by LC-MS. It was hoped that Brucella specific proteins could be detected in the lymph nodes of infected cattle, and that this information could be compared against the proteome of the in-vitro cultivated isolate of UK 18/03-211.

The experiment was designed to provide proof of concept that LC-MS techniques could be used on clinical material in immunomics applications, and to test the sensitivity of the technique for detection of Brucella specific proteins. Unfortunately, Brucella specific proteins were not determined in the protein preparations from the selected infected lymph nodes, and therefore proteome comparisons from in vivo and in vitro isolated Brucella were not possible. Nevertheless, host derived proteomic differences between infected and uninfected samples were noted using this method. Significantly fewer proteins were identified in the preparations from infected lymph nodes compared to uninfected controls (1186 ± 136 proteins in uninfected LN preparations, and only 844 ±113 for infected lymph nodes), and analysis of the respective proteomes suggested focussed expression of immune system associated proteins in the infected group. A considerable number of proteins identified in the infected LN sample are associated with immune function and in particular with MHC expression, antigen processing and presentation, and the generation of bacteriocidal peptides. In contrast, the proteome of the uninfected LNs is more diverse. Although considerable further research would be necessary to validate these findings it is possible that the proteome profile observed in the infected LN samples represents that of an active LN. When LN cells are activated to deal with an infection, phagocytosis, and antigen presentation functions are up-regulated. Therefore, proteins associated with these processes are expected to become more abundant in the proteome. Conversely, other non-critical cellular functions may be diminished or remain at a basal level of expression in order to conserve energy. Since the LC-MS protein identification technique is linked to the abundance of peptides in the test sample, the ‘resting’ or uninfected LN are likely to have a more comprehensive and consequently more variable proteomic profile than LN engaged in the combat of infection. Our data conforms to this hypothesis with Brucella infected LNs presenting a distinctly ‘activated’ proteome. Further research in this area may reveal important diagnostic markers or improve understanding of host-pathogen interaction.

Definition of the ‘immunome’ for various infectious diseases (representing a snapshot of host and pathogen proteomes under defined conditions) is being pursued by a number of research groups (For review see Schonbach, 2003). Ultimately, the proteomics technologies have found utility for identification of novel diagnostic markers, drug discovery, or improved understanding of host-pathogen interaction. Notably, use of LC-MS and 2-DE for measuring inflammatory responses has recently been described by Verhoeckx et al (2004). In this study, microarrays, 2-D gel electrophoresis and LC-MS methods were used to monitor the transcriptome, proteome and metabolome of macrophages activated through exposure to LPS. Techniques were found to be sensitive enough to allow the discrimination of distinct immunomic profiles from activated cells exposed to different anti-inflammatory compounds. Similarly, Pitarch et al (2006) have used 2-DE approaches to identify serum proteins which are specifically represented in the immunome of patients with systemic candidiasis. These publications suggest that with appropriate validation and optimisation of the 2-DE or LC-MS techniques, further studies of

SID 5 (Rev. 3/06) Page 12 of 18

clinical material could reveal whether components of the host proteome are specific to Brucella spp., infection or induced by contact with any pathogen. Furthermore, comparison of the active proteome profiles of different hosts and different Brucella species may reveal responses restricted to particular combinations of host and pathogen that reflect the reported ‘host-preferences or host-specificities’. In conclusion, although we were unable to ‘unlock the secrets of host-specificity’ using the 2-DE approaches outlined in the initial project proposal, LC-MS techniques and expertise have been developed and experimental design has been modified such that future research projects could be better focussed towards this fundamental question.

Objective 02:To follow experimental B. suis infection in pigs in order to identify biomarkers of infection which could form the basis of future diagnostic tools.

Experimental approaches

The diagnosis of porcine brucellosis is problematic, not only through poor performance of tests when applied to individual pigs, but also through inherent specificity problems resulting in cross-reactive antibodies produced following infection with other microbes frequently found in pigs (Wrathall et al., 1993). Collectively, these problems have led to the conclusion that “none of the conventional serological tests has been shown to be entirely reliable in routine diagnosis in individual pigs” (OIE, 2004). This study was designed to permit concurrent evaluation of existing (serological) and novel (IFN or serum biomarker profiling) techniques for the accurate diagnosis of porcine brucellosis.

