The DAU cluster: a comparative analysis of 18 RHD alleles, some forming partial D antigens

Abstract

Background

The Rh system is the most complex and polymorphic blood group system in humans with more than 460 alleles known for the RHD gene. The DAU cluster of RHD alleles is characterized by the single nucleotide change producing the p.Thr379Met amino acid substitution. It is called the DAU-0 allele and has been postulated to be the primordial allele, from which all other alleles of the DAU cluster have eventually evolved.

Study design and methods

For 2 novel DAU alleles, the nucleotide sequences of all 10 exons as well as adjacent intronic regions, including the 5’ and 3’ untranslated regions (UTR), were determined for the RHD and RHCE genes. A phylogenetic tree for all DAU alleles was established using the neighbor-joining method with Pan troglodytes as root. Standard hemagglutination and flow cytometry tests were performed.

Results

We characterized 2 DAU alleles, DAU-11 and DAU-5.1, closely related to DAU-3 and DAU-5 respectively. A phylogenetic analysis of the 18 known DAU alleles indicated point mutations and interallelic recombination contributing to diversification of the DAU cluster.

Conclusions

The DAU alleles encode a group of RhD protein variants, some forming partial D antigens known to permit anti-D in carriers; all are expected to cause anti-D alloimmunization in recipients of red cell transfusions. The DAU alleles evolved through genomic point mutations and recombination. These results suggest that the cluster of DAU alleles represent a clade, which is concordant with our previous postulate that they derived from the primordial DAU-0 allele.

Background

The D antigen, encoded by the RHD gene, is the second most immunogenic and clinically significant blood group antigen, next only to A and B antigens.¹ Genetic rearrangements in the RHD gene, such as gene conversions and point mutations, led to a great variety of RHD alleles, more than 460 of which have been identified to date. They encode variants with either a normal or an altered RhD protein expression on the red blood cell (RBC) surface.^2–6

In phylogenetic analysis, a group of sequences defined by a common DNA mutation are known as a “cluster”. We have applied the term “cluster” to describe, so far, 4 groups of phylogenetically related RHD alleles.^7–10 A cluster of alleles may represent more specifically a clade, if they originated from a common ancestral allele, known as the primordial allele. ^11,12 The DAU-0 has been the proposed primordial allele of the DAU cluster.⁷ It is characterized by the single nucleotide polymorphism (SNP) c.1136C>T (p.Thr379Met) in exon 8 of the RHD gene.⁷ Only 9 DAU alleles have formally been published between 2002⁷ and 2009,^10,13–15 with documented anti-D in carriers of DAU-3⁷ and DAU-4,¹⁶ although several more DAU allele candidates have since accrued in online repositories.^2,17,18

The DAU-3 partial D allele is defined by 2 non-synonymous mutations (p.Val279Met and p.Thr379Met) in the RHD exons 6 and 8 and comes in a haplotype with an RHce allele lacking further characterization.^7,13,15 The DAU-3 allele has previously been associated with an anti-D immunization in a carrier.⁷ In 2012, the nucleotide sequence of a DAU-3 allele with the additional non-synonymous mutation p.Ala85Val has been deposited (GenBank accession number HE965768.1) in trans to the RHD*Ψ allele, but no serologic data were reported.

The DAU-5 partial D allele is defined by 3 non-synonymous mutations (p.Phe223Val, p.Glu233Gln and p.Thr379Met) in the RHD exons 5 and 8 and is associated with an RHce allele lacking further characterization.^4,10,14,15 DAU-5 is reported to be a recombinant allele between the DAU-0 and DV type 1 alleles^4,14 and has not been associated with anti-D alloimmunizations in carriers. In 2014, the nucleotide sequence of a DAU-5 allele with the additional synonymous mutation p.Ile374Ile has been deposited (GenBank accession number HG918112.1), but no serologic data were reported.

In the present study, we describe these 2 new DAU alleles that are closely related to DAU-3 and DAU-5. We also investigated the phylogenetic relationship among the 18 known DAU alleles and the distribution of their amino acid substitutions in the RhD protein. DAU-0 was confirmed to be the primordial allele for all of them.

