and Immunity (2000) 1, 325–329  2000 Macmillan Publishers Ltd All rights reserved 1466-4879/00 $15.00 www.nature.com/gene Identification of one (CR1) polymorphisms in West

JM Moulds1, L Kassambara2, JJ Middleton1, M Baby2, I Sagara2, A Guindo2, S Coulibaly2, D Yalcouye2, DA Diallo2, L Miller3 and O Doumbo2 1University of Texas-Houston Medical School, Houston, TX, USA; 2University of Mali, Bamako, Mali, W. Africa; 3National Institute of Allergy and Infectious Disease, NIH, Bethesda, MD, USA

Complement receptor one (CR1) is a ligand for the rosetting of infected red cells with uninfected cells. Since CR1 exhibits three known polymorphisms, we studied European-Americans (n = 112) and African-Americans (n = 330) and Malians (n = 158) to determine if genetic differences existed in an area endemic for that could offer a survival advantage. The frequencies of Knops group McC(b+) and Sl(a−) were greatly increased in Africans vs Europeans. Although the frequency of McC(b+) was similar between Africans from the USA or Mali, the Sl(a−) was significantly higher in Mali (39% vs 65%, respectively). There was an increased frequency of the largest size (250 kD) of CR1 in Mali, but this did not differ significantly from the USA (P = 0.09). Both cohorts of Africans had higher expression of red cell CR1 than European-Americans but this showed little difference between the USA and Mali groups. Thus, the most important CR1 polymorphism relevant to rosetting of malaria infected cells appears to be the Knops blood group. Genes and Immunity (2000) 1, 325–329.

Keywords: malaria; blood groups; complement receptors

Introduction found in a study of African-Americans.5 Although there appears to be genetic regulation of E-CR1, low CR1 copy Complement receptor one (CR1) is a found numbers can also be an acquired phenomenon in malig- on most peripheral blood cells including red blood cells. nant and inflammatory disorders.6–8 Its major function is the adherence to and removal of In 1991, two groups independently identified the immune complexes, however, it also has complement Knops blood group system on CR1.9,10 It has been shown 1 regulatory activity. Human CR1 exhibits three known to be independent from the other known CR1 polymor- polymorphisms including: molecular weight (structural), phisms, ie, molecular weight or red cell expression. The level of red cell expression, and the Knops blood group. Knops blood group system began with the description There are four molecular weight variants that differ in of anti-Knops (Kna), in a transfused Caucasian woman.11 size by approximately 30 kD yet possess similar Factor I Presently it consists of seven including the 2 co-factor activity. The molecular weights for these vari- allelic pairs Kna/Knb (Knops), McCa/McCb (McCoy), ants determined by SDS-PAGE using non-reduced con- Sla/Vil (Swain-Langley/Villien) and Yka (York). McCb = = = ditions are: CR1*1 190 kD, CR1*2 220 kD, CR1*3 and Vil have not been officially recognized by the Inter- = 3 160 kD and CR*4 250 kD. Holers et al showed that the national Society of Blood Transfusion12 as they lack com- difference in molecular weight was due to differing sizes plete biochemical and genetic studies for inclusion in the of mRNA and not post-translational processing. The CR1 system. To date, none have been identified at the molecu- size polymorphism is not unique to red cells but also lar level. The serological null phenotype is known as the exists on other cells such as and eosinophils. ‘Helgeson phenotype’ which results from very low CR1 The number of CR1 copies on erythrocytes (E-CR1) can copy numbers on the red cell.13 vary as much as ten-fold between individuals. In Euro- The importance of CR1 in malaria became evident pean-Americans, a Hind II RFLP has been identified when Rowe et al14 reported that CR1 was a ligand to which correlates with high (H ) or low (L allele) which the parasite receptor, ie, PfEMP-1, bound when 4 expression. However, a similar correlation was not Plasmodium falciparum infected red cells rosetted with uninfected cells. Red cells with low CR1 copy numbers did not rosette or rosetted poorly. In addition, red cells Correspondence: Joann M Moulds, Department of Microbiology and that typed as Sl(a−) formed fewer rosettes than Sl(a+) red Immunology, MCP Hahnemann University School of Medicine, 2900 cells. In an earlier study,15 the Sl(a−) phenotype was Queen Lane, Room G44, Philadelphia, PA 19129–1096, USA. found with an increased frequency among African- E-mail: [email protected] Americans compared to European-Americans. Since This research was supported by grants from the National Institutes rosetting had previously been correlated with severity of of Health RO1 AI 42367 (JMM) and the National Blood Foun- 16 14 a dation (JMM). disease, Rowe et al suggested that the Sl mutation Received 20 December 1999; revised 25 January 2000; accepted 8 might protect against severe malaria in African derived February 2000 populations. Thus, we studied a population in West CR1 polymorphisms in West Africa JM Moulds et al 326 Table 1 frequencies for CR1 structural variants. Correspond- Table 2 Comparison of E-CR1 copy numbers between European- ing molecular weight of the is noted in parenthesis Americans and Africans