IFN production in response to stimulation with pathogen specific antigens as a tool for diagnosis has previously been investigated for cattle brucellosis and tuberculosis. Although these assays are technically complex they have may have value as confirmatory tests in the face of false positive serological reactions (FPSR). Commercially available porcine IFN ELISAs were evaluated in this study in order to evaluate the utility of this test for diagnosis of brucellosis in swine.

Ciphergen SELDI chip technology is used to profile protein composition of biological materials. The technique is relatively low resolution but is useful for elucidating differences in protein profiles (biomarkers). Definitive identification of individual markers must then be pursued through alternative methods. The objective of this investigation was to use the SELDI technology to elucidate potential Brucella specific biomarkers in the sera of infected pigs. Ultimately these disease specific biomarkers could be exploited for diagnosis of disease. A secondary objective was to assess whether any biomarkers observed in porcine infection were preserved in an alternative host / model system (the mouse).

To assure known providence of test material, Gottingen mini-pigs™ were challenged with Brucella suis or Yersinia enterocolitica O;9 under experimental conditions. Four groups of mini-pigs were used in the study: (a) High dose B. suis infection (challenged intra-conjunctivally with 1 X 107 B. suis biovar 1 (strain 01-5744) CFU / pig. 25 l dose volume)(b) Low dose B. suis infection (challenged intra-conjunctivally with 1 X 105 B. suis biovar 1 (strain 01-5744) CFU / pig. 25 l dose volume)(c) Y. enterocolitica O:9 infection (challenged orally with 1 X 1010 Y. enterocolitica O:9 (strain YE 234/02) CFU / pig). (d) PBS (sham) / Uninfected controls (sham intra-conjunctival challenge with PBS. 25 l dose volume)

Whole blood, and serum samples were taken from these animals at regular intervals post-infection and standard serodiagnostic tests, recall cytokine tests and bacteriological isolation data were performed in order to monitor the infection status. A selection of serum samples collected during this study were made available for biomarker analysis.

For the secondary objective, adult female BALB/c mice were challenged 1 X 10 6 CFU B. suis strain 01-5744 / mouse (intra-peritoneal inoculation, 100 l volume). A control group of mice were inoculated with an equivalent volume of PBS. At weekly intervals post-infection three mice per group were exsanguinated and the serum from this procedure was used for Biomarker analysis.

The selected porcine sera were examined using a range of Ciphergen biomarker chips (IMAC30 CU, Q-10, and CM-10) and under a range or conditions (denatured or non-denatured samples, alternative elution buffers, pH ranges of binding buffers, fractionation of sera), to determine the best conditions for stability of biomarker profile and maximum generation of peaks. This work demonstrated that when using non-denatured unfractionated sera the IMAC 30–Cu chips revealed the greatest number of peaks and thus these chips were selected for subsequent biomarker analyses. All suitable porcine material and murine material collected during the experimental infection studies was subsequently analysed using these conditions.

SID 5 (Rev. 3/06) Page 13 of 18

In addition to the specified Ciphergen™ Biomarker profiling approach, attempts were made to narrow the focus of the investigation toward cytokine profiles associated with infection. For this study, a small selection of porcine plasma samples generated from standard IFN- assays, were selected for assessment using the Perbio™/ Endogen™ Searchlight™ Cytokine array system. This multiplex cytokine array assay was used to detect and quantify concentrations of IFN, IL-1b, IL-2, IL-4, IL-6, IL-8, IL-10, IL-12p70, and TNF in the plasma samples from Brucella and Yersinia infected pigs or matched negative controls.

