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 23

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 VM02144 - CSA6611

2. Project title

Identification of biomarkers and development of tests to detect administration of illegal and veterinary drugs in livestock

3. Contractororganisation(s)

HFL LtdNewmarket RdFordham     CambridgeshireCB7 5WW               

54. Total Defra project costs £ 737,712(agreed fixed price)

5. Project: start date................ 01 May 2004

end date................. 30 April 2008

SID 5 (Rev. 3/06) Page 2 of 23

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.This final report for period 2004-2008, project VM02144, summarises the work conducted towards the identification of biomarkers and development of tests to detect administration of illegal and veterinary drugs in livestock.The objectives of the project were set out as follows:

1. Use of in vitro systems to identify biomarkers using transcriptomics and proteomics2. Confirm that the biomarkers identified in the in vitro study are applicable in vivo when the animal

is challenged by administration of unauthorised β-agonists and anabolic steroids3. Development of assays to biomarkers based upon antibodies generated through phage display

technology4. Application of the assays to analysis of biological samples from animals challenged with growth

promoting agents and to “normal” samples.

The following results were obtained by the end of the project:

Optimisation of in vitro systems for biomarker production. Detection of putative biomarkers from transcriptomics and proteomics studies from challenges to cell lines with hormonal growth

Eight cell models (human LPS-treated macrophages, U937 cells, HepG2 cells, T47D breast carcinoma cell line, prostate-derived myofibroblastic cell line WPMY-1, rat skeletal muscle cell lines L6 and L8, and mouse C2C12) What’s the biological relevance of these cells lines as models? were investigated by transcriptomics and proteomics technology for suitability for

LPS-treated macrophages and U937 cells were challenged with four β-adrenoceptor agonists (clenbuterol, formoterol, salbutamol, and zilpaterol). A microarray experiment on LPS-treated macrophages revealed the differential expression of genes, encoding proteins belonging to the intercrine α/β- family, IL-6, transforming growth factor, amphiregulin and type I interferon families. In the proteomics study on LPS-treated U937 challenged with zilpaterol in combination with propranolol (β-adrenoceptor antagonist) only one secreted protein (macrophage inflammation protein) was identified by 2D gel separation and nano liquid chromatography tandem mass spectrometry (nLC-MS/MS) analysis. Proteomics data did not correspond to the proteins revealed in transcriptomic experiment. How do they explain this discrepancy? Were they actually using the same cells for RNA and protein samples?

The effects of anabolic steroid (AS) treatment were investigated on human (HepG2, T47D, WPMY-1), rat (C6 and C8) and mouse (C2C12) cells. The proteins reported in the literature as biomarkers of AS treatment (C1 esterase inhibitor and sex hormone binding protein, SHBG) were monitored in culture media of HepG2 cells by ELISA. Testosterone, nandrolone, boldenone and

SID 5 (Rev. 3/06) Page 3 of 23

stanozolol did not alter the C1 esterase inhibitor release. In addition, SHBG was not detected in the experiment. The microarray experiment did not reveal the expected up-regulation of genes encoding C1 esterase inhibitor, SHBG or aromatase. Why? Were the same cell lines used in the literature they reviewed? How did the positive controls fare?

Rat, L6 and L8, and mouse C2C12 skeletal muscle cell lines were investigated for their response to the steroids named above. The effect of insulin-like growth factor 1 (IGF-1) release into culture medium was investigated by ELISA. IGF-1 level appeared to be too low to detect by this method. Moreover, high concentrations of the anabolic steroids inhibited proliferation of cells.

Human T47D cell line containing an endogenously expressed androgen receptor and a reporter plasmid with the luciferase gene cloned downstream of the androgen response element, was exposed to three anabolic steroids to investigate whether they could activate the androgen receptor pathway. Microarray experiments did not reveal differentially regulated genes after exposure of T47D cells to AS compared to the untreated cells. The direct proteomic approach revealed the presence of a few high abundant proteins, mainly serum proteins, in culture alongside some interfering unknown proteins. What about the luciferase results?

Finally, the WPMY-1 cell line was investigated in the presence of a synthetic androgen receptor agonist R1881 and nandrolone. Neither R1881 nor nandrolone was able to induce a significant increase in proliferation of WMPY-1 cells. I am unclear about what the objective of this experiment was.

These in vitro studies did not reveal any protein biomarkers to be taken forward for further immunoassay development and an in vivo proteomics study was considered a priority.

Administration of β-agonists and anabolic steroids to animals Plasma samples were collected following the administration of a beta agonist, clenbuterol

hydrochloride (Ventipulmin® injection solution and Ventipulmine® syrup, two i.m. injections, 1.15 µg/kg body weight (BW), oral administration for a further 21 days, 20 µg/kg BW), and an AS nandrolone phenylpropionate (Nandrolin®, i.m., single, three injections weekly, 5 mg/kg BW) to five male (castrates) Holstein Friesian calves (one was administerd with clenbuterol and four animals were treated with nandrolone). Blood samples were collected for 62 days in total including pre-dose samples, administration samples, and withdrawal period samples up to 31 days after the final dose. The administration study was performed at Royal Veterinary College, London (years 2006-2007). Collection of administered animal tissue samples was transferred to CSL for incurred tissue projects.

Control drug-free samples were obtained from two untreated study animals and from a group of lactating cows (dairy farm, 59 animals). Why were the lactacting cows used as well as the control calves?

Development and optimisation of proteomic techniques for the analysis of blood samples from appropriate species

A number of the cell culture supernates (LPS-treated U937 cells, β-adrenoceptor agonists) and 12 bovine plasma samples (control and post administration) were analysed to evaluate the levels of inflammatory cytokines (TNF-α, IL-8) by ELISA. The high concentration of inflammatory cytokines was confirmed for cell supernates, however there was no evidence for their presence in plasma. The use of human ELISA kits for the detection of proteins reported in transcriptomic and proteomic studies (β-adrenoceptor agonists) did not reveal any presence of the proteins under consideration. Thus, these in vitro studies did not reveal any protein biomarkers which could be taken forward for further immunoassay development and an in vivo study was considered a priority. How good is the cross reactivity of the ELISA kits to bovine cytokines otherwise?

Five control plasma samples (administration study) and five plasma from each of the clenbuterol and nandrolone treated animals were analysed by 2D-gel (DIGE) in combination with nLC-MS/MS. The total number of protein spots on the DIGE map was 420 in the nandrolone vs control dataset and 437 in the clenbuterol vs control dataset. 18 candidate spots generated from DIGE gels were subsequently sequenced and identified as complement component C9, haemoglobin, complement factor H, fibrinogen (α- and γ-chains and fragments), annexin A5, actin, apolipoproteins A-IV and E. The expression level of fibrinogen was found down-regulated in plasma of challenged with clenbuterol and nandrolone animals. The results were obtained on a limited number of samples and could not be considered as statistically valid. Why was the number of samples too low to be statistically significant? Therefore, further studies on ex vivo administration samples were required to confirm these findings.

In total 98 plasma samples (34 control, 31 from nandrolone treated, and 33 from clenbuterol treated animals, first administration study) were analysed by SELDI-MS. The protocol for sample preparation and analysis was established at HFL. A difference in the protein expression profile of control and treated cattle plasma was detected, and as such it was concluded that future work should be continued and involve the identification of protein biomarkers of anabolic administration to cattle by mass spectrometry, development of a method of analysis, and, finally, sample analysis

SID 5 (Rev. 3/06) Page 4 of 23

of normal samples alongside the experimental samples. In parallel with in vitro experimental work and the first administration study, the protocols for

production of antibodies by phage display technology were established and applied to a trial protein. The technology was transferred from Domantis to HFL. Despite the considerable amount of work that had been undertaken to try and identify biomarkers by the original methods proposed, the lack of identification of any protein biomarkers from the in vitro work made the development of ELISAs required to form the basis of any drug screening process within project time very unlikely.

Targeted multiplexing analysis of bovine plasma proteins, without a preliminary protein separation step, based on MRM transitions determined in 2D-LC-MS/MS experiments, was developed. The final method utilised 60 transitions to monitor 41 bovine plasma proteins in a 52 min analytical run.

The final protocol for nLC/MS/MS developed includes the following steps: plasma protein trypsin digestion, separation of tryptic peptides by liquid chromatography with subsequent identification by tandem mass spectrometry. Transfer of the protein digestion stage to a 96-well plate format resulted in a shorter sample preparation time.

This study shows that proteins present in plasma at a high levels (µg/mL) can be detected in a single analytical run without preliminary separation.

The method is amenable to automation and, in conjunction with stable isotope labelled standard peptides could be used as a quantification method.

Investigation of in vivo expression of biomarkers identified from normal population

80 control drug free plasma and 200 PA plasma samples were analysed using 5 µl of neat plasma for protein digestion. 80 bovine plasma samples were analysed alongside 16 QC samples in one run with a total run time of 3.5 days per plate.

The ability of the method to detect differentially expressed proteins was assessed by the analysis of drug-free plasma samples derived from study animals (young male castrates cattle) and lactating cows.

Several proteins were found up-regulated in plasma of lactating cows compared to the group of male (castrates) cattle: histidine-rich protein, complement component C9, α-1-acid glycoprotein, apolipopretein E, vitamin K dependent protein S, while prothrombin and C4b binding protein were found up-regulated in young animals.

There were no differences in the protein profile of the young control animals and those administered with nandrolone.

Future studies on animals from an age group used for meat production will validate the test suitability to monitor the biomarkers of growth promoter abuse.

The method could be applied to other matrices and tissues, e.g. saliva, to monitor the changes in protein expression and could provide a basis for the development of a large-scale screening method.

To the best of our knowledge, this study represents the first successful application of nLC-MS/MS for the analysis of the bovine proteome as a first step in biomarker identification of growth promoter abuse.

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).

INTRODUCTION

SID 5 (Rev. 3/06) Page 5 of 23

Effective enforcement of the current EU prohibition (directive 96/22/EC) of use of hormonal growth promoters in livestock requires the development of efficient and cost-effective screening and confirmation techniques to support surveillance programmes.

The detection and confirmation of the administration of natural steroid hormones to cattle and other food producing animals poses a significant analytical challenge and requires a detailed knowledge of both the endogenous steroid profiles in the biological fluids under study and the metabolism of the endogenous hormones of interest. It is relatively common, for example, for a parent drug to be rapidly converted to one or two major metabolites, in which case there is much to be said for targeting screening and confirmatory activities on the metabolites in question.

The relatively recent advent of illegal use of combined drug formulations, ‘anabolic cocktails’, exacerbates the difficulties of drug detection. In such circumstances, the concentrations of individual components are significantly lower than those resulting from established treatments and thus making the potential for laboratory detection by conventional screening methods (chromatographic/spectrometric techniques or immunochemical approaches) very remote.

Attention has thereby focused on developing novel testing methodology applying principles based on the functional activity of the administered substance assuming that the individual drugs with the same pharmacological action administered in a “cocktail” should elicit a similar response at a cellular level. For the purpose of monitoring these perturbations, the ’omics’ technology for the biomarkers detection and identification was applied. The ‘omics’ approach allows the study of the biological process as a whole, by analysing all transcripts, proteins, or metabolites in a cell, tissue, or organism. The partnership of HFL Ltd, TNO Laboratories (The Netherlands), and Domantis (UK) combined considerable experience of molecular biology, phage display technology, immunoassay development and mass spectrometry to develop a method of biomarkers identification to detect administration of illegal and veterinary drugs. The results returned from the participating laboratories have been used to find the appropriate strategy for biomarkers detection. Due to the lack of identification of any protein biomarkers from the in vitro work, it was not possible to start the production of antibodies relevant to any biomarkers. Thus, several other avenues of a biomarker discovery were explored and resulted in the development of a method based on simultaneous targeted monitoring of a set of proteins in cattle blood by mass spectrometry I am unclear about what proteins are being tested?. Studies with β-agonists and anabolic steroids were used to evaluate this approach. HFL has undertaken a development of the methodology and samples analysis.

