Evolution of the human RH (rhesus) blood group genes: a 50 year old prediction (partially) fulfilled
Evolution of the human RH (rhesus) blood group genes: a 50 year old prediction (partially) fulfilledB. Carritt*, T. J. Kemp and M. Poulter
MRC Human Biochemical Genetics Unit, University College, London, Wolfson House, 4 Stephenson Way, London NW1 2HE, UK
Received January 8, 1997;Revised and Accepted March 21, 1997
DDBJ/EMBL/GenBank accession nos U66340, U66341
Almost exactly 50 years ago, R.A.Fisher and R.Race proposed a model for the evolution of the RH (rhesus) genes in which the less common haplotypes were derived from the commoner ones by recombination, and in which the gene order was D-C-E. No direct evidence bearing on this model was available then, and has not been until now. Here we present evidence for non-reciprocal intergenic exchange (gene conversion) occurring once in human history to generate the common RHCE allele, Ce. We have also used new polymorphisms to construct haplotypes which suggest that intragenic recombination played a major role in the generation of the less common haplotypes, but only if RHD lies 3' of RHCE, i.e. the order is C-E-D. We provide both genetic and physical evidence supporting this arrangement.
The Rh blood group, popularly referred to as Rhesus, is second only to the ABO system in its importance in transfusion medicine. Although the Rh system is highly polymorphic, and comprises at least 44 distinct antigens, clinically the most significant polymorphism is due to the presence or absence of the RhD antigen on red cells. In addition to the problems caused by incompatibility for RhD type between transfusion donor and recipient, incompatibility between a multiparous RhD- mother and her unborn child can result in a severe immune reaction leading to neonatal hemolytic disease (HDN) or even intrauterine death. Incompatibility for the other products of the RH locus, the C-series and E-series antigens, can also occasionally give rise to hemolytic reactions.
The Rh antigens are carried on three non-glycosylated transmembrane proteins that are encoded in only two genes, RHD and RHCE (1 -5 ). Alternative mRNA splicing is probably responsible for the production of two distinct polypeptides from the single RHCE gene (6 ). Lack of D antigen expression is usually due to the absence of the entire RHD gene from the genome of RhD- individuals (7 ), whereas variant forms of the C and E antigens are generated by base substitutions in RHCE (5 ,8 ). A single C to G transition in exon 5 of RHCE results in a polypeptide that reacts with anti-e, rather than anti-E, antisera. Alleles of RHCE that express the C antigen differ from those expressing c at six nucleotide positions, one in exon 1 and five in exon 2. Only four of these result in amino acid substitutions, and that at nucleotide 307 (serine to proline, residue 103) is most consistently associated with C or c antigenicity, respectively. There are, in addition, rare variants of C, c, E and e, whose molecular basis is, for the most part, unknown.
In the presence and absence, respectively, of RHD, the four alleles of RHCE make up all possible haplotypes, although there are marked differences in their relative frequencies, both within and between populations (see Table 1 ). In 1946 Fisher and Race (9 ) proposed a model for the evolution of the RH polymorphism in which the less common haplotypes (CDE, Cde, cdE and CdE) are generated and maintained by recombination from those found at higher frequency. The frequency ratios of the supposed parental and recombinant haplotypes suggested that the arrangement D-C-E was most likely under the crossing-over hypothesis.
At the time, Fisher and Race assumed that the three series of antigens are encoded in three separate genes, whereas it is now known that there are only two; recombination between the sites encoding C/c and E/e would therefore be intragenic, and occur between exons that are separated by ~30 kb (10 ). Moreover, the imperfect homology between genomes that have and those that lack an RHD gene may impede pairing between them, and reduce the possibility of recombination in heterozygotes. There has been one suggestion (11 ), but no proven case, of recombination between RHD and RHCE within a single pedigree.
We now present evidence which suggests a probable origin for the common alleles of RHCE and which lends compelling support for the crossing over model for the origin of the less common RH haplotypes. However, our evidence only supports crossing over where the order of antigen determining sites is C-E-D, and this arrangement was confirmed by direct analysis of genomic DNA.