Materials and Methods

Study subjects

EDTA-anticoagulated blood samples were obtained from the patients with written informed consent. The DNA was extracted using a BioRobot EZ1 workstation with EZ1 DNA blood kit (Qiagen, Valencia, CA).

Immunohematology

Hemagglutination tests were performed by standard tube and anti-IgG gel matrix testing with licensed reagents (Ortho, Raritan, NJ). Several monoclonal anti-D from RhD typing kits were used to establish the epitope patterns (D Screen; Diagast, Loos, France; and Advanced Partial RhD typing kit; Alba Bioscience, Edinburgh, UK). Additional monoclonal anti-D were BS226 and BS232 (both IgM, Seraclone anti-D (RH1); Bio-Rad, Dreieich, Germany), LDM1, RUM-1 and TH28 (all IgM, DiaClon ABO-Confirmation for Patients; Bio-Rad), P3x61 (IgM, Seraclone Anti-CDE (RH2, 1, 3); Bio-Rad) and TH28 (IgM) and MS26 (IgG) (microtiter plate, Galileo Neo; Immucor, Norcross, GA, USA). Antibody screening and identification were done with gel matrix (rabbit anti-IgG; Micro Typing Systems, Pompano Beach, FL, USA).

Flow cytometry

The D antigen density was estimated by flow cytometry (FACSCalibur; Becton Dickinson, Heidelberg, Germany) with 4 monoclonal anti-D as described previously¹⁹ (Birma D6 and BRAD 3; International Blood Group Reference Laboratory, Bristol, UK; and BS221 and H41; Bio-Rad, Dreieich, Germany). Cryopreserved RBCs with a D+C+E-c+e+ phenotype expressing 14,000 D antigens per RBC was used as reference, which had previously been calibrated by a published workshop standard.²⁰

RHD molecular screening

Initial RHD genotyping was done with kits (BAGene Weak D-TYPE and Partial D-TYPE; BAG Health Care, Lich, Germany).

RHD and RHCE sequencing

The RHD and RHCE genes were sequenced at NIH^21,22 or at Linz²³ as previously described. The nucleotide sequences of all 10 exons as well as the adjacent intronic regions including the 5’ and 3’ untranslated regions (UTR) were determined for both genes. Zygosity testing for the RHD gene was done at NIH by restriction fragment length polymorphism (RFLP)²⁴ and quantitative fluorescence polymerase chain reaction (QF-PCR)²⁵, while in Linz a hybrid Rhesus Box assay was applied (RBC-Ready Gene ZygoFast; Inno-train Diagnostic, Kronberg, Germany).

RH Sequence analysis

Nucleotide sequences were aligned and compared with the RHD (NG_007494.1) and RHCE reference sequences (NG_009208.3). All variations are described according to current mutation nomenclature guidelines,²⁶ ascribing the A of the first ATG translational initiation codon as nucleotide +1 in the mRNA coding region of RHD (NM_016124.4) and RHCE (NM_020485.4). Multiple sequence comparisons were carried out (MUSCLE, v3.8 with default settings).²⁷

Database mining for DAU alleles

The human RhesusBase² and NCBI GenBank²⁸ genetic sequence databases were searched for RHD alleles fitting the definition of the DAU cluster.⁷ One allele (GenBank accession number EU557240) harboring a codon insertion (GTG) immediately following the start codon (ATG) in addition to p.Thr379Met was excluded from the study, as no 5’UTR or corroborating information could be obtained since its release in 2008.

Reference red cell genotyping

At the BloodCenter of Wisconsin (BCW), genomic DNA was extracted from patient samples and evaluated for SNPs, insertions and deletions associated with non-RhD antigens, including C, E, c and e, and the 7 Rh variant antigens, such as partial C, partial c, partial e, V, VS, hrB and hrS,²⁹ or for partial D (BAGene Partial D-TYPE). To resolve 2 different variant RHD alleles in samples, called RHD compound heterozygotes, the coding sequence of the RHD gene was sequenced in full length³⁰ with RHD-specific intron amplification primers. Results of all samples, sent between February 1, 2013 and February 15, 2016 by 20 outside institutions, were collated for reference red cell genotyping. Comparable data sets were established at the institutions in Linz and Springe.