SCIENTIFIC OBJECTIVES

The objectives of the project were set out as follows:

1. Use of in vitro systems to identify biomarkers using transcriptomics and proteomics2. Confirm that the biomarkers identified in the in vitro study are applicable in vivo when the animal is

challenged by administration of unauthorised β-agonists and anabolic steroids3. Development of assays to biomarkers based upon antibodies generated through phage display

technology4. Application of the assays to analysis of biological samples from animals challenged with growth

promoting agents and to “normal” samples.

The primary Milestones for the project were determined as follows:

1/01 Optimisation of in vitro systems for biomarker production1/02 Detection of putative biomarkers from transcriptomics and proteomics studies from challenges to cell lines with hormonal growth promoters2/01 Completion of administrations of beta agonists and anabolic steroids to animals2/01b Collection of administered animal tissue samples for transfer to CSL for incurred tissue project. Additional milestone agreed 21/04/20062/02 Development and optimisation of proteomic techniques for the analysis of blood samples from appropriate species2/03 Investigation of in vivo expression of biomarkers identified from in vitro studies2/04 Identification (including Mass Spectrometric ID) of protein biomarkers from the in vitro and in vivo studies3/01 Establishment of protocols for production of single domain antibodies3/02 Production of single domain antibodies for at least one biomarker identified by transcriptomics3/03 Production of single domain antibodies where necessary to key biomarkers identified from in vitro and in vivo studies3/04 Production of immunoassays for up to eight of the key biomarkers4/01 Completion of analysis of ‘normal’ and post administration samples for targeted biomarkers using the developed immunoassays.

SID 5 (Rev. 3/06) Page 6 of 23

In vitro studies performed to fulfil the Objective 1 did not reveal any protein biomarkers suitable for immunoassay development and the further work as originally projected (Milestones 2/03, 3/01, 3/02, 3/03, and 3/04) was considered unlikely to lead to meaningful results. A new proposal was suggested to use a different analytical technique with the outcome expected to meet the former Milestone 2/04. Re-scoped primary Milestones were approved on 08.01.2007 as follows:

1/01 Detection of putative biomarkers from proteomic studies (SELDI-MS) using samples from administrations of beta agonists and anabolic steroids to cattle1/02 Identification (including Mass Spectrometric ID) of protein biomarkers from the in vivo studies1/03 Completion of further administrations of beta agonists and anabolic steroids to animals1/03a Collection of administered animal tissue samples for transfer to CSL for incurred tissue project. Additional milestone agreed 01/04/20081/04 Investigation of in vivo expression of biomarkers identified from further in vivo studies1/05 Investigation of in vivo expression of biomarkers identified from normal population.

EXTENT TO WHICH OBJECTIVES WERE MET:

1. Completed. An overall analysis including all in vitro models tested did not reveal any protein biomarker of anabolic treatment. However, some important recommendations and conclusions were drawn from the analysis: what were they?. Re-scoped primary milestones (1/01 – 1/05) were agreed.

2-3. Completed in accordance with the re-scoped milestones4. Completed.

METHODS AND RESULTS

This final report summarises the work conducted and describes the application of different analytical approaches towards the identification of biomarkers of β-agonist and anabolic steroid administrations to cattle.

Objective 1. Use of in vitro systems to identify biomarkers using transcriptomics and proteomics

To select a suitable in vitro model, a literature search was performed to gather information on the biological action of beta-agonists in the U937 cell line inflammation model and on cell models responsive to anabolic steroid treatment.1, 2

The Application of β-Agonists to in vitro Systems

It was previously reported, that the beta-adrenoceptor agonist clenbuterol potently suppresses the lipopolysaccharide (LPS) induced release of the proinflammatory cytokines TNFα (tumor necrosis factor alpha) and IL-6 (interleukin 6) both in vitro and in vivo. To investigate this effect further, four beta-adrenoceptor agonists, clenbuterol, formoterol, salbutamol and zilpaterol, were incubated with LPS-treated macrophages in a concentration range from 10-4 to 10-9 M for 6 hours. Control substances were included in the experiment to increase the specificity of the beta-agonist specific marker genes (what control substances were used?). To verify the efficacy of all test substances in reducing the LPS-induced TNFα release, the TNFα concentrations in the culture media were determined using ELISA. The concentration of each individual beta-agonist that reduced the LPS-induced TNFα release by at least 75% was chosen for further experiments. Cell viability determination using a MTT ([3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide]) assay showed that none of the test substances affected cell viability.

To investigate whether beta-adrenoceptor agonist drugs induce a particular set of genes in treated cells a transcriptomics study was performed. The total RNA of treated and untreated cells was isolated, transcribed to cDNA and labelled with either Cy3 Dye (LPS-stimulated cells without drug treatment) or Cy5 Dye (LPS-stimulated cells treated with a test substance). The labelled cDNAs were hybridised to a human DNA microarray to analyse the expression of more than 16,000 genes. The dye ratios were calculated for each spot and analysed using multivariate data analysis (principal component analysis (PCA) and principal component discriminant analysis (PC-DA)) in order to find differentially regulated genes that were specific for beta-agonist exposure. Genes encoding for secreted proteins found within the 30 most differentially regulated genes were specifically selected as potential biomarker targets and included one down-regulated gene and eight up-regulated genes.1

Since a human array was used during the experiments, a bioinformatic analysis of the up-regulated genes was performed to determine the similarity of the human protein products compared to bovine proteins. Human protein products for these genes were identified and compared with the corresponding bovine proteins. A similarity search was performed against primary protein sequence databases, and secondary protein pattern databases were searched for any conserved motifs. Several protein characteristics such as isoelectric point (pI),

SID 5 (Rev. 3/06) Page 7 of 23

hydrophobicity index, molecular weight and composition of each protein were also evaluated. It was established that the genes encoded proteins belonging to the intercrine alpha/beta family, IL-6, transforming growth factor, amphiregulin and type I interferon families. A database search revealed seven bovine analogues for reported human proteins with homology ranging from 50% to 92%.Following the results obtained in the transcriptomics study, proteomics was used to study the beta-adrenoceptor agonists influence on the cells. U937 monocytes were differentiated into macrophages and treated with either LPS, LPS and zilpaterol or LPS with zilpaterol and the beta-2-adrenoceptor antagonist propranolol. The culture medium was collected and the proteins were precipitated, labeled with CyDye Fluor dyes and compared using 2-dimensional gel electrophoresis (DIGE) over a pH range of 4 to 7. The 2D gels were analysed, and spot ratios were calculated using pattern recognition tools (PCA, PC-DA) to identify protein spots that were differentially expressed. Differentially expressed proteins were identified using in-gel digestion followed by nano-LC-MS/MS (nano liquid chromatography tandem mass spectrometry). Of the proteins significantly up- and down-regulated in culture media only one was identified as a secreted protein, macrophage inflammatory protein-1 beta (MIP-1 beta), which is known to be involved in the inflammatory response. Other proteins identified were enzymes involved in the maintenance of the cell or proteins responsible for protein folding or development. The proteins identified by 2D gel did not correspond to the proteins revealed on the DNA microarray. Bioinformatic analysis of the proteins reported from the in vitro study and comparison with the corresponding homology to bovine proteins narrowed the possible candidate list to granulocyte chemotactic protein 2 (GCP-2) and macrophage inflammatory protein-1 alpha (MIP-1α). Human GCP-2, detected in the transcriptome (top hit reported2), is homologous to the bovine protein for 67% of its primary structure, whereas human MIP-1α (detected in proteome) possesses 72% homology to the bovine protein.

To confirm that MIP-1 beta was also a marker protein for exposure to other beta-agonists, U937 macrophages were exposed to LPS with zilpaterol, clenbuterol, formoterol and salbutamol separately. The effect of the four different beta-agonists on the production of MIP-1 beta was investigated using ELISA. Was the up-regulation confirmed?

The Application of Anabolic Steroids to in vitro Systems

To select a suitable cell system to investigate the effects of anabolic steroid administration a literature search for human cell lines that are affected by treatment with androgenic anabolic steroids and testosterone was performed. As a result, a number of publications were found describing the effect of testosterone on HepG2 cells leading to the release of C1 esterase inhibitor and sex hormone binding globulin in the culture media alongside elevated aromatase activity3. The experiments in these publications were first reproduced by determination of the marker proteins concentration in the culture media. The concentrations of anabolic steroids (AS) (nandrolone, boldenone and stanozolol) that could be used without loss of cell viability were established by the MTT assay. To investigate whether these steroids had an effect on C1 esterase inhibitor secretion, HepG2 cells were incubated with selected AS at a concentration of 10-7 mol/L, whilst incubation with testosterone at different concentrations (10-9-10-4 mol/L) served as a positive control. The 24 h supernates were collected and C1 esterase inhibitor activity was analysed by ELISA. It was found that the presence of testosterone did not alter the C1 esterase inhibitor release compared to the control (So the positive control did not work). In addition, nandrolone, boldenone, stanozolol did not appeared to affect C1 esterase inhibitor release2. An elevated level of sex hormone binding globulin was not detected in the experiments with HepG2 cells incubated with different concentrations of testosterone (10-9-10-5 mol/l) in the presence, or absence, of foetal calf serum.

Since the results did not match the published data, an additional microarray experiment was performed using HepG2 cells that were exposed to three different concentrations of the anabolic steroids: nandrolone, boldenone and stanozolol (10-8, 10-6 and 10-4 mol/L) at two different time points (6 hours and 24 hours). The total RNA was isolated from the exposed and control cells, transcribed to cDNA and labelled with Cy3 dye (control cells) or Cy5 dye (exposed cells). The cDNA was subsequently hybridised to a microarray chip as described above and the dye ratios were analysed. It was found that the expression of a small number of genes was up-regulated, of which only one gene, the human alpha satellite 3 junction DNA sequence, was up-regulated for all three of the anabolic steroids. The DNA sequence of this gene encodes a putative open reading frame of an unknown protein. The up- or down-regulation in expression of most of the genes was only observed at one concentration and at one time point, indicating that the gradient in the chosen concentration was too steep to show a gradual increase or decrease in expression. The expected up-regulation in expression of C1 esterase inhibitor, SHBG and aromatase was not observed and, therefore, it was concluded that HepG2 was not a suitable cell system to investigate the effects of androgen anabolic steroids on protein expression.

Since anabolic steroids affect muscle growth, skeletal muscle cell lines from rat (L6 and L8) and mouse (C2C12) were investigated for their response to the steroids named above at concentrations of 10 -11-10-5 mol/L. The effect of insulin-like growth factor-1 (IGF-1) release into the culture medium of proliferating and differentiated cells was analysed by ELISA. Levels of IGF-1 appeared to be too low to detect using this method. In addition, the effect on cell proliferation was determined by incorporation of [3H] thymidine. Nandrolone, boldenone and stanozolol did not have a positive effect on proliferation in the three skeletal muscle cell lines after 48 h and 72 h incubation.

SID 5 (Rev. 3/06) Page 8 of 23

Moreover, high concentrations of the AS inhibited proliferation, which was most likely caused by a toxic effect on the cell lines.