Gene frequency CR1*1 CR1*2 CR1*3 CR1*4 European-Americans African-Americans Malians (190 kD) (220 kD) (160 kD) (240 kD) (n=65) (n=60) (n=158)

Tienequebougou Average 284 389 410 (n = 46) 0.76 0.16 0.06 0.02 Std. Dev. 130 125 173 Bamako Median 311 362 382 (n = 40) 0.81 0.14 0.03 0.03 Bancoumana (n = 53) 0.84 0.08 0.06 0.02 Total Mali = derived groups, the average levels of E-CR1 were higher (n 139) 0.80 0.13 0.05 0.02 than in White subjects (Table 2). The Mali data had a African-Americans = (n = 330)a 0.84 0.12 0.04 Ͻ0.01 somewhat broader range of values (s.d. 173) when com- European- pared to the other two groups (s.d. = 125 and 130). Americans (n = 112)a 0.87 0.11 0.02 Ͻ0.01 Knops blood group The results for the Knops blood grouping are shown in aReported in Moulds et al.20 Table 3. Although the McCb is rare among European-Americans it was found in ෂ40% of both Africa to determine whether any of the CR1 polymor- African-Americans and West Africans from Mali. Of the phisms were more frequent in Africans as compared to 11 families studied, three were informative for the African or European-Americans. McCoy (Figure 2). In families three and 20 the mother carries the McC(b+) haplotype which is passed to the child while in family eight the paternal haplotype is Results McC(b+). These families provide evidence that McCb is

CR1 molecular weight The gene frequencies for the four molecular weight vari- Table 3 Phenotype frequencies of the Knops blood group antigens in unrelated European-Americans and Africans ants of CR1 are shown in Table 1. The gene frequencies for CR1*1, CR1*2 and CR1*3 occurred with the same fre- Phenotype Kn(a+) McC(a+) McC(b+) Sl(a−) quency in Africa as in the USA. Of note, the CR1*3 allele (160 kD) has been found more often in African- 17 Tienequebougou Americans than any other population tested. The larg- (n = 46) 99% 95% 44% 72% est size of CR1 (250 kD, CR1*4) was found in 4.6% of Bamako Malians as compared to 1% of African-Americans (P = (n = 40) 100% 93% 53% 72% 0.09). We also noted that not all CR1 structural variants Bancoumana = were equally expressed (Figure 1) but this differential (n 53) 100% 82% 56% 68% expression did not correlate with any particular CR1 Total Mali (n = 139) 100% 89% 49% 70% size allele. African-Americans (n = 150) 99% 90% 44% 39% E-CR1 expression European-Americans E-CR1 was quantified in 65 European-Americans, 60 (n = 112) 98% 98% 1% 1% African-Americans and 158 Malians. In both African-

Figure 1 Differential expression of CR1 structural is noted using equal amounts of sample in all lanes. Lane 1 shows low expression of the CR1*4 allele. Lane 3 is a donor with the Helgeson phenotype that is the result of low E-CR1. This allele is inherited from the mother shown in lane 2. Lanes 4–8 are unrelated Malian donors having normal expression of E-CR1.

Genes and Immunity CR1 polymorphisms in West Africa JM Moulds et al 327 the allele to McCa and is inherited in a Mendelian domi- nant manner. Of greater interest perhaps, is the incidence of Sla. The Sl(a−) phenotype is rare in European-Americans (Ͻ1%) but found in ෂ39% of African-Americans. However, all three sites in West Africa had a high frequency of the Sl(a−) phenotype ranging from 68–72% suggesting that this mutation may have a survival advantage. The Helge- son or null phenotype occurred in 2% of the Mali donors which is similar to other populations. Red cells having the Helgeson phenotype type as negative for all of the Knops antigens and possess low E-CR1 numbers.

