Blood Journal
Leading the way in experimental and clinical research in hematology

The association of the glycophorin C exon 3 deletion with ovalocytosis and malaria susceptibility in the Wosera, Papua New Guinea

  1. Sheral S. Patel,
  2. Rajeev K. Mehlotra,
  3. William Kastens,
  4. Charles S. Mgone,
  5. James W. Kazura, and
  6. Peter A. Zimmerman
  1. 1 From the Division of Geographic Medicine, Case Western Reserve University, and University Hospitals of Cleveland, Cleveland, OH; and Papua New Guinea Institute of Medical Research, Goroka, Papua New Guinea.


Erythrocyte polymorphisms, including ovalocytosis, have been associated with protection against malaria. This study in the Wosera, a malaria holoendemic region of Papua New Guinea, examined the genetic basis of ovalocytosis and its influence on susceptibility to malaria infection. Whereas previous studies showed significant associations between Southeast Asian ovalocytosis (caused by a 27– base pair deletion in the anion exchanger 1 protein gene) and protection from cerebral malaria, this mutation was observed in only 1 of 1019 individuals in the Wosera. Polymerase chain reaction strategies were developed to genotype individuals for the glycophorin C exon 3 deletion associated with Melanesian Gerbich negativity (GPCΔex3). This polymorphism was commonly observed in the study population (GPCΔex3 frequency = 0.465, n = 742). Although GPCΔex3 was significantly associated with increased ovalocytosis, it was not associated with differences in either Plasmodium falciparumor P vivax infection measured over the 7-month study period. Future case-control studies will determine if GPCΔex3 reduces susceptibility to malaria morbidity.


The geographic overlap between malaria and red blood cell (RBC) disorders led Haldane to hypothesize that many polymorphisms in the human genome have arisen by natural selection to protect from severe malaria infection and thereby increase reproductive fitness of populations living in malaria endemic regions.1In Papua New Guinea, Southeast Asian ovalocytosis, caused by a 27–base pair (bp) deletion in the anion exchanger 1 protein gene (AE1Δ27), is observed in many coastal malaria holoendemic areas and is associated with protection from cerebral malaria.2 3 While AE1Δ27 has not been reported in residents of the malaria holoendemic Wosera region of Papua New Guinea,2 ovalocytic RBCs are common.

Additional polymorphisms characterizing the human population in the Wosera include an exon 3 deletion of the integral membrane sialoglycoprotein glycophorin C (GPC).4-6 This deletion (GPCΔex3) changes serologic phenotypes of the Gerbich (Ge) blood group system in Melanesians.7 Because GPC is involved in maintaining the integral RBC membrane lattice, the association between GPCΔex3 and ovalocytosis was tested. Furthermore, because GPCΔex3 is distributed within a malaria holoendemic region, we tested the association between GPCΔex3 and susceptibility to malaria infection.

Study design

Study population and malaria

The study was conducted in the Wosera region of Papua New Guinea, where all 4 human Plasmodium species are transmitted year-round.8 Blood samples were collected monthly from permanent residents (median age, 17 years; range, 1-86 years) of 6 villages within the Wosera from July 1998 to January 1999. The human investigations institutional review boards of Case Western Reserve University, University Hospitals of Cleveland, and the Papua New Guinea Medical Research Advisory Committee approved all protocols.

Malaria and red blood cell morphology

Thick and thin films stained with 4% Giemsa were prepared at the time of blood collection (Figure 1A-D).8 Malaria parasites were identified by light microscopy. Parasite densities were recorded as the number of parasites per 200 leukocytes (average 8000 leukocytes per microliter).8

Thin smears were examined by light microscopy for the proportion of ovalocytes (erythrocytes with length-width ratio more than 1:1) without knowledge of genotypic results.2 Because previous studies have shown that elliptocytes are rare in the Wosera, the distinction between ovalocytes and elliptocytes (RBC with length:width more than 2:1) was not made.9 Blood smears from North Americans were prepared using a modified Wright stain (Diff-Quick Stain Set, Dade-Behring, Newark, DE). One reader reviewed all of the blood smears. Two additional observers independently reviewed a subset of smears.