Since nandrolone, boldenone, stanozolol show structural similarity with testosterone, a human breast carcinoma cell line T47D was selected, containing an endogenously expressed androgen receptor and a reporter plasmid with the luciferase gene cloned downstream of the Androgen Response Element (ARE). The cell line was exposed to three AS to investigate whether they could activate the androgen receptor pathway 3. When the Androgen Receptor has bound its ligand the complex migrates to the nucleus where it interacts with the ARE and initiates transcription of the luciferase gene. The T47D transfected cells were incubated with AS for three different time periods (24 h, 31 h, and 48 h). Cells were lysed and luciferase activity was measured. All three AS initiated expression of the luciferase gene, indicating that they interacted with the androgen receptor and activated the pathway in these cells. Maximum expression was obtained after 24 h incubation at a concentration of 10 -5 -10-6

mol/L, and a transcriptomic microarray experiment was performed to determine differential expression of genes. The cells were exposed to previously selected concentrations of nandrolone, boldenone, and stanazolol (10 -5, 10-5

and 10-6 mol/L, respectively), and also to β-agonist clenbuterol (10-7 mol/L) and glucocorticoid dexamethasone (10-7 mol/L). Microarray data were subjected to multivariate analysis (PCA). Transcriptomic experiments did not reveal differentially regulated genes after exposure of T47D cells to anabolic steroids compared to the untreated cells.

In order to determine whether a direct proteomics approach would be more successful, culture medium of T47D cells exposed to nandrolone was analysed by both 1D and 2D gel electrophoresis. Both 1D as well as 2D gel analysis revealed the presence of a few high abundant proteins (mainly serum proteins in T47D culture) and interfering proteins that made the identification of differentially expressed proteins in the T47D culture medium a very difficult and time consuming task.

In a final attempt to find a suitable cell system to identify in vitro protein markers for AS exposure, the responsiveness of the androgen receptor expressing prostate-derived myofibroblastic cell line WPMY-1 was investigated. Based on cell optimisation results, a concentration of 0.1% FBS was selected for proliferation experiments with the anabolic steroids. The cells were treated with nandrolone and R1881 (a synthetic androgen receptor agonist) at a concentration range of 0 nM to 104 nM in the presence of 0.1% FBS, however, neither R1881 nor nandrolone was able to induce a significant increase in proliferation of WPMY-1 cells.

In addition, the effect of nandrolone and R1881 on nuclear translocation of AP-1 c-jun in WPMY-1 cells was investigated. The cells were exposed to the test substances at a concentration range of 0 nM to 104 nM in RPMI1640 or DMEM media supplemented with 0.1% BSA for 96 hours. The nuclear extracts were prepared and the nuclear translocation of AP-1 c-jun was measured using the TransAM DNA-binding ELISA (what time points were used for AP-1, c-Jun lysate collection?). It was found that the response of the WPMY-1 cells was insufficient to use this cell line as a model system.

Based on the above described findings it was decided to stop with the search for an in vitro cell system for the identification of secreted proteins as markers of anabolic steroid exposure. Instead, efforts were made to optimize analytical methods for the identification of protein markers in plasma samples of nandrolone treated cattle.

This work was previously discussed in greater detail in reports submitted to DEFRA during the project1,2,3 (Years 2005-2006).

Objective 2. Confirm that the biomarkers identified in the in vitro study are applicable in vivo when the animal is challenged by administration of unauthorised β-agonists and anabolic steroids

Two administration studies were carried out to collect blood samples from animals challenged with beta agonists and anabolic steroids (primary milestone 2/01 and re-scoped milestone 1/03).

The first administration study was set up on a small group of animals consisted of three 56-70 days old on Day 0 male (castrates) Holstein Friesian breed calves. Animal 290 was administered with a beta agonist, clenbuterol hydrochloride (Ventipulmin® injection solution and Ventipulmine® syrup, Boehringer Ingelheim), according to the following scheme: Day 0: intramuscular administration (am and pm) at a dose of 1.15 µg/kg body weight (BW), Days 1-21: oral administration at a dose of 20 µg/kg BW adjusted to the BW changes. Animal 283 was administered intramuscularly with the dose of 5 mg/kg BW of the anabolic steroid, nandrolone phenylpropionate (Nandrolin®, Intervet), on Days 7, 14, and 21. Animal 284 was used as a naïve control. During the study, all three animals showed signs of a nasal discharge and were administered with Alamycin (three doses on Days 37, 39, and 42) and a single dose of Flunixin on Day 37. Blood was collected by jugular venepuncture and processed to obtain serum and plasma samples. Blood samples were collected for 62 days in total including pre-dose samples, administration samples, and withdrawal period up to 31 days after the final dose.

SID 5 (Rev. 3/06) Page 9 of 23

The second administration study was set up using similar group of five males (castrates), 69-83 days old on Day 0, of the same breed, Holstein Friesian. Four animals (numbers 662, 667, 669, and 670) were administered with nandrolone phenylpropionate on Days 0, 7, and 14 at the dose of 5 mg/kg/bodyweight according to the scheme applied in the first study. One animal (665) was kept untreated and is referred further as a control animal. Blood samples processed as plasma (K3EDTA treated vacutainer®) were collected for 61 days in total.

All blood sampling was completed prior to animal feed or any administration of test materials at approximately the same time of the day on each sampling occasion. All samples were stored frozen in multiple aliquots and all experiments were performed on samples which had been through a single freeze/thaw cycle.

In addition to plasma, urine samples were collected in the second administration study to determine steroids occurrence.

In both administration studies, the experimental conditions were kept as similar as possible. Animals were individually housed in multipurpose animal accommodation, where all animals were allowed both natural and artificial light and provided with the same type of feed and water. Feed consumption and weight gained by the animals were monitored during both studies. All animals were euthanased on Study Day 42 and tissue samples were collected immediately post mortem as required (primary milestone 2/01b and re-scoped milestone 1/03a). The results obtained in the in vitro study on the LPS-induced inflammation model of U937 human cell line in the presence or absence of beta-agonists led to the discovery of several potential protein biomarkers. Bioinformatic analysis of the proteins narrowed the possible candidate list to GCP-2 and MIP-1α. ELISA kits (R&D Systems Europe Ltd) for detection of these recombinant human proteins were purchased for comparative analyses of the cell culture and bovine samples.

Taking into account that LPS treatment of U937 cells is responsible for the inflammation cytokine release, e.g. TNF-α, the cell culture supernates provided by TNO and 12 bovine plasma samples (blank and post administration samples) were analysed for the presence of inflammatory cytokines TNF-α and IL-8. Data obtained in ELISA confirmed the high concentration of TNF-α in cell culture supernates and high level of IL-8. However, there was no evidence of elevated level for TNF-α or IL-8 in bovine plasma samples (why would there be? I am unclear about what the stimulation was?) The same set of samples was analysed for the presence of GCP-2 and MIP-1α by the human ELISA kits. However, there was no visible evidence for the presence of GCP-2 or MIP-1α. The lack of determination could be due to either the high specificity of the antibodies raised against particular epitopes of human proteins that may be different for human and bovine analogues despite a relatively high similarity of primary structure, or the concentration of the proteins under consideration is too low in bovine samples to be detected, or there is no basal level of inflammation in cattle1. Could you test some recombinant bovine GCP-2 and MIP-1 with your ELISA?

These in vitro studies (objective 1) did not reveal any protein biomarkers which can be taken over for further immunoassay development (objective 3) and an in vivo proteomics study (primary milestones 2/02-2/04) was considered a priority.

The Development of Proteomic Techniques

In order to determine changes in protein expression after beta-agonists and AS administration to cattle, plasma samples from different periods of the study including pre-administration samples (Day -3), samples collected during the administrations (Days 3 and 14), and samples collected during the withdrawal period (Days 23 and 28) were subjected to proteomic analysis. Plasma is the best source of proteins in the body, however the proteomic analysis of plasma in search of low abundant proteins is obscured by the level of high abundant proteins (22 proteins constitute 99% of the protein content of plasma). Special pre-treatment protocols are required to remove interfering proteins. To find the best strategy for high abundant protein depletion from bovine plasma, especially bovine serum albumin (BSA) depletion, the following methods were investigated including the ProteoPrep Blue albumin depletion kit (Sigma), the Swell gel kit (Pierce), Albumin and IgG removal kit (Amersham), acetonitrile precipitation, the MARS column (Agilent), and the IgY-BSA spin column (Beckman Coulter). Analysis of depleted bovine plasma on 1D and 2D gels revealed that except for the IgY-BSA spin column none of the kits and protocols was able to remove BSA from bovine plasma completely. The lack of the BSA depletion could be due non optimised protocols designed for human samples. Subsequently, the best BSA depletion kit (IgY-BSA spin column) was used in combination with the Agilent MARS column to pre-treat bovine plasma prior to analysis on 2D gels. Optimal buffer composition and focussing times were established and the reproducibility of each step and the whole final protocol was determined. The final protocol consisted of BSA depletion using the IgY-BSA spin column (Beckman Coulter) followed by removal of high abundant plasma proteins (transferrin, IgG, IgA, antitrypsin, fibrinogen, and haptoglobin) by the human MARS HPLC column (Agilent). After desalting and concentration of the unbound fractions by trichloroacetic acid precipitation and subsequent reconstitution of proteins in lysis buffer, proteins were labelled with CyDye DIGE fluorophores and run on a 2D gel using a pH

SID 5 (Rev. 3/06) Page 10 of 23

range of 3-10 and a gradient of 9%-15%. These gels resulted in good quality images suitable for further analysis by Progenesis software (Progenesis workstation, PG240 v2006).

Following the pre-treatment method development, 15 bovine plasma (five time points for control and five time points for clenbuterol and nandrolone treated) were examined using DIGE technology. The protein content of pre-treated plasma was determined by a modified Bradford method and all samples were labelled with two different DIGE fluorophores, Cy3 Dye and Cy5 Dye. The internal standard, a mixture of five controls and five administered with individual drug samples, was prepared and labelled with Cy2 Dye. Three different samples (labelled with Cy3 Dye, Cy5 Dye, and internal control) were mixed together and run on one gel. Ten gels were run for each set of samples providing a total of 20 gels for further analysis.

To compare the proteomic profile of the control and treated plasma from experimental animals, in silico analysis of the DIGE images was performed. The gels were scanned with a laser scanner (Typhoon 9400) at three different settings corresponding to the different DIGE fluorophores giving three images per gel. Comparative analysis of gels was performed using software, which allowed the estimation of normalised intensity volume for each protein within the same gel and matched the spots present within the set of gels belonging to an individual subject. The acquired datasets containing the expression ratios of each spot in each sample was further analysed by multivariate data analysis. Two datasets to compare control vs nandrolone treated plasma and control vs clenbuterol treated plasma were subjected to principal component analysis (PCA). The Cy3 Dye and Cy5 Dye samples were used as separate samples in the analyses instead of taking the average value because of the small number of samples (only one animal per group). The number of protein spots (variables) was similar in both datasets. 420 discreet gel spots (variables) were monitored in the nandrolone vs control dataset, and 437 gel spots were monitored in the clenbuterol vs control dataset.

The PCA revealed a group separation between the gels obtained for both groups of comparison, however there was no single opinion as to whether the difference observed was caused by the administration or by biological variations between animals because of the large number of variables compare to limited number of samples in the experiment. A time trend was observed in the control group which did not result from the administration of drugs and could be caused by an aging effect as the young animals (8-10 wks old) were submitted to the long (over 7.5 wks) experiment.

Finally, principal component discriminant analysis for a dataset including measurements at one specific time point was performed. However, the cross validation of this model required more samples, thus the results of this type of analysis were considered preliminary and further analysis on an adequate data set was necessary to confirm the initial results (how did this consideration escape experimental planning?).

Based on pattern recognition results, a few spots were selected and their importance was verified univariately in the original data set. For each analysis, the spot numbers with the highest absolute loadings were identified. Protein expression levels were visualised as histograms representing the ratio of protein spot intensity of the sample to the matched spot intensity of the internal standard on the DIGE gel. The ratios are expressed as the mean of two expression ratios (Cy3 Dye/Cy2 Dye and Cy5 Dye/Cy2 Dye).

