Glycophorin C (Gerbich Antigen Blood Group) and Band 3 Polymorphisms in Two Malaria Holoendemic Regions of Papua New Guinea

Abstract

The geographic overlap between the prevalence of erythrocyte polymorphisms and malaria endemicity is thought to be an example of natural selection on human populations. In Papua New Guinea (PNG), the Gerbich-negative phenotype is caused by an exon 3 deletion in the glycophorin C gene (GYPCΔex3) while heterozygosity for a 27-base pair deletion in the SLC4A1 gene (anion exchanger 1 or erythrocyte membrane protein, band 3), SLC4A1Δ27, results in Southeast Asian ovalocytosis. Two geographically and ethnically distinct malaria endemic regions of PNG (the Wosera [East Sepik Province] and Liksul [Madang Province]) were studied to illustrate the distribution of two prominent deletion polymorphisms (GYPCΔex3 and SLC4A1Δ27) and to determine if the genetic load associated with SLC4A1Δ27 would constrain independent assortment of GYPCΔex3 heterozygous and homozygous genotypes. The frequency of the GYPCΔex3 allele was higher in the Wosera (0.463) than Liksul (0.176) (χ²; P < 0.0001). Conversely, the frequency of the SLC4A1Δ27 allele was higher in Liksul (0.0740) than the Wosera (0.0005) (χ²; P < 0.0001). No individuals were homozygous for SLC4A1Δ27. In 355 Liksul residents, independent assortment of these two deletion polymorphisms resulted in 14 SLC4A1Δ27 carriers heterozygous for GYPCΔex3 and one SLC4A1Δ27 carrier homozygous for GYPCΔex3 (Fisher’s exact test; P = 0.8040). While homozygosity for SLC4A1Δ27 appears to be nonviable, the GYPCΔex3 allele is not lethal when combined with SLC4A1Δ27. Neither mutation was associated with altered susceptibility to asymptomatic Plasmodium falciparum or P. vivax infection. While these erythrocyte polymorphisms apparently have no effect on blood-stage malaria infection, their contribution to susceptibility to clinical malaria morbidity requires further study.

INTRODUCTION

Malaria has been one of the most important environmental influences of natural selection in human populations [1]. The protective effect of various erythrocyte polymorphisms against severe disease caused by Plasmodium falciparum infection has been suggested by the epidemiologic association of these mutations in populations where malaria currently or historically has high endemicity [1–5]. Since children who are susceptible to severe malaria die before reproductive age, alleles that offer protection from severe disease would be expected to rise in frequency in these populations. Identifying these alleles could determine which children are at high risk for malaria morbidity while expanding knowledge of the mechanisms of disease and protection from malaria.

The erythrocyte polymorphism referred to as Southeast Asian ovalocytosis (SAO) is found in malaria endemic regions of Papua New Guinea (PNG) [6]. It is strongly associated with heterozygosity for a 27-base pair deletion of the anion exchanger 1 or erythrocyte membrane protein, band 3 gene (SLC4A1Δ27) on chromosome 17 [7,8]. Band 3 is an integral red blood cell membrane protein that plays an important role in maintaining the erythrocyte cytoskeleton and shape, and functions as the principal exchanger of bicarbonate for chloride in the process of removing carbon dioxide from tissues [9–11]. While associations between heterozygosity for SLC4A1Δ27 and prevalence and parasitemia of asymptomatic blood-stage P. falciparum and P. vivax infections may be equivocal, the selective advantage of this mutation is demonstrated by its powerful protection against cerebral malaria [12,13]. This advantage is balanced by presumed nonviability in the homozygous state [6,13].

Another integral red blood cell membrane protein polymorphism found in malaria holoendemic regions of Papua New Guinea is the Gerbich-negative phenotype caused by an exon 3 deletion in the glycophorin C gene (GYPCΔex3) on chromosome 2 [14–19]. Our recent studies in the Wosera region of PNG have shown that these residents have ovalocytic red blood cells but do not possess the SLC4A1Δ27 mutation [20]. In this population, the proportion of ovalocytes on blood smear progressively increases with the presence of GYPCΔex3 so that individuals who are homozygous for the deletion have a greater percentage of ovalocytes than heterozygotes [20]. When malaria infection status was examined at monthly intervals over a 7-month period, we found that blood-stage P. falciparum or P. vivax infection was not affected by GYPCΔex3 [20]. These findings parallel those for SLC4A1Δ27 and susceptibility to blood-stage malaria infection.

While the prevalence of SLC4A1Δ27 and GYPCΔex3 has been correlated with malaria endemicity, the distribution of, and genotype–genotype interactions between these two mutations in the Wosera and Liksul regions of PNG has not been investigated [6,14].

Both SLC4A1 and GYPC are integral red blood cell membrane proteins. While it has been suggested that homozygosity for SLC4A1Δ27 is lethal, we were interested in learning whether the combination of SLC4A1Δ27 and GYPCΔex3 alleles would have a similar deleterious effect. In order to test this, we first determined the distribution of SLC4A1Δ27 and GYPCΔex3 in two distinct populations residing in different malaria endemic areas of PNG. Second, we determined whether the presence of one polymorphism influenced the distribution of the other. Would the apparent genetic load of SLC4A1Δ27 constrain independent assortment of GYPCΔex3 heterozygous and homozygous genotypes? Finally, we determined the relationship of SLC4A1Δ27 and GYPCΔex3 genotypes to asymptomatic malaria infection status.

MATERIALS AND METHODS

Study Population

Cross-sectional studies were conducted in villages within the Wosera (East Sepik Province) and Liksul (Madang Province) areas of Papua New Guinea. These are historically isolated populations from two different linguistic groups that are separated by 350 km of rugged terrain [21–23]. There is little migration into and out of either region. Intra-village and intervillage marriages are equally common. Marriages within patrilineal clans are considered taboo. Nearly all of the residents of the study villages volunteered to participate. All four human Plasmodium species are transmitted year round in both areas [21–23]. Blood samples were collected from Wosera residents in July 1998 and from Liksul residents in May through August 2000. A subset of the samples included in our previous study in the Wosera was used [20]. 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. Informed consent was obtained from study participants.

Malaria Infection Status

Thick and thin blood films were prepared with 4% Giemsa stain and read by light microscopy [21]. Parasite densities were recorded as the number of parasites per 200 leukocytes (average 8,000 leukocytes/μl) [21].

Genotyping for Band 3 and Glycophorin C Polymorphisms

Blood was collected in EDTA 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 and glycophorin C was performed as previously described [7,20]. All Wosera genotypes are a subset of data presented in our previous study in the area [20].

Statistical Analysis

Categorical variables were analyzed by the chi-square or Fisher’s exact test. Continuous variables were analyzed by the Wilcoxon or Kruskal–Wallis test. The Statistical Analysis Systems version 8.1 (SAS, Inc., Cary, NC) software package was used.

RESULTS

Frequency of the SLC4A1Δ27 and GYPCΔex3 in the Wosera (East Sepik Province) and Liksul (Madang Province)

Only one person of 976 (0.1%) tested from the Wosera carried SLC4A1Δ27 (Table I). In Liksul, 54 of 365 (14.8%) residents carried the mutation. Consistent with previous studies, no homozygous SLC4A1Δ27 individual was identified [6,20]. The allele frequency of SLC4A1Δ27 was significantly higher in Liksul than the Wosera (0.0740 vs. 0.0005, respectively; Fisher’s exact, P < 0.0001). In the Wosera, 152 of 705 (21.6%) participants were homozygous for GYPCΔex3. In Liksul, 11 of 357 (3.1%) individuals were GYPCΔex3 homozygotes (Table II). The allele frequency for GYPCΔex3 was significantly higher in the Wosera than Liksul (0.463 versus 0.176, respectively; χ² = 161; P<0.0001). In both the Wosera and Liksul, these frequencies were in Hardy–Weinberg proportions (χ² = 0.0151, P = 1 and χ² = 0.0018, P = 1, respectively).

TABLE I.

Band 3 Genotype in the Wosera and Liksul Areas^*

Band 3 genotype	Wosera (n = 976)	Liksul (n = 365)
	n (%)	n (%)
wt/wta	975 (99.9)	311 (85.2)
wt/Δ27b,c	1 (0.1)	54 (14.8)

^*All Wosera genotypes are a subset of data presented in Patel et al. [20].

^aHomozygous wild type individuals.

^bHeterozygotes for band 3 deletion.

^cNo homozygote for the band 3 deletion was identified.

TABLE II.

Glycophorin C Genotype in the Wosera and Liksul Areas^*

Glycophorin C genotype	Wosera (n = 705)	Liksul (n = 357)
	n (%)	n (%)
wt/wta	204 (28.9)	242 (67.8)
wt/Δex3b	349 (49.5)	104 (29.1)
Δex3/Δex3c	152 (21.6)	11 (3.1)

^*All Wosera genotypes are a subset of data presented in Patel et al. [20].

^aHomozygous wild type individuals.

^bHeterozygotes for glycophorin C exon 3 deletion.

^cHomozygotes for glycophorin C exon 3 deletion.

Distribution of Band 3 and Glycophorin C Genotypes in Liksul

Because the SLC4A1Δ27 and GYPCΔex3 allele frequencies were both elevated in Liksul, we were able to investigate the interactions between these two erythrocyte membrane protein deletion polymorphisms. For this analysis the distribution of SLC4A1 and GYPC alleles was examined in 355 Liksul residents (Table III). Based on SLC4A1Δ27 and GYPCΔex3 allele frequencies and assuming independent assortment, the expected number of individuals calculated for each genotype combination was similar to the number observed in the study population (Fisher’s exact, P = 0.8040). In addition, as predicted by allele frequencies, 15 individuals carried both the SLC4A1Δ27 and GYPCΔex3 polymorphisms.

TABLE III.

Observed and Expected Distribution of Glycophorin C and Band 3 Genotypes in 355 Liksul Residents

Glycophorin C genotype	Band 3 genotype
	wt/wta	wt/Δ27b
	Observedc (expected)d	Observed (expected)
wt/wte	203 (207)	39 (34)
wt/Δex3f	88 (88)	14 (14)
Δex3/Δex3g	10 (9)	1 (1)

^aHomozygous wild type individuals.

^bHeterozygotes for band 3 deletion.

^cNumber of individuals observed with genotype.

^dNumber of individuals predicted to have genotype.

^eHomozygous wild type individuals.

^fHeterozygotes for glycophorin C exon 3 deletion.

^gHomozygotes for glycophorin C exon 3 deletion.

Association of Band 3 and Glycophorin C Genotypes to Malaria Infection Status

Next, we determined whether the presence of both SLC4A1 and GYPC polymorphisms would influence susceptibility to asymptomatic blood-stage malaria infection. Baseline characteristics of the two study populations were similar with respect to age, gender, and hemoglobin levels. In the Wosera (n = 1,690), the median age was 17.5 years, with 53% females and median hemoglobin 10.7 g/dl. In Liksul (n = 1,024), the median age was 18 years, with 48% females and median hemoglobin 11.1 g/dl. The prevalence of P. falciparum infection by blood smear was also similar (20.2% in the Wosera vs. 21.8% in Liksul). The prevalence of P. vivax infection was twice as high in the Wosera (17.7%) versus Liksul (8.8%).

The prevalence and density of P. falciparum and P. vivax infections was examined in relation to different SLC4A1 and GYPC genotypes in Liksul (Table IV). For both P. falciparum and P. vivax, different genotypic combinations did not influence the mean log₁₀ transformed parasitemia or prevalence rates in the population. In both populations, glycophorin C genotype alone did not influence P. falciparum and P. vivax parasitemia and prevalence rates.

TABLE IV.

Glycophorin C, Band 3, and Susceptibility to Malaria Infection in Liksul, Papua New Guinea

Genotype		n	P. falciparum		P. vivax
Glycophorin C	Band 3		Mean parasitemiaa	Prevalence (%)	Mean parasitemiaa	Prevalence (%)
wt/wtb	wt/wtb	203	0.5619 ± 1.0914	46 (22.7)	0.2687 ± 0.7702	24 (11.8)
wt/wt	wt/Δ27c	39	0.3652 ± 1.0086	5 (12.8)	0.2227 ± 0.7168	4 (10.3)
wt/Δex3d	wt/wt	88	0.7803 ± 1.3428	24 (27.3)	0.2396 ± 0.6944	10 (11.4)
wt/Δex3	wt/Δ27	14	1.1840 ± 1.4470	6 (42.9)	0.1152 ± 0.4310	1 (7.1)
Δex3/Δex3e	wt/wt	10	0.3226 ± 0.6800	2 (20.0)	0.2807 ± 0.8876	1 (10.0)
Δex3/Δex3	wt/Δ27	1	0	0 (0)	0	0 (0)
		P value	0.1880f	0.2352g	0.9894f	1.0000g

^aLog₁₀ parasites/μl blood.

