Emergence of FY*A^null in a Plasmodium vivax-endemic region of Papua New Guinea

Abstract

In Papua New Guinea (PNG), numerous blood group polymorphisms and hemoglobinopathies characterize the human population. Human genetic polymorphisms of this nature are common in malarious regions, and all four human malaria parasites are holoendemic below 1500 meters in PNG. At this elevation, a prominent condition characterizing Melanesians is α⁺-thalassemia. Interestingly, recent epidemiological surveys have demonstrated that α⁺-thalassemia is associated with increased susceptibility to uncomplicated malaria among young children. It is further proposed that α⁺-thalassemia may facilitate so-called “benign” Plasmodium vivax infection to act later in life as a “natural vaccine” against severe Plasmodium falciparum malaria. Here, in a P. vivax-endemic region of PNG where the resident Abelam-speaking population is characterized by a frequency of α⁺-thalassemia ≥0.98, we have discovered the mutation responsible for erythrocyte Duffy antigen-negativity (Fy[a−b−]) on the FY*A allele. In this study population there were 23 heterozygous and no homozygous individuals bearing this new allele (allele frequency, 23/1062 = 0.022). Flow cytometric analysis illustrated a 2-fold difference in erythroid-specific Fy-antigen expression between heterozygous (FY*A/FY*A^null) and homozygous (FY*A/FY*A) individuals, suggesting a gene-dosage effect. In further comparisons, we observed a higher prevalence of P. vivax infection in FY*A/FY*A (83/508 = 0.163) compared with FY*A/FY*A^null (2/23 = 0.087) individuals (odds ratio = 2.05, 95% confidence interval = 0.47–8.91). Emergence of FY*A^null in this population suggests that P. vivax is involved in selection of this erythroid polymorphism. This mutation would ultimately compromise α⁺-thalassemia/P. vivax-mediated protection against severe P. falciparum malaria.

Substantial overlap between malaria endemicity and erythroid polymorphisms suggests that these genetic conditions protect against malaria (1, 2). Whereas mechanisms governing “balanced polymorphism” presumably constrain the increase in frequency of mutations conferring homozygous lethal phenotypes [HbS, sickle cell anemia (3)], other mutations have increased to fixation, including erythroid Duffy blood group-negativity (Fy[a−b−]) in Africans (4) and α⁺-thalassemia in Melanesians (5).

Molecular genetic studies on the Duffy (FY) blood group antigen receptor for chemokines (alternative gene name DARC) have identified mutations underlying the major serologically defined antigens of the Fy blood group system, Fy^a, Fy^b (6), Fy^bweak (7), and the Fy-negative phenotype (Fy[a−b−], GATA-1 promoter mutation; ref. 8). Individuals homozygous for this GATA-1 mutation fail to express Fy specifically on erythroid cells, providing the molecular basis for Fy(a−b−). In Africans the GATA-1 mutation is haplotypically associated with 625A, encoding Fy^b (FY*B^null). Because Fy(a−b−) prevents Plasmodium vivax from invading host erythrocytes (9–11) and completing its complex life cycle, it is not surprising that vast regions of Africa inhabited by Fy(a−b−) human populations are devoid of this malaria parasite (12). Although the specificity of the parasite–receptor interaction suggests that selective pressure by the pathogen drove the allele to fixation, the reported lack of significant mortality from P. vivax infection counters this hypothesis. Alternatively, debate suggests that preexisting high frequency Fy(a−b−) prevented P. vivax from becoming established in Africa (13).

The primary mutations responsible for α⁺-thalassemia in Melanesians are 3.7- and 4.2-kilobase deletions that eliminate one of the two tandemly repeated α-globin genes on human chromosome 16 (5). In contrast to Fy(a−b−), the mechanism by which α⁺-thalassemia protects against malaria is unknown (14). Recent studies in Melanesian populations have demonstrated an increased susceptibility to uncomplicated malaria among young α⁺-thalassemic children. It is further proposed that this genetic condition may facilitate P. vivax infection to act later in life as a natural vaccine against severe P. falciparum malaria (14, 15).

As in other areas in the South Pacific, P. vivax causes significant morbidity and coexists with the other three human malaria parasites in Papua New Guinea (PNG). P. vivax is responsible for 25% of malaria cases in the East Sepik Province in PNG; P. falciparum (55%), P. malariae (20%), and P. ovale (0.04%) are also holoendemic in this area (16). Resident human populations have been phenotypically characