Anti-glycan IgM repertoires in newborn human cord blood

Abstract

Natural antibodies are an innate-like subset of serum antibodies involved in host defense, tumor surveillance, homeostasis, and autoimmunity. Defining the natural antibody repertoire is critical for identifying biomarkers, developing vaccines, controlling and preventing autoimmunity, and understanding the development and organization of the immune system. While natural antibodies to protein antigens have been studied in depth, little is known about natural antibodies to carbohydrate antigens. To address this, we profiled IgM from umbilical cord blood and matched maternal sera on a glycan microarray. Since standard methods to detect maternal contamination in cord serum did not have sufficient sensitivity for our study, we developed a highly sensitive microarray-based assay. Using this method, we found that over 50% of the cord samples had unacceptable levels of maternal contamination. For the cord samples with high purity, anti-glycan IgM antibodies were prevalent and recognized a broad range of non-human and human glycans. Using principal component analysis and hierarchical clustering, cord IgM repertoires showed a high degree of similarity with each other but were distinct from maternal IgM repertoires. Our results demonstrate that many anti-glycan antibodies in human serum are natural antibodies and provide new insights into the development of anti-glycan antibody repertoires.

Fig 1. Relationship between cord IgA level and Pearson coefficients for cord-maternal IgM profiles.

A) IgA levels in cord samples. B) Plot of IgA level (y-axis) versus Pearson correlation of cord and maternal IgM signals measured on the array (x-axis). Solid dots represent samples with low IgA level and circles represent samples with high IgA level. Linear regression was fitted using samples with high IgA level. Also shown are representative scatter plots of IgM signals for cord serum (x-axis) versus matched maternal serum (y-axis) for a contaminated cord sample (C) and pure cord sample (D).

Fig 2. Evaluation of maternal contamination.

Two pairs of cord-maternal sera were used. Pair #3 had dissimilar cord and maternal IgM profiles (top panel). Pair #11 had highly similar cord and maternal IgM profiles (bottom panel). Cord samples were spiked in with 0.01%, 0.1% and 1% of their corresponding maternal sera; IgM antibodies in the resulting mixture was assayed with our glycan microarray. X-axis: IgM signals in maternal sera (in Log2 scale); Y-axis: The change of IgM antibody signals in maternal-spiked cord relative to cord alone (in Log2 scale). Dotted line indicates a 3-fold cutoff.

Fig 3. Impact of 0.1% maternal contamination on cord IgM antibody profiles.

The bar graph compares RFU signals for 30 array components before and after addition of 0.1% maternal serum sample M4 into cord sample C4. All signals below 150 are set to 150 (floor = 150 RFU).

Fig 4. Effects of adding maternal serum to cord serum samples.

Cord sera were profiled with and without adding 0.1% matched maternal serum; IgM antibodies were profiled on glycan microarray. X-axis: IgM signals in maternal (in Log2 scale); Y-axis: The change of antibody signals in maternal-spiked cord relative to cord alone (in Log2 scale). Dotted line indicates a 3-fold cutoff. Panel A: Cord samples with high purity (i.e. many points above the threshold); panel B: cord samples estimated to have 0.03–0.1% maternal contamination level (i.e. a few points above the threshold); panel C: cord samples estimated to have >0.1% maternal contamination level.

Fig 5. Heatmap of cord IgM signals.

Each row represents a cord sample and each column represents an antigen on the microarray. Antigens are grouped by family. The magnitude of IgM signals is denoted by different colors: black (no/low signal), yellow (signals with medium intensity) and red (signal with high intensity). A positive signal is defined if the fluorescent intensity is 2-fold above background (≥ 8.8 at Log2 scale).

Fig 6. Principle component analysis of cord IgM vs. maternal IgM.

PCA analysis of cord and maternal RFU signals on the array. Blue are cord samples. Red are maternal samples.

Fig 7. Clustering of quantile normalized cord and maternal IgM.

Hierarchical clustering was carried out for both samples and array components using average linkage and Euclidean distance metric. Cord and maternal IgM separated into two distinct groups. Data for each serum sample is presented in rows. Samples are labeled with “C” for cord or “M” for maternal followed by the sample number. Data for each glycan is shown in columns.