Discussion The different CR1 molecular weight variants are the pro- ducts of four different mRNA rather than varying glycos- ylation.3 It had been proposed that they arose through the duplication and of long homologous regions of a primordial CR1 gene. Wong et al18 showed that the smaller form of CR1 (FЈ or CR1*3) had a reduced capacity to bind immune complexes (ICs). It had been reported that this small form of CR1 was a genetic risk factor for SLE;19 however, other studies have found no difference in the structural allele frequencies between SLE patients and matched controls.20 We found that CR1*3 occurred with similar gene frequencies in both African-Americans and Malians. Presumably, the larger forms (S or CR1*2 and CR1*4) which have more -binding sites would have an enhanced ability to remove ICs. Although SLE is uncommon in Mali (D. Diallo, unpublished data), it is unknown whether the CR1 structural alleles are protec- tive genetic factors in this population. The increased fre- quency of the larger forms of CR1 in the West African population may be in response to infectious diseases including viruses, bacteria and parasites. In addition to the size variant, the number or CR1 mol- ecules on the red cell can affect its ability to bind ICs. We found that the average E-CR1 among Africans was higher than European-Americans. In the latter group, the expression polymorphism is linked to a Hind III restric- tion fragment length polymorphism located in an intronic region of the CR1 gene.4 A recent report by Herrera et al5 suggests that there is little correlation between this RFLP and CR1 expression in African-American donors. The actual genetic elements controlling red cell expression have yet to be identified in either population. Screening of 2091 unrelated blood donors of unknown race using the original anti-Kna yielded only four Kn(a−) individuals.11 Molthan21 found 1.2% of random African- Americans to be Kn(a−). In Mali, two donors typed as Kn(a−), however, one had the Helgeson, ie, null pheno- type, and can be considered a false negative. The second Mali donor was also negative for the low frequency allele Knb suggesting the presence of a third allele or an amorph. Molthan15 had previously speculated that a third allele might be present at this locus among African-Americans. 22 Figure 2 Malian families exhibiting inheritance of the McCoy anti- Molthan and Moulds tested 948 European-Americans gens. and 624 African-Americans from Philadelphia and found that McCa was a high frequency antigen in both groups (98.5% and 96.7%, respectively). In our study, the frequency of McCa was lower in Bancoumana than at the other two sites but this was due to the increased fre- quency of its allele McCb. Otherwise, the McCoy

Genes and Immunity CR1 polymorphisms in West Africa JM Moulds et al 328 frequencies were similar between the African-Americans were separated by SDS-polyacrylamide gel and Malians. The majority of donors used by Rowe et al14 electrophoresis using a 3% stacking gel and a 5% separat- for the rosetting studies were McC(b−), therefore, the ing gel at 150V for 1 h. The proteins were electrophoret- effect of this mutation on rosetting is unknown at this ically transferred to nylon membrane and blocked over- time. night in 5% milk powder/PBS. The blots were incubated Of the Malian families obtained for the study of inherit- for 1 h at 24°C with a monoclonal anti-CR1 (J3D3) then ance, we found three to be informative for the McCoy washed three times in PBS/Tween. This was followed by antigens (Figure 2). McCa and McCb were co-dominantly a 1-h incubation with goat anti-mouse IgG conjugated to expressed, ie, McC(a+b+), not only in the families but also horseradish peroxidase. The bands were visualized using in the entire African population studied and are allelic. a chemiluminescence method (ECL, Amersham, USA) Combining the USA and Malian data for the McCoy and compared to known molecular weight markers. locus results in the following gene frequencies: McCa = 0.72 and McCb = 0.28 among Africans. These studies pro- Determination of E-CR1 copy number vide data for the eventual inclusion of McCb in the Knops E-CR1 copy number was determined using modified ver- blood group. sion of previously described ELISA.13 Briefly, membrane Other antigens have been reported to be part of the ghosts were solubilized in NP 40 and 100 ␮l was added Knops blood group including: Sla or McCc, Vil or McCd, to a microtiter plate previously coated with anti-CR1 McCe, McCf, etc.15 Molthan reported that 49% of the 371 (J3D3) and blocked with 0.05% BSA. The plates were random African-American donors from Pennsylvania incubated for 2 h at 24°C and washed in PBS/0.05% that she tested were McCc−, ie, Sl(a−), but in our study of Tween. Next, 100 ␮l of biotinylated anti-CR1 (E11) was African-Americans (from Houston, TX, and Birmingham, added, incubated for 2 h at 24°C and washed as before. AL, USA) the frequency was only 39%. This differed sig- Finally, 100 ␮l of strepavidin were added and the plates nificantly from the Malian donors where 70% were Sl(a−). incubated at 24°C for 20 mins. followed by three washes. Furthermore, less than 1% of European-Americans have Color development utilized a hydrogen peroxide: tetra- the Sl(a−) phenotype. If indeed the Sl(a−) mutation pro- methyl benzidine substrate (Pierce Chemical, USA). The

