Cancer biomarkers defined by autoantibody signatures to aberrant O-glycopeptide epitopes

Abstract

Autoantibodies to cancer antigens hold promise as biomarkers for early detection of cancer. Proteins that are aberrantly processed in cancer cells are likely to present autoantibody targets. The extracellular mucin MUC1 is overexpressed and aberrantly glycosylated in many cancers; thus, we evaluated whether autoantibodies generated to aberrant O-glycoforms of MUC1 might serve as sensitive diagnostic biomarkers for cancer. Using an antibody-based glycoprofiling ELISA assay, we documented that aberrant truncated glycoforms were not detected in sera of cancer patients. An O-glycopeptide microarray was developed that detected IgG antibodies to aberrant O-glycopeptide epitopes in patients vaccinated with a keyhole limpet hemocyanin-conjugated truncated MUC1 peptide. We detected cancer-associated IgG autoantibodies in sera from breast, ovarian, and prostate cancer patients against different aberrent O-glycopeptide epitopes derived from MUC1. These autoantibodies represent a previously unaddressed source of sensitive biomarkers for early detection of cancer. The methods we have developed for chemoenzymatic synthesis of O-glycopeptides on microarrays may allow for broader mining of the entire cancer O-glycopeptidome.

Fig. 1. Microarray platform for detection of MUC1 O-glycopeptide specific antibodies. Immunization of cancer patients with GalNAc-MUC1 break tolerance eliciting glycopeptide specific antibodies

A, GalNAc glycosylated variants of MUC1 (6Tn, 9Tn, and 15Tn) were synthesized chemoenzymatically using recombinant polypeptide GalNAc-transferases (GalNAc-T2, -T4, and -T11) (yellow boxes indicate the position of GalNAc in each of the three MUC1 repeats in the produced 60-mer MUC1 (glyco)peptides). Further elongation was carried out by β3GnT6, c1Gal-T, or ST6GalNAc-T1 creating truncated core 3, core 1, and STn structures respectively. MUC2 glycopeptides were synthesized by similar methods. Glycopeptides and control structures were printed on NHS-activated hydrogel slides in quadruplicates at three different concentrations. The glycopeptide array was validated by incubation with glycan and glycopeptide specific MAbs detected by Cy3-conjugated secondary antibodies and expressed as fluorescence intensity visualized in column diagram. Anti-MUC1 peptide MAb HMFG2 recognized both non-glycosylated (designated M1) and all glycoforms of MUC1 confirming efficient printing of all MUC1 compounds. The glycopeptide specific MAb 5E5 recognizes the immunodominant Tn/STn-MUC1 epitope -GSTAP- with one GalNAc residue in the threonine (6Tn-MUC1) or both the serine and threonine (9/15Tn-MUC1). In contrast the carbohydrate hapten antibodies recognizing Tn (GalNAcα1-O-Ser/Thr) MAb (5F4), STn (NeuAcα2-6GalNAcα1-O-Ser/Thr) MAb (3F1) react regardless of peptide backbone as shown by reactivity with the corresponding glycoforms of MUC2 (designated M2) and the Tn and STn glycosylated mucins AOSM and OSM. Finally the anti-T (Galβ1-3GalNAcα1-O-Ser/Thr) MAb reacts specifically with T-MUC1. B, 3D-column diagram (above) and dot-plot presentation (below) showing results of analysis of human sera from a clinical trial of a MUC1 glycopeptide vaccine in breast cancer patients (n=20) immunized with 106-mer MUC1 tandem repeat peptide with 25 Tn O-glycans conjugated to KLH. The immunogen corresponds to the 15Tn MUC1 60-mer glycopeptide used in this study with Tn residues at all five potential O-glycosylation sites in the 20-mer tandem repeated peptide. The array was incubated with diluted (1:25) serum followed by incubation with IgG and IgM specific secondary Cy3 conjugated antibodies. Results from all twenty subjects analyzed for IgG shown in rows 1–20 with glycopeptides as indicated. Pre-vaccination sera showed essentially no IgG antibodies, while vaccination with the 25Tn MUC1 vaccine induced IgG antibodies specifically reactive with the 9Tn and 15Tn MUC1 glycopeptides but not unglycosylated MUC1. Results from one subject analyzed for IgM is shown in last row. Pre-immune and vaccinated patients have abundant IgM antibodies to Tn, STn, and T carbohydrate haptens. This underscores the importance of selective detection of IgG responses in order to identify true combined glycopeptide epitopes. C, Summary of the results with indication of the number of patients tested positive for each target out of the total number of tested patients. Values above 3 times the SD of the average values obtained with sera before vaccination were considered positive. A total of 18 out of the 20 subjects vaccinated showed induction of IgG auto-antibodies to 9Tn and 15Tn MUC1 glycopeptides. These glycopeptides have in common two O-glycans in the immunodominant -GSTAP- epitope suggesting that this is the reactive epitope. Partial reactivity was seen with STn-MUC1 and core 3 MUC1 in 5 and 7 immunized patients, respectively. No reactivity was seen with Tn and STn MUC2 glycopeptides or AOSM and OSM. The graphic presentation of MUC1 glycoforms shows one 20 amino acid MUC1 tandem repeat sequence (HGVTSAPDTRPAPGSTAPPA) with glycans using symbols as in legend to Supplementary Fig. S1.

Fig. 2. Detection of cancer-induced auto-antibodies to MUC1 O-glycopeptide epitopes by microarray analysis

A, 3D-column diagram of IgG antibody reactivity in sera from healthy controls (n=39) and newly diagnosed patients with breast (n=26), ovarian (n=20) and prostate (n=10) cancer (Supplementary Table S1). Specific IgG responses were detected in identified breast, ovarian and prostate cancer patients towards Tn-MUC1, STn-MUC1, T-MUC1 or truncated core 3 O-glycopeptide epitopes. Only very few healthy subjects and cancer patients had IgG antibodies to AOSM and OSM, which are considered Tn and STn hapten antibodies. B, Dot-plot analysis of the antibody responses for each glycopeptide target. C, Summary of results for each MUC1 glycopeptide with indication of glycan structures and positions within the MUC1 tandem repeat. Values were considered positive if above 3 times the SD of the average value obtained with sera from healthy individuals. A total of five out of the 26 breast cancer patients showed induction of IgG auto-antibodies to either Tn, STn, T or core 3 MUC1 glycopeptides. The same reactivity pattern was seen in ovarian cancer patients with five out of 20 patients demonstrating IgG auto-antibodies to one or more of the MUC1 glycopeptides. In contrast 4 out of 10 prostate cancer patients showed induction of auto-antibodies to STn, T and, core 3 MUC1 glycopeptides. In the majority of individuals no reactivity was seen with Tn and STn MUC2 glycopeptides or AOSM and OSM.