Genotyping for AE1 and GPC polymorphisms

Blood was collected in ethylenediaminetetraacetic acid vacutainer tubes and stored at −70°C until DNA extraction was performed with the QIAmp96 DNA blood kit (Qiagen, Valencia, CA). Genotyping of band 3 was performed as previously described.10 New polymerase chain reaction (PCR) genotyping strategies for GPC are described in Figure1E-G.

Fig. 1.

Representative blood smears and GPC genotyping results.

(A) North American control. Arrows indicate examples of ovalocytes: (B) Melanesian, Southeast Asian ovalocytosis (heterozygous for AE1Δ27); (C) Melanesian homozygous for GPCΔex3; and (D) Melanesian heterozygous for GPCΔex3. GPC genotyping; lane 1: homozygous wt (wt/wt); lane 2: GPCΔex3 heterozygote (wt/Δ); lane 3: GPCΔex3 homozygote (Δ/Δ); lane 4: North American, wt/wt genotype. Because exon 2 and exon 3 of the GPC gene are highly homologous, 2 reactions are needed to delineate all 3 genotypes. All PCR amplifications were performed as previously described.12(E) To assess the presence of exons 2 and 3, primers (GPCup 5′-CAGATTCTTGTCCTCTGTTCACAG-3′ and GPCdn 5′-TCAAAACCACCTCTGAGGGAGAG-3′) annealing to conserved sequence around exons 2 and 3 were used (thermocycling program: 94°C for 30 seconds, 68°C for 30 seconds, and 72°C for 30 seconds [× 40]). PCR products were subjected to electrophoresis on a 4% 5:1 GTG NuSieve:LE agarose (FMC Bioproducts, Rockland, ME) gel. A 264-bp band (exon 2) is detected in wt/wt, wt/Δ, and Δ/Δ. A 240 bp band (exon 3) is detected in only wt/wt and wt/Δ. The heteroduplex product is created by hybridization between complementary strands of exon 2 and 3 amplicons. (F) To identify individuals with the wt allele, primers annealing to a 51-bp intron 2 insert (GenBank AF342984) and downstream intron 2 polymorphism (GenBank M24627)5 (GPC349up 5′-GGAAACTGCCGTGACTTCAGA-3′ and 1679dn 5′-CATGGTCCTTGCAACAGTTGC-3′) were used to amplify a 1376-bp product (thermocycling program: 94°C for 30 seconds, 60°C for 30 seconds, and 72°C for 90 seconds [× 40]). PCR products were subjected to electrophoresis on a 1.5% 5:1 GTG NuSieve:LE agarose gel. (G) To identify the GPCΔex3 allele and differentiate between wt/wt and wt/Δ, primers annealing to a 51-bp intron 2 insert and downstream intron 3 polymorphism (GenBank M24628)5 (GPC349up and GPC1680dn 5′-AGGTTAGAATCATACCCCAGG-3′) were used. Because the 3.4-kb deletion of the GPC gene includes exon 3, this PCR produces a 1377-bp band. To ensure that absence of the 1376-bp (D) and 1377-bp (E) bands was not caused by heterogeneity in PCR master mix, an additional primer, GPCdn, along with GPC349up, amplified conserved regions downstream of exon 2. The 155-bp product produced in all genotypes by GPC349up and GPCdn is not shown. All gels were stained with a 1:10 000 dilution of SYBR Gold (Molecular Probes, Eugene, OR), and products were visualized using a Storm 860 (Molecular Dynamics, Sunnyvale, CA).

Statistical analysis

Categorical variables were analyzed by the χ2 test and continuous variables by the Wilcoxon or Kruskal-Wallis test. Statistical Analysis Systems version 8.1 software package (Cary, NC) was used.

Results and discussion

Ovalocytes in the Wosera

To assess the relationship between ovalocytosis and Melanesian ancestry, 13 North Americans and 199 individuals from the Wosera were compared. The frequency of ovalocytes per 1000 RBCs was significantly higher in residents of the Wosera than North Americans (median, 292 ± 131.4; median, 24 ± 22.6, respectively, Wilcoxon,P < .0001). Representative blood smears from a North American Caucasian and 3 Melanesians with different GPC and band 3 genotypes are illustrated in Figure 1A-D.


