Individual recombinant repeats of MUC16 display variable binding to CA125 antibodies

Immunoaffinity-free chromatographic purification of ovarian cancer biomarker CA125 (MUC16) from blood serum enables mass spectrometry characterization

The enrichment of trace proteins from human fluid samples is of great importance in diverse clinical and industrial applications. In clinical diagnostics, such enrichment may enable detection of trace proteins that serve as biomarkers of disease. Affinity-based approaches, such as immunoaffinity pulldown, are widely used to enrich trace proteins, but this strategy relies on the availability and performance of antibodies that act on all proteoforms in an unbiased manner. Our prior work to characterize MUC16 (the mucin protein that carries the ovarian cancer biomarker CA125) by mass spectrometry successfully overcame the reliance on affinity-based enrichment and was used to enrich this biomarker from ascites of individual ovarian cancer patients, however, this strategy was not demonstrated on clinically relevant volumes of serum, a biofluid that is more accessible than ascites. The present work developed a non-affinity-based chromatographic method to enrich MUC16 from serum. The enriched MUC16 sample was further processed using a Midi Top 14 abundant protein depletion column. Peptides identified using bottom-up proteomics yielded 1-8% coverage of MUC16. Additionally, MUC16 was detected in samples containing less than the clinical cut-off level of CA125 (35 U mL-1), suggesting that this strategy of enrichment and bottom-up proteomics can enable analysis of CA125 from the serum of individuals with early-stage ovarian cancer and those whose tumors express CA125 (MUC16) at low levels.

A Revised Molecular Model of Ovarian Cancer Biomarker CA125 (MUC16) Enabled by Long-read Sequencing

The biomarker CA125, a peptide epitope located in several tandem repeats of the mucin MUC16, is the gold standard for monitoring regression and recurrence of high-grade serous ovarian cancer in response to therapy. However, the CA125 epitope along with several structural features of the MUC16 molecule are ill defined. One central aspect still unresolved is the number of tandem repeats in MUC16 and how many of these repeats contain the CA125 epitope. Studies from the early 2000s assembled short DNA reads to estimate that MUC16 contained 63 repeats.Here, we conduct Nanopore long-read sequencing of MUC16 transcripts from three primary ovarian tumors and established cell lines (OVCAR3, OVCAR5, and Kuramochi) for a more exhaustive and accurate estimation and sequencing of the MUC16 tandem repeats.The consensus sequence derived from these six sources was confirmed by proteomics validation and agrees with recent additions to the NCBI database. We propose a model of MUC16 containing 19-not 63-tandem repeats. In addition, we predict the structure of the tandem repeat domain using the deep learning algorithm, AlphaFold.The predicted structure displays an SEA domain and unstructured linker region rich in proline, serine, and threonine residues in all 19 tandem repeats. These studies now pave the way for a detailed characterization of the CA125 epitope. Sequencing and modeling of the MUC16 tandem repeats along with their glycoproteomic characterization, currently underway in our laboratories, will help identify novel epitopes in the MUC16 molecule that improve on the sensitivity and clinical utility of the current CA125 assay.

SIGNIFICANCE

Despite its crucial role in clinical management of ovarian cancer, the exact molecular sequence and structure of the biomarker, CA125, are not defined. Here, we combine long-read sequencing, mass spectrometry, and in silico modeling to provide the foundational dataset for a more complete characterization of the CA125 epitope.

Strategies for Mitigating Commercial Sensor Chip Variability with Experimental Design Controls

Surface plasmon resonance (SPR) is a popular real-time technique for the measurement of binding affinity and kinetics, and bench-top instruments combine affordability and ease of use with other benefits of the technique. Biomolecular ligands labeled with the 6xHis tag can be immobilized onto sensing surfaces presenting the Ni2+-nitrilotriacetic acid (NTA) functional group. While Ni-NTA immobilization offers many advantages, including the ability to regenerate and reuse the sensors, its use can lead to signal variability between experimental replicates. We report here a study of factors contributing to this variability using the Nicoya OpenSPR as a model system and suggest ways to control for those factors, increasing the reproducibility and rigor of the data. Our model ligand/analyte pairs were two ovarian cancer biomarker proteins (MUC16 and HE4) and their corresponding monoclonal antibodies. We observed a broad range of non-specific binding across multiple NTA chips. Experiments run on the same chips had more consistent results in ligand immobilization and analyte binding than experiments run on different chips. Further assessment showed that different chips demonstrated different maximum immobilizations for the same concentration of injected protein. We also show a variety of relationships between ligand immobilization level and analyte response, which we attribute to steric crowding at high ligand concentrations. Using this calibration to inform experimental design, researchers can choose protein concentrations for immobilization corresponding to the linear range of analyte response. We are the first to demonstrate calibration and normalization as a strategy to increase reproducibility and data quality of these chips. Our study assesses a variety of factors affecting chip variability, addressing a gap in knowledge about commercially available sensor chips. Controlling for these factors in the process of experimental design will minimize variability in analyte signal when using these important sensing platforms.