In an earlier study (12 ), we sequenced a 5.5 kb region that includes exon 2 and parts of introns 1 and 2 of both RHD and the Ce allele of RHCE. We found that these two genes are completely identical over the 4.26 kb that includes exon 2. In the 304 and 935 bp of sequence flanking this region, the two genes are 96% homologous, a figure close to that previously determined for their coding regions (1 -5 ). We have now sequenced the equivalent region of a ce allele of RHCE and find that the converse relationship holds, i.e. that ce is 96% homologous to Ce and RHD over the 4.26 kb that includes exon 2, and except for a single base change, identical to Ce in the 5' and 3' flanking sequence (see Fig. 1 ).
The association of both C and c with the E and e alleles of RHCE, both in the presence and the absence of RHD (Table 1 ) may indicate intra- and/or intergenic recombination, as first suggested by Fisher and Race (9 ). In order to investigate this we sought additional polymorphisms within the RH genes so that more extensive haplotypes could be constructed.
We identified a simple sequence repeat polymorphism in intron 8 of the RHCE gene which displays autosomal co-dominant inheritance (Fig. 3 a). A maximum of two alleles was detected in any individual, confirming the absence of a PCR product from RHD. A survey of 88 RHD+ and 68 RHD- haplotypes revealed that this polymorphism is associated with RHD status (Table 2 ). All D-negative samples, representing all haplotypes, have allele 2, or in two cases the rare allele 1, whereas all D-positives had alleles 3 or 4. The association between alleles of an RHCE intron 8 polymorphism and the presence or absence of RHD, and the less marked association of RHD status with serologically defined C- or E-type (Table 1 ) suggests that RHD is closer to intron 8 than to exons 2 or 5 of RHCE, i.e. that RHD lies to the 3' side of RHCE.abc
There is an AflIII site 3' of exon 2 in the Ce allele which is absent from the ce allele (Fig. 1 ). A second AflIII site, 627 bp 3' of the first, is polymorphic in RHD (starred in Fig. 1 ); ~70% of RHD genes resemble Ce in having both these sites, whereas the remainder have only one. The distribution of the different RHD AflIII alleles in the different RH haplotypes is not random (Table 2 ). The Ce and ce alleles are associated with an RHD with two AflIII sites and cE and CE are associated with an RHD with a single site (examples shown in Fig. 3 b). AflIII alleles of RHD are therefore strongly associated with serologically defined E- and e-antigen type, and are markedly less so with C type. As E/e antigenicity is encoded 3' of C/c, this association is also consistent with a 3' location for RHD.
We sought to obtain direct evidence for the arrangement of the RHD and RHCE genes by analysis of an RH yeast artificial chromosome (YAC) clone (13 ). This clone is ~450 kb and contains both RHCE and RHD sequences. We assayed for the presence of exons 1 and 10 of both genes. At their 3' ends, the two genes were distinguished using a locus-specific duplex PCR designed on exon 10 3' untranslated sequence of each gene (12 ). The RH YAC clearly contains the 3' end of both genes (Fig. 4 , right). Using the intron 1 and 2 assays described above, we showed the presence of both RHD and RHCE introns 1 and 2 (data not shown). The RHCE exon 2 was typical of those expressing c-antigen. Exons 1 of the two genes were distinguished by the presence of a HindIII site in RHCE 1200 bp 5' of exon 1 which is absent from RHD (7 ,13 and our unpublished sequence data). PCR products were generated using primers that flank the RHCE-specific HindIII site and digested with HindIII. As shown in Figure 4 (left) the YAC contains exon 1 sequences from RHD only. This suggests that the 5' end of RHCE is terminal in the YAC, exon 1 having been excluded during cloning. The YAC was isolated from a library of EcoRI-digested DNA, and, by restriction mapping an RHCE cosmid, we located an EcoRI site ~4.5 kb 3' of exon 1. A PCR reaction using RHCE intron 1 antisense primers, together with a YAC arm sense primer (data not shown), confirmed this site as a probable cloning site. These results therefore support the arrangement C-E-D.
