Getting to the bottom of biomarkers

A new research paper sheds light on the way antibodies distinguish between different but closely related biomarkers – in this case protein fragments which reveal information about the condition of the human body. This new understanding could enable pharmaceutical companies to develop new technologies for quickly diagnosing and treating fatal diseases

All diseases exhibit one or more molecular species, often proteins, which act as unique identifiers or biological markers. Detection of these biomarkers, among other closely related compounds, is a powerful and much sought after diagnostic capability. These biomarkers are detected by immunoassays – a test which mixes a substance (eg blood, urine) with antibodies, which bind to the protein if it is present. The antibodies can then be measured to identify the level of the biomarker, which in turn indicates the presence and extent of an illness.

Antibodies bind with high specificity to one protein molecule or a limited group of molecules (eg hormones), which is why antibodies can be used to test for specific biomarkers. Problems arise when they bind to groups of similar hormones that are associated with normal functions thereby leading to false positives and unreliable information.

The highly sought solution is ‘intelligent selection’ of antibody-specific interaction sites on hormones that may differ from similar sites in other hormones by just one molecule – a so-called point mutation.

This research focused on hCG (human chorionic gonadotropin), a hormone normally produced during pregnancy. A subunit of hCG – hCG? – is secreted by some cancers, meaning detection can give early warning of tumours.

hCG is very similar to other reproductive hormones, known as LH and FSH, which are always present in the body. Reliable discrimination of hCG among these other hormones is a significant challenge and unreliable results are inevitable unless very high-fidelity tests can be developed.

The research was partly carried out with a view to proving a concept which would help Mologic, a UK biotechnology SME to develop new diagnostic tools. Mologic, along with Pepscan Presto, a Dutch specialty-peptide company worked together on the binding data measurements and the experimental design. The National Physical Laboratory (NPL) provided the accurate experimental measurement of molecular structure, and IBM and the University of Edinburgh developed and ran powerful computer models to refine the experimental interpretation with structural data at atomic-level detail.

Mologic determined which linear sequential epitopes were recognised by various antibody preparations, a library of overlapping peptides representing the entire sequence of hCG? was created in several wells.

The first well was endowed with an array of identical immobilised peptides consisting of amino acids 1-12, the second well carried peptides consisting of amino acids 2-13, the third 3-14 etc. In conjunction with further libraries constructed with longer peptides, these libraries were used to determine which host species (eg mouse, rabbit, or sheep) were able to produce antibodies that recognised the linear sequence of the ?3-loop. From these findings a panel of candidate monoclonal antibodies with binding specificity for the ?3-region were selected.

The binding of antibodies to each peptide was tested in an Enzyme-linked immunosorbent assay (ELISA) at Pepscan.

The covalently linked peptides were incubated with primary antibody 8G5 (obtained from sheep) diluted in blocking solution. After the wells had been washed, the peptides were incubated with a 1/1000 dilution of secondary antibody peroxidase to detect and quantify the binding of 8G5 antibodies. One peptide gave notable binding responses. This peptide was found to be identical to a LH fragment, with the only difference being an atomic detail in the side chain of one amino acid.  Therefore, the two peptides (one from hCG and one from LH) were selected for comparative biophysical and molecular dynamics studies to reveal the structural rationale of the activity.

After a further washing step, a peroxidase substrate was added to provide florescence. Colour development was measured by means of an NPL charge coupled device (CCD) – a camera and an image processing system.

The Biotechnology team at NPL provided expertise in an array of spectroscopic techniques. These techniques can be utilised to study the structure of proteins as structural changes can have a major impact on their activity, stability and toxicity, and consequently can compromise the efficacy and shelf life of products.

NPL characterised peptides in situ using two techniques, for which it has some of the world’s leading facilities. First measurements were carried out using circular dichroism (CD) spectroscopy. CD in the far UV region (180–260nm) provides information regarding different forms of regular secondary structure found in proteins. (In the near UV region it can provide detailed fingerprints of tertiary structures, DNA-protein interactions and can also be very useful in the comparison of batches of pharmaceuticals.)

Next, Fourier Transform Infrared (FTIR) Spectroscopy was used to facilitate the structural analysis of the proteins in different chemical environments. This technique is needed to analyse the structure of proteins at higher concentrations than CD and gives more detailed information about specific conformers and their relative ratios in a given peptide population.

Once measurements had been taken of the peptides they were passed on to IBM’s Watson Research Centre and the University of Edinburgh physics department to create models using molecular dynamics (MD) – a computer simulation of physical movements of atoms and molecules.

MD simulations offer powerful support to experimental programmes in biomolecular structure. In principle, MD simulations can provide “ultimate” detail concerning individual molecular movement at full atomic resolution; thus, they can be used to answer specific questions about the properties of a system often more readily than experiments on the actual system, provided forces are accurately described and the sampling of configurations is sufficiently extensive.

In order to analyse the structure of the peptide using molecular dynamics simulations, the team made use of techniques known as Ramachandran plots and bond probability distribution functions. The former allows for the assignment of secondary structure motifs based on dihedral angles, while providing a clear illustration of the conformational space explored during the course of the simulation. The latter allows for the probability distribution of selected atomic contacts as a function of inter-atomic separation to be assessed.

The analysis was carried out on IBM’s machines, which are some of the most powerful modelling machines available, and some of the most advanced algorithms.

Combined, the data revealed that hCG epitope adopts a very specific spatial arrangement, a b-turn conformation, which stabilises it into an active form. This is this form that is recognised by the antibodies. In contrast, the LH fragment does not have the same structure and cannot fix in space thus failing to bind the antibodies.

Professor Paul Davis, Chief Scientific Officer of Mologic, commented that the project “was a great collaborative effort, and it stands as a fine example of what can be achieved when motivated scientists work together openly across boundaries.”

The immunoassay antibodies bind to a tiny part of the hormone called an epitope. Hormones are made up of amino acids, with epitopes making up less than 10 of these blocks. These epitope regions are critical for the molecular-level recognition process and small differences in structure in these regions can have a major effect on the process of recognition and antibody binding.

The team showed how very subtle, atomic level characteristics define the antibody selectivity in closely related epitopes of different proteins. They identified that specific antibodies are highly selective in immunoassays and can distinguish between hCG? and closely related LH fragments.

Understanding these structural differences is likely to be the origin of the observed selectivity in the full hormones. Armed with this knowledge, scientists can develop intelligent epitope selection to achieve clinically relevant assay performance. This means reliable tests can be developed to identify the presence of different hormones – in this case the presence of hCG? which indicates cancer, as opposed to LH, which is always present.

The advances described in this research will enable the development of further immunoassays to identify other biomarkers from similar groups. Pharmaceutical companies could use this to develop new technologies for diagnostics and clinical disease treatments, for example tests for tumours as part of routine screenings.

This work answers one of the big questions in distinguishing biomarkers which are critical for identifying and treating serious diseases. NPL hopes this breakthrough will underpin the development of a range of new diagnostic techniques and treatments.