Results

Differential diagnosis of porcine brucellosis and yersiniosis following experimental infection

Figure 2 summarises the results of standard serology and IFN assay for Brucella and Yersinia infected animals during the acute phase of infection. The data illustrates the difficulties associated with sensitive and specific diagnosis of porcine brucellosis. Variation in responses of individual animals causes considerable problems in both the cELISA and IFN ELISA, with at least one animal from each group at each time giving contrasting results to that of its cohorts. At the population level, the cELISA was able to detect Brucella infection from four weeks post-infection, but the poor specificity of this assay is apparent in that Yersinia infected animals are also detected positive in cELISA. The data presented in Figure 2 shows that the detection of positive responses in the Yersinia group begins to decline over the time of the investigation in comparison to the results obtained from the matched Brucella group. Ciphergen ProteinChip™ Biomarker discovery

o PORCINE BIOMARKER IDENTIFICATION VIA CIPHERGEN TECHNOLOGY

The preliminary assessment of a small panel of sera from the porcine experimental infections revealed statistically significant variation in the quantity of certain ‘biomarkers’ between the Brucella and Yersinia infected populations, but did not elucidate biologically relevant biomarkers that could be exploited for diagnosis. The statistical analysis performed by the proprietary Biomarker Discovery software from Ciphergen relies upon non-parametric T tests to deduce p values for population means. Although statistically significant differences were evident at a population level the marker was present in both Yersinia and Brucella infected animals (and in negative animals) and therefore could not be used as a discriminatory tool for individual samples. In addition the method was unable to discriminate populations on a profile basis (ie: a Yersinia spectra is not notably different to a Brucella spectra or a negative spectra). Representative images of specific areas of the profile where potential biomarkers were observed are shown in Figure 3. The traces (figure 3 (i)) indicate an area of the profile (around 12-13 kDa mass range) from the low mass range (1500 – 15000 Da) optimised reading, generated from average readings of over 200 hits to each spot. The data indicates the presence of a peak at approximately 12.6 kDa that is more abundant in the two Brucella infection derived samples than the Yersinia derived sample. Figure 3 (ii) illustrates profile derived principal component analysis (PCA) plots for a potential biomarker (profile peak) derived from a selection of sera isolated throughout this project. Separation of the Brucella and non-Brucella population is observed in this data. However, restricting the sample set to known Brucella and Yersinia infected sera does not permit population level discrimination (data not shown). Similarly, a preliminary comparison of murine Brucella infected and uninfected sera did not reveal evidence of any infection specific biomarker. Examples of protein profile traces for high mass optimisation (10,000 – 50,000 Da) are not included in the report as very little evidence of proteins in this range was observed.

o LONGITUDINAL BIOMARKER IDENTIFICATION

Since we were unable to elucidate unique biomarkers in the first phase of the study the secondary objective to determine if such markers were preserved in the mouse model was not possible. In terms of profile comparison, differences exist between pig and mouse, but these do not relate to infection group. Therefore an alternative strategy was adopted for this phase where the data from each group was reassessed in longitudinal terms – ie: does the profile alter across the time course of infection? For this analysis the samples from each group at each time point were pooled and the variation in SELDI profile over time for each group was assessed. This was done for both porcine and murine samples. Infection and temporal changes to the profiles were not determined in either the murine or porcine samples. Figure 4 shows the longitudinal profile of markers from the murine studies. Notably, there are no immediately apparent markers to indicate the progression of a specific infection.

Alternative ‘biomarker’ profiling techniques

A selection of supernatants (n = 24) harvested post-stimulation for the IFN assay were also tested in using the Perbio™ SearchLight Cytokine array technology. This system involves simultaneous detection of nine porcine cytokines in a 96 well plate format Cartesian array, read by cooled LCC camera. The data is presented in Figure 5, which illustrates cytokine concentration [pg/ml] for each of the nine cytokines detected in the assay, for each of

SID 5 (Rev. 3/06) Page 14 of 18

the four groups of animals under investigation (Brucella high dose infection, Brucella low dose infection, Yersinia infection, uninfected). The data indicates detection of specific IFN responses in both the high and low dose Brucella infection groups, and also demonstrates differential expression of IL-6 and IL-8 by Brucella infected animals in comparison to animals known to be brucellosis free. Figure 6 highlights the IL-8, IL-6 and IFN results for a smaller panel of samples (Brucella and Yersinia infected, cELISA seropositive), and shows significant differences in the production of IFN and IL-6 between the two infection groups.

Discussion and future work

The experimental infection studies from both mini-pigs and mice provided a known provenance sample set with which to investigate novel approaches for improving diagnostic specificity. The infection of two dissimilar hosts with the same strain of Brucella also facilitated investigation of the ‘host-preference’ of B. suis for pigs.

