In-Depth Characterisation of Heavily Glycosylated Proteins and Their Potential as Colorectal Cancer Biomarkers Valeriia Kuzyk (Presenter) Vrije Universiteit Amsterdam Presenter Bio: Completed Bachelors degree in Biology and Master degrees in Biochemistry and Bioinformatics. Research interests: glycomics, proteomics, mass-spectrometry, capillary electrophoresis, biomarker research, method development.

Authors: Valeriia Kuzyk (1,2), Guinevere S.M. Kammeijer (2), Rob Haselberg (1), Manfred Wuhrer (2), Govert W. Somsen (1)
(1) Vrije Universiteit Amsterdam, Division of Bioanalytical Chemistry, Amsterdam, The Netherlands, (2) Leiden University Medical Center, Center for Proteomics and Metabolomics, Leiden, The Netherlands

Protein glycosylation is known to rapidly change in disease, as well as in cancer progression. Therefore, heavily glycosylated tumor-derived circulating proteins hold a great potential to perform as “disease ambassadors” in the bloodstream, enabling more sensitive and specific diagnostics. However, proper characterization is challenging due to low quantity in serum and the high heterogeneity of the glycan distribution. Here, we present a comparison between different analytical platforms for the in-depth characterization of two glycoproteins (carcinoembryonic antigen and laminin B1). Further studies will examine if altered glycosylation profiles of these glycoproteins can be used as prognostic tools for colorectal cancer.

Introduction

Aiming for better diagnostic tools in cancer, researchers target the changes in expression of circulating proteins. However, these changes may occur due to various reasons and often do not provide the direct evidence for disease progression. That results in poor sensitivity and specificity of ELISA-based diagnostic assays. The ELISA test on serum carcinoembryonic antigen (CEA) concentration is currently used in clinics and, predominantly, is only conclusive in the late stages when the disease is hardly curable. To add, the test is incapable to determine the disease prognosis and tunor malignancy potential. Glycosylation changes dramatically in various diseases, especially in cancer [1]. Moreover, glycosylation variations described so far are specific for the type and the stage of cancer. With that in mind, we expect the glycoproteins, that are overexpressed by tumor tissue, to reflect its biology by the glycan pattern. The more glycosylation sites are present, the higher the chance is for changes in glycosylation to be found and detected. If these proteins are shed into circulation, they may serve as “tumor markers” in liquid biopsy tests. The proteins that are chosen as CRC candidate biomarkers in this study, are the aforementioned CEA and laminin B1 (LAMB1). CEA is a recognized serum CRC biomarker, that is mostly produced by the tumor. It carries 28 potential N-linked glycosylation sites. LAMB1 is an integral part of the structural scaffolding for the tissue growth. It is produced by many tissues, including carcinomas, and is overexpressed in CRC [2]. Moreover, its glycosylation (11 potential N-glycosylation sites) hasn’t been explored so far.

Methods

In this work we present several analytical methods that provide in-depth information about the N-linked glycosylation profiles of heavily glycosylated proteins (CEA and LAMB1). We have tested several affinity purification protocols by spiking CEA or LAMB1 standards into cell secretomes and serum. NHS-terminated magnetic beads covered with protein-specific nanobodies yielded the highest recovery rates (>60%). N-glycans were either released with PNGase-F (for glycan analysis) or proteins were digested by combining the enzymes trypsin and chymotrypsin for CEA and trypsin and GluC for LAMB1 (for glycopeptide profiling). The overall released N-glycan patterns were assessed by MALDI-MS and HILIC-MS/MS mass-spectrometry based analytical techniques. HILIC-MS/MS resolved a larger number of glycoforms (50 versus 66 N-glycans for LAMB1, and 84 versus 113 for CEA), whereas MALDI-MS was able to discriminate between α2,3 and α2,6- sialylated glycans due to an additional derivatization step. To get the information about the glycan site-specific distribution we tuned the combined protease digestion to result in single N-glycosylation site per peptide, making the proper data interpretation possible. We have performed the glycopeptide analysis with capillary electrophoresis coupled to mass spectrometry by electrospray ionization with a sheathless interface (CE-ESI-MS) and the addition of a dopant enriched nitrogen gas [3]. This extremely sensitive technique allowed us to reach a lower limit of detection - starting concentration of 250 ng/mL (prior to the capturing step).

