The RHCE*CeRN allele: molecular junction characterization and frequency distribution in Sub-Saharan African populations

Abstract

Background

The CeRN (RHCE*02.10) allele was described as resulting from hybrid RHCE-D-CE genes, involving either exon 4 alone or exon 4 and part of exon 3. No solution allows to distinguish between the two reported CeRN alleles. The objectives of this study were to determine the existence of both alleles, their genetic sequence and their frequencies.

Materials and methods

We investigated 17 heterozygous CeRN samples and 10 homozygous CeRN samples, and described the junction between the RHCE and RHD genes. Analysis combined classical PCR associated with Sanger sequencing, and Oxford Nanopore NGS applied to long-range PCR.

We also explored CeRN allelic frequency in sub-Saharan populations from the 1000 Genomes Project and the International Genome Sample Resource.

Results

One CeRN allele sequence was observed, involving exon 4 alone with an RHCE-RHD junction approximately 400 base-pairs upstream of exon 4.

We observed the CeRN allele in Gambian ethnic groups, with a maximum allelic frequency of 6.0% in the Fula group.

Discussion

The CeRN allele involving exon 4 and part of exon 3 was not observed here, supporting at least a low frequency, as our sample size limited the power to investigate this putative second CeRN variant.

The homogeneity of the CeRN sequence observed in DNA samples and the high allelic frequency are consistent with a single genetic founder event in the Fula ancestral group.

Nearly 20% of individuals in The Gambia have abnormal hemoglobin. Knowledge of CeRN genetic architecture and frequency may have significant implications for more precise molecular diagnostics and transfusion therapy.

The probabilistic pipeline developed here, based on subset of samples for which both DNA and WGS data were available, showed that WGS low-coverage data can be used to investigate complex genetic variants in population studies.

INTRODUCTION

The RHCE gene, encoding the Rh blood group antigens C/c and E/e, exhibits remarkable genetic diversity across populations¹. Many alleles have been described, notably in individuals of African origin, leading to significant rates of Rh alloimmunization and challenging blood transfusion management especially for therapy of Sickle Cell Disease (SCD) patients².

Among the numerous variants, the CeRN allele (RHCE*02.10) represents significant clinical implications for transfusion medicine. This RHCE-D-CE hybrid allele results in a RH:-46, RH:54 phenotype, with a severe alteration of the expression of RhC and Rhe antigens. This can still trigger anti-C or anti-e alloimmunization in susceptible recipients because of their genetic characteristics (HLA/Cytokine polymorphisms), inflammatory status (like SCD patients), and/or iterative transfusion needs (oncohematological conditions)³.

The CeRN allele is responsible for most of the weak RhC and/or Rhe phenotypes in patients or donors analyzed in the immunohematology laboratories in southern France and the Reunion Island (Indian Ocean)⁴, justifying specific attention in its characterization. Moreover, patients with weak RhC and/or Rhe, mostly with Sub-Saharan ancestry, have an additional immunization risk associated to antigenic donor/recipient mismatch and immunogenicity, i.e., K (KEL), Fya (FY), Jkb (JK), and S (MNS)^5,6.

Due to limitations in conventional serological testing, accurate identification of the CeRN variant necessitates molecular approaches, such as BioArray⁴ or TaqMan PCR⁶. Although very efficient and reliable, these approaches are not capable of discriminating between the two CeRN alleles initially reported, RHCE*02.10.01 and RHCE*02.10.02, respectively involving either exon 4 alone or exon 4 and part of exon 3⁷.

Discriminating DNA-based protocols to accurately characterize RHCE variants, particularly RHCE-RHD hybrid genes, are challenging to set up because of the homology of RHCE and RHD genes and the repeated sequences within both genes, especially in the intronic regions.

Whether both alleles exist, and in which proportion, may be of importance in terms of transfusion management. The coding polymorphism in exon 3 may affect the conformation of the gene product and modify protein epitope(s). Moreover, the existence of two alleles may imply two molecular events, with different genetic backgrounds, and potentially different regulatory region polymorphisms leading to a modulation of expression levels.

In addition, little data is available concerning the origin and frequency of the CeRN allele, except empiric immunohematology laboratories observations describing its presence in individuals with sub-Saharan origins^2,6.