^bHomozygous wild type individuals.

^cHeterozygotes for band 3 deletion.

^dHeterozygotes for glycophorin C exon 3 deletion.

^eHomozygotes for glycophorin C exon 3 deletion.

^fKruskal–Wallis test.

^gχ² analysis.

DISCUSSION

Epidemiologic evidence suggests that erythrocyte polymorphisms have arisen in malaria endemic regions to protect an individual from severe malaria [1,2,4]. We studied genetic polymorphisms of two integral red blood cell membrane proteins that have reached a high frequency in malaria endemic regions of Papua New Guinea. Our results show that the allele frequencies for both SLC4A1Δ27 and GYPCΔex3 polymorphisms differed in two ethnically and geographically distinct populations. Consistent with previous reports, SLC4A1Δ27 was rare in the Wosera and no SLC4A1Δ27 homozygous individuals were identified in either population [6,20]. SLC4A1Δ27 is considered to be a balanced polymorphism, where the selective advantage against severe malaria for heterozygotes outweighs the disadvantage of lethality in homozygous persons at the population level [12,13]. In contrast, while allele frequencies for GYPCΔex3 were higher in the Wosera than Liksul, the distribution of these alleles was in Hardy–Weinberg proportions in both populations and suggests that GYPCΔex3 alone does not confer a selective disadvantage.

While both SLC4A1 and GYPC play an important role in red blood cell morphology and function, the genes that encode these proteins are located on different chromosomes [7,19,20]. This would suggest that the SLC4A1Δ27 and GYPCΔex3 alleles should assort independently, if the apparent genetic load represented by SLC4A1Δ27 is not increased past a viable threshold by the addition of another significant erythrocyte membrane protein polymorphism.

In Liksul, the SLC4A1Δ27 and GYPCΔex3 allele frequencies were high, thus allowing us to examine the interactions of both band 3 and glycophorin C genotypes. The observed and expected frequencies of these genotypes were similar, indicating that SLC4A1Δ27 and GYPCΔex3 are independently distributed as would be expected under Mendel’s law of independent assortment. This finding suggests that, despite the fact homozygosity for SLC4A1Δ27 appears to represent a genetic load that is not viable, neither heterozygosity nor homozygosity for GYPCΔex3 leads to a readily apparent deleterious effect. It is interesting to note that similar findings with regard to SLC4A1Δ27 and α-globin deletions (located on chromosome 16 and associated with α-thalassemia) have been reported in earlier studies in Madang [24].

The relationship of glycophorin C or band 3 mutations to malaria susceptibility has been investigated [13,20,24]. Neither mutation alone has been shown to significantly influence asymptomatic asexual parasitemia by P. falciparum or P. vivax [13,20,24]. Our preliminary analysis of the relationship of both GYPCΔex3 and SLC4A1Δ27 to malaria susceptibility in Liksul revealed no association with the prevalence of P. falciparum or P. vivax prevalence or parasite density. These results parallel findings of other red blood cell polymorphisms where genotypic differences are observed at the level of severe malaria morbidity with no discernable effect on susceptibility to blood-stage infection [13].

In summary, we found that the frequency of GYPCΔex3 and SLC4A1Δ27 differs in the two study populations. These mutations are independently distributed and heterozygosity for SLC4A1Δ27 does not influence heterozygosity or homozygosity for GYPCΔex3. In addition, we did not observe an association between these erythrocyte membrane protein deletion mutations and alterations in the prevalence or density of asymptomatic blood-stage malaria infection. A prospective longitudinal study examining both parasitemia and malaria morbidity in children would better elucidate a potential protective role of GYPCΔex3 and SLC4A1Δ27 in PNG and other Melanesian populations.