Results

The overall glyco(peptide) profile of the CRC tumor derived CEA standard revealed known cancer related signatures, such as high level of branching in complex N-glycans (38% of all complex type glycans), abnormally high fucosylation (83% of all glycan species), presence of sialyl Lewis X/A antigens and high-mannose glycans, as well as truncated glycan species and paucimannose glycans. Moreover, the distribution of the N-glycan species appeared to be site-specific with some glycosylation sites offering more microheterogeneity than others and some potential N-glycosylation sites appearing to be non-glycosylated. Furthermore, some glycopeptides could not be mapped on the canonical CEA sequence, suggesting possible mutations and/or splice variants, that can also bear a potential diagnostic value [4]. The glycoprofile of non-tumor-derived LAMB1 standard did not reveal any cancer-related features – with the prevalence of the complex type structures and moderate amount of fucosylation and bisection. It should be noted, that to the best of our knowledge, this is the first N-glycan profile of LAMB1 reported and it will serve as a reference for its cancer variant characterization.

Conclusions & Discussion

To summarize, we have established the capturing of heavily glycosylated proteins (CEA and LAMB1) from cell secretome, with some extra optimization steps needed for more protein-rich biofluids (e.g. abundant proteins depletion and more extensive washing protocols). We have explored several analytical platforms to identify the N-linked glycosylation on low-abundant serum proteins. Each method seems to be complementary with each other, for example, MALDI-MS appeared to be the most promising platform regarding high-throughput analysis but might not provide the necessary sensitivity and resolution, that HILIC-MS is capable of. Furthermore, glycopeptide analysis with CESI-MS/MS provides information about site-specific N-glycan distribution, offering great sensitivity and resolving power. Depending on the research question, each of these platforms may be a good choice to assess the N-linked glycosylation of putative serum candidate biomarkers. In the nearest future, we expect to apply the developed workflows on CRC cells and secretomes with further attempt to link glycosylation changes to disease progression. In case of success, these findings will be tested and validated on CRC patient cohorts.

This work was done within AIMMS PhD training program and is a part GlyCoCan consortium (Marie Curie European Training Network, Horizon 2020 programme under grant agreement number 676421).

1. Holst, S., Wuhrer, M., & Rombouts, Y. (2015). Glycosylation Characteristics of Colorectal Cancer. Advances in Cancer Research Glycosylation and Cancer, 203-256. doi:10.1016/bs.acr.2014.11.004

2. Lin, Q., Lim, H. S., Lin, H. L., Tan, H. T., Lim, T. K., Cheong, W. K., . . . Chung, M. C. (2015). Analysis of colorectal cancer glyco-secretome identifies laminin β-1 (LAMB1) as a potential serological biomarker for colorectal cancer. Proteomics, 15(22), 3905-3920. doi:10.1002/pmic.201500236

3. Kammeijer GSM, Kohler I, Jansen BC, et al. Dopant Enriched Nitrogen Gas Combined with Sheathless Capillary Electrophoresis–Electrospray Ionization-Mass Spectrometry for Improved Sensitivity and Repeatability in Glycopeptide Analysis. Analytical chemistry 2016;88(11):5849-56.

4. Hatakeyama, K., Wakabayashi-Nakao, K., Ohshima, K., Sakura, N., Yamaguchi, K., & Mochizuki, T. (2013). Novel protein isoforms of carcinoembryonic antigen are secreted from pancreatic, gastric and colorectal cancer cells. BMC Research Notes, 6(1), 381. doi:10.1186/1756-0500-6-381