To identify the proteins of interest, the protein spots were excised from the 2D gels, S-alkylated and digested with trypsin, separated and analysed by nLC-MS/MS. Proteins were identified by searching MS/MS spectra against the SwissProt database using the Mascot search engine (Mascot 2.1 UK SP1 server AID 2130, Matrix Science). Nanobore chromatography was performed on an Ultimate nano-LC (nLC) system (LC Packings, Netherlands) with 300 µm ID x 0.5 mm PepMap C18 trap column (LC Packings, Netherlands). Tryptic peptides were separated in aqueous acetonitrile gradient with 0.1% formic acid in 45 min. The nano-LC column was coupled to a LTQ ion trap MS (Thermo Electron, San Jose, Ca, USA) via a nano-electrospray interface. Electrospray was generated by applying 1.0 kV to the electrospray pico tip (20 µm ID, 10 µm tip ID, distal coated, New Objective, Cambridge, MA, USA) via a Pt wire; ions were introduced into the mass spectrometer through a heated capillary kept at 200°C. The ion trap was operated in data-dependent mode, selecting top two ions for MS/MS scans at 35% collision energy units.

Five out of 11 proteins were identified in the experiment with plasma from the clenbuterol treated animal and control, while 13 out of 16 proteins were identified in the other experiment comparing nandrolone treated samples vs control plasma. Proteins were identified using the Mascot search engine against the SwissProt database with the following search parameters: peptide mass tolerance of 0.8 Da, MS/MS tolerance of 1.2 Da, with one allowed misscleavage, applying enzyme trypsin, mammals taxonomy, carbamidomethyl fixed modification on cystein, possible oxidation on methionine, and pyro-glu modification on N-terminus of glutamine and glutamic acid as variable modifications.

The proteins identified in the clenbuterol experiments were mostly upregulated in the control samples and were identified as complement component C9 and haemoglobin. Four spots were reported as the haemoglobin beta

SID 5 (Rev. 3/06) Page 11 of 23

subunit (identified as human with 84% homology to the known sequence of bovine haemoglobin beta chain). The haemoglobin beta subunit belongs to the globin family of proteins and is found in red blood cells.

For the nandrolone vs control plasma comparison, the proteins identified were actin, complement component 9 and complement factor H, and the components of the clotting cascade including fibrinogen α-chain and fragments of the α- and γ-chains of fibrinogen, annexin A5, and apolipoproteins A-IV and E. The only non-secreted protein identified was actin which is ubiquitously expressed in all eukaryotic cells. Its presence in the plasma of the nandrolone treated animals at a higher level could be associated with inflammation or necrosis processes. Complement component 9 is a component of membrane attack complex of complement pathway while factor H is an activator of alternative pathway. The expression levels of complement component C9 and complement factor H in the plasma of the nandrolone treated animal were below the expression level of both proteins in the control animal. The presence of fibrinogen, the component of the coagulation cascade, in the samples could reflect the sample preparation process, e.g. different pre-processing time or time when samples were kept at room temperature before long term storage that could lead to differences in protein patterns. Annexin A5 belonging to the clotting cascade was found to be down regulated in the nandrolone treated plasma. Complement factor H, as well as complement component C9 described above, are expressed in hepatocytes and secreted in plasma.

Plasma is one of the most difficult matrices for proteomic studies. Because blood proteins concentrations are evenly distributed over all orders of magnitude, a new set of proteins become abundant after the next round of depletion no matter how many fractionation steps will be used. Moreover, as it is demonstrated in the case of fibrinogen, the depletion step is not 100% effective and often it is not possible to deplete high abundant protein from plasma completely, e.g. because of the presence of various protein isoforms. In the case of fibrinogen the immunoaffinity column designed for human proteins was used for bovine plasma. This could explain the high concentration of fibrinogen detected at the last stage of separation process. However, to reduce plasma complexity and thus increase dynamic range prior the identification, the fractionation steps are necessary despite of the cost and time they required.

Other concerns raised during the analysis of the plasma proteome arose from uncertainties at each step of the proteome workflow, e.g. limited protein concentration range, accidental modifications in sample preparation, shifting of spots in gels that may result in missing values and false protein identification.

The protein markers found in this study are high abundant plasma protein and are known house-keeping proteins. These results were obtained on a limited number of samples and are not statistically valid. Therefore, to conclude whether these proteins could be used as markers of illegal use of beta-agonist or anabolic steroids in cattle further studies need to be conducted.

Objective 3. Development of assays to biomarkers based upon antibodies generated through phagedisplay technology

In parallel with in vitro experimental work and the first administration study, the protocols for production of antibodies by phage display technology were established and applied to a trial protein, BSA. The isolation of BSA specific antibodies involved repeated rounds of panning (selection of antibodies on the basis of their affinity against purified antigen adsorbed on the surface). The presence of the anti-BSA antibodies was confirmed by ELISA. The phage display technology for antibody production was transferred to HFL Ltd, however it has not been possible to start the production of relevant antibodies due to the lack of identification of the protein biomarkers from the in vitro part of the study.

Despite the considerable amount of work that had been undertaken to try and identify biomarkers by the original methods proposed, the lack of identification of any protein biomarkers from the in vitro work made the development of ELISAs required to form the basis of any drug screening process very unlikely.

Initially, the EDTA plasma samples collected from administered cattle were analysed for the presence of free drugs by GC-MS/MS (nandrolone presence) and LC-MS/MS (clenbuterol presence). The plasma concentration of nandrolone in samples collected 31 days after the final dose remained above the concentration of 2 ng/ml. The level of free clenbuterol after treatment for 21 days was in the range of 30-55 ng/ml for the duration of the administration period, which tailed off to an undetectable level of clenbuterol 14 days after the final dose. Free drugs (nandrolone or clenbuterol) were not detected in the plasma of the control animal.

Following the results obtained by using the techniques set out in the original proposal, it was decided prudent to re-evaluate the proteomics workflow to determine if it was appropriate for the future work of this project. Thus the several other avenues of a biomarker discovery pipeline were explored which leaded to re-evaluation of primary milestones 3/01- 4/01.

Detection of Putative Biomarkers using SELDI-MS

SID 5 (Rev. 3/06) Page 12 of 23

Surface Enhanced Laser Desorption Ionization Mass Spectrometry (SELDI-MS), has been used for protein pattern development and biomarker identification for several diseases including alcoholism4, breast cancer5, and colorectal cancer6. SELDI-MS is a powerful technique that uses a combination of chromatography and mass spectrometry to study complex protein mixtures. It consists of selective protein separation by retention on various chromatographic chip surfaces (hydrophobic, anionic, cationic, and metal binding)7 and their subsequent analysis by laser desorption/ionization mass spectrometer. In addition to qualitative information determined by the existence of peaks, the method provides the proteins peak heights, therefore containing quantitative information. The resulting MS profiles are compared using bioinformatic techniques to determine the molecular weight to charge ratio of proteins that are most significantly up- or down-regulated between different sample populations. Characterisation of the biomarkers was possible by purifying the proteins at the molecular weight of choice and sequencing the amino acid structure by liquid chromatography mass spectrometry (LC-MS). The resulting spectra would be used to identify the biomarker by searching through protein databases to find a matching bovine protein or a protein from another species that had a high degree of homology.

SELDI analysis allows the detection and evaluation of proteomic patterns in a wide range of protein masses that includes the low molecular weight (below 20 kDa) proteins. By contrast, 2D electrophoresis typically assays proteins within the mass range of 20 - 250 kDa and pI of 3 - 11.

The plasma samples were analysed by SELDI-MS to determine the changes in the spectral patterns of proteins in treated and control plasma. SELDI experiments were performed on ProteinChip® arrays (Ciphergen Biosystems, Inc.) where each array contained eight identical spots that were used to process samples in parallel. Briefly, plasma samples were incubated on chips derivatised with a protein-fractionation resin. Following the incubation step, the unbound compounds were washed away and the chip’s surface was coated with an energy-absorbing matrix. Mass spectra were acquired by using laser ionisation and time of flight (TOF) mass spectrometry. The output from the instrument represents a list of clustered peak intensities that purport to reflect the protein concentration in plasma. Statistical analysis was further applied to the data set to distinguish those peaks whose intensities significantly differed between investigated cases.

To assess the reproducibility of SELDI-MS for analysis of proteins in plasma, internal control samples (IC) were prepared from 59 individual dairy cattle samples collected during the autumn 2006. All samples were treated and stored the same way as the study samples. In total 98 plasma samples were prepared from the first administration study to submit to SELDI-MS analysis including 34 samples derived from the control drug-free animal (284), 31 samples from the nandrolone treated animal (283) and 33 samples from the clenbuterol treated animal (290).

The total protein concentration in plasma from each individual sample was determined in a colorimetric assay (Dojindo Molecular Technologies Inc., NBS Biologicals, UK) and adjusted to 40 mg/mL prior to globular protein depletion by the addition of acetonitrile to plasma (6). The depleted samples were dried under reduced pressure (GeneVac, UK) and stored below -20°C.

In a screening experiment, each sample was applied to three different array surfaces (Ciphergen Biosystems Inc.) to find the optimal array for protein binding. The various chip/array surface chemistries used in the experiments are listed below. The CM10 ProteinChip arrays incorporate a carboxylate chemistry and act as a weak cation exchanger. The Q10 arrays incorporate a quaternary amine chemistry and serve as a strong anion exchanger. The H50 arrays bind proteins through reversed phase or hydrophobic interaction chromatography.

For ProteinChip array binding, the spots on the H50 arrays were preloaded twice with 5 µl of 50% AcN, dried for 60 min at RT, and equilibrated with 5 µl of binding buffer (10% AcN, containing 0.1% TFA) prior to sample loading. The CM10 and Q10 arrays were equilibrated twice with 5 µl binding buffer (Ammonium acetate buffer (pH 4.5), 50 mM, containing 0.01% Triton X-100 for the CM10 arrays and Tris buffer (pH 7.5), 50 mM, containing 0.01% Triton X-100 for the Q10 arrays). 10 µL of ultra pure water was used to reconstitute each sample of depleted cattle plasma followed by the addition of 10 µl of 8M urea-2% CHAPS. For the H50 array ultra pure water was used to reconstitute the samples. After 30 min incubation on ice, the samples were diluted with 100 µl of the specific binding buffer. 2 µL of sample in the corresponding binding buffer was added per array spot in accordance with the chip map. The arrays were incubated in a humidity chamber for 60 min (CM10 and Q10 arrays) and for 30 min (H50 arrays). All steps were performed at RT. After incubation, the CM10 and Q10 arrays were washed three times with binding buffer (two washing steps for H50 array) followed by two water rinses. A sinapinic acid solution (SPA) as an energy absorbing matrix was prepared according to the recommendations of the manufacturer (saturated solution in 50% acetonitrile containing 0.5% trifluoroacetic acid). After the arrays were dried, 0.5 µl SPA solution was applied. This step was repeated once and the ProteinChip Arrays were stored at nominally 4°C protected from light and moisture until further use.

The CM10, Q10 and H50 arrays were read on the ProteinChip System Series 4000 Reader (Ciphergen Biosystems Inc., USA) equipped with pulsed nitrogen laser (337 nm, 4 ns pulse width). The system was internally calibrated with the measured mass within 0.01% of the true mass and 95% confidence (0.005%CV) with a focus

SID 5 (Rev. 3/06) Page 13 of 23

mass of 12,000 Da in the range of 3,500 Da to 40,000 Da. External calibration of the instrument was performed using All-in-one protein standards (Bio-Rad Laboratoires) on the normal phase chip array, NP-20 array.

Each study sample was analysed in single. The inter-assay reproducibility of SELDI-MS was determined by analysis of IC samples as an average coefficient of variance (CV).