The objectives of this study were to determine the existence of both alleles, their genetic sequence and their frequencies.

Thanks to projects like the 1000 Genomes Project⁸ and the International Genome Sample Resource (IGSR)⁹, both DNA samples and in silico genetic data from worldwide human populations are accessible to the scientific community and can be used to track down a specific allele and screen its frequency in different populations. We thus explored CeRN allelic frequency in sub-Saharan populations by analyzing 355 sub-Saharan DNA samples from the 1000 Genomes Project⁸ (Table I).

Table I.

CeRN allele frequencies in Sub-Saharan populations

Population	Source	Code	Location	No.	Method	CeRN (No.)	Uncertain (No.)	CeRN (Fq%)
Mende	1000G	MSL	Sierra Leone	98	TaqMan	0	-	0
Yoruba		YRI	Nigeria	138	TaqMan	0	-	0
Mandinka		GWD	The Gambia	119 120	TaqMan WGS	7 5	- 2	2.9 2.1–2.9
	IGSR			100	WGS	4	2	2.0–3.0
Jola				100	WGS	0	0	0
Fula				100	WGS	10	2	5.0–6.0
Wolof				100	WGS	5	0	2.5

Sub-Saharan population name, code, location and number of samples (No.) included in the study. Samples were from the 1000 Genomes Project⁸ or the International Genome Sample Resource (IGSR)⁹. Method type (TaqMan and/or WGS data) and source (1000G: 1000 Genomes Project or IGSR) are given. The number of CeRN alleles observed (CeRN [No.]) is given. For WGS data analysis, the number of CeRN alleles refers to confident prediction whereas the number of samples categorized as uncertain (Uncertain [No.]) is separated. CeRN allelic frequency (Fq %) is accordingly given as range for WGS data.

We investigated 10 heterozygous CeRN samples and 10 homozygous CeRN samples from French blood donors analyzed by the French Blood Center (Marseille) and 7 heterozygous CeRN samples from the 1000 Genomes Project⁸ to describe the junction(s) of the segmental DNA exchange between the RHCE and RHD genes.

Further investigation was conducted based on IGSR resources from the “Gambian Genome Variation Project”⁶: with i) validation of an analysis pipeline to match DNA-in silico data and ii) its application to WGS data of 520 sub-Saharan samples.

MATERIALS AND METHODS

Characterizing the CeRN sequence

Sample collection

The molecular immunohematology laboratory of the French Blood Center (Etablissement Français du Sang PACA Corse) in Marseille (France) centralizes blood donor serological issues from different French immunohematology laboratories⁴. Twenty samples for which CeRN status was characterized by TaqMan PCR⁶ were explored in this study: ten samples displayed one CeRN allele and ten samples were homozygous. Five samples were included as controls and displayed the following phenotypes: D-C-E-c+e+ (No.=2); D-C+E-c+e+; D+C-E-c+e+; D+C-E+c+e−; D+C+E-c-e+.

Sanger and next-generation sequencing

All DNA samples were first screened for the CeRN presence by TaqMan PCR⁶ and the positive samples were explored with the following molecular biology protocols.

Gene references were RHD (accession number NG_007494.1), RHCE (accession number NG_009208.3), RHD exons (accession number NM_016124.6), RHCE exons (accession number NM_020485.8), rs associated to the CeRN allele and their frequency in gnom AD (Table II)^1,7,10. All nucleic acid bases are given according to ISBT nomenclature, which is reversed/complemented as compared to NCBI.

Table II.

