Protein biomarker-based screening for detection of recombinant bovine somatotropin abuse in dairy cows

Recombinant bovine somatotropin (rbST) is a 22 kDa proteohormone, which can be used to increase milk production in dairy cows. It has been marketed since 1994 and while its use in food production is approved in several countries, such as the US, it is banned in the EU since 2000. To enforce the ban on rbST in the EU and to control for ‘rbST-free’ –labelling in the US, detection methods are required that identify whether rbST has been used. Existing rbST detection methods focus on the detection of rbST itself in bovine serum. The recombinant form of the hormone has one amino acid exchanged at the N-terminus of the protein. RbST can therefore be potentially discriminated from the endogenous bST by mass spectrometric methods. Other methods employ sandwich enzyme-linked immunosorbent assays (ELISAs) with antibodies having a higher affinity to rbST than bST. These methods, however mainly lack sensitivity, reproducibility or selectivity for rbST and are therefore not widely applied. Hence, no method has been implemented so far to monitor rbST abuse in dairy farming. Screening methods developed for veterinary drug residue control in the EU have to perform according to Commission Decision 2002/657/EC and have to identify at least 95 % of the treated animals. An alternative approach for rbST abuse detection is the analysis of rbST-dependent biomarkers. A biomarker is defined as an indicator of normal physiological processes, pathogenic processes, or pharmacological responses to a therapeutic intervention. Therefore, the levels of rbST-dependent protein biomarkers are either up- or downregulated after administration of rbST and rbST-specific biomarker profiles can be used to detect its abuse. RbST exerts similar physiological actions in the cow’s body as the endogenous bST. Therefore, proteins involved in the regulatory circuit of bST have been chosen as candidate biomarkers, such as insulin-like growth factor-1 (IGF-1), IGF binding protein 2 (IGFBP2) and osteocalcin. Additionally to that, the administration of rbST induces anti-rbST antibodies in the cow’s body, which can be detected as biomarkers. This approach is according to the growth hormone (GH) abuse detection in sports doping control, where solely protein biomarker profiles are used to identify the abuse. <strong>Chapter 2 </strong>introduces protein biomarkers and how biomarkers can be used in sports doping and veterinary control to detect the abuse of illegal substances. The advantages of using biomarkers are that the biological effect of a substance usually lasts longer than the substance itself can be detected and therewith, the window of detection is expanded. Moreover, since different substances exert similar effects on physiological machineries for growth or production enhancement, biomarker-based-detection methods have the potential to detect a whole class of substances. Furthermore, low-dose mixtures of different banned substances, which might escape from direct detection of each individual substance used, could be still detected by the combined effect they exert. In this chapter, protein biomarker-based detection strategies are discussed against generic challenges in biomarker discovery and method development. <strong>Part I </strong>of the thesis concerns biomarker analysis in serum and plasma samples from cattle, which are analysed using laboratory-based equipment.A triplex flow cytometric immunoassay (FCIA), which combines the detection of three rbST-dependent biomarkers, <em>viz. </em>IGF-1, IGFBP2 and anti-rbST antibodies is demonstrated in <strong>Chapter 3</strong>. Serum samples from treated and untreated dairy cows from a single animal study were analysed using this triplex FCIA. Characteristic treatment-dependent responses for all three individual biomarkers were shown. These results were combined using the statistical model <em>k</em>-nearest neighbours (kNN). This model discriminated rbST-treated from untreated cows with a truepositive rate of 89.1 % and a true-negative rate of 97.7 %. This triplex FCIA was further extended with the biomarker osteocalcin and the resulting fourplex FCIA was used for biomarker profiling in serum samples from rbST-treated and untreated cows from two independent rbST treatment studies. In <strong>Chapter 4</strong>, different data analysis approaches were tested with the aim to detect the highest possible number of true-positive samples. The statistical model kNN was used on all 11 possible biomarker combinations and the combination of the biomarkers osteocalcin and endogenously produced antibodies against rbST proved to be very reliable and correctly predicted 95 % of the samples of treated cows starting from the second rbST injection until the end of the treatment period and even thereafter. With the same biomarker combination, only 12 % of the samples of untreated animals appeared false-positive. This reliability meets the requirements of Commission Decision 2002/657/EC for screening methods in veterinary control. It can be expected that rbST-dependent biomarkers also show a response upon other treatments. Therefore in <strong>Chapter 5</strong>, the fourplex FCIA for rbST abuse detection was applied to bovines treated with steroids, such as estradiol, dexamethasone and prednisolone. Each treatment resulted in a specific plasma biomarker profile for IGF-1, IGFBP2, osteocalcin and anti-rbST antibodies, which could be distinguished from the profile of untreated animals. Therefore, the fourplex biomarker FCIA is, apart from rbST, also capable of detecting treatment with other growth-promoting agents and clearly shows the potential of biomarker profiling as a screening method in veterinary control. <strong>Part II </strong>of the thesis focusses on protein biomarker analysis in milk samples and the change from laboratory-based to on-site analysis. In <strong>Chapter 6</strong>, the detection of anti-rbST antibodies in raw milk samples was demonstrated, which discriminated rbST-treated from untreated cows with a 67 % true-positive and 94 % true-negative rate. The laboratory-based assay was also applied to simulated tank milk and pasteurized milk samples. Using milk as a sample matrix for detection has the advantages of non-invasive sampling, and for tank milk analysis at the farm only one milk sample is needed to screen the whole farm for rbST (ab)use. As a next step in <strong>Chapter 7</strong>, this assay was translated to an on-site pre-screening platform including a cellphone. Using this on-site platform, samples can be tested at the point where they were taken. Only samples that are suspect are transported to a laboratory for further analysis. To this end, a cellphone-based fluorescence imaging platform for the detection of anti-rbST antibodies in milk extracts was developed, which is based on a microsphere fluorescence immunoassay. After performing the assay, the fluorescence is excited by UV LEDs embedded in a dedicated cellphone attachment and the emitted fluorescence light is imaged by the cellphone camera. The fluorescence micro-images were analysed using a custom-developed Android application running on the same cellphone and milk samples from rbST-treated and untreated cows were discriminated. Also in milk samples, the simultaneous detection of several biomarkers is advantageous as they can increase the confidence of a positive finding. Therefore in <strong>Chapter 8</strong>, a protein microarray-based platform for multiple rbST biomarker detection on a cellphone is presented, which detects anti-rbST antibodies and IGF-1 in milk samples. The 48 microspots on the microarray were labelled with Quantum Dots depending on the biomarker levels in the sample. Quantum Dot fluorescence was detected by the cellphone camera and the same opto-mechanical attachment as in <strong>Chapter 7 </strong>and images were analysed by custom software. RbST-treated clearly showed a treatment-dependent biomarker profile in milk that could be discriminated from the profile of untreated cows. Future research should focus on the simultaneous detection of different targets of interest in milk samples, such as hormones, allergens, antibiotics, contaminants and other substances, all at the same time using the microarray platform on the cellphone. Moreover, sample handling can be facilitated by the use of pre-fabricated microfluidic devices including all required assay reagents. With the work presented in this thesis, screening for rbST abuse in serum and milk becomes possible: in the laboratory and on-site. The future implementation of these testing platforms for rbST abuse detection is a major leap forward concerning the enforcement of the rbST ban in the EU and concerning the value of protein biomarker-based approaches in veterinary control.