A novel change in the ABO gene was found in a healthy Chinese person with weak A antigen expression, and laboratory tests found the same exon 2-skipping pattern in that variant and two previously reported changes.
The case began with discordant forward and reverse ABO typing. The person’s cells showed no direct reaction with anti-A, anti-B or anti-A,B reagents, while anti-H reactivity was strong. Reverse testing showed weak A1 and stronger B reactions, and adsorption-elution confirmed weak A. The overall pattern was classified as Ael, a weak A phenotype, with naturally occurring anti-A1.
Long-range PCR and Nanopore sequencing identified a novel ABO*A1.02-like allele, c.29-2A>G, alongside ABO*O.01.02. No other exonic or intronic variants were detected in the haplotype.
The same pattern in a cell assay
The researchers then used splice-site prediction tools and a minigene assay, a laboratory construct used to inspect RNA splicing. They compared a wild-type construct with three mutant constructs: the novel c.29-2A>G variant and the reported c.29-3C>G and c.29-10T>A variants. Sanger sequencing verified the constructs before they were transiently transfected into HEK293T cells.
Computational predictions pointed to a common outcome: exon 2 skipping, with a 70-base deletion, for all three variants. The c.29-10T>A change was also predicted to activate a novel acceptor site, adding 8 bases. The tools showed that the canonical S3 signal was lost or weakened for the tested variants, while an S2 signal appeared only with c.29-10T>A.
The cell assay supported those predictions. On an RT-PCR gel, the wild-type product measured 238 bp, compared with 168 bp for each mutant; c.29-10T>A also produced a 246-bp product. Capillary electrophoresis measured 236.29 bp for wild type and 167.85 bp for c.29-2A>G and c.29-3C>G, with an additional 245.19-bp transcript for c.29-10T>A. No normal mutant transcript was detected. Sanger sequencing confirmed exon 2 skipping in every mutant and an additional 8-bp intronic retention in c.29-10T>A, corresponding to the shared 70-bp deletion.
That shared molecular pattern offers a possible explanation for the attenuated A antigen expression seen in the case. But the study did not analyze endogenous ABO transcripts, so the minigene results do not establish how the variants are spliced in the person.
A tentative map of variant sensitivity
Beyond the three tested variants, the researchers ran computational saturation mutagenesis across positions 1 to 20 upstream of the acceptor. The scan found that T substitutions outside positions 1 to 3 and all substitutions at position 15 preserved wild-type splicing. Among predicted disruptive changes, exon 2 skipping was common, while some variants created novel acceptor or alternative donor events.
The resulting pattern depended on both position and the substituted base. Positions 1 and 2 were most sensitive, position 3 depended on which base was introduced, position 4 was tolerant, positions 5 to 14 were variable, and positions 15 to 20 returned toward baseline. T and C substitutions generally retained higher predicted stability than A and G substitutions, with T slightly above C.
A separate regression analysis of 10 wild-type T loci found positive associations between positional distance and prediction scores for T>A and T>G changes. The reported goodness-of-fit values were 0.8140 for T>A and 0.9119 for T>G. For T>C, the correlation was significant in ASSP and SpliceRover but not NetGene.
What the evidence does not yet show
The limits are substantial. The human component consisted of one person, and fresh blood was unavailable for endogenous ABO RT-PCR and flow-cytometric measurement of antigen density. The authors also say the computational dataset was too small to establish a generalizable model.
Taken together, the findings support a shared splicing pattern among the three tested variants, but they do not establish that c.29-2A>G alone accounts for the Ael phenotype or that the position-and-base framework will hold broadly. More variants and functional assays will be needed to test that proposal.
Paper data and sources
Original title: Three ABO 3′ splice-site variants including a novel one converge on a common splicing defect to attenuate blood group antigen expression
Authors: Guan H, Yang Y, Li J et al.
Journal/Repository: Frontiers in immunology
Status: Peer-reviewed
First online: 2026-10-19
DOI: Not available
Original paper · Full text