SNPs associated to the CeRN allele with allele frequency in the gnomAD¹,⁷,¹⁰

rs number	Nucleotide change	Exon	Amino-acid change	Position in hg 38	Position in hg 37	Allele frequency in Africans (e−04)	Allele frequency in non-Africans (e−04)
rs35109888	RHCE/RHD, c.455C/A	3	p.Thr152Asn	chr1:25402627	chr1:25729118	89.63	6.686
rs1132761 (ISBT)	RHCE/RHD, c.505C/A	4	p.Leu169Met	chr1:25392123	chr1:25718614	1.805	0
rs1020280601
rs1132762 (ISBT)	RHCE/RHD, c.509G/T	4	p.Arg170Met	chr1:25392119	chr1:25718610	2.106	39.16
rs987753117
rs1053349	RHCE/RHD, c.514T/A	4	p.Phe172Ile	chr1:25392114	chr1:25718605	1.815	0
rs1053350	RHCE/RHD, c.544A/T	4	p.Thr182Ser	chr1:25392084	chr1:25718575	59.74	3.443
rs1384157219	RHCE/RHD, c.577A/G	4	p.Lys193Glu	chr1:25392051	chr1:25718542	0	9.097
rs1053354	RHCE/RHD, c.594T/A	4	p.Asn198Lys	chr1:25392034	chr1:25718525	0	1.320
rs141398055	RHCE/RHD, c.602G/C	4	p.Arg201Thr	chr1:25392026	chr1:25718517	4.782	0.562

The CeRN sequence was explored by independent PCR reactions performed using Platinum SuperFi II DNA Polymerase (Invitrogen, Fisher Scientific France, IllkirchGraffenstaden, France) according to manufacturer recommendations. Sequencing reactions were performed either by Sanger sequencing technology using PCR primers as sequencing primers, or by Next Generation Sequencing (NGS) using Oxford Nanopore GridION (flow cell R10.4.1) (Eurofins Genomics, Ebersberg, Germany).

PCR primers were designed using Primer3 software (https://primer3.ut.ee/) to amplify different fragments of the RHCE and/or RHD gene within the intron 2-intron 4 region (Figure 1 and Table III).

Figure 1. Schematic representation of the primer’s localization within the intron 2-intron 4 region of the RHD and RHCE genes and according to CeRN specific polymorphisms.

Primer and probe characteristics of allele-specific molecular groove binder (MGB) assay from Tournamille et al.⁶ are also represented.

Table III.

Primer characteristics used to sequence DNA segment from intron 2 to intron 4 of the CeRN allele

Name	Sequence 5′>3′	Specificity	Localization	Product length	Sequencing technology
RHD-IVS3-F_A	AAGGACTATCAGGGCTTGCC	RHD	Intron 3	445	Sanger
RHCE-IVS4-R_A	TGAGTCTCTCCTTTCTAGTGGA	RHCE	Intron 4
RHCE-IVS3-F_Q1	CCGGCCTGACTTGCATTTTA	RHCE	Intron 3	1,773	Sanger
RHCE-IVS4-R_Q1	TAGCCTCTCCCAGCAGTCTA	RHCE	Intron 4
RHDCE-IVS3-F_J	GTCTGGTTGCTTCATGTTGC	RHD/RHCE	Intron 3	977	Sanger
RHDCE-ex4-R_JJ	CTCCGTTCCCTCGGGTAGAG	RHD/RHCE	Exon 4
RHDCE-IVS2-F_K	AGTAAGGTGCCCAACAGTGT	RHD/RHCE	Intron 2	10,838	Oxford
RHDCE-ex4-R_K	TGGGTATCGTTGCTGTCTGA	RHD/RHCE	Exon 4
RHCE-IVS3-F_N	AAAAGAGGCCGGGCACAG	RHCE	Intron 3	449	Sanger
RHD-ex4-R_O	TGTGCATCATGTTCATGTGGT	RHD	Exon 4
Forward primer RHD-ex4	GACTACCACATGAACATGAT	RHD	Exon 4	454/455	TaqMan4
Forward primer RHCE-ex4	ACTACCACATGAACCTGAG	RHCE	Exon 4
Reverse primer RHCE-IVS4	CCCTACCTTCTTCACTTGTCCG	RHCE	Intron 4
MGB probe RHD-ex4	6-FAM-AGGATAAAGATCAGACAGC	RHD	Exon 4
MGB probe RHCE-ex4	6-VIC-AATGATCAGAGAGCAAC	RHCE	Exon 4

Primers and probes characteristics of allele-specific molecular groove binder (MGB) assay from Tournamille et al.⁶ are also given. PCR product lengths are given in base pairs.