Using the results obtained in the present study, together with data on an SphI RFLP in intron 6 reported by Huang et al. (14 ), we can construct a limited number of haplotypes. These are shown in Table 2 . The relationships between them strongly argue, as suggested by Fisher and Race (9 ), for recombination in the generation of those haplotypes that are found at lower frequencies in all populations, namely, Cde, cdE, CdE and CDE. One of these events, that which generates the rare CdE haplotype, requires recombination between two recombinant haplotypes, Cde and cdE. A scheme that is consistent with all the data is shown in Figure 5 a. Because of the large chromosomal segments involved, reciprocal recombination, rather than conversion, seems more likely. Whether these were once-only events, or recurrent events maintaining low frequency haplotypes, as suggested by Fisher and Race (9 ), is not clear. But it may be noted that in three of these proposed schemes the reciprocal product is cDe. Interestingly, Huang et al. (14 ) found cDe donors both with and without an intron 6 SphI restriction site. This results in two different haplotypes, one of which (c, e, intron 6 SphI+, intron 8 allele 3 or 4, RHD AflIII +) is consistent with its being the reciprocal product of either the CDe/cDE or the CDe/cde exchange.
The scheme proposed in Figure 5 a is dependent on the chromosomal orientation 5'-C-E-D-3'; alternative orders require multiple crossovers. This conflicts with the arrangement proposed by Fisher and Race (9 ), and widely accepted since, and with kinship mapping data averaged over major racial groups (15 ); we have no explanation for this inconsistency. However, although there is conclusive molecular evidence placing the genetic determinants of C-series antigens 5' of those determining the E-series (5 ,8 ), there has been no direct evidence for the placement of RHD. Our analysis of a YAC containing both RH genes strongly supports the arrangement C-E-D.
Exon 2 of the C alleles of RHCE is identical to exon 2 of RHD, even at the position that encodes C antigen (5 ,8 ). Why RHD does not therefore direct the synthesis of a polypeptide with C antigen specificity is not known, although some rearranged RHD genes with intact exons 2 may encode an antigen with weak C activity (13 ). The complete identity between RHD and the C alleles of RHCE in the non-coding region surrounding exon 2 strongly suggests that the Ce allele arose as a result of non-reciprocal exchange of RHD sequences into the ce allele of RHCE. Our finding that the C allele has the same structure in all haplotypes indicates that this was most likely a once-only event in human evolutionary history. This event can be dated with some accuracy, since RHD and the C alleles are identical over 4.26 kb, but within 1239 bp of 5' and 3' flanking sequence, the Ce allele differs from c alleles and from RHD by one nucleotide, a substitution which must therefore have occurred after the appearance of the Ce allele. Assuming a nucleotide substitution rate of 4 * 10-9 per nucleotide per year, this suggests Ce first appeared some 0.2 million years ago. The 4.26 kb region appears to act as a recombination hot-spot, as it was found to be the target for the introduction of additional 3' RHD sequence in the chimeric RHCE/RHD genes in five unrelated donors with the very rare D- - phenotype (12 ).
Intergenic exchange seems to be a common event in the RH genes. In addition to those resulting in the D- - phenotype (12 ), recombinant RHD genes have been found in the Cdes phenotype (13 ) and in the so-called category (partial) D phenotypes (16 ,17 ). The latter are a relatively heterogeneous group, which have varying amounts of RHD sequence replaced by homologous RHCE sequence, although, after allowing for ascertainment bias, these appear not to be targeted to any great extent. It is possible that these intergenic events represent rare errors in a process that serves to maintain homogeneity within gene families.
