Hunting for the elusive target antigen in gestational alloimmune liver disease (GALD)

Abstract

The prevailing concept is that gestational alloimmune liver disease (GALD) is caused by maternal antibodies targeting a currently unknown antigen on the liver of the fetus. This leads to deposition of complement on the fetal hepatocytes and death of the fetal hepatocytes and extensive liver injury. In many cases, the newborn dies. In subsequent pregnancies early treatment of the woman with intravenous immunoglobulin can be instituted, and the prognosis for the fetus will be excellent. Without treatment the prognosis can be severe. Crucial improvements of diagnosis require identification of the target antigen. For this identification, this work was based on two hypotheses: 1. The GALD antigen is exclusively expressed in the fetal liver during normal fetal life in all pregnancies; 2. The GALD antigen is an alloantigen expressed in the fetal liver with the woman being homozygous for the minor allele and the father being, most frequently, homozygous for the major allele. We used three different experimental approaches to identify the liver target antigen of maternal antibodies from women who had given birth to a baby with the clinical GALD diagnosis: 1. Immunoprecipitation of antigens from either a human liver cell line or human fetal livers by immunoprecipitation with maternal antibodies followed by mass spectrometry analysis of captured antigens; 2. Construction of a cDNA expression library from human fetal liver mRNA and screening about 1.3 million recombinants in Escherichia coli using antibodies from mothers of babies diagnosed with GALD; 3. Exome/genome sequencing of DNA from 26 presumably unrelated women who had previously given birth to a child with GALD with husband controls and supplementary HLA typing. In conclusion, using the three experimental approaches we did not identify the GALD target antigen and the exome/genome sequencing results did not support the hypothesis that the GALD antigen is an alloantigen, but the results do not yield basis for excluding that the antigen is exclusively expressed during fetal life., which is the hypothesis we favor.

Fig 1. Overview of the three-pronged approach.

A three-pronged approach was used to investigate the two hypotheses addressing the identity of the target antigen of GALD antibodies. Not all samples were investigated in all 3 experimental approaches. No convincing leads were identified from the 3 different experimental approaches.

Fig 2. Exome analysis pedigree.

The pedigrees of the three families included in this study for the exome analysis are shown. DNA from the fathers and mothers was used in the exome analysis as well as DNA from the child II.8. For children II.6 and II.7 the mother was treated with IVIG during pregnancy. The diagnosis of GALD was clinically verified for the affected and deceased children.

Fig 3. Immunoprecipitaion from Huh-7.

Normalized log₂ intensities of Primary Biliary Cirrosis (PBC) antigens in IPs with Huh-7 cell line.

Fig 4. Immunoprecipitation from human fetal liver.

Normalized log₂ Label Free Quantification (LFQ) intensities of PBC antigens in IPs with fetal liver material.

Fig 5. Principle components 1 and 2 of PCA of MS results of IPs with Huh-7 cell line.

The PBC sample and the GALD samples were prepared with EDTA while the healthy samples were prepared with citrate.

Fig 6. Differential protein abundance in GALD plasma vs healthy controls in Huh-7 cell line.

A) Volcano plot depicting differential abundance of proteins in IPs with GALD plasma vs healthy plasma with Huh-7 cell line. Proteins with Benjamin-Hochberg adjusted p-value (FDR) < 0.05 in orange: Complement C3 (C3), Complement C4-A (C4A), Complement C4-B (C4B), Monocarboxylate transporter 1 (SLC16A1), Tropomyosin beta chain (TPM2), Complement C1q subcomponent subunit B (C1QB). B) Normalized log₂ LFQ intensities of significantly differentially abundant proteins in 3 conditions GALD, healthy and PBC plasma. Number of unique peptides in brackets. LogFC = log Fold Change.

Fig 7. Differential protein abundance in GALD plasma vs healthy controls in fetal liver.

A, B, C) Volcano plots depicting differential abundance of proteins in IPs with GALD plasma vs healthy plasma in IP1, IP2 and IP3 respectively. Proteins with Benjamin-Hochberg adjusted p-value (FDR) < 0.05 in orange: Ig lambda chain V-II region TOG (P01704) and Carbamoyl-phosphate synthase 1, mitochondrial (CPS1). Proteins with log₂ Fold Change > 2 are labeled with gene name or UniProt ID. Plotted below are average log₂ Fold Change of proteins present in either GALD plasma or healthy controls and in at least 2 samples and the lower 0.1 percentile of normalized LFQ intensities in the respective IP. Top 5 proteins (Fold Change) are labeled. D) Normalized log2 LFQ intensities of CPS1 in 3 conditions GALD, healthy and PBC plasma in 3 IPs. LogFC = log Fold Change. https://github.com/HenriettaHolze/GALD_proteomics/blob/main/scripts/fetal_liver_tissue_analysis.Rmd.