tects patients from severe malaria, then one would expect reaction was stopped with 0.5 M H2SO4 and the color an increased frequency of the Sl(a−) phenotype in a pure read at 450 nm in a microtiter-plate spectrophotometer. African population. Certainly, there is precedence for The E-CR1 number was extrapolated from a standard these observations within the Duffy (Fy) blood group. curve using a donor of known copy number (determined The Fy(a−b−) phenotype occurs in 60% of African- during the VIIth International Complement Genetics Americans but 100% of West African populations; red Workshop).24 cells with this type are unable to be invaded by Plasmod- 23 ium vivax. Determination of Knops blood group phenotypes The CR1 protein appears to play a major role in the Anti-Kna, -McCa, -McCb and -Sla were obtained from mul- rosetting of Plasmodium falciparum infected red cells to tiply transfused patients who had been immunized uninfected cells. This phenomenon has been correlated through transfusion to the CR1 polymorphisms. These with the increased severity and mortality of falciparum antibodies were initially identified by AABB Reference malaria. Thus, it might be expected that several Laboratories and confirmed by us. The same antisera mutations would occur in this gene to impede the para- were used to type the samples in this study. When poss- site. Our data suggest that the blood group mutation, ible ABO compatible sera were used and for all others − Sl(a ) may offer a protective advantage from severe sera absorbed free of A and B isoagglutinins were util- malaria. However, since these are the first ever studies ized. Test red blood cells were washed and re-suspended of CR1 polymorphisms in Africa, further testing of other to a 4% concentration using PBS. Fifty microliters of the African populations needs to be performed. cell suspension was added to 75 ␮l of each antibody specificity, incubated 1 h at 37°C followed by an anti- Materials and methods globulin test using rabbit polyspecific antisera. The The appropriate Human Use Committees approved the results were read macroscopically using an agglutination research protocol at both the University of Texas Health viewer and scored on a scale of 0–4+. Science Center and the University of Mali. After obtaining consent, ACD blood samples were drawn from 112 White controls, 330 random Black controls from Acknowledgements across the USA and 158 donors from three sites in Mali. The authors would like to acknowledge the technical Tienequebougou is a small village of Bambaran ethnicity, assistance of John J Moulds MT(ASCP)SBB and Dr Rob- Bamako is a large metropolitan area having mixed eth- ert Gwadz. nicity and Bancoumana is a village primarily of Malinke ethnicity. The Malian cohort included 11 families that were studied for inheritance patterns but the related References = donors (n 19) were excluded from gene frequency cal- 1 Birmingham DJ. Erythrocyte complement receptors. Crit Rev culations. Red cell ghosts were immediately prepared on Immunol 1995; 15: 133–154. site and kept frozen at −80°C until tested and serological 2 Seya T, Holers VM, Atkinson JP. Purification and functional studies were performed within 48 h. analysis of the polymorphic variants of the C3b/C4b receptor (CR1) and comparison with H, (C4bp), and decay accelerating Determination of molecular weight factor (DAF). J Immunol 1985; 135: 2661–2667. Red cell ghosts were prepared using hypotonic lysis and 3 Holers VM, Chaplin DD, LeyKam JF, Gruner BA, Kumar V, the membrane proteins were solulibilized in SDS. The Atkinson JP. Human complement C3b/C4b receptor (CR1)