Phylogenetic analysis

A possible phylogenetic tree for DAU alleles was developed, based on the RHD coding sequence and the presence of its associated RHCE allele. Each single nucleotide substitution was counted as one event. Clustering of the described DAU alleles was done manually. Sequences from chimpanzees (Pan troglodytes Rh-like protein IIR, GenBank accession number L37050.1)³¹ were used for external rooting, as previously described for RHD.^7,9,10,32

Computational modeling of RhD protein and amino acid substitutions

The 3D structure for the RhD protein was modeled from the crystal structure of RhCG protein (Protein Data Bank accession code 3HD6)³³ using SWISS-MODEL.³⁴ Stereochemical quality and accuracy of the predicted RhD model was analyzed using Ramachandran plot analysis,³⁵ ProSA^28,29 and ProQ.³⁶ The distances between the C-alpha atoms of the amino acids in the DAU alleles and a line traversing the central pore of the RhD protein were estimated.

Polymorphism Phenotyping algorithm (PolyPhen-2)³⁷, Sorting Intolerant From Tolerant (SIFT)³⁸, Protein Variation Effect Analyzer (PROVEAN)³⁹ and Screening for Non-Acceptable Polymorphisms (SNAP2)⁴⁰ were used to predict the functional impact of amino acid substitutions on RhD protein structure.

Nomenclature

New DAU alleles described in this study were named following the nomenclature in the human RhesusBase.² DAU-0 to DAU-7 had been named previously.^3,10,13–15 A new allele that differed by a non-synonymous substitution from any previously described DAU allele was denoted by a new number, such as DAU-8; whereas a new allele that differed by a synonymous substitution from a previously described allele, for example DAU-5, was designated as a subtype and denoted with a decimal, such as DAU-5.1. The numbers were in chronological order of deposition in any public database.

Results

We defined 2 novel DAU alleles in 3 patients (Table 1). A 32 year old African female patient in Linz carried the DAU-5.1 allele. A 50 year old African American male with bladder cancer at NIH and another 32 year old African pregnant patient in Salzburg carried the DAU-11 allele. The DAU-5.1 allele is probably occurring in one haplotype with the recently published, rare RHCE*ce48C, 105T allele^41,42 (in cis on 1 chromosome) with the RHCE*ce reference allele in trans. The DAU-11 allele (NIH sample) could have either the RHCE*ce254G (RHCE*ceAG) or RHCE*733G allele in one haplotype (Table 1).

Table 1.

Molecular basis of DAU alleles described in this study

RHD Allele	Nucleotide Substitution in RHD gene	Effect on Protein sequence	Exon involved	Observed samples
							RHD genotype		RHCE genotype (observed RHCE alleles)	GenBank accession number
				Patients (n)	Ethnicity	Phenotype	Zygosity	RHD allele in trans
DAU-5.1	667T>G	F223V	5	1	African	D+C-E-c+e+	Hemizygous	None	RHCE*ce48C, 105T RHCE*ce	HG918112.1
	697G>C	E233Q	5
	1122C>T	I374I	8
	1136C>T	T379M	8
DAU-11	254C>T	A85V	2	1	African American	D+C-E-c+e+	Hemizygous	None	RHCE*ce254G RHCE*733G	KU248927.1
	835G>A	V279M	6
	1136C>T	T379M	8
				1	African	D+C-E-c+e+	Compound heterozygous	RHD*Ψ	Unknown (DNA supply exhausted)	HE965768.1

Immunohematology

All 3 patients were found in routine D antigen typing by discordant results with 2 different anti-D reagents (Table S1). Antibody screening and direct antiglobulin results were negative for all 3 samples. The D antigen density was approximately 6200 per RBC for DAU-5.1 and 2400 for DAU-11 (Table S2). The DAU-11 (NIH sample) reacted in variable strength with all 25 monoclonal anti-D tested (Table S3).