Previous studies reported an absence of AE1Δ27 in the Wosera (n = 216).2 Consistent with these findings, only 1 of 1019 residents from the Wosera had this mutation (Figure 1B). The PCR product from this individual was cloned and sequenced, verifying that this individual carried AE1Δ27 (not shown).10

GPC genotype in the Wosera

GPC genotyping was performed on 742 individuals (Figure 1E-G). The first reaction, screening for the presence or absence of the wild-type (wt) GPC allele where exons 2 and 3 are both present (Figure 1E), identified homozygous GPCΔex3 individuals (lane 3). The second (Figure 1F) and third reactions (Figure 1G) amplified GPC sequence within the wt or GPCΔex3 alleles, respectively, and allowed homozygous wt individuals (Figure 1F-G, lanes 1 and 4) to be distinguished from heterozygous individuals (Figure 1F-G, lane 2). Allele frequencies for GPC wt and GPCΔex3 were 0.535 and 0.465, respectively. Genotyping showed 211 (28.4%) of 742 individuals as homozygous wt, 372 (50.1%) of 742 as heterozygotes, and 159 (21.4%) of 742 as GPCΔex3 homozygotes. This distribution is in Hardy-Weinberg equilibrium, indicating that GPCΔex3 does not confer a selective disadvantage. This is in contrast to AE1Δ27, a balanced polymorphism, where the disadvantage of lethality in the homozygous form is outweighed by the selective advantage against severe malaria for heterozygotes.2

GPC genotype and ovalocytosis

The association between ovalocytosis and GPC genotype was evaluated in 134 individuals who did not carry AE1Δ27. The wt individuals (n = 32) had the lowest proportion of ovalocytes per 1000 RBCs (median, 238 ± 115.1). Heterozygous individuals (n = 52) had a higher proportion of ovalocytes (median, 297 ± 103.8), while homozygotes (n = 49) had the highest of all 3 genotypes (median, 312 ± 145.9). In a comparison of all 3 genotypes, the proportion of ovalocytes was significantly associated with GPCΔex3 (Kruskal-Wallis, P = .021). Individual comparisons among the 3 genotypic groups showed significant differences by a one-sided Wilcoxon test (wt/wt vs wt/GPCΔex3, P = .0392; wt/wt vs GPCΔex3/GPCΔex3, P = .0045). These results suggest that GPCΔex3 contributes to ovalocytosis in the Wosera. When erythrocyte morphology was compared between homozygous wt individuals from the Wosera and North Americans, the former had a significantly increased ovalocyte frequency (Wilcoxon, P < .0001). This suggests that altered RBC morphology in the Wosera is a heterogenous condition caused by additional unknown mutations in RBC membrane proteins, such as protein 4.1 and spectrin as well as environmental or nutritional factors.

GPC genotype and infection status

The prevalence of infection with P falciparum orP vivax determined by blood smear has been examined in relation to serologic Ge antigen status in one published study of 266 people.11 This study observed a lower combined smear positive rate for P falciparum and/or P vivaxinfection in Ge-negative individuals, suggesting that Ge antigen negativity protects against infection.11 To examine the relationship between GPC genotype and susceptibility to malaria infection more rigorously, we studied a larger population over 7 months. Results for genotype and infection status were available for 325 to 696 individuals at each of 7 monthly intervals. This analysis showed that the prevalence or density of P falciparum andP vivax infection was not significantly different for individuals in the 3 GPC genotypic groups at any time (Table1). These results parallel findings of other RBC polymorphisms, such as AE1Δ27, where genotypic differences are associated with reduced susceptibility to severe malaria morbidity with no effect on susceptibility to infection.3 The relationship of GPCΔex3 to malaria morbidity in young children, the age group most susceptible to the clinical phenotype, requires further study.

View this table:
Table 1.

Glycophorin C genotype and infections over time


We thank the residents of the Wosera for participating in this study. We thank Chloe Hill for reviewing blood smears.


  • Sheral S. Patel, Pediatric Infectious Diseases, Div of Geographic Medicine, School of Medicine W123, 2109 Adelbert Rd, Cleveland, OH 44106; e-mail: ssp6{at}

  • Supported by National Institutes of Health grants AI-36478-07 and AI-07024.

  • The publication costs of this article were defrayed in part by page charge payment. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 U.S.C. section 1734.

  • Submitted January 30, 2001.
  • Accepted July 30, 2001.


View Abstract