The 3′ side of the 4^th exon was specifically explored by a PCR using an RHD-specific forward primer in intron 3 and an RHCE-specific reverse primer in intron 4 (RHD-IVS3-F_A and RHCE-IVS4-R_A). The 5′ junction of the CeRN sequence was explored by a PCR using RHD/RHCE specific primers, in intron 3 and in exon 4 (RHDCE-IVS3-F_J and RHDCE-ex4-R_JJ). Both PCR were analyzed by Sanger sequencing.

Supplemental analyses were performed to encompass both junctions: a PCR using an RHCE-specific forward primer in intron 3 and an RHCE-specific reverse primer in intron 4 (RHCE-IVS3-F_Q1 and RHCE-IVS4-R_Q1) followed by Sanger sequencing; and a second PCR using RHD/RHCE specific primers in intron 2 and in exon 4 (RHDCE-IVS2-F_K and RHDCE-ex4-R_K) sequenced using Oxford Nanopore technologies.

Finally, CeRN allele presence was confirmed by a CeRN specific PCR, with a forward primer located before the 5′ junction in the RHCE sequence and a reverse primer located in exon 4 of RHD (RHCE-IVS3-F_N and RHD-ex4-R_O).

Sanger sequencing results were aligned using CodonCode Aligner v10.0.2 (CodonCode Corporation, Centerville, MA, USA) and NGS sequencing results were aligned with PolyPheme v3.0 (Xegen, Gemenos, France) and visualized with Tablet v1.12.02.08¹¹.

Determining CeRN allelic frequency

Sample collection

Three hundred fifty-five DNA samples from African populations speaking different languages from the 1000 Genomes Project were analyzed: Mandinka in The Gambia (No.=119), Mende in Sierra Leone (No.=98) and Yoruba in Nigeria (No.=138) (Table I)⁸. All DNA samples were first screened for CeRN presence by TaqMan PCR⁶ and the positive samples were explored with the analyses formerly described.

Five hundred and twenty samples available as low coverage Whole Genome Sequencing (WGS) from the “Gambian Genome Variation Project” (International Genome Sample Resource [IGSR]⁹) were included in the study. Individuals were from 4 Gambian populations: Mandinka, Jola, Fula and Wolof (Table I).

Low-coverage WGS data analysis

WGS data was aligned to Human Genome reference hg 38 (GCA_000001405.15_GRCh38) with Bowtie2 software¹². CeRN positions of interest (No.=19, Table IV) were studied using samtools mpileup command line¹³. A dedicated script allowed to extract for each position, the total number of bases observed, number of bases corresponding to the reference, number of bases corresponding to the variant and number of other bases.

Table IV.

Position-specific weighting coefficients obtained by ridge logistic regression analysis of the 119 Mandinka individuals

Hg 38 Position	rs number	SNP position	Exon/Intron	Reference	Variant	Coefficient
chr1.25402627	rs35109888	c.455	Exon 3	C	A	1.7931
chr1.25392562	rs1355017134	IVS3-421	Intron 3	G	A	0.2919
chr1.25392556	rs1229366898	IVS3-415	Intron 3	G	A	0.6935
chr1.25392552	rs1162087139	IVS3-411	Intron 3	A	G	0.6473
chr1.25392299	rs141277768	IVS3-158	Intron 3	A	G	−3.5018
chr1.25392256	rs200586526	IVS3-115	Intron 3	G	A	−0.6097
chr1.25392205	rs1343374800	IVS3-64	Intron 3	C	T	−1.0921
chr1.25392194	rs1430270140	IVS3-53	Intron 3	T	G	−1.2004
chr1.25392189	rs770920427	IVS3-48	Intron 3	T	del	−1.3285
chr1.25392123	rs1020280601	c.505	Exon 4	C	A	−5.5699
chr1.25392119	rs987753117	c.509	Exon 4	G	T	−3.8494
chr1.25392114	rs1053349	c.514	Exon 4	T	A	−3.2594
chr1.25392084	rs1053350	c.544	Exon 4	A	T	−4.0689
chr1.25392051	rs1384157219	c.577	Exon 4	A	G	−6.0137
chr1.25392034	rs1053354	c.594	Exon 4	T	A	−2.7717
chr1.25392026	rs141398055	c.602	Exon 4	G	C	−2.9084
chr1.25391965	rs112970979	IVS4+29	Intron 4	A	C	0.4029
chr1.25391900	rs556663345	IVS4+94	Intron 4	C	G	0.3738
chr1.25391862	rs1337396001	IVS4+132	Intron 4	C	T	5.6825

Weight coefficients are shown for each position (ridge logistic regression analysis using the glmnet package [v. 4.1–10] in R [v. 4.4.0]¹⁴). Positive coefficient values are predictive of the reference sequence (RHCE), and negative values are predictive of the variant sequence (CeRN). Values close to 0 have little predictive value for the outcome, whereas values above 1 or below −1 have positive association. In bold bold are the variants characteristic of the CeRN sequence.

Pipeline validation for CeRN analysis

The 119 Mandinka individuals from 1000 Genomes Project, for which both DNA analysis results and WGS data were available, were used to calculate a weight coefficient for each position by a ridge logistic regression analysis using the glmnet package (v 4.1–10) in R (v 4.4.0)¹⁴. A decision threshold for classification was determined from the receiver-operating characteristic (ROC) curve of the training model, corresponding to the Youden-optimal cutoff that maximized discrimination between positive and negative samples. To account for uncertainty in borderline predictions, an “uncertain zone” was defined, bordered by two probability cutoffs: the 5^th percentile of predicted probabilities among positive samples and the 95^th percentile among negative samples. Samples with predicted probabilities within this intermediate range were considered indeterminate and excluded from confident classification. Leave-one-out cross-validation (LOOCV) was performed to evaluate model performance and select optimal parameters with final position-specific weighting coefficients derived from the regularized model trained on the complete dataset (Table IV).

These coefficients were then applied to WGS data (No.=520) within a ridge-regularized logistic regression framework to estimate, for each individual, the probability of carrying a CeRN allele.

Homozygous samples were defined as displaying >90% of reads mapped to the variant positions defining the CeRN allele Table V)^15,16.

Table V.

SNPs associated to the CeRN allele (this study)

rs number	Nucleotide change	Exon/Intron	Amino-Acid change	Position in hg 38	Position in hg 37
rs141277768	RHCE/RHD, IVS3-158 A/G	Intron 3	N/A	chr1:25392299	chr1:25718790
rs200586526	RHCE/RHD, IVS3-115 G/A	Intron 3	N/A	chr1:25392256	chr1:25718747
rs1343374800	RHCE/RHD, IVS3-64 C/T	Intron 3	N/A	chr1:25392205	chr1:25718696
rs1430270140	RHCE/RHD, IVS3-53 T/G	Intron 3	N/A	chr1:25392194	chr1:25718685
rs770920427	RHCE/RHD, IVS3-48 T/del	Intron 3	N/A	chr1:25392189	chr1:25718680
rs1020280601	RHCE/RHD, c.505 C/A	Exon 4	p.Leu169Met	chr1:25392123	chr1:25718614
rs987753117	RHCE/RHD, c.509 G/T	Exon 4	p.Arg170Met	chr1:25392119	chr1:25718610
rs1053349	RHCE/RHD, c.514 T/A	Exon 4	p.Phe172Ile	chr1:25392114	chr1:25718605
rs1053350	RHCE/RHD, c.544 A/T	Exon 4	p.Thr182Ser	chr1:25392084	chr1:25718575
rs1384157219	RHCE/RHD, c.577 A/G	Exon 4	p.Lys193Glu	chr1:25392051	chr1:25718542
rs1053354	RHCE/RHD, c.594 T/A	Exon 4	p.Asn198Lys	chr1:25392034	chr1:25718525
rs141398055	RHCE/RHD, c.602 G/C	Exon 4	p.Arg201Thr	chr1:25392026	chr1:25718517

CeRN allelic frequency in each population was subsequently calculated.

RESULTS

One CeRN allele is observed

Among the 355 DNA samples from the three sub-Saharan populations, seven individuals from the Western Divisions of The Gambia were screened positive for CeRN allele by TaqMan PCR⁴; no homozygous CeRN sample was observed (Table I).

These seven heterozygous CeRN DNA samples were studied in all subsequent analyses along with the 10 heterozygous CeRN and 10 homozygous CeRN DNA samples from the molecular immunohematology laboratory of the French Blood Center (Etablissement Français du Sang PACA Corse) in Marseille (France).

The RHD-RHCE junction of CeRN sequence occurs at the end of exon 4

The CeRN sequence was first explored at the 3′ side of the 4^th exon by two independent PCR and following Sanger sequencing.

Using RHD-specific forward primer in intron 3 and an RHCE-specific reverse primer in intron 4 (RHD-IVS3-F_A and RHCE-IVS4-R_A), all samples displaying a CeRN allele had the same sequence, with exon 4 being that of RHD and the beginning of intron 4 being that of RHCE. No difference was observed between all CeRN sequences (Online Supplementary, Figure S1). Using an RHCE-specific primers from intron 3 to intron 4 (RHCE-IVS3-F_Q1 and RHCE-IVS4-R_Q1), all CeRN homozygous samples displayed the same sequence, confirming that CeRN allele is composed of exon 4 from RHD and intron 4 from RHCE. Analysis of samples displaying one CeRN allele further confirmed these results, as they displayed a heterozygous signal in exon 4 (corresponding to RHD and RHCE) and a homozygous signal in intron 4 (corresponding to RHCE).

The RHCE-RHD junction of CeRN sequence occurs 400 base-pairs upstream of exon 4

Next the CeRN sequence was explored at the 5′ side of the 4^th exon using different pairs of RHD/RHCE specific primers.

Using primers from intron 3 to exon 4 (RHDCE-IVS3-F_J/RHDCE-ex4-R_JJ), all CeRN homozygous samples displayed similar Sanger sequencing results, with an RHD specific sequence until 170 base-pairs upstream of exon 4, characterized by 5 SNPs (rs770920427 (RHCE/RHD, IVS3-48 T/del); rs1430270140 (RHCE/RHD, IVS3-53 T/G); rs1343374800 (RHCE/RHD, IVS3-64 C/T); rs200586526 (RHCE/RHD, IVS3-115 G/A); rs141277768 RHCE/RHD, IVS3-158 A/G); a common sequence to RHCE and RHD from 170 to 400 base-pairs upstream of exon 4, and an RHCE specific sequence characterized by 3 SNPs: rs1355017134 (RHCE/RHD, IVS3-421 G/A); rs1229366898 (RHCE/RHD, IVS3-415 G/A); rs1162087139 (RHCE/RHD, IVS3-411 A/G), preceded by a poly-A stretch (rs1211993147; RHCE/RHD, I VS3-422 2 6A/14A) ( Table V, Figure 1 and Online Supplementary Figure S2).

Analysis of heterozygous samples was not possible as a deletion in the RHD sequence (rs770920427, RHCE/RHD, IVS3-48 T/del) is responsible for a double signal in Sanger sequencing traces obtained with reverse primer; those obtained with the forward primer were not readable either at the 3′side of the poly-A stretch.

Sanger sequencing of the PCR with primers from intron 3 to intron 4 (RHCE-IVS3-F_Q1 / RHCE-IVS4-R_Q1) confirmed the RHCE-RHD junction location.

A third PCR, encompassing exon 3 (RHDCE-IVS2-F_K/ RHDCE-ex4-R_K) sequenced using Oxford Nanopore technologies sequencing, further confirmed these results (Online Supplementary Figure S3).

All samples were confirmed using the final CeRN specific PCR (RHCE-IVS3-F_N / RHD-ex4-R_O), all CeRN samples were positive for this PCR, whereas control samples were negative (data not shown).

No CeRN sequence encompassing a part of the 3^rd exon of RHD was observed

Oxford Nanopore technologies sequencing results were further explored to analyze the CeRN sequence for any RHD signal in exon 3 (rs35109888 (RHCE/RHD, c.455 C/A)) as previously described^1,7. All reads displaying an RHD signal at this position corresponded to the RHD gene sequence in exon 3 (Online Supplementary Figure S4), and no CeRN sequence, characterized by the above rs in intron 3 presented an RHD base at rs35109888 (RHCE/RHD, c.455 C/A).

The sequence of CeRN validated by these analyses was submitted to GenBank (accession number PV563567).

CeRN is observed in Fula, Mandinka and Wolof in The Gambia

CeRN haplotype frequency has not yet been described in the literature and allele frequencies of its constitutive SNPs in gnomAD do not allow to predict it (Table II) (Rouillac et al. 1996; Gassner et al. 2022; Chen et al. 2024). CeRN frequency in samples from The Gambia, Sierra Leone and Nigeria was retrieved from DNA analysis (No.=355), WGS data analysis (No.=520) or both (No.=119) (Table I).

Among the 355 DNA samples, 7 individuals from the Western Divisions of The Gambia displayed a CeRN allele at heterozygous state, i.e., an allelic frequency of 2.9%. No individual from Sierra Leone or Nigeria displayed any CeRN allele.

Weight coefficients for each position, obtained by the comparison of WGS data with DNA-based results are shown in Table IV, and were used to interpret WGS data. Quality Score of the WGS data displayed a median of 35.98 [23.49–40.30] (data not shown). WGS data analysis results (No.=520) are shown in Online Supplementary Table SI, displaying the probability of carrying a CeRN allele (positive, negative or uncertain) and the percentage of reads mapped to the variant positions defining the CeRN used to predict homozygous samples.

CeRN allelic frequency in each population was subsequently calculated and are displayed in Table I. Within the Gambian population, Jola samples displayed no CeRN allele, Wolof samples displayed a predicted CeRN allelic frequency of 2.5%; Mandinka individuals displayed a predicted CeRN allelic frequency of 2.0–3.0%, the overall allelic frequency (DNA and WGS data) in this group being thus 2.5–2.9%. The Fula population displayed the highest predicted CeRN allelic frequency (5.0–6.0%). All CeRN positive samples were heterozygous, except in the Fula population where one homozygous sample was characterized.

DISCUSSION

Two distinct CeRN variants, one involving exon 4 alone and another involving exon 4 and part of exon 3, were reported⁷, both associated with an alteration of the expression of RhC and Rhe antigens.

We present in this study the characterization of the genetic architecture of one CeRN allele. The molecular analysis of twenty-seven DNA samples from French blood donors and Gambian individuals harboring an identical CeRN allele permitted to establish the exact junction points between the RHCE and RHD genes. The observed CeRN variant features an RHCE-RHD junction approximately 400 base-pairs upstream of exon 4, with an RHD sequence extending through exon 4, followed by a transition back to the RHCE sequence at the beginning of intron 4.

We were not able to observe the second CeRN allele involving exon 4 and part of exon 3. The fact that we did not observe this latter allele does not necessarily question its existence but does support its low frequency. It should be stressed however that our sample size limits the power to investigate a second, rarer CeRN variant.

As far as its distribution is concerned, our study reveals a maximum CeRN allelic frequency of 6.0% in the most western region of Africa. Among the five populations included in the study (Mandika, Jola, Fula and Wolof in The Gambia; Mende in Sierra Leone and Yoruba in Nigeria), CeRN allele was observed in Fula, Mandinka and Wolof populations of The Gambia.

Eight main ethnic groups and languages are presents in The Gambia, where Mandinka is the largest ethnic group. This language is spoken by around 11 million people in The Gambia, the southern region of Senegal (Casamance), Mali, Guinea and Guinea Bissau. The Mandinka ethnic group is distinct from Fula (or Fulani, or Peul), Wolof and Jola (or Diola) who have distinct historical and geographical origins. Peul is spoken in many countries in Western African and Western Sahel, concerning approximately 60 million individuals; thus, the geographic location of Fula and Mandinka populations overlap in some countries. Furthermore, Wolof is the largest ethnic group in Senegal, a country which literally surrounds The Gambia^17–20. Despite a geographical overlap, the populations of Western Africa originated from distinct genetic history. A recent genome-wide survey has revealed genetic signatures of a population bottleneck followed by a strong founder effect especially in the Fula population²¹. As no biological benefit supports positive selection of the CeRN allele, the homogeneity of the sequence observed in DNA samples and the high allelic frequency are consistent with a single genetic founder event in the Fula ancestral group. In Mandinka and Wolof populations, the CeRN allele would have circulated as the result of admixture, as interethnic marriage rate was reported to be of 19.8% in The Gambia¹⁷. In the Jola group, the absence of CeRN allele is supported by earlier genomic-wide analysis showing that Jola individuals cluster together without any individuals from other ethnic groups of The Gambia^22–24. It should be pinpointed however that most of the samples used in their study were also from the Gambia Genome Variation Project analyzed in the present study. A larger survey is needed to refine the presence and the frequency of this allele in the populations living in or nearby The Gambia and Senegal.

Knowledge of CeRN genetic architecture and frequency has significant implications for transfusion medicine and more precise molecular diagnostics, particularly for managing patients with Sickle Cell Disease who require chronic transfusion therapy²⁵. In The Gambia, a recent study reported that nearly 20% of individuals had abnormal hemoglobin²⁶. One could extrapolate that with an allelic frequency of 0.06 (q), expected homozygous CeRN frequency would be 0.0036 (q²). Thus, given that 20% of the Gambian population could require transfusion, 0.07% of the global population might be at risk for global immunization (anti-C, anti-e, anti-RH46), because of CeRN homozygous status. Also, more than 2% (2pq) of the population might be at risk for specific alloimmunization because of CeRN heterozygous status, depending on their phenotype in trans (anti-C in case of partial C in trans and anti-e in case of partiale in trans).

Molecular typing of RHCE and RHD genes, even more their hybrid product, is particularly challenging. Here, we used classical PCR associated with Sanger sequencing, and NGS Oxford Nanopore technology applied to long-range PCR. The complementarity of both approaches, combining the accuracy of the first with the ability of long-read sequencing of the second, provides an unprecedented solution to analyze hybrid products of these homologous genes.

To complete our exploration of the CeRN allele distribution in sub-Saharan populations, we analyzed low-coverage WGS public data.

Low-coverage genome sequencing allows to access a large sample size, and is thus very informative for population studies²⁷. However, low-coverage sequencing may fail to identify variants, i.e., leading to missing information, because of reduced coverage. Also, because of reduced depth, low-coverage sequencing may lead to miscalled heterozygous variants, and consequently underevaluate variant frequency¹⁶. On the contrary, the uncertainty of base calls, i.e., the probability of a base call being wrong, estimated by the Quality Score, was very low in this study (above Q30 [1 in 1,000]) and greatly reduced the risk of overestimating variant frequency. Because of its limitations, low-coverage sequencing is not adapted for whole sequence analysis of highly polymorphic genes such as RHD and RHCE. Here, we specifically targeted a limited number of positions to limit false positive results (miscalled variants) and developed a probabilistic pipeline based on a subset of samples for which both DNA and WGS data were available to present a range of allelic frequency with predicted probabilities, which allows appreciation of false negative results.

Our results could not explain however the variable C and/or e antigen intensities obtained with serological analyses in individuals harboring the CeRN allele observed in our immunohematology laboratories (data not shown), especially for homozygous samples for which no RHCE haplotype in trans could account for differential antigen expression. Such variable serological results might be due to different methodologies, i.e., Microplate or Gel column, combined with various assays, kits or antibodies, as well as with human interpretation. Unrevealed cis- or trans-acting polymorphic regulators, notably in RHD and RHAG genes, may also interfere with RhCE expression^28–30.

CONCLUSIONS

In this study, we precisely characterized the molecular architecture of the CeRN allele, defining the exact junctions between the RHCE and RHD genes. The probabilistic pipeline developed here, based on subset of samples for which both DNA and WGS data were available, showed that WGS low-coverage data can be used to investigate complex genetic variants in population studies. Our findings suggest that this allele most likely arose from a single founder event in the Fula ancestral group and subsequently spread through population admixture. The relatively high frequency observed in certain Gambian ethnic groups carries important implications for transfusion medicine, particularly in regions where hemoglobinopathies necessitate chronic transfusion therapy. Overall, these results highlight the importance of implementing extended molecular typing approaches and conducting larger population-based studies to better estimate CeRN prevalence and improve prediction of alloimmunization risk in affected populations.

Data availability statement

All data supporting the findings of this study are available within the paper and its Supplementary Information and openly available from the corresponding Author upon reasonable request.