Diagnosis

The use of the cELISA served to indicate the current problems of FPSR and diagnostic sensitivity for porcine brucellosis. The IFN assay was evaluated in this study for it’s ability to discriminate Brucella and Yersinia infected pigs. The IFN assay results indicate superior specificity of diagnosis compared to cELISA: none of the Yersinia animals were detected as positive in this assay during the course of the investigation (100% diagnostic specificity for IFN assay). However, the sensitivity of the assay is lacking on the following fronts. Firstly, the Brucella infected animals are not detected as positive in the IFN assay until week five of the investigation, representing a delay of one week in the time from infection to detection in comparison to the cELISA. This delay in detection time increases the risk of infection becoming established within a herd. Secondly, the variability of IFN responses for individual animals is significant with animals reporting positive one week and negative (unresponsive) the next. This means the test has poor predictive power for diagnosis of infection in individual animals.

During the lifetime of this project a similar experimental infection study was undertaken in collaboration with Jungersen et al (2006), showing high titre serological responses from Brucella infected pigs persisting throughout the 21-week study period, and the occurrence of FPSR in Y. enterocolitica infected pigs up to nine weeks post-infection. The conclusion from these investigations was that the relatively short duration of response observed in Yersinia infected animals could be exploited for diagnosis when dealing with suspect reactors (through appropriate timing of repeat sampling). Whilst this process may reduce the risk of slaughtering FPSR animals, it is both expensive and time consuming and more effective diagnostic tests are still required to reduce the need for re-testing. Our initial investigations suggest that IFN assay can improve the specificity of diagnosis but improvements are needed in the sensitivity of the assay before it can be reliably used as a diagnostic tool. In on-going research project SE0310 we aim to adapt the IFN assay to alternative technology platforms (luminex bead arrays, and fluorescence based immunoassay) with enhanced analytical sensitivity. Notably, work to evaluate the performance of the IFN assay in a field situation has revealed some decrease from the 100% specificity observed in experimental infection studies (Nieslen et al, 2006). In conclusion, with appropriate improvements in analytical sensitivity the IFN assay from porcine whole blood has the potential to be a useful confirmatory test in the face of FPSR. Furthermore, adjustments to the composition of the stimulatory antigen cocktail may be able to improve the diagnostic specificity of the assay and stimulate responses from a greater proportion of infected animals. Both of these options are under investigation in project SE0310.

SELDI biomarker profiling analysis failed to reveal unique biomarkers specific for Brucella infection in either the pig or the mouse models. Potential differences in the abundance or concentration of particular peaks or markers were indicated, which could form the basis of further investigations to formally identify markers relevant to diagnosis. However, a closer look at the raw data (data not shown) indicates that although statistically significant differences in the peak intensity (relative abundance of the protein) are observed, the range of intensities for each identified biomarker measured in the two different populations (Brucella infected versus other groups) overlap. Thus suggesting that whilst population level discrimination may be possible using this marker, individual animal disease status could not be reliably predicted.

To use the biomarker discovery software accurately requires multiple samples from multiple individuals to accurately represent a population, but the panel of material used in the pig experiment only consisted of four animals per group sampled at different intervals post-infection. A large panel of positive material (from real infections?) and negatives from the field may be a better sample set with which to look for diagnostically relevant markers. Unfortunately, during the lifetime of this project an appropriate panel of serum from naturally infected pigs was not available.

Longitudinal analyses of the murine sample set did not reveal any temporally restricted biomarkers. Data from existing serological techniques (in this study and similar reports in literature) confirms distinct and reliable kinetics for immunoglobulin generation following infection, and therefore we know that the serum profile of infected animals should change over time. However, using the SELDI approach under the conditions described

SID 5 (Rev. 3/06) Page 15 of 18

(unfractionated, non-denatured sera, IMAC-Cu-30 chips) we were unable to detect any differences in the profile of infected versus uninfected mice in the mass region of 180 kDa (the average molecular weight of IgG). The failure to observe IgG (or peaks at the appropriate mass for IgG) may be explained by the generally poor resolution of profile observed in this mass range, and suggests that further optimisation of sample preparation may be beneficial. It is possible that fractionation of sera could be used to improve the resolution of technique and therefore reveal differences (biomarkers present in low concentrations), that are currently obscured by the presence of high abundance proteins in the whole serum samples.