DAU alleles

We collated the 18 known DAU alleles characterized by harboring the c.1136C>T single nucleotide polymorphism encoding p.Thr379Met (Fig. 1 and Table S4). They differed by 1 or more additional missense (non-synonymous) or silent (synonymous) substitutions dispersed throughout the length of the RHD coding sequence (CDS).² Almost all alleles were originally described in individuals with an African ethnic background,² presented as D+C-E-c+e+ phenotypes (Table S5), and hence all occurred in a Dce haplotype. There were 18 nucleotide substitutions encoding 14 non-synonymous and 4 synonymous substitutions in the RHD CDS (Fig. 1). Most SNPs in the CDS had been listed in the dbSNP database, but 4 SNPs were novel (Table S4). Information about the variations in the non-coding regions (5’-UTR, introns and 3’-UTR) was lacking for many DAU alleles, while their association with distinct RHCE alleles has been shown for some of them (Table S6).

Figure 1. Known DAU alleles.

The mutations in RHD gene exons are shown for the 18 known DAU alleles. The RHD allele comprise 10 exons each (yellow boxes). Non-synonymous (solid lines) and synonymous nucleotide substitutions (dotted lines) depict differences to the RHD reference sequences (NM_016124.4).

Clinical patient samples

Within 4 years, Milwaukee has received requests for reference testing by red cell genotyping in 2257 patient samples (Table 2).^41,43,44 Among the 379 samples tested for partial D, 155 (41%) were confirmed as partial D. Among those, 75 (48%) patients carried at least one DAU allele, of which 8 were shown to carry a DAU-4 or DAU-5 allele hemizygously. Springe analyzed 3147 patient referrals within 8 years and Linz 1271 within 4 years. The distribution of RHD alleles detected differed between the US and European centers. For 52 DAU samples from Milwaukee, the associated RHCE alleles were identified (Table S7), all being concordant with published associations (Table S6).

Table 2.

DAU alleles among samples tested by red cell genotyping at regional reference laboratories

	Patient samples (n) at reference laboratories
Red cell genotyping procedure and result	German Red Cross Springe	BloodCenter of Wisconsin	Austrian Red Cross Linz
Any procedure
Total, including weak D screening	3147 *	2257	1271
Screening test for partial D
Partial D and normal D confirmed	79	379 †	532
RHD*01 (normal) only	n.d.	224	306
Any partial D	78	155	226
Partial D test result
Partial D allele other than DAU	56	80	219
Any DAU allele	22	75	7
DAU test result
DAU allele not specified	1	34	2
DAU-0, 1, 2 or 3 allele hemizygous	n.a.	28	n.a.
DAU-0 allele hemizygous	3	n.a.	n.a.
DAU-2 allele hemizygous	4	n.a.	n.a.
DAU-4 or 5 allele hemizygous	n.a.	8	3
DAU-4 allele hemizygous	2	n.a.	n.a.
DAU-6 allele hemizygous	1	n.a.	2
DAU compound heterozygous	11 ‡	5 ¶	0
Time frame	8 years	4 years	4 years

^*includes 954 blood donor samples

^†Partial D analysis using kit (n = 370) or nucleotide sequencing (n = 9)

^‡5 DAU-0/RHD*01; 1 DAU-0/RHD*Ψ; 1 DAU-0/DIIIa-CE(4–7)-D;^40,42,43 1 DAU-1/RHD*Ψ; 1 DAU-3/RHD*01; 1 DAU-3/weak D type 4.2; and 1 DAU-3/ DIIIa-CE(4–7)-D^40,42,43

^¶2 DAU-0/DAU-5; 1 DAU/weak D type 4.2; 1 DAU-0/weak D type 41; and 1 DAU-5/RHD*01

n.d. – not determined, n.a. – not applicable

Phylogenetic analysis of DAU alleles

We parsed the new DAU alleles in our previously published phylogenic trees.^7,10,32 The 8 DAU alleles DAU-0.1,-0.2, -1, -3, -6, -9, -12 and -13 could have originated by single synonymous or non-synonymous substitutions in the DAU-0 allele (Fig. 2). The 4 other DAU alleles DAU-2, -5.1, -10 and -11 could have originated through mutation in the previously established DAU alleles.^7,10,13,14 The DAU-14 allele was likely a result of interlocus gene conversion between the DAU-0 allele and the exon 2 of an RHCE allele. The 4 remaining DAU alleles DAU-4, -5, -7 and -8 could have originated by single recombination events between 2 RHD alleles (Fig. 2).