Genes and Immunity CR1 polymorphisms in West Africa JM Moulds et al 329 mRNA polymorphism that correlates with the CR1 allelic mol- testing for CR1-related (Knops/McCoy/Swain-Langley/York) ecular weight polymorphism. Proc Natl Acad Sci 1987; 84: blood group antigens: negative and weak reactions are caused 2459–2463. by variable expression of CR1. Vox Sang 1992; 62: 230–235. 4 Wilson JG, Murphy EE, Wong WW, Klickstein LB, Weis JH, Fea- 14 Rowe JA, Moulds JM, Newbold CI, Miller LH. P. falciparum ron DT. Identification of a restriction fragment polymorphism rosetting mediated by a parasite-variant erythrocyte membrane by a CR1 cDNA that correlates with the number of CR1 on protein and complement-receptor 1. Nature 1997; 388: 292–295. erythrocytes. J Exp Med 1986; 164: 50–59. 15 Molthan L. Expansion of the York, Cost, McCoy, Knops blood 5 Herrera AH, Xiang L, Martin SG, Lewis J, Wilson JG. Analysis group system: the new McCoy antigens McCc and McCd. Med of complement receptor type 1 (CR1) expression on erythrocytes Lab Sci 1983; 40: 113–121. and of CR1 allelic markers in Caucasian and African American 16 Rowe A, Obeiro J, Newbold CI, Marsh K. Plasmodium falciparum populations. Clin Immunol Immunopath 1998; 87: 176–183. rosetting is associated with malaria severity in Kenya. Infect 6 Currie MS, Vala M, Pisetsky DS, Greenberg CS, Crawford J, Immun 1995; 63: 2323–2326. Cohen HJ. Correlation between erythrocyte CR1 reduction and 17 Klickstein LB, Moulds JM. CR1. In: Morley BJ, Walport MJ (eds). other blood proteinase markers in patients with malignant and The Complement Facts Book. Academic Press: New York, pp inflammatory disorders. Blood 1990; 75: 1699–1704. 136–145. 7 Walport MJ, Lachmann PJ. Erythrocyte , 18 Wong WW, Farrell SA. Proposed structure of the FЈ allotype of immune complexes, and the rheumatic diseases. Arthritis Rheum human CR1. Loss of a C3b binding site may be associated with 1988; 31: 153–158. altered function. J Immunol 1991; 146: 656–662. 8 Hara T, Kojima A, Fukuda H et al. Levels of complement regu- 19 Van Dyne S, Holers MV, Lublin DM et al. The polymorphism latory proteins CD35 (CR1), CD46 (MCP) and CD55 (DAF) in of the C3b/C4b receptor in the normal population and in human haematological malignancies. Br J Haematol 1992; 82: patients with systemic erythematosus. Clin Exp Immunol 368–373. 1987; 68: 570–579. 9 Moulds JM, Nickells MW, Moulds JJ, Brown MC, Atkinson JP. 20 Moulds JM, Reveille JD, Arnett FC. Structural polymorphisms The C3b/C4b receptor is recognized by the Knops, McCoy, of complement receptor 1 (CR1) in systemic lupus ery- Swain- Langley, and York blood group sera. J Exp Med 1991; thematosus (SLE) patients and normal controls of three ethnic 173: 1159–1163. groups. Clin Exp Immunol 1996; 105: 302–305. 10 Rao N, Ferguson DJ, Lee SF, Telen MJ. Identification of human 21 Molthan L. The serology of the York-Cost-McCoy-Knops red erythrocyte blood group antigens on the C3b/C4b receptor. blood cell system. Am J Med Tech 1983; 49: 49–56. J Immunol 1991; 146: 3501–3507. 22 Molthan L, Moulds J. A new antigen, McCa (McCoy), and its 11 Helgeson M, Swanson J, Polesky HF. Knops-Helgeson (Kna), a relationship to Kna (Knops). Transfusion 1978; 18: 566–568. high frequency erythrocyte antigen. Transfusion 1970; 10: 737– 23 Miller L. Impact of malaria on genetic polymorphism and gen- 738. etic diseases in Africans and African Americans. Proc Natl Acad 12 Daniels GL, Anstee DJ, Cartron JP et al. Blood group termin- Sci 1994; 91: 2415–2419. ology 1995. ISBT working party on terminology for red cell sur- 24 Moulds JM, Brai M, Cohen J et al. Reference typing report for face antigens. Vox Sang 1995; 69: 265–279. complement receptor 1 (CR1). Exp Clin Immunogenet 1998; 15: 13 Moulds JM, Moulds JJ, Brown MC, Atkinson JP. Antiglobulin 291–294.

Genes and Immunity