In conclusion, the SELDI profiling approach adopted for this study was unable to reveal potential diagnostic markers of infection, or markers restricted to Brucella infection in different hosts. Further work on fractionated samples (and with a larger sample set) may provide the means to address these questions. However this would be a significant project in itself and require the fulltime attention of a dedicated in-house SELDI specialist, and considerable investment in specialist equipment. Although the practical technique for sample handling is simple and relatively rapid, data interpretation and assay optimisation require considerable input and expertise. This work was undertaken through collaborations with Ciphergen™ SELDI users at Imperial College and with technical support through Ciphergen™ (Guildford, UK) as part of a technology demonstration workshop. These collaborators are dedicated specialists and their input has been invaluable for this project. There are no plans at present to pursue SELDI based biomarker discovery for brucellosis diagnosis, although techniques and contacts have been established that could be utilised for this purpose in future research projects, if necessary.

The Perbio cytokine array investigations indicated that Brucella infection in pigs prompts production of IFN-, IL-8 and IL-6 from whole blood in response to Brucellergene™. A specifically discriminatory cytokine profile based upon all 12 analytes in the assay could not be demonstrated between Brucella and Yersinia animals, but measurable differences in IL-6, IL-8 and IFN production were observed. It is possible that future investigations into the antigen specific production of IL-6, IL-8 and IFN may prove relevant to infection specific diagnosis, and this avenue is being pursued in project SE0310. Multiplex measurement of IL-8, IL-6 and IFN in the serum and plasma of infected animals (pigs, sheep or cattle) is being developed in on-going projects. Notably, the preliminary investigation described here demonstrated that using a multiplex immunoassay system did not demonstrate any loss in sensitivity in IFN detection compared to the monoplex ELISA. Overall, the investigation of host derived biomarkers of infection identified IFN (in monoplex and multiplex ELISA assays), IL-6 and IL-8 (multiplex assay tested) as potentially useful and specific markers of porcine brucellosis. Recent publications indicate that the transcripts for these cytokines are also elicited in response to Brucella infection in the mouse (Paranavitana et al 2005), suggesting that this immune profile is preserved in disparate species infections. Further work is underway to determine the value of these findings for diagnosis.

References

Chain PS, Comerci DJ, Tolmasky ME, Larimer FW, Malfatti SA, Vergez LM, Aguero F, Land ML, Ugalde RA, Garcia E. Whole-genome analyses of speciation events in pathogenic Brucellae. Infect Immun. 2005 Dec;73(12):8353-61.

Coldham NG, Woodward MJ. Characterization of the Salmonella typhimurium proteome by semi-automated two dimensional HPLC-mass spectrometry: detection of proteins implicated in multiple antibiotic resistance. J Proteome Res. (2004). 3(3):595-603.

DelVecchio, V. G., Wagner, M. A., Eschenbrenner, M., Horn, T. A., Kraycer, J. A., Estock, F., Elzer, P., Mujer, C. V. (2002). Brucella proteomes--a review. Vet Microbiol , 90, (1-4), 593-603.

Dowsey., A.W., Dunn, M.J., Yang, Guang-Zhong The role of bioinformatics in two-dimensional gel electrophoresis. Proteomics.(2003) 3, 1567-1596.

Fey, S.J., and Larson, P.M. 2D or not 2D. Curr. Opin. Chemical Biology (2001). 5, 26-33.

Halling SM, Peterson-Burch BD, Bricker BJ, Zuerner RL, Qing Z, Li LL, Kapur V, Alt DP, Olsen SC. Completion of the genome sequence of Brucella abortus and comparison to the highly similar genomes of Brucella melitensis and Brucella suis. J Bacteriol. 2005 Apr;187(8):2715-26.

Jungersen G, Sorensen V, Giese SB, Stack JA, Riber U. (2006) Differentiation between serological responses to Brucella suis and Yersinia enterocolitica serotype O:9 after natural or experimental infection in pigs. Epidemiol Infect. 134(2):347-57.