Figure 2. Phylogeny of alleles in the DAU cluster.

A phylogenetic tree of the DAU cluster is shown for the 18 known alleles. For each evolutionary step, the event is indicated; the depicted distances of the alleles are arbitrary, as previously described for RHD⁷ The extended molecular phylogenetic analysis of RHD alleles delineated 4 clusters: the Eurasian D cluster with the consensus RHD (NM_016124.4) and 3 African clusters designated DIVa, DAU, and weak D type 4. ^7–10 “Gene conversion” denotes a gene conversion in RHD using RHCE as template, if not mentioned otherwise. In this genealogy, DIII type 5 is assumed to be derived from the “basal” DIVa cluster rather than DIII type 4, because DIII type 4 is a rare allele caused by a recombination of an allele of the DIVa cluster with Eurasian RHD⁶ In this analysis, the RHCE allele polymorphisms were not considered, and the actual phylogeny may be even more complex. However, the typically associated RHCE allele is indicated for each RHD allele, if known (see Tables S5 and S6). ce* indicates a ce-like allele which may frequently be a variant.

Predicted effect of non-synonymous substitutions

The PolyPhen-2, SIFT, PROVEAN, SNAP2 and INPS bioinformatic programs predicted deleterious structural changes induced by the non-synonymous p.Arg114Trp, p.Phe179Leu, p.Leu181Pro, p.Val247Leu and p.Val279Met substitutions (Table S8). The 14 non-synonymous substitutions were distributed along the whole length of the RHD CDS without any apparent clustering (Fig. 3).

Figure 3. Model of RhD protein in the red cell membrane.

The RhD protein consist of 417 amino acids (circles). The first amino acid is lacking from the mature protein in the membrane. The extracellular Rh vestibule (inverted black arc) is in part bordered by amino acids of loops 3 and 4.^3,4 There are 9 exon boundaries in the RHD cDNA as reflected in the amino acid sequence (black bars).⁵ All known amino acid substitutions encoding DAU alleles are labeled (colored circles). The 4 synonymous SNPs cause no amino acid change (grey). The other SNPs are non-synonymous and cause amino acid changes that are predicted to affect the RhD protein structure (red) or to be neutral (blue). The p.Thr379Met amino acid change (yellow ring), defining the DAU cluster, is predicted to having no effect on the RhD protein structure (neutral).

Comparative homology modeling of the RhD protein

The template-based homology model of RhD protein was consistent with the model proposed on the basis of computational hydropathy plots⁴⁵ (Fig. 4). The model comprised 408 amino acids from Lys4 to Pro411 and lacked 9 residues (3 in the N terminus and 6 in the C terminus). An analysis of the stereochemistry using RAMPAGE software showed all the main chain atoms falling within the generously allowed region of the Ramachandran plot³⁵ with 394 residues in the most favored region (96.5%), 10 residues in the additionally allowed region (2.5%), and 4 residues in the outlier region (1%). The ProSA-web z-score of the model was −6.2 (Fig. S1), a value within the range of other experimentally determined protein structures of the same size.^28,29 ProQ results predicted LGscore and MaxSub as 6.66 and 0.63, respectively, indicating a very good model.³⁶

Figure 4. Molecular structure of RhD protein.

The side view of a homology model of the RhD protein is depicted as it is situated in the RBC membrane; the top faces the RBC surface and bottom the RBC inside (A). The view to the RhD protein from the inside of an RBC is depicted as it is embedded in the RBC membrane (B). Panel A is rotated 90° relative to panel B. Most amino acid substitutions (red and blue) occur in the transmembraneous helices (grey ribbons) rather than the extracellular loops (black lines). The position of the central pore is indicated by the white rod ( ) and the open circle (○). The 14 non-synonymous amino acid substitutions are predicted to either affect (red) or not affect the RhD protein structure (blue). The p.Thr379Met amino acid change is predicted to having no effect on the RhD protein structure (yellow).