For the detection of changes in the protein profile, plasma samples of the treated animals and drug-free animal were compared. Peaks were auto-detected by Ciphergen Biosystems Data Manager with a minimal peak threshold of 20% of all spectra and a signal to noise ratio of 5.0 valley depths for the first pass. The peak clusters were exported to Biomarker Patterns (BPS) (Ciphergen Biosystems Inc., USA) and further analysed by Statistica v. 7.0 (StatSoft Inc, USA) software package.

A scatter graph was used to analyse the association between two variables (e.g. study day and peak intensity) within the same group of samples. Regression was used to find out whether the changes in one variable caused the changes in the other. A comparative statistical t-test was used to compare two sets of normally-distributed data. A two-tailed test for unpaired data was used to analyse treated vs control data. Three or more data sets of data were analysed using ANOVA. To help with the visualization of the results, Principal Component Analysis (PCA) was applied to the data set.

For analysis of depleted plasma on the CM10 array, the following settings were chosen: laser energy 4000 nJ, 7 shots/pixel, mass range from 3500 Da to 40 000 Da. The control group consisted of 25 samples. The pre-administration samples included 8 samples in total derived from both the pre-treated animals, which were used to determine the different patterns between individuals before any treatment. The post-administration samples included 31 samples obtained in respond to anabolic steroid treatment and 30 samples obtained after beta-agonist administration.

Analysis on the CM10 array resulted in a list of 58 m/z values of clustered peaks, with 20 significant m/z values (p<0.05). The t-test, including all 58 clusters, revealed 8 peak clusters significantly differ between two animals prior to the drug treatment. After drug administration, the difference in 14 m/z values was significant for the nandrolone treated animal vs control, 9 of these peaks appeared in the spectrum after administration. Three peaks were up-regulated (1.4- to 2.0-fold) and six peaks were down regulated (1.3- to 4.6-fold) in nandrolone treated animal. Four of these reported peaks represent a significant m/z value (p<0.05) and show the difference in the spectral patterns between nandrolone treated and control groups. The following m/z values in SELDI-MS spectra were found useful to distinguish between case (nandrolone treatment at a growth promoting dose) and control: up-regulated peaks at 9673 (2-fold) m/z and at 9956 (1.8-fold) m/z, down regulated peaks of 10493 (2-fold) m/z and of 16426 (4.6-fold) m/z.

Comparison of the clenbuterol treated animal vs control animal yielded a list of 25 m/z values that were different before any treatment. 16 new different peaks were found in the SELDI spectra after clenbuterol administration. Six m/z values were significant (p<0.05) and showed the difference (treated vs control): up-regulated at m/z 6853 (1.3-fold) and at m/z 21220 (1.2-fold), down regulated at m/z 8783 (1.4-fold), at m/z 14005 (1.7-fold), at m/z 27570 (1.3-fold) and at m/z 37780 (2.3-fold).

The SELDI-MS spectra of depleted plasma samples on Q10 array were recorded at the following conditions: laser energy 3000 nJ, 7 shots/pixel, mass range of 2500 Da to 40000 Da. The control group consisted of 34 samples. The pre-administration samples included 8 samples in total derived from both the pre-treated animals. The post-administration samples included 26 samples obtained in respond to nandrolone treatment and 29 samples obtained after clenbuterol administration.

For analysis of depleted plasma on H50 array, the following settings were chosen: laser energy 4000 nJ, 6 shots/pixel, mass range from 3500 Da to 40 000 Da. The control group consisted of 34 samples. The pre-administration samples included the same samples as were used for CM10 and Q10 arrays. The post-administration samples included 27 samples obtained in respond to anabolic steroid treatment and 31 samples obtained after beta-agonist administration.

The data on the spectral patterns obtained in the SELDI-MS experiments by comparison the results obtained from treated vs control animals are presented in the table below.

Table 1 Differences in the spectral patterns of depleted plasma proteins determined by SELDI-MS in treated animals (283 and 290) vs control animal (284)

Control Animal

CM10 array (number of peaks)283 (pre-), number of peaks

290 (pre-), number of peaks

283 (nandrolone),number of peaks

290 (clenbuterol),number of peaks

284 8 25 14 16Q10 array (number of peaks)

SID 5 (Rev. 3/06) Page 14 of 23

284 25 19 9 21H50 array (number of peaks)

284 8 27 19 27

Using this approach a difference in protein expression profiles in cattle plasma was detected and as such it was concluded that further work on the project should be continued in line with the re-scoped primary milestones 1/02-1/05 suggested at the project review meeting on 4th June 2007. This included the identification of protein biomarkers of anabolic administration to cattle by mass spectrometry, development of an analytical method and, finally, sample analysis of normal (untreated) samples alongside the experimental samples.

The identification of proteins is the major challenge in proteomics. Several novel technologies have been developed which consist of the separation of proteins, digestion with specific proteases such as trypsin and identification of the resulting tryptic peptides by tandem mass spectrometry. Tandem mass spectrometry incorporates two stages of mass selection: a first stage (MS1) selecting the mass of the intact analyte, or parent ion, and a second stage (MS2) selecting a specific fragment of the parent ion. Following these selection steps, a multiple reaction monitoring (MRM) assay to detect several parent ions can be performed in one experiment. This approach can provide information required for an analyte structure elucidation, and in combination with appropriate stable-isotope labelled internal standard, it will provide an absolute quantification of analyte concentration.

Recently the MRM assay has been applied to the measurements of specific peptides in the tryptic digest of human plasma (Anderson) and for horse plasma protein mapping8. MRM assays appear to provide an efficient specific assay platform for protein biomarker validation that can be used for lower abundance proteins following an additional step to enrich target proteins.

Following the strategy used for the identification of horse plasma proteins, a targeted proteomics method using a 2D-LC-MS/MS analysis was developed for bovine plasma proteins combining an initial discovery phase using an information-dependent acquisition (IDA mode) with a subsequent determination (MRM mode). Briefly, proteins from pooled bovine plasma of untreated control animals and treated with nandrolone animals were subjected to tryptic digest and separated by off-line LC on a strong cation exchange (SCX) column. The fractions collected were analysed further by reverse phase LC-MS/MS on an Ultimate 3000 LC system (Dionex, San Francisco, CA, USA) coupled to a 4000 QTrap hybrid triple quadrupole linear ion trap mass spectrometer (Applied Biosystems/MDS Sciex, Concord, ON, Canada) operating in full-scan mode. The data obtained were used to prepare a list of peptides and corresponding proteins in both types of sample by searching MS/MS spectra against the SwissProt database using Mascot search engine (Mascot 2.1 UK SP1 server AID 2130, Matrix Science) based on the charge state and most abundant y-ion m/z value under the electrospray ionization conditions with collisional peptide fragmentation. Peptide identification from IDA experiments allowed the generation of an MRM method that could be implemented to analyze the same sample in a subsequent run, triggering MS/MS scans at any MRM signal.

The MRM method was constructed using the following parameters: tryptic peptide (precursor ion) mass to charge ratio (m/z) and charge state (z), characteristic y-type fragment ion m/z ratio, collision energy required for fragmentation to obtain fragment ions, and chromatographic retention time (RT) of peptides. The m/z ratios were selected in accordance with the signal intensity of MS/MS taking into account that each fragment ion should contain at least four amino acids with m/z ratio higher than the m/z ratio of the precursor ion (where possible). Collision energy settings were optimized in a separate experiment where the intensity of y-ion fragments was greater at lower collision energy. The final MRM method was constructed by merging the data for all peptides.

IDA experiments were performed on pooled bovine plasma obtained during the first administration study (nandrolone phenylpropionate administration, one animal, six time points) and from control animals (two animals, 7 time points). Plasma proteins were digested with trypsin (Trypsin Gold, MS Grade, Promega) that specifically cleaves peptide bonds at the carboxyl side of arginine and lysine residues. Plasma proteins were reduced and alkylated prior to trypsin digestion to prevent protein aggregation and to make the enzyme digestion more effective. To reduce the disufide bonds of plasma proteins (cysteine residues), the diluted bovine plasma (1:10 in Ammonium Bicarbonate Buffer (AMBIC), 50 mM) was incubated with equal volume of 100 mM solution of dithiothreitol in AMBIC) for 1 h at 60°C. To prevent disulfide-bond reformation, the sulphydryl groups generated were alkylated with an excess of iodoacetamide at room temperature in the dark for 30 min with subsequent quenching of the reactions by exposure to light. The digestion step was conducted in solution by adding 250 µg/mL trypsin in 50 mM acetic acid (1.8 µg per 5 µL of neat plasma) to the mixture from previous steps. The enzymatic reaction was conducted overnight at 37°C. Finally, the reaction mixture was added to 1% formic acid (equal volume compared to neat plasma used in the experiment), mixed and centrifuged at 16100 rcf (5415R Centrifuge, Eppendorf) to remove any sediment.

Separation of the peptide digest was achieved on a 5 µm, 300 Å, 4.6 mm ID x 100 mm PolySULFOETHYL A™ SCX column (The Next Group Inc, MA, USA) on an Agilent HP 1100 HPLC system coupled to a DAD detector

SID 5 (Rev. 3/06) Page 15 of 23

and equipped with Chemstation A.08.04 software. The SCX column was washed overnight with 40 mM EDTA-sodium salt solution to minimize the concentration of Fe3+ ions in the system and conditioned with three gradient cycles of 1M ammonium acetate solution containing 10% acetonitrile (Solution B) and 25 mM ammonium acetate solution containing 10% acetonitrile (solution A) prior to analysis. Gradient elution at a constant flow rate of 1.0 mL/min was performed as follows: 100% solvent A for 10 min as starting conditions, linear increase of solvent B to 30% for 20 min, with subsequent increase of solvent B to 100% over the next 20 min, hold for 10 min, and step decrease of solvent B to 0% for 10 min. The samples were loaded manually through the injection loop (Rheodyne, IDEX Corp, IL, USA) as a mixture with solvent A (12:1, 2 mL) and the total run time was 70 min. 40 fractions containing peptides from protein digest of diluted bovine plasma were collected and analysed by LC-MS/MS in the IDA experiment.

Separation of tryptic peptides was achieved on a reverse-phase 3 µm, 75 µm x 150 mm PepMap 100 (C18) nanocolumn (Dionex) with a peptide cap trap PepMap 100 (C18) 5 µm, 300 µm x 5 mm (Dionex) following a method previously developed at HFL9. The HPLC system was an Ultimate 3000 LC system coupled to a 4000 QTrap hybrid triple quadrupole linear ion trap mass spectrometer, controlled by Chromeleon v.6.70 SP5 Build 1914 (Dionex Corp, USA) software. The column temperature was maintained at 30°C. Positive ion electrospray was performed using a NanoSpray II1 XYZ stage and a Microionspray II module with pneumatic assistance. An uncoated 20 µm id Silica Tip (New Objective, Cambridge, MA, USA) was used with a voltage of 2.2 kV applied. Gradient elution at a constant flow rate of 300 nL/min was performed as follows: 100% solvent A (0.1% formic acid in 2% acetonitrile) hold for 3 min as starting conditions, linear increase of solvent B (0.1% formic acid in 90% acetonitrile) from 0% to 5% for 2 min, further increase of solvent B from 5% to 60% over 55 min, and further to 90% solvent B in 10 min, hold for 12.5 min, step decrease of solvent B to 0% for 10 sec, and equilibration in 100% of solvent A for the following 13 min. The injection volume was 5 µl and the total run time was 95 min. All peaks were integrated using Analyst v.1.4.1 software (Applied Biosystems, Canada).

