Abstract
Natural antibodies (nAbs) are immunoglobulins that arise without prior exposureto exogenous antigens, providing immediate defense against pathogens and contributing to
tissue homeostasis, clearance of apoptotic cells, and regulation of inflammation. Despite
these roles, nAbs remain understudied relative to conventionally adaptive antigen-specific
antibodies, partly due to their polyreactivity, low affinity, and complex regulation. Among
nAbs, phospholipid-reactive antibodies constitute a key subset recognizing endogenous
lipid antigens, including phosphatidylcholine (PtC), an essential phospholipid commonly
found in cell membranes. Previous sequencing of PtC-reactive (PtC+) antibodies using the
VH12 heavy chain revealed distinct skewing of light chain gene usage. Studying PtC+
nAbs is difficult because their polyreactivity and low affinity makes distinguishing specific
binding from artifact difficult in traditional immunoassays. Computational methods offer
alternative models. Structural modeling and in silico simulations enable prediction and
visualization of antibody-antigen interactions to decipher modes of binding, affinities, and
specificity determinants of PtC+ nAbs. Here we present evidence that KV4-91 KJ gene
usage skews away from [KJ1] due to not forming an accessible binding site and towards
[KJ5] in receptor editing deficient mice because it creates a nAb with a higher binding
affinity to PtC. Lastly, we show that antibody-PtC binding is directed towards the
phospholipid headgroup. Computational approaches can reveal patterns in antibody
repertoires, identify key structural motifs, and guide future experimental validation.