The central pore of the modeled RhD protein coincided with the crystal structure of the RhCG protein model (Fig. 4).³³ We estimated the distance between the C-alpha atom of each amino acid and the central pore of the RhD protein (Fig. 4): there was no statistically significant difference between the 14 amino acids in DAU variants (12.4 Å ± 5.8 Å; mean ± SD) and the remaining 394 amino acids (14.8 Å ± 5.8 Å; p>0.05, Mann-Whitney U-test, 2 sided).

SNPs at CpG sites

Nucleotide substitutions are known to occur frequently at CpG sites, which are defined by a cytosine followed by a guanine in the linear nucleotide sequence along its 5’ → 3’ direction. We found 38 CpG sites in the 1254 nucleotides of the RHD CDS. Among the 18 mutated positions in DAU alleles, including the primordial DAU-0 allele (Table S4), 4 sites represented C>T transition in CpG sites (Fig. S2). The mutations at the CpG sites in the DAU alleles were overrepresented (4 of 38 compared to 18 of 1254; p<0.01, Fisher’s exact test, 2 sided). The CpG site mutation at position 201 was excluded from this calculation because it likely resulted from a gene conversion event with RHCE that also comprised the position 203.

Discussion

The 14 non-synonymous DAU mutations were found to be dispersed over the entire RhD protein with no evidence of clustering at specific sites. All non-synonymous DAU mutations occurred inside the red cell membrane (Fig. 3) with the only exception of the previously described DAU-1 (p.Ser230Ile).⁷ The recurrent p.Thr379Met mutation in exon 8 (Fig. 1) may represent a possibly neutral amino acid substitution that became originally fixed in an isolated African population.

The serology of DAU phenotypes (Table S1 and S2) exemplified the potential relevance of testing the D antigen in the clinical routine with 2 different anti-D monoclonals,⁴⁶ not mandatory in the US, but widely applied in Europe for 2 decades. Current serologic routine procedures detected DAU variants (Table 2 and Table S1). The choice of the right monoclonal anti-D reagents will obviously determine that the clinically relevant D variants⁴⁷ are preferentially recognized and forwarded to red cell genotyping (Table 2). The distribution of RHD alleles detected differed much between the US and European centers, which can be explained by differences in the populations, the routine serologic screening procedures and the approaches to the molecular work-up, which are yet to be standardized. A more detailed immunohematologic workup is still possible for many DAU variants (Table S5), which are primarily needed to determine the clinical relevance of distinct alleles. These in vivo data may also elucidate the molecular mechanisms integrating proteins into cell membranes.

Various RHD alleles are associated with frequent anti-D alloimmunization, especially in chronically transfused patients such as patients with hemoglobinopathies.^14,48 Knowledge of RH alleles, their phylogeny and prevalence will aid in identifying the clinically relevant RHD alleles occurring in patient samples by high throughput technologies, such as next generation sequencing (NGS).⁴⁹ Haplotypes can refer to the specific combination of alleles at different locations on a single chromosome.⁵⁰ At a given RH gene locus, the 1 RHD and 1 RHCE allele represent 1 haplotype. Distinct RHD alleles have been documented to accompany distinct RHCE alleles, each combination thus constituting a unique haplotype. We sequenced the RHCE gene in many different DAU samples (Table S6 and S7)^51–55 and identified the most probable RHCE allele associated with a given DAU allele as a haplotype (Fig. 2).

The DAU-5.1 allele harbored the p.Ile374Ile substitution in combination with the 3 previously described DAU-5 mutations (p.Phe223Val, p.Glu233Gln and p.Thr379Met), while the DAU-11 allele harbored the p.Ala85Val substitution in combination with the 2 previously described DAU-3 mutations (p.Val279Met and p.Thr379Met). The 2 new DAU-5.1 and DAU-11 phenotypes were both found to express a lower D antigen density than their parent DAU phenotypes, with 6236 and 2483 D antigens per RBC respectively. The D antigen densities for these DAU-0, DAU-3 and DAU-5 phenotypes have been reported to be 15,285,⁷ 10,879⁷ and 10,131 D antigens per RBC (Table S2 and Table S5).

