Autoantibody panels for improved disease diagnosis

Michael McAndrew, Colin Wheeler and John Anson

studies have identified The need for associated with many types of cancer1,2 3 4 Our increasing knowledge of the complex including, cancers of the lung , breast , head and neck5, colon6, ovary7 and nature of molecular interactions has 8 enabled not only a better understanding of prostate , offering the potential of improved physiological and pathological processes, diagnosis of these important cancers. It is but also the identification of biological believed that autoantibodies are generated markers (biomarkers) that define a particular through over-expression, mutation, release state or condition. Molecular biomarkers are of proteins from damaged tissues, mis- now used across many disciplines and can folding or mis-presentation of proteins which be any molecule, part of a molecule or even leads to their recognition by the immune a particular configuration that is both system. detectable and measurable, where the amount, appearance or other property is indicative of a particular biological state. Benefits of biomarkers: Biomarkers have many applications x Earlier disease detection ² prior including the diagnosis of disease, the to clinical symptoms identification of a pre-disposition towards a given disease and the identification of x Higher sensitivity and specificity patients who would benefit from a given ² more accurate than existing therapy. tests

Until now, most diagnostic tests have been x Accessible ² easy to collect from based on single biomarkers; however, blood enabling cost-effective basing a clinical decision on a single biomarker can lead to a significant level of assays false positives. Increasingly, multiplexing of biomarkers (i.e. signatures or panels) is being used to provide improved sensitivity and specificity for the diagnosis and The autoantibody advantage characterisation of disease. Unlike most other proteins found in serum, Autoantibodies as early disease markers autoantibodies are stable, abundant, highly specific, easily purified from serum, and are Developing practical and robust diagnostic readily detected with well-validated panels of biomarkers requires a dependable secondary reagents2. Since these reagents system for detecting and measuring them. bind to the constant regions of have several properties which immunoglobulins, antibodies of the same make them excellent indicators of disease class which recognise different antigens can VHH³%HQHILWVRIDXWRDQWLERG\ELRPDUNHUV´  be detected simultaneously, supporting high and their detection forms the basis of many levels of multiplexing and enabling higher in vitro diagnostic tests. throughput and faster biomarker In addition to producing antibodies against identification than other technologies. foreign molecules, the immune system generates antibodies to self-proteins Due to their inherent amplification within the ³DXWRDQWLERGLHV´  LQ UHVSRQVH WR PDQ\ immune system, autoantibodies are pathological processes. For example, relatively abundant and are easily measured, making them ideal for the 1 detection of disease at an early stage when antigens with as little as 5 µl of undiluted other potential biomarkers may be serum. Detection of autoantibody binding by undetectable. In diseases such as established and well characterised and cancer, secondary antibodies also enables autoantibodies can be detected several identification of specific classes. years prior to the presentation of symptoms.

Autoantibody biomarkers with high Discovery platform technologies sensitivity and specificity can be identified often involves accurately and easily utilising samples comparison of large datasets to identify obtained with minimally invasive techniques. novel markers that can be used to Based on these properties, development of differentiate between case and control autoantibody-based diagnostic assays could rapidly advance the diagnosis and treatment groups. Given the requirement for statistical of disease. rigor, biomarker studies may incorporate large numbers of samples, collected and assayed over extended timeframes. It is Identification of autoantibodies for biomarker discovery therefore important that technical variation attributable to the platform itself is kept to a Characterisation of autoantibodies in serum minimum. using traditional methods such as ELISA can be slow, relatively low throughput and Currently, the most powerful platforms for require substantial amounts of purified screening for novel autoantibody antigen. ELISAs use large volumes of biomarkers are based on protein or peptide serum, particularly when profiling against antigen arrays9. Such arrays are available multiple antigens and generally only single through a number of different sources, and antigens are assayed per well. Because of can offer fast, high-throughput and cost- these limitations, ELISAs are not readily effective identification of autoantibodies. To amenable to high-throughput screening of ensure the identification of sensitive and hundreds or thousands of antigens for specific biomarkers, it is imperative that biomarker discovery. careful consideration is given to the choice of array. Protein arrays from alternative An ideal platform for autoantibody discovery sources differ in a number of ways, the most would have the following attributes: important being the protein content on the array and how that content is attached to x Assay performance equal or greater than the slide surface. ELISA x Facility to multiplex hundreds of antigens Protein content per sample in one assay The majority of commercially available x Non-hypothesis-driven discovery protein arrays comprise random collections x Facility to detect all immunoglobulin of proteins with no particular relevance to isotypes specific disease areas. A more targeted and x Requirements for minimal amounts of statistically powerful approach to protein serum per assay array design is to carefully select proteins according to their classifications (e.g. Based on these criteria, addressable protein kinases, signalling, transcription factors) and microarrays are the ideal platform for the the disease of interest (e.g. infectious identification of autoantibody biomarkers. disease, cancer, autoimmune disorders). Such arrays enable immediate identification of the cognate binding partners for specific Protein attachment and conformation autoantibodies in sera and, with current The target epitopes of many spotting densities, enable the simultaneous autoantibodies1,10 are conformation- screening of hundreds or even thousands of 2 dependent structures, termed discontinuous epitopes. Recognition and binding of the antibody to its cognate epitope is determined by the three-dimensional folding of the antigen on which the epitope is displayed. Labelled anti-human antibody for Proteins are by nature diverse in size, detection structure and function, therefore immobilisation on a solid support such that Autoantibody they retain their native structure presents a challenge. Conventional protein immobilisation methods fail to address this Correctly folded functional protein problem because the attachment is random and non-specific. This can compromise the specificity, and therefore effectiveness, of BCCP Tag arrays11 because:

Biotin x The protein may be denatured by being immobilised on a surface x 7KH SURWHLQ¶V K\GURSKRELF FRUH UHJLRQV Figure 1: The unique BCCP tag conserves may be exposed leading to non-specific the native protein conformation enabling the interactions discovery of autoantibody biomarker panels x The discontinuous epitopes themselves exhibiting maximum sensitivity and may be destroyed, causing false specificity. negatives as study size increases beyond small pilot An innovative solution to the challenge of stages and between-assay variability needs attaching proteins to the array surface in to be minimised over prolonged periods of their native form is the use of a unique time and large sample numbers.

BCCP tag12 (Figure 1). Outsourcing biomarker discovery and analysis The BCCP tag ensures the correct three- dimensional presentation of the The generation of raw data from protein discontinuous epitope is maintained. arrays is however only one part of the Importantly, it also minimises exposure of biomarker discovery process. Biomarker WKH HSLWRSHV¶ FRUH K\GURSKRELF UHJLRQV discovery is one of the most active areas in which is crucial in order to prevent non- biomedical research today. Despite specific binding that is frequently the source substantial numbers of putative biomarkers being published in the literature, few ever of the false positive results13,14,15. translate to the clinic despite initially promising results16. Many of the reasons Reproducibility behind this relate to poor study design, Replicates of each protein on the array inherent bias and over-fitting of data17, 18. It provide statistical robustness and allows is clear that good study design, exemplary identification of technical issues such as sample collection and robust statistical printing anomalies, surface variations and analysis are essential for reliable biomarker sample variability. The number of protein identification. replicates spotted on the array varies but having 3 or more replicates per protein For this reason, many scientists and ensures accurate, robust data can be companies choose to outsource their obtained even if one of the protein spots is biomarker discovery and validation projects compromised. This is particularly important to a specialist service provider. Such

3 providers work either in partnership or on an ³DV-QHHGHG´ EDVLV WR GHOLYHU KLJK-quality, high-throughput biomarker datasets. One of ³1RQ-invasive serum sampling is the the advantages of outsourcing biomarker future of cancer diagnostics. By discovery is the ease with which companies detecting autoantibodies in serum can access highly specialised experts and using a novel functional protein technology with state-of-the art processes, microarray, the OGT approach can methods and statistical analysis techniques. improve both the specificity and

VHQVLWLYLW\RIWKHVHWHVWV´ Outsourcing your biomarker discovery project to Oxford Gene Technology Norman J. Maitland, Professor of

Molecular and Director of Founded by Professor Ed Southern in 1995, the Yorkshire Cancer Research Oxford Gene Technology (OGT) has a long (YCR) Cancer Research Unit history of innovation in biomarker discovery, with unrivalled experience in the development and use of microarrays.

The Discovery Array ² a unique protein array platform OGT delivers high-quality results 2*7 KDV GHYHORSHG D XQLTXH ³IXQFWLRQDO OGT brings extensive expertise in SURWHLQ´ DUUD\ WHFKQRORJ\ ZKLFK XWLOLVHV biomarker discovery study design, array correctly folded proteins and has the ability fabrication and analysis of array data to to display native, discontinuous ensure that the best possible data is epitopes2,12,19,20,. Based on the BCCP tag obtained from precious clinical samples22. (Figure 1), the OGT Discovery Array has Our high-throughput workflow produces been developed to identify specific and high-quality results quickly and incorporates sensitive autoantibody biomarkers. over 140 critical quality measures to provide confidence that results are valid. The array content represents multiple functional and disease pathways, many of We can help you design an informative which have been previously reported to autoantibody discovery programme for your elicit autoimmune responses in cancer disease area of interest ² from concept to patients, thereby increasing the probability completion. Our experienced team, of discovering clinically relevant innovative technology and purpose-built autoantibodies. Protein content can also be facilities ensure that your programme is added to further enhance discovery designed and carried out to the highest capability for a specific disease. standards.