Proteins from the IDA experiment were identified using the Mascot search engine against the SwissProt database (release 26.07.07, 276256 sequences) with the following search parameters: peptide mass tolerance of 1.6 Da, fragment mass tolerance of 0.8 Da, without any restrictions on missed cleavage sites, applying enzyme trypsin, mammals taxonomy, carbamidomethyl fixed modification on cystein, possible oxidation on methionine, and deamidation on glutamine and asparagine. Following these search criteria, 193 proteins were reported. Using the same search parameters in the Mascot search against the NCBI non-redundant database (release 20.07.07, 5303346 sequences), 219 proteins were reported. The protein list generated by Mascot was evaluated for the biological significance of the hits reported, the number of missed cleavage sites of selected peptides, Mascot peptide score, length of the selected peptides, confidence value, and an intensity of b- and y-ions signals of the fragment ions. The second search round was performed with more stringent parameters against the bovine proteins, no missed cleavages and a peptide score above 35 revealed a list of 55 candidate proteins from which 348 MRM transitions were selected for further LC-MS/MS method development.

To identify the proteins of interest, pooled bovine plasma samples were digested with trypsin as described above, separated and analysed by nLC-MS/MS. Nanobore chromatography was performed using the same reverse phase column and instrumentation as in the IDA experiment described above. Tryptic peptides were separated in aqueous acetonitrile gradient with 0.1% formic acid in 52 min. Following the identification procedure, 35 proteins out of 55 targeted in the plasma protein fractions were detected. This could be due to insufficient chromatographically separation of peptides and due to an ion suppression caused by high abundant ions. In order to confirm the identification of proteins that present at lower concentration, the same experiment was repeated with acetonitrile depleted plasma prior to digestion with trypsin10. The presence of an additional 7 proteins was confirmed following this procedure. In total 60 MRM transitions from 41 bovine plasma proteins were selected for further optimisation of the assay.

In order to observe progressive fragmentation of all target peptides in the MS/MS scan, a dependent scan was carried out at five different collision energies (30, 35, 40, 45, and 50 eV). Source dependent parameters were as follows: ion spray voltage 2.7 kV, source temperature 175°C, collision gas high, curtain gas, 20V. This experiment was performed by operating the system in IDA mode where the survey scan was an MRM containing 60 transitions from 41 proteins. The dwell time for each peptide was optimised and was set up in the range of 25-50 milliseconds at selected collision energy.

MRM transitions for 41 detected bovine proteins are documented in Table 1 (Appendix) where protein name, Swiss-Prot ID, tryptic peptide sequence, mass/charge values, collision energy (CE), retention time (RT), molecular weight (MW), and biological function are shown.

The majority of these proteins are found in human11 and horse8 plasma, however, some additional proteins were included (actin, collectin-43, complement component C7, conglutinin, lactotransferrin, pigment epithelium-derived factor, and vitamin K-dependent protein S). The proteins identified in bovine plasma can be classified by their known biological activity as follows: transport proteins including general, lipid and metal-ion transport, oxygen transport, retinol and vitamin D transport, blood coagulation proteins, proteins regulating the immune response,

SID 5 (Rev. 3/06) Page 16 of 23

and proteins involved in various type of cell motility. However, some proteins, e.g. apolipoprotein A-II, kininogen, display multiple biological activities. The molecular weight of proteins detected by nLC-MS/MS is in the range of 10.7 kDa (Apolipoprotein C-II) to 248.98 kDa (Coagulation Factor V) which covers the detection range of both DIGE and SELDI methods. The protein abundance within the samples varies from serum albumine (g/L) down to prothrombin (normal human plasma concentration of approximately 1 µg/ml12).

The results obtained from the MRM analysis were compared to the proteome plasma pattern from the control and treated samples generated by using DIGE analysis. The total number of protein spots on the DIGE map was 420 in the nandrolone vs control dataset and 437 in the clenbuterol vs control dataset. 18 candidate spots generated from DIGE gels were also subsequently sequenced and identified as complement component C9, haemoglobin, complement factor H, fibrinogen (α- and γ-chains and fragments), annexin A5, actin, apoliporpoteins A-IV and E. Good correlation between proteins identified by DIGE analysis and by nLC-MS/MS analysis was observed: only one protein, annexin A5, was not identified by MRM targeted approach. Moreover, the latter approach allows identifying of 41 bovine plasma proteins vs 9 proteins in DIGE in a single run.

Thus, targeted multiplexing analysis of bovine plasma proteins based on MRM transitions determined in two-dimensional LC-MS/MS experiments provides a powerful tool to investigate the expression levels of plasma proteins which might serve as an indicator of the administration of illegal or veterinary drugs in livestock.

Objective 4. Application of the assays to analysis of biological samples from animals challenged with growth promoting agents and to ‘normal’ samples

To further investigate the expression levels of plasma proteins after the administration of growth promoting agents to cattle, plasma samples from control animals and after AS administration were analysed by nLC-MS/MS. 80 individual control samples from which 42 samples were from untreated animals (administration study: samples from two control animals and pre-administration samples from six animals) and 38 samples from lactating cows (Cedar Farm, Reading, UK) were analysed. In total, 200 bovine plasma including samples from five nandrolone treated animals were analysed by the method developed. The expression levels of plasma proteins were compared in groups by using basic statistics in combination with multivariate statistics (PCA).

To increase sample throughput, protein digestion of 5 µl neat bovine plasma in 96-well format plates (0.2 ml, PP, Thermo Fast PCR plate, Abgene) was carried out, which allowed a shorter sample preparation time. Following digestion, the samples were analysed by nLC-MS/MS using 52 min run time. An individual batch (80 samples) required 3.5 days for analysis.

Quality control (QC) samples were used during each assay run (16 QCs per plate at fixed positions to monitor batch to batch variations). QC samples were prepared from pooled plasma of control and treated animals and are a good representation of study samples. QC precision was calculated considering the peak area of each peptide observed and showed significant variations from batch to batch including row to row variations on the 96-well plate (1.15%CV to 142%CV for different peptides). Therefore, the sample peak areas were normalised to the values obtained for the QCs in each row (10.4 h instrument run) and these normalised values were used for further statistical analysis. The high variations observed for the QC samples on each plate could be due to the nature of the detection method. Mass spectrometry measures a ratio of mass to charge; therefore the degree of ionisation depends on the conditions during the analysis and is determinative for signal intensity. Nevertheless, the comparison of relative abundances of identical peptides in similar samples is possible despite limitations in analytes quantification without an internal standard.

Of note, we focused on data from fibrinogen because it was found in the DIGE experiment to be down-regulated in treated animals compared to control drug-free animal. Fibrinogen molecules are constructed out of two sets of disulfide-bridged α-, β-, and γ-chains. It shows many biological functions and yields a variety of degradation products due to enzymatic degradation (at least three distinctive spots were identified in our DIGE experiments). Using nLC-MS/MS detection of fibrinogen α-, β-, and γ-chains in control drug-free plasma, no differences between relative peak area of the corresponding peptides in our administration samples and samples collected from lactating cows were detected.

Another remarkable aspect of the protein profile of lactating cows’ plasma was that certain proteins were up-regulated compared to our study group of male castrates cattle, e.g. histidine-rich protein, complement component C9, α-1-acid glycoprotein, apolipopretein E, vitamin K dependent protein S while prothrombin and C4b binding protein were found up-regulated in young animals. Therefore, based on the analysis of protein profile, the control group animals could be separated into two groups corresponding to the age and gender of the animals. The difference in protein profile between these two groups could be also explained by the differentiation of immune status of the animals. It shows that the nLC-MS/MS assay developed allows identifying the variation in protein profile thus determines the assay principle was working.

SID 5 (Rev. 3/06) Page 17 of 23

To compare the expression levels of plasma proteins after growth promoting agents administration to cattle, plasma samples obtained from control animals and after AS administration (200 samples) were analysed by nLC-MS/MS. The acquired data set contained the relative peak area of each monitored peptide in each sample. To identify differentially expressed proteins in samples derived from treated and control animals, a t-test was applied to the data set. The difference in protein profile was observed at different time points during the administration study between treated and control animals. However, due to the large data volume, it was difficult to see patterns and relationships in the data set by performing t-test. The data were further analysed by multivariate data analysis – PCA to determine a correlation between variables if any. This analysis did not reveal any group separation between both groups of comparison (control vs treated). However, a significant correlations in the range of 30% to 75 % (p<0.05) was observed between peptides from the same protein and also for proteins possessing the similar biological functions thus confirmed the assay principle. A few peptides from proteins presenting at low concentration level did not show correlation in this statistical test.

As a first approach to multiplexing bovine plasma protein analysis by nLC-MS/MS the study provided a general overview of plasma proteins levels in young (10-16 weeks old) animals. The lack of group separation between treated and control samples based on 41 proteins analysis could be explained in terms of high synthetic rates in young animals where there is a high conversion of feed into animal tissues. In this study, the high abundant plasma proteins were monitored and it was found that these proteins were expressed at a constant level that could be linked to the unchanged general health status of a drug treated animal. There were individual variations of protein expressions at different time points of the study but it was not possible to identify a general trend in changes of investigated protein profile in drug administered cattle. There are several publications resulting from the BioCop project (http://www.biocop.org/) where similar proteins to the ones detected by the nLC-MS/MS method were reported as biomarker candidates (alpha-2-antiplasmin, serotransferrin, endopin-1, and alpha-1-antitrypsin precursors, and also fetuin A and L-lactate dehydrogenase B) which could be related to the treatment of animals with growth promoters. However, because the results were obtained from pooled samples corresponding to only two time points within an administration, there is insufficient statistical evidence to conclude anything further at this stage13.

There were significant changes in the proteomics field since the time the project was initiated. To achieve the project targets several technologies for proteome analysis were on trial resulted in an analytical method for simultaneous analysis of plasma proteins without preliminary separation of proteins. There is a possibility of future technology development that will allow analysis of low abundant proteins to investigate the changes of proteins at low concentration levels that could lead to discovery of biomarkers of growth promoter abuse.

CONCLUSION

Eight cell models (human LPS-treated macrophages, U937 cells, HepG2 cells, T47D breast carcinoma cell line, prostate-derived myofibroblastic cell line WPMY-1, rat skeletal muscle cell lines L6 and L8, and mouse C2C12) were investigated by transcriptomics and proteomics technology for suitability for production of biomarkers in response to hormonal growth promoter treatment.

LPS-treated macrophages and U937 cells were challenged with four β-adrenoceptor agonists (clenbuterol, formoterol, salbutamol, and zilpaterol). A microarray experiment on LPS-treated macrophages revealed the differential expression of genes encoding proteins belonging to the intercrine α/β- family, IL-6, transforming growth factor, amphiregulin and type I interferon families. In the proteomics study on LPS-treated U937 challenged with zilpaterol in combination with propranolol (β-adrenoceptor antagonist) only one secreted protein (macrophage inflammation protein) was identified by 2D gel separation and nLC-MS/MS analysis. Proteomics data did not correspond to the proteins revealed in the transcriptomic experiment.

The effects of anabolic steroid (AS) treatment were investigated on human (HepG2, T47D, WPMY-1), rat (C6 and C8) and mouse (C2C12) cells. The proteins reported in the literature as biomarkers of AS treatment (C1 esterase inhibitor and sex hormone binding protein, SHBG, and aromatase) were monitored in culture media of HepG2 cells by ELISA. Testosterone, nandrolone, boldenone and stanozolol did not alter the C1 esterase inhibitor release. In addition, SHBG was not detected in the experiment. The microarray experiment did not reveal the expected up-regulation of genes encoding C1 esterase inhibitor, SHBG or aromatase.

Rat, L6 and L8, and mouse C2C12 skeletal muscle cell lines were investigated for their response to the steroids named above. The effect of insulin-like growth factor 1 (IGF-1) release into the culture medium was investigated by ELISA. IGF-1 level appeared to be too low to detect by this method. Moreover, high concentrations of the anabolic steroids inhibited proliferation of cells.