The p.Ala85Val amino acid substitution observed in DAU-11 is predicted to reside in the transmembrane region of the RhD protein.⁵ Alanine, hydrophobic like valine but smaller, is a much better helix-forming residue.⁵⁶ Because position 85 resides in the middle of the 3^rd helix (Fig. 3), the disruptive effect by Valine on the helix structure was predicted to be stronger and this perturbation of the helix may hamper lodging of the RhD protein in the RBC membrane.⁵⁷ Because it is in direct contact with the lipid bilayer, the substitution may also affect the tertiary interactions and stabilization of the RhD protein (Fig. 4).⁵⁸ A different nucleotide substitution (c.254C>G; GenBank accession number HE613970.1) at the same codon position causing an p.Ala85Gly substitution expressed even less D antigens with 618 D antigens per RBC.⁵⁸ Glycine, the smallest amino acid but hydrophilic, may more strongly disrupt the RhD folding, lipid membrane integration or interaction with other proteins of Rhesus complex.⁵⁸

The potential impact of non-synonymous nucleotide substitutions on protein expression has recently been well illustrated in vitro for the Dombrock blood group system.⁵⁹ The p.Ile374Ile synonymous nucleotide substitution observed in DAU-5.1 is an excellent in vivo example that synonymous substitutions are also neither random nor neutral. The much reduced D antigen density of DAU-5.1 as compared to DAU-5.0 can be explained on the basis of well-documented molecular effects, such as changing mRNA splicing,⁶⁰ mRNA folding,⁶¹ codon usage bias,⁶² and RNA-RNA interactions, all influencing gene function.⁶³ According to the codon usage database,⁶⁴ ATC coding for Isoleucine (I - Ile) in the normal RhD protein is used in humans 1.3-fold more frequently than the ATT coding for Ile in DAU-5.1. In the reference RHD CDS, the ATC codon is utilized 15 times and the alternate ATT codon used 7 times, a 2.1-fold difference. Translation efficiency and protein folding can be disturbed by this codon bias mechanism.

We used the RhCG protein (Protein Data Bank accession code 3HD6) as template for our homology modeling of the RhD protein (Fig. 4). RhCG is the protein with known crystal structure, most homologous to RhD. However, due to low sequence similarity between RhD and RhCG proteins (33.1%), refinement in the accuracy of RhD protein modeling will be possible.⁶⁵ The 14 non-synonymous mutated positions were not found to be clustered around the central pore of the RhD protein (Fig. 4) and thus may not directly affect the yet unknown function of the central pore. The 5 bioinformatic programs predicted deleterious effects for 5 out of the 14 amino acid substitutions (Table S8); their damaging effect may involve destabilizing the RhD protein, its integration in the RBC membrane or its interaction with other proteins in the Rh complex.

It has previously been proposed that ancestral African populations were structured.⁶⁶ Hence mutations arising in isolated populations were prevented from recombining with one another, and differentiated haplotypes emerged with very little recombination between lineages.⁶⁷ Later, local selective pressures might have favored the spread of different alleles and haplotypes in the populations of distinct geographic areas, such as Eurasia and Africa. The primordial allele of DAU cluster, DAU-0, may have originated in such an isolated population, probably as a premeiotic mutation, where it became fixed.⁶⁸ Premeiotic mutations pass through meiosis and recombination; during these events, the ancestral DAU-0 allele was joined to different RHCE alleles while accumulating additional nucleotide substitutions, forming a variety of new, often more than 1, RH haplotypes (Fig. 2).

Our analysis prominently indicates the role of interallelic recombination in the evolution of DAU alleles, a conclusion based on the observation of 5 shared substitutions between at least 10 different DAU alleles. The present study supports the previous postulate⁷ that the 18 known DAU alleles evolved through random mutation in the primordial DAU-0 allele or through recombination among DAU and other RHD alleles.