The high degree of homology between the coding regions of the RHCE and RHD genes is consistent with an ancestral gene duplication. This homology breaks down very abruptly in the 3' untranslated region of the RH genes, suggesting that this part of the gene was not involved in the event. The ~4% divergence over the coding region would suggest, assuming an average nucleotide substitution rate of 4 * 10-9 per nucleotide per year, that duplication occurred some 10 million years before the present (BP). This timing is consistent with the finding that the gorilla and chimpanzee are unique among the anthropoid apes in expressing homologues of both the human D and c antigen (18 ). As our analysis of the C alleles strongly suggests it arose by non-reciprocal intergenic exchange of RHD sequences into a ce allele 0.2 million years BP, it follows that RHD is an earlier duplicate of a ce allele (or vice versa) which diverged by ~4% before the exchange. The human lineage thus starts with the haplotype cDe. This is consistent with its very high incidence in black Africans and their descendents (0.4-0.5 compared to <0.1 elsewhere; see Table 1 ). a
DNA was extracted from samples typed locally for Rh. We also used unrelated DNA samples from those CEPH families for whom Rh types are available. All haplotypes were distributed across all racial groups. The numbers of each haplotype examined is shown in Table 2 .
DNA sequencing of Ce and RHD genomic clones was as described (12 ). Intron 1-exon 2-intron 2 sequence of the ce allele was determined on cloned PCR products from cde/cde donors, using primers designed on the Ce/RHD sequence. We used the Sequenase v2.0 kit (Amersham International, UK), as recommended. The region sequenced in RHD and in the Ce and ce alleles of RHCE, delimited by vertical arrows in Figure 1 , extends from 1349 5' to 3972 bp 3' of exon 2 (numbering refers to the Ce allele, as both RHD and the ce allele have insertion/deletion differences). The region of identity between RHD and the Ce allele extends from 1034 bp 5' to 3037 bp 3' of exon 2. Sequence differences between genes were dated following Jukes and Cantor (24 ).
Many of the differences between the RHD, Ce and ce sequences give rise to restriction site changes. We used restriction sites that define the 5' boundary of the region of RHD/Ce homology. These are (i) two TaqI sites 5' of exon 2 in RHD only, which reduce the 606 bp PCR product to 306, 203 and 97 bp and (ii) a HinfI site, also 5' of exon 2, in RHD and Ce only. There are two additional invariant HinfI sites, so that all products yielded fragments of 211 and 172 bp; the residual fragment is reduced by HinfI to 711 and 349 bp in RHD and in C alleles of RHCE. All these sites were detected in PCR products generated using the same sense primer viz; GTTTAAATCTTGGCTGTAGGC, with the following antisense primers: TaqI, CAGCTTGAGCTCCAGAACG, annealing at 58oC; HinfI, ATGAAGAGGTTGAAGGCCAC, annealing at 54oC (all reactions described here, unless stated, are based on 94oC and 72oC denaturing and extending, respectively, with the specified annealing temperature, all for 1 min). The PstI polymorphism that defines the 3' end of the region of identity between RHD and the Ce allele has been described before (25 ,26 ).
There are AflIII sites at 325 and 952 bp 3' of exon 2 in C alleles. The latter site is absent from all c alleles; the former is polymorphic in RHD (starred in Fig. 1 ). The following primers: CATCTCCCCACCGAGC (sense) and CCTAAGATGCCTAGGATGAC (antisense), annealing at 54oC for 1 min and extending at 72oC for 1.5 min generate a 1295 bp PCR product from all genes. That from C alleles and from some RHD genes is reduced by AflIII to 627, 531 and 137 bp, that from other RHD genes to 1158 and 137 bp, and that from the c alleles to 764 and 531 bp. From the cDE (R2) haplotype significant amounts of undigestible heteroduplex are formed, visible in lane 3 of Figure 3 . The intron 8 polymorphism is a simple repeat structure in RHCE of the form (A4T)n. It was amplified using the flanking primers CTGGATGTGGTGGCAGGC (sense) and GACAGTGTCTCAACAGAAATC (antisense), annealing at 59oC, and was analyzed by polyacrylamide gel electrophoresis on 6% denaturing sequencing gels.