Human breast carcinoma T47D cell line containing an endogenously expressed androgen receptor and a reporter plasmid with the luciferase gene cloned downstream of the androgen response element, was exposed to three anabolic steroids to investigate whether they could activate the androgen receptor pathway. Microarray experiments did not reveal differentially regulated genes after exposure of T47D cells to AS compared to the

SID 5 (Rev. 3/06) Page 18 of 23

untreated cells. The direct proteomic approach revealed the presence of a few high abundant, mainly serum proteins, in culture alongside some interfering unknown proteins.

Finally, an androgen receptor expressing prostate-derived myofibroblastic cell line WPMY-1 was investigated in the presence of a synthetic androgen receptor agonist R1881 and nandrolone. Neither R1881 nor nandrolone was able to induce a significant increase in proliferation of WMPY-1 cells.

Based on the above described findings, it was decided to stop the search for an in vitro cell system for the identification of secreted proteins as markers of anabolic steroid exposure. Instead, efforts were made to optimise analytical methods for the identification of protein markers in plasma samples of drug administered cattle.

Plasma samples were collected following the administration of a beta agonist, clenbuterol hydrochloride (Ventipulmin® injection solution and Ventipulmine® syrup, two intramuscular injections, 1.15 µg/kg body weight (BW), oral administration for a further 21 days, 20 µg/kg BW), and an AS nandrolone phenylpropionate (Nandrolin®, i.m., three injections weekly, 5 mg/kg BW) to five male (castrates) Holstein Friesian calves (one was administered with clenbuterol and four animals were treated with nandrolone). Blood samples were collected for 62 days in total including pre-dose samples, administration samples, and during the withdrawal period samples up to 31 days after the final dose.

Control drug-free samples were obtained from two untreated study animals and from a group of lactating cows (dairy farm, 59 animals) and used to establish the expression of plasma proteins compared to drug-treated study animals.

A number of the cell culture supernates (LPS-treated U937 cells, β-adrenoceptor agonists) and 12 bovine plasma samples (control and post administration) were analysed to evaluate the levels of inflammatory cytokines (TNF-α, IL-8) by ELISA. The high concentration of inflammatory cytokines was confirmed for cell supernates, however there was no evidence for their presence in plasma. The use of human ELISA kits for the detection of proteins reported in transcriptomic and proteomic studies (β-adrenoceptor agonists) did not reveal any presence of the proteins under consideration. Thus, these in vitro studies did not reveal any protein biomarkers which could be used for further immunoassay development and the in vivo study was considered a priority.

In parallel with in vitro experimental work and the first administration study, protocols for the production of antibodies by phage display technology were established and applied to a trial protein. Despite the considerable amount of work that had been undertaken to identify biomarkers by the original methods proposed, the lack of identification of any protein biomarkers from the in vitro work made the development of ELISAs required to form the basis of any drug screening process impossible.

Five control plasma samples (administration study) and five plasma from each clenbuterol and nandrolone treated animals were analysed by 2D-gel (DIGE) in combination with nLC-MS/MS. The total number of protein spots on the DIGE map was 420 in the nandrolone vs control dataset and 437 in the clenbuterol vs control dataset. 18 candidate spots generated from DIGE gels were subsequently sequenced and identified as complement component C9, haemoglobin, complement factor H, fibrinogen (α- and γ-chains and fragments), annexin A5, actin, apoliporpoteins A-IV and E. The expression level of fibrinogen was found to be down-regulated in plasma from the animals challenged with clenbuterol and nandrolone. Evaluation of the results obtained from the 2D gel electrophoresis methods led to the conclusion that the technique was not sensitive enough for low molecular weight, low abundance proteins. Although several changes in protein markers were observed these were considered to be intrinsic components of blood, possibly generated through the sample collection process. In addition, the differences appeared to be animal specific, and were observed between the control and treated animals using pre-administration samples as well as post-administration samples. The results were obtained on a limited number of samples and could not be considered as statistically valid. Therefore, further studies on ex vivo administration samples were required to confirm these findings.

Other concerns raised during the DIGE analysis of the plasma proteome arose from uncertainties at each step of the proteome workflow, e.g. limited protein concentration range, accidental modifications in sample preparation, shifting of spots in gels that may result in missing values and false protein identification. DIGE is a low throughput and time-consuming process (3-4 days per run), that involves many steps and requires a high level of laboratory skill to obtain good results. It is also not a good technique for the analysis of extremely acidic, basic or hydrophobic proteins such as membrane-bound proteins, and also smaller proteins and peptides ( <15 kDa) and it is relatively expensive compared to traditional 2D gel.

To overcome the limitations of DIGE, SELDI-MS analysis was applied to a reasonably large numbers of administration samples. In total 98 plasma samples (34 control, 31 from nandrolone treated, and 33 from clenbuterol treated animals) were analysed by SELDI-MS. The protocol for sample preparation and analysis was established at HFL. A difference in the protein expression profile of control and treated cattle plasma was detected and, as such it was concluded that future work should be continued and involve the identification of

SID 5 (Rev. 3/06) Page 19 of 23

protein biomarkers of anabolic administration to cattle by mass spectrometry, development of a method of analysis, and finally, sample analysis of normal samples alongside the experimental samples. The main advantage of SELDI-MS is its potential for high-throughput. In addition, by using the respective chemical properties of each array it is possible to focus the analyses to either negatively or positively charged proteins or to target specifically metal binding or phosphoproteins. Of particular use is the hydrophobic array (H50) that facilitates the analyses of hydrophobic proteins, such as membrane-associated proteins, which are notoriously difficult to analyse by gel based assays. However, SELDI-MS was not suitable for the identification of the proteins of interest.

Several other avenues of a biomarker discovery pipeline were explored and resulted in the development of a method based on the simultaneous targeted monitoring of a set of proteins in cattle blood by mass spectrometry.Targeted multiplexing analysis of bovine plasma proteins, without a preliminary protein separation step, based on MRM transitions determined in 2D-LC-MS/MS experiments, was carried out. The final method utilised 60 transitions to monitor 41 bovine plasma proteins in a 52 min analytical run. The final protocol for nLC/MS/MS developed includes the following steps: trypsin digestion of plasma proteins, separation of tryptic peptides by liquid chromatography with subsequent identification by tandem mass spectrometry. Transfer of the protein digestion stage to a 96-well plate format resulted in a shorter sample preparation time.

80 control drug free plasma samples including 40 samples from 7 young male animals and 40 samples from individual lactating cows and 200 PA plasma samples were analysed using 5 µl of neat plasma for proteindigestion. 80 bovine plasma samples were analysed alongside 16 QC samples in one run with a total run time of 3.5 days per plate.

Several proteins were found to be up-regulated in plasma of lactating cows compared to the group of male (castrates) cattle: histidine-rich protein, complement component C9, α-1-acid glycoprotein, apolipopretein E, vitamin K dependent protein S, while prothrombin and C4b binding protein were found up-regulated in young animals.

Multivariate statistical analysis (PCA) did not reveal any group separation between the control and treated animals. Calves have a high conversion of feed into animal tissue. There were individual variations in protein expression but it was not possible to identify a general trend in changes of the investigated protein profiles in drug administered cattle.

To summarise the project outcomes:

The application of nLC-MS/MS and MRM analysis enabled a number of bovine proteins of varying molecular weights and at concentrations of µg/mL to be monitored

This study shows that proteins present in plasma at high levels (µg/mL) can be detected in a single analytical run without preliminary separation

The method is amenable to automation and, in conjunction with stable isotope labelled standard peptides, could be used as quantification method

The ability of the method to detect differentially expressed proteins was assessed by the analysis of drug-free plasma samples derived from study animals (young male castrates cattle) and lactating cows

There were no differences in the protein profile of the young control animals and those administered with nandrolone

Future studies on animals from an age group used for meat production will validate the test suitability to monitor the biomarkers of growth promoter abuse

The method could be applied to other matrices and tissues, e.g. saliva, to monitor changes in protein expression and could provide a basis for the development of a large-scale screening method

To the best of our knowledge, this study represents the first successful application of nLC-MS/MS for the analysis of the bovine proteome as a first step in biomarker identification of growth promoter abuse

References1. TNO Report V5922/02, 20052. TNO Report V5922/03, 20053. TNO Report V5922/04, 2006

SID 5 (Rev. 3/06) Page 20 of 23

4. F. Nomura, T. Tomonaga, K. Sogawa, T. Ohashi, M. Nezu, M. Sunaga, N. Kondo, M. Iyo, H. Shimada, T. Ochiai, Proteomics (2004), 4, 1187–1194

5. Y. Hu, S. Zhang, J. Yu, J. Liu, S. Zheng, The Breast (2005), 14, 250–2556. J.Y.M.N. Engwegen, H.H. Helgason, A. Cats, N. Harris, J.M.G. Bonfrer, J.H.M. Schellens, and J.H.

Beijnen, World J Gastroenterol. (2006), 12 (10), 1536-1544.7. G. Reddy and E. A. Dalmasso, J Biomed Biotechnol. (2003), 4, 237-241.8. C. Barton, P. Driver, R. Kay, P. Teale, BSPR 2007 Conference, Poster number 109. R.G. Kay, B. Gregory, P.B. Grace, S. Pleasance, Rapid Commun. Mass Spectrom.(2007), 21, 2585–

259310. R.G. Kay, P.R. Brown, J. Roberts, J.S. Boateng, G. Ball, C.S. Creaser, S.Y. Yang, S. Beech, G.

Goldspink, P. Teale, International Mass Spectrometry Conference (2006) Poster MoP 224.11. L. Anderson, C. L. Hunter, Molecular & Cellular Proteomics (2006) 5.4, pp. 573-588.12. L. Anderson et al, Molecular & Cellular Proteomics (2004) 3-4, pp. 311-326.13. M. Mooney et al, Proceedings of the EuroResidue VI Conference Egmond aan Zee, The Netherlands,

(2008) Oral presentation O22, v.1, pp.159-164.

AppendixTable 1. MRMs transitions designed and used for the detection of 41 proteins in bovine plasma.

Number Protein nameSwiss-Prot Peptide Sequence MS1/MS2

CE, eV RT,min MW, kDa Biological function

1

AMBP protein [Contains: Alpha-1-microglobulin; Inter-alpha-trypsin inhibitor light chain P00978 TVEACNLPIVQGPCR 857.3/926.5 45 19.5 39.2

Inhibitor of trypsin, plasmin, and lysosomal granulocytic elastase; alpha-1-migroglobulin occurs as a monomer and also in complexes with IgA and albumin

2 Actin, cytoplasmic 1 P60712 VAPEEHPVLLTEAPLNPK 652.0/568.3 35 19.4 41.7Involved in various type of cell motility

3Alpha-1-acid glycoprotein Q3SZR3 AIQAAFFYLEPR 713.3/971.5 35 23.5 23.2

Modulate the activity of immune system (acute phase reaction)

4 Alpha-2-antiplasmin P28800 LCQDLGPGAFR 617.3/604.3 30 22 54.7Inhibitor of plasmin, trypsin, and chymotrypsin

4 Alpha-2-antiplasmin P28800 LGPPSEEDYAQPSSPK 851.4/515.3 30 24.4 54.7Inhibitor of plasmin, trypsin, and chymotrypsin

5Alpha-2-HS-glycoprotein P12763 TPIVGQPSIPGGPVR 737.9/1064.6 45 19.2 38.4

Regulator of inflammatory response, promoter of endocytosis, suggested to have lymphocyte stimulating properties

6 Antithrombin-III P41361 DIPVNPMCIYR 689.3/839.4 35 22.2 52.3

Blood coagulation (serine protease inhibitor, as well as trombin and factors IXa, Xa, and Xia.