Each protein on the OGT Discovery Array is Visit www.ogt.co.uk to learn more about printed in quadruplicate to give a measure our protein array technology and see of the technical variability of printing and how the OGT approach can enhance assay. Each array has multiple controls to your biomarker discovery projects. allow comparison of arrays within assays, between assays and between assay References operators. 1. Tan, E.M., and Zhang, J. (2008) Monitoring and control of all critical Autoantibodies to tumor-associated parameters within a study enables assay antigens: reporters from the immune accuracy and reliability. The reproducibility system. Immunol Rev 222, 328 of the platform is exceptionally high21*, which enables the screening of statistically 2. Gnjatic, S. et al (2009) Seromic analysis meaningful numbers of clinical samples of antibody responses in non-small cell lung over extended timescales. cancer patients and healthy donors using conformational protein arrays. J Immunol * CV <2.0% inter-assay after normalisation Methods 341(1-2), 50

4 3. Diesinger, I. et al (2002) Toward a more aluminum hydroxide. Immunol Methods 200 complete recognition of immunoreactive (1-2), 99 antigens in squamous cell lung carcinoma. Int J Cancer 102, 372 15. Butler, J.E., Navarro, P., and Sun, J. (1997) Adsorption-induced antigenic 4. Disis, M.L. et al (1994) Existent T-cell and changes and their significance in ELISA and antibody immunity to HER-2/neuProtein in immunological disorders. Immunol Invest 26 patients with breast cancer. Cancer Res 54, (1-2), 39 16 16. Ptolemy, A.S. (2010) Biomarker 5. Carey, T.E. et al (1983) Antibodies to translation and ² human squamous cell carcinoma. Context, reality and a call for enhanced Otolaryngol Head Neck Surg 91, 482 validation. Current Pharmacogenomics and Personalized Medicine 8, 171 6. Scanlan, M.J. et al (1998) Characterization of human colon cancer 17. Zolg, W. (2006) The proteomic search antigens recognized by autologous for diagnostic biomarkers: lost in antibodies. Int J Cancer 76, 652 translation? Molecular & cellular 5, 1720 7. Chatterjee, M. et al (2006) Diagnostic markers of ovarian cancer by high- 18. Ransohoff, D.F. (2005) Bias as a threat throughput antigen cloning and detection on to the validity of cancer molecular-marker arrays. Cancer Res 66, 1181 research. Nature Reviews Cancer 5, 142

8. Wang, X. et al (2005) Autoantibody 19. Boutell, J.M. (2004) Functional protein signatures in prostate cancer. N Engl J Med microarrays for parallel characterisation of 353, 1224 mutants. Proteomics 4(7), 1950

9. Casiano, C.A., Mediavilla-Varela, M., and 20. Blackburn, J.M. and Hart, D.J. (2005) Tan, E.M. (2006). Tumor-associated antigen Fabrication of protein function microarrays arrays for the serological diagnosis of for systems-oriented proteomic analysis. cancer. Mol Cell Proteomics 5, 1745 Methods Mol Biol 310, 197

10. Welin Henriksson, E. et al. (1999) Key 21. Shoner, A et al (2009) Robust-Linear- residues revealed in a major conformational Model Normalization To Reduce Technical epitope of the U1-70K protein. PNAS 96 Variability in Functional Protein Microarrays. (25), 14487 Journal of Proteome Research 8, 5451

11. Cha, T., Guo, A., and Zhu, X.Y. (2005) 22. Conrad, D.F. et al (2010) Origins and Enzymatic activity on a chip: the critical role functional impact of copy number variation of protein orientation. Proteomics 5(2), 416 in the human genome. Nature 7289, 704

12. Koopmann, J.O., McAndrew, M.B. and %ODFNEXUQ -0   LQ ³3URWHLQ 0LFURDUUD\V´ &KDSWHU  HG 6FKHQD 0 (Jones and Bartlett) Oxford Gene Technology Diagnostic Biomarkers 13. Yoshida, H., Imafuku, Y., and Nagai, T. The Southern Centre (2004) Matrix effects in clinical Oxford Industrial Park immunoassays and the effect of preheating Mead Road and cooling analytical samples. Clin Chem Yarnton Lab Med 42(1), 51 Oxford OX5 1QU United Kingdom 14. Houen, G. and Koch, C.J, (1997) A non- T: +44 (0) 1865 845000 denaturing enzyme linked immunosorbent F: +44 (0) 1865 848684 assay with protein preadsorbed onto E: [email protected]

7KLVGRFXPHQWDQGLWVFRQWHQWVDUH‹2[IRUG*HQH7HFKQRORJ\,3/LPLWHG$OOULJKWVUHVHUYHG2*7ŒDQG2[IRUG*HQHTechnology are 5 trademarks of Oxford Gene Technology IP Limited.