Lavigne JP, Vergunst AC, Bourg G, O'Callaghan D. The IncP island in the genome of Brucella suis 1330 was acquired by site-specific integration.Infect Immun. 2005 Nov;73(11):7779-83.

SID 5 (Rev. 3/06) Page 16 of 18

Nielsen K, Smith P, Yu W, Nicoletti P, Jungersen G, Stack J, Godfroid J. (2006) Serological discrimination by indirect enzyme immunoassay between the antibody response to Brucella sp. and Yersinia enterocolitica O:9 in cattle and pigs. Vet Immunol Immunopathol. 109(1-2):69-78.

Molloy MP, Herbert BR, Walsh BJ, Tyler MI, Traini M, Sanchez JC, Hochstrasser DF, Williams KL, Gooley AA. (1998) Extraction of membrane proteins by differential solubilization for separation using two-dimensional gel electrophoresis. Electrophoresis. 19 (5):837-44.

Paranavitana C, Zelazowska E, Izadjoo M, Hoover D.Interferon-gamma associated cytokines and chemokines produced by spleen cells from Brucella-immune mice.Cytokine. 2005 Apr 21;30(2):86-92

Pitarch, A., Jimenez, A., Nombela, C., Gil, C. (2006). Decoding serological response to Candida cell wall immunome into novel diagnostic, prognostic, and therapeutic candidates for systemic candidiasis by proteomic and bioinformatic analyses. Mol Cell Proteomics. 5, (1), 79-96.

Rafie-Kolpin, M., Essenberg, R., Wyckoff II, J.H. Identification and comparison of macrophage induced proteins and proteins induced under various stress conditions in Brucella abortus. Infect. Immun. (1996) 64, (12), 5274-5283.

Rajashekara G, Glasner JD, Glover DA, Splitter GA. Comparative whole-genome hybridization reveals genomic islands in Brucella species. J Bacteriol. 2004 Aug;186(15):5040-51

Schonbach C. From immunogenetics to immunomics: functional prospecting of genes and transcripts. Novartis Found Symp. 2003;254:177-88.

Verhoeckx, K. C., Bijlsma, S., Jespersen, S., Ramaker, R., Verheij, E. R., Witkamp, R. F., van der Greef, J., Rodenburg, R. J. (2004). Characterization of anti-inflammatory compounds using transcriptomics, proteomics, and metabolomics in combination with multivariate data analysis. Int Immunopharmacol. 4, (12), 1499-1514.

Vizcaino N, Cloeckaert A, Zygmunt MS, Fernandez-Lago L. Characterization of a Brucella species 25-kilobase DNA fragment deleted from Brucella abortus reveals a large gene cluster related to the synthesis of a polysaccharide. Infect Immun. 2001 Nov;69(11):6738-48.

Wagner MA, Eschenbrenner M, Horn TA, Kraycer JA, Mujer CV, Hagius S, Elzer P, DelVecchio VG. (2002). Global analysis of the Brucella melitensis proteome: Identification of proteins expressed in laboratory-grown culture. Proteomics; 8, 1047-60

Wrathall AE, Broughton ES, Gill KP, Goldsmith GP. Serological reactions to Brucella species in British pigs. Vet Rec. 1993 May 1;132(18):449-54.

References to published material9. This section should be used to record links (hypertext links where possible) or references to other

published material generated by, or relating to this project.

SID 5 (Rev. 3/06) Page 17 of 18

Papers

Cutler, S. and Whatmore, A. Progress in understanding brucellosis. (2003) The Vet. Rec., Nov 22, 641-642.

Cutler, S.J., Whatmore, A.M. and Commander, N.J. Brucellosis – new aspects of an old disease. (2005) Journal of Applied Microbiology, 98:1270-1281.

Cutler, S.J. Brucellosis – New paradigms for a classical pathogen. (2006) Culture (Oxoid), 27(1): 1-3.

Cutler, S.J. and Cutler, R.R. Brucellosis: The most common zoonotic disease? (2006) The Biomedical Scientist, April 2006, 1-5.

SID 5 (Rev. 3/06) Page 18 of 18