7 Apolipoprotein A-I P15497 LLDNWDTLASTLSK 788.8/935.5 35 24.2 30.3Lipid binding (cholesterol) and transport

7 Apolipoprotein A-I P15497 QGLLPVLESLK 590.4/785.5 30 27.8 30.3Lipid binding (cholesterol) and transport

8 Apolipoprotein A-II P81644 TQEELTPFFK 620.4/1010.5 30 21.3 11.2

Lipid binding and transport, HDL metabolism, positive regulation of IL-8 biosynthesis, antimicrobial activity

8 Apolipoprotein A-II P81644 AGTDLLNFLSSFIDPK 869.9/1168.6 35 29.7 11.2

Lipid binding and transport, HDL metabolism, positive regulation of IL-8 biosynthesis, antimicrobial activity

9 Apolipoprotein A-IV Q32PJ2 LVPFATELHER 656.3/855.4 30 20 43.0

Lipid binding and transport (HDL), required for activation of lipoprotein lipase by ApoC-II

9 Apolipoprotein A-IV Q32PJ2 LGEVSTYTDDLQK 734.9/1070.5 30 19.2 43.0

Lipid binding and transport (HDL), required for activation of lipoprotein lipase by ApoC-II

10 Apolipoprotein C-III P19035 DALSSVQESQVAQQAR 858.8/1144.6 30 18.4 10.7Lipid degradation and transport (VLDL)

11 Apolipoprotein E Q03247 FGPLVEQGQSR 609.3/704.3 30 19.8 35.98Lipid binding and transport, can serve as a ligand for LDL

11 Apolipoprotein E Q03247 LAVYQAGASEGAER 711.3/776.4 30 21.9 35.98Lipid binding and transport, can serve as a ligand for LDL

12Beta-2-glycoprotein 1 (Apolipoprotein H) P17690 FTCPLTGLWPINTLK 880.9/685.4 45 25.3 38.3

Blood coagulation, prevents activation of coagulation by binding to phospholipids on the surface of damaged cells, binds to negatively charged substances (heparin, phospholipids)

13C4b-binding protein alpha chain Q28065 ALCLKPEIEYGR 724.8/863.4 35 18.8 68.9

Immune response, control of the classical pathway of compliment activation

13C4b-binding protein alpha chain Q28065 ELSCTSSGWSPAVPQCK 947.4/532.3 45 19 68.9

Immune response, control of the classical pathway of compliment activation

14 Coagulation factor V Q28107 DLASGLIGLLLICK 743.4/816.5 35 29.1 248.98

Blood coagulation, activate (with factor Xa) prothrombin to thrombin

15 Collectin-43 P42916 SSAENEAVTQLVR 702.3/786.5 30 21.1 33.6Immune response, binding to various sugars

16 Complement C3 Q2UVX4 SDLDDDIIPEEDIISR 922.8/958.5 40 23 187.3

Immune response, inflammatory, causes histamine release from mast cells and basophilic leukocytes

SID 5 (Rev. 3/06) Page 21 of 23

16 Complement C3 Q2UVX4 NYAGVFTDAGLTLK 735.4/965.5 30 22.3 187.3

Immune response, inflammatory, both classical and alternate complement pathways, causes histamine release from mast cells and basophilic leukocytes

17

Complement C4 [Contains: Complement C4 alpha chain; C4a anaphylatoxin; Complement C4 γ-chain P01030 TLEIPGNSDPNIIPEGDFK 1028.5/692.3 45 22.3 101.9

Immune response, classical pathway, mediator of local inflammatory processes

18Complement component C7 Q29RQ1 SLEGPSTFLCSSSLK 806.9/681.3 30 18.6 93.1

Immune response, complement alternate pathway, membrane attack complex

19Complement component C9 Q3MHN2 AIEDYINEFSVR 728.3/864.5 45 23.3 62.0

Immune response, complement alternate pathway, final components of membrane attack complex

19Complement component C9 Q3MHN2 LVPIYDLIPVK 635.4/847.5 30 24.5 62.0

Immune response, complement alternate pathway, final components of membrane attack complex

20 Complement factor B P81187 YGLVTYATEPK 621.3/809.4 30 19 85.4Immune response, complement alternate pathway

20 Complement factor B P81187 GIPEFYDYDVALVR 828.9/1113.6 30 23.8 85.4Immune response, complement alternate pathway

21 Complement factor H Q28085 TPVILNGQAVLPK 675.4/939.6 30 20.2 140.4Immune response, complement alternate pathway

22 Conglutinin P23805 AGVTGPSGAIGPQGPSGAR 818.9/826.4 30 22.5 38.0

Calcium dependent lectin-like protein, binds to immune complexes through complement C3b component

22 Conglutinin P23805 AVLFPDGQAVGEK 665.8/900.4 30 23.7 38.0

Calcium dependent lectin-like protein, binds to immune complexes through complement C3b component

23

Fibrinogen alpha chain [Contains: Fibrinopeptide A] P02672 GLIDEVDQDFTSR 747.8/868.4 35 23.1 67.0

Blood coagulation, cofactor in platelet aggregation

23

Fibrinogen alpha chain [Contains: Fibrinopeptide A] P02672 TGLAPEFAALGESGSSSSK 898.4/1453.7 40 21 67.0

Blood coagulation, cofactor in platelet aggregation

24

Fibrinogen beta chain [Contains: Fibrinopeptide B] P02676 TSSSTFQYITLLK 744.8/1025.6 35 22.6 67.0

Blood coagulation, cofactor in platelet aggregation

24

Fibrinogen beta chain [Contains: Fibrinopeptide B] P02676 EDGGGWWYNR 620.3/638.3 30 20.7 67.0

Blood coagulation, cofactor in platelet aggregation

25Fibrinogen gamma-B chain P12799 IQLEDWNGR 565.8/889.4 30 19.5 50.2

Blood coagulation, cofactor in platelet aggregation

26 Fibronectin P07589 WLPSSSPVTGYR 675.3/1050.5 35 19.6 249.6

Binding to cell surface, collagen heparin, fibrin, DNA, actin, cell shape maintenance, wound healing, opsonisation, cell motility

26 Fibronectin P07589 SSPVVIDASTAIDAPSNLR 956.9/1330.7 50 21.2 249.6

Binding to cell surface, collagen heparin, fibrin, DNA, actin, cell shape maintenance, wound healing, opsonisation, cell motility

27 Gelsolin Q3SX14 AQPVQVAEGSEPDSFWEALGGK 1151.5/1206.6 30 23.5 80.7

Binding to actin preventing monomer exchange (blocking or end-capping) and to fibronectin

28Hemoglobin subunit alpha P01966 VGGHAAEYGAEALER 765.3/1108.5 30 18.7 15.2 Oxygen transport

29Hemoglobin subunit beta P02081 LLVVYPWTQR 637.8/850.4 30 22.6 15.9 Oxygen transport

30Histidine-rich glycoprotein P33433 APLPFPPPGLR 581.3/636.4 30 22 44.5

The physiological role is not yet known, may mediate the contact activation phase of intrinsic blood coagulation cascade

31

Inter-alpha-trypsin inhibitor heavy chain H4 Q3T052 AAAQEQYSAAVAR 668.3/994.5 30 16.1 101.5 Acute phase reaction

32 Kininogen*P01044(I), P01045(II) RPPGFSPFR 530.8/807.4 30 18.4

68.9 (I)68.7(II)

Total protease inhibitor, blood coagulation, mediator of inflammation, cardioprotective, smooth muscle contraction

32 Kininogen*P01044(I), P01045(II) LISDFPETTSPK 667.9/759.4 30 19

68.9 (I)68.7(II)

Total protease inhibitor, blood coagulation, mediator of inflammation, cardioprotective, smooth muscle contraction

33 Lactotransferrin P24627 CLQDGAGDVAFVK 690.3/863.5 30 24.9 78.1Antimicrobial, metal ion (Fe3+) binding, ion transport

34Pigment epithelium-derived factor Q95121 LQSLFDAPDFSK 684.3/1126.5 30 22.5 46.2

Potent inhibitor of angiogenesis, neurotrophic, does not inhibit serine protease

35Plasma retinol-binding protein P18902 DPSGFSPEVQK 595.8/600.3 30 20.4 21.1

Retinol binding, transport, as a complex with retinol interacts with transhyretin

36 Plasminogen P06868 VSPYVPWIEETMR 803.8/1061.5 40 23.6 91.2

Blood coagulation, activator of complement zymogens (C1, C5), inactivates by alpha-2-antiplasmin

36 Plasminogen P06868 EQSVQEIPVSR 636.3/1014.6 35 21.4 91.2

Blood coagulation, activator of complement zymogens (C1, C5), inactivates by alpha-2-antiplasmin

37 Prothrombin (coagulation factor II)

P00735 LGEDPDPDAAIEGR 727.8/1040.5 30 18.5 70.5 Blood coagulation, homeostasis, inflammation, and wound healing, protease, cleaves bonds after Arg and

SID 5 (Rev. 3/06) Page 22 of 23

Lys

37Prothrombin (coagulation factor II) P00735 SGIECQLWR 574.8/891.4 30 19.8 70.5

Blood coagulation, homeostasis, inflammation, and wound healing, protease, cleaves bonds after Arg and Lys

38Serotransferrin (transferrin) Q29443 CGLVPVLAENYK 681.9/933.5 30 21.2 77.6

Iron binding transport protein, regulation of cell proliferation

38Serotransferrin (transferrin) Q29443 TYDSYLGDDYVR 733.8/837.4 35 19.8 77.6

Iron binding transport protein, regulation of cell proliferation

39 Serum albumin P02769 HPEYAVSVLLR 642.3/1049.6 30 19.6 69.3

Transport protein (water, metal ions, fatty acids,hormones, bilirubin, and drugs), regulator of blood osmotic pressure

39 Serum albumin P02769 VPQVSTPTLVEVSR 756.4/1088.6 45 19.8 69.3

Transport protein (water, metal ions, fatty acids,hormones, bilirubin, and drugs), regulator of blood osmotic pressure

40Vitamin D-binding protein Q3MHN5 VLDQYIFELSR 691.9/1170.6 30 23.2 53.3

Vitamin D transport, immune response

40Vitamin D-binding protein Q3MHN5 FPDATETDLQELVAK 838.9/915.5 35 19.1 53.3

Vitamin D transport, immune response

41Vitamin K-dependent protein S P07224 YLGCLGSFR 536.7/579.3 30 20.2 75.1

Blood coagulation (anticoagulant)

*peptide fragment selected corresponds to two different proteins P01044 Kininogen-1 light chain) and P01045 (Kininogen-2 light chain).

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.Project Manager Dr Jane Roberts has given several presentations at various international meetings and conferences that have included substantial reference to the Biomarkers project. Including:

Performance enhancement in sport: New Threats and New Solutions. Institute of Biology lecture, Royal Veterinary College, London, UK - September 2007

Gene Doping, Designer drugs, proteins: New Threats, New Solutions. Oral Presentation Istanbul, Turkey - September 2005 Association of Racing Chemists Conference.

  Screening Biomarkers as indicators of Drug Abuse. Oral Presentation, Philadelphia, USA -

January 2006 Racing consortium Proteomics Workshop  

Surrogate Markers for Transgene Expression – Global Gene Expression Analysis. Oral Presentation, Stockholm, Sweden – December 2006 WADA Gene Doping meeting 

Biomarkers in Clinical Development. Oral Presentation, HFL, UK – March 2006 Institute of Clinical Research seminar

We have also had several internal review meetings where the project has been discussed with external contacts from organisations such as Cambridge University and Imperial College London.

SID 5 (Rev. 3/06) Page 23 of 23