The RH YAC has been briefly described before (13 ). It was isolated after screening the partial EcoRI library of Anand et al. (27 ) by PCR using exon 1 primers as described (8 ); a positive signal was consistently obtained at coordinate 38A-A10. It is ~450 kb. We used the locus-specific exon 10 primers described earlier (12 ), which is a duplex reaction in which a single sense primer, designed on exon 10 coding sequence common to both genes, is matched with RHCE- and RHD-specific antisense primers designed on 3' untranslated sequence. The RHCE product is 133 bp and the RHD product is 291 bp. We also devised an assay to distinguish exons 1 of the two genes, which made use of the observation that exon 1 lies on a HindIII fragment that is 2.2 kb in RHD and 2.0 kb in RHCE (7 ,13 ,24 ). We cloned this fragment from an RHD genome, sequenced it and designed primers that yielded a 1.97 kb product containing the RHCE-specific HindIII site. Digestion with HindIII yields a 150 bp fragment from RHCE only. The primers were: sense, CCTGATCTAGGCTACAGT; antisense, ACTGTTCCAATGAACTCTC, annealing at 56oC.
We thank Prof. D.A.Hopkinson and Dr Patricia Tippett for helpful discussions, and Drs Tippett and Geoff Daniels for providing serologically typed blood samples.
1 Arce, M.A., Thompson, E.S., Wagner, S., Coyne, K.E., Ferdman, B.A. and Lublin, D.M. (1993) Molecular cloning of RhD cDNA derived from a gene present in RhD-positive, but not D-negative individuals. Blood82, 651-655MEDLINE Abstract
2 Avent, N.D., Ridgewell, K., Tanner, M.J.A. and Anstee, D.J. (1991) cDNA cloning of a 30 kDa erythrocyte membrane protein associated with Rh (Rhesus) blood group expression. Biochem. J.271, 821-825
3 Le Van Kim, C., Cherif-Zahar, B., Raynal, V., Cherrier, C., Cartron, J-P. and Colin. Y. (1992). Molecular cloning and a primary structure of the human blood group RhD polypeptide. Proc. Natl. Acad. Sci. USA89, 10925-10929MEDLINE Abstract
4 Cherif-Zahar, B., Bloy, C., Le Van Kim, C., Blanchard, D., Bailly,P., Hermand, P., Salmon, C., Cartron, J-P. and Colin, Y. (1991) Molecular structure and protein structure of a human blood group Rh polypeptide. Proc. Natl. Acad. Sci. USA87, 6243-6247.
5 Simse, S., de Jong, C.A.M., Cuijpers, H.T.M., Westers, T.M., Overbeeke, M.A.M., Goldschmeding, R., van der Schoos, C.E. and von dem Borne, A.E.G.K. (1994) Sequence analysis of cDNA derived from reticulocyte mRNAs coding for Rh polypeptides and demonstration of E/e and C/c polymorphisms. Vox. Sang.67, 203-209.
6 Le Van Kim, C., Cherif-Zahar, B., Raynal, V., Mouro, I., Lopez, M., Cartron, J-P. and Colin, Y. (1992) Multiple Rh mRNA isoforms are produced by alternative splicing. Blood80, 1074-1078.MEDLINE Abstract
7 Colin, Y., Cherif-Zahar, B., Le Van Kim, C., Raynal, V., Van Huffel, V. and Cartron J-P. (1991) Genetic basis of the RhD positive and RhD negative blood group polymorphism as determined by Southern analysis. Blood78, 2747-2752MEDLINE Abstract
8 Mouro, I., Colin,Y., Cherif-Zahar, B., Cartron, J-P. and Le Van Kim, C. (1993) Molecular genetic analysis of the human Rhesus blood group system. Nature Genet. 5, 62-65MEDLINE Abstract
9 Fisher, R.A. and Race, R.R (1946) Rh gene frequencies in Britain. Nature (London) 157, 48-49.
10 Cherif-Zahar, B., Le Van Kim, C., Rouillac, C., Ranal, V., Cartron, J-P. and Colin, Y. (1994) Organization of the gene (RHCE) encoding the human blood group RhCcEe antigens and characterizing of the promoter region. Genomics 19, 69-74.
11 Steinberg, A.G. (1965) Evidence for crossing-over or mutation at the Rh locus. Vox. Sang. 10, 721-724.MEDLINE Abstract
12 Kemp, T.J., Poulter, M. and Carritt, B. (1996) A recombination hot-spot in the Rh genes revealed by analysis of unrelated donors with the rare D- - phenotype. Am. J. Hum. Genet. 59, 1066-1073MEDLINE Abstract
13 Blunt, T., Daniels, G. and Carritt, B (1994) Serotype switching in a partially deleted RHD gene. Vox. Sang. 67, 397-401MEDLINE Abstract
14 Huang C-H., Reid M. E., Chen, Y., Coghlan, G. and Okubo, Y. (1996) Molecular definition of red cell Rh haplotypes by tightly linked SphI RFLPs. Am. J. Hum. Genet. 58, 133-142.
15 Morton, N.E. and Simpson, S.P. (1983) Kinship mapping in multilocus systems. Hum. Genet. 64, 103-104.MEDLINE Abstract
16 Rouillac, C., Colin, Y., Hughes-Jones, N.C., Beolet, M., D'Ambrosio, A.M., Cartron, J-P. and Le Van Kim, C. (1995) Transcript analysis of D category phenotypes predicts hybrid Rh D-CE-D proteins associated with alteration of D epitopes. Blood85, 2937-2944.MEDLINE Abstract
17 Mouro, I., Le Van Kim, C., Roulliac, C., van Rhenen, D.J., Le Pennec, P.Y., Bailly, P., Cartron, J-P. and Colin, Y. (1994) Rearrangements of the blood group RhD gene associated with the DVI category phenotype. Blood83, 1129-1135.MEDLINE Abstract
18 Socha, W.W. and Ruffie J. (1983) The rhesus system. In Blood Groups of Primates: Theory, Practice and Evolutionary Meaning. Alan R. Liss, New York. pp 75-90.
19 Haldane, J.B.S. (1942) Selection against heterozygosis in Man. Ann. Eugenics 11, 333-340.
20 Hogben, L. (1943) Mutation and the Rhesus reaction. Nature152, 721-722.
21 Fisher, R.A., Race, R.R. and Taylor, G.L. (1944) Mutation and the Rhesus reaction. Nature153, 106.
22 Li, C.C. (1953) Is the Rh facing a crossroad? A critique of the compensation effect. Am. Naturalist87, 257-261.
23 Feldman, M.W., Nabholz, M. and Bodmer, W.F. (1969) Evolution of the Rh polymorphism: A model for the interaction of incompatibility, reproductive compensation and heterozygote advantage. Am. J. Hum. Genet. 21, 171-193. MEDLINE Abstract
24 Jukes, T.H. and Cantor, C.R. (1960) Evolution of protein molecules. In H.N. Munro (ed.) Mammalian Protein Metabolism. Academic Press, New York.
25 Carritt, B., Blunt, T., Avent, N., Daniels, G. and Steers, F. (1993) Rh null phenotypes are not due to a gross deletion and can occur on different genetic backgrounds. Ann. Hum. Genet.7, 273-279.
26 Poulter, M., Kemp, T.J. and Carritt, B.(1996) DNA-based RH typing: simultaneous determination of Rh C and D status using the polymerase chain reaction. Vox. Sang. 70, 164-168.MEDLINE Abstract
27 Anand, R., Riley, J.H., Butler, R., Smith, J.C and Markham, A.F. (1990) A 3.5 genome equivalent multi access YAC library: construction, characterization, screening and storage. Nucleic Acids Res. 18, 1951-1956.MEDLINE Abstract
*To whom correspondence should be addressed. Tel: +44 171 380 7415; Fax: +44 171 387 3496; Email: ben@galton.ucl.ac.uk
-->
This page is maintained by OUP admin. Last updated Mon May 12 18:10:04 BST 1997. Part of the OUP Journals World Wide Web service.
Copyright
Oxford University Press, 1996
