Epigenetic Dissection of Human Blood Group Genes Reveals Unknown Regulatory Elements and Detailed Characteristics of KEL and Other Loci

Accurate blood group typing is essential for compatible blood transfusions and can often be achieved with the combination of serology and genotyping technologies. Currently, most genotyping platforms focus on variants in coding regions. In addition, variants in the non-coding region that can disrupt regulatory motifs in the promoter or enhancer regions are also important for blood group typing, as demonstrated in ABO, FY and XG systems for example. These blood group regulatory regions have yet to be fully characterized for all genes. Recently, we systematically explored the GATA1-dependent blood group regulome in silico and found 193 candidate regulators in 33 blood group genes. As proof of concept, we resolved in vitro the GATA1-dependent mechanism underlying the Helgeson phenotype of the Knops system, where there is very low expression of Complement Receptor 1 (CR1) on red blood cells.

Aiming to improve the precision of our global blood group gene regulator prediction, we expanded the analysis to include preexisting epigenetic datasets such as ChIPseq with erythroid transcription factors (TFs; GATA1, KLF1, RUNX1, NFE2), histone modifications (H3K27ac, H3K4me1, H3K4me3), and ATAC-seq.

In total, 96 candidate regions were identified where at least 2 TFs were bound. Four genes (CR1, EMP3, ABCB6 and ABCC4) showed co-localization of ChIPseq peaks for all 4 TFs within their introns 37, 4, 1 and 1, respectively. Functional analysis with luciferase assay of these potential enhancers with their respective promoters showed different levels of enhancement.

Among the more well-known and clinically important blood groups, the KEL proximal promoter was shown to co-localize GATA1 and KLF1. Previous studies suggested the presence of 3 GATA1-binding sites and an Sp1 site, but KLF1 was novel to our knowledge. Moreover, decreased KEL expression has been reported in a KLF1-null human and in mouse knockouts. Candidate GATA1 sites near the KLF1 motif were observed, with the GATA1 motif closest to the KLF1 motif having the highest motif score in JASPAR (939 vs. 913 for the 5’ and 851 for the 3’ GATA1 motifs using GATA1 logotype MA0035.3). In light of this, and the importance of KEL for transfusion and fetomaternal incompatibility, we carefully examined the KEL promoter region.

A decrease in JASPAR score was used to select two natural variants that disrupt TF binding at the GATA1 and KLF1 motifs compared to the wild-type (wt): rs1204402835:A>G for GATA1 (939>815); rs1442994069:G>A for KLF1 (924>801, KLF1 logotype: MA0493.2). We showed binding of KLF1 and GATA1 to the −35 to +1 bp region (NM_000420.3) with electrophoretic mobility shift assay (EMSA) and mass spectrometry (MS) using synthetic probes and nuclear extract from HEL cells. Disruption of the KLF1 and/or GATA1-binding motifs with the aforementioned natural variants led to the loss of bands which, compared to the wt, represented nuclear extract binding to the probes in the EMSAs. The decreased abundance of bound TFs was confirmed by MS (−2.38 for KLF1 and −1.76 for GATA1, log₂-fold changes compared to wt). Luciferase assays of the KEL promoter exhibited the highest activity (set to 100%) with wt GATA1 and KLF1 motifs and omitting Sp1 and 3’-side GATA1 motifs (Figure 1). The whole region including Sp1 and 3’ GATA1 sites gave 44% activity whilst the 3’ GATA1 alone or with Sp1 site was <10% (similar to vector control). The promoter construct containing the 5’ GATA1 site and the central GATA1 and KLF1 sites showed the highest activity, but reduced activity was observed with disrupted GATA1 motif at 44% and KLF1 motif at 48%. Activity decreased to 17% when both variants were present (Figure 1).

In summary, we used a combined bioinformatics and experimental approach to unravel regions with potential to regulate blood group expression and explain weak phenotypes that may escape detection in clinical practice. Our studies are limited by the number of available datasets for relevant TFs but were able to detect regions where these TFs co-localize in CR1, EMP3, ABCB6 and ABCC4. Furthermore, we characterized the clinically important KEL blood group gene promoter in greater detail indicating that the central GATA1 motif is the most crucial. For the first time, we described that KLF1 and GATA1 jointly enhance KEL transcription. Finally, our updated pipeline provides multiple candidate regulatory regions to be explored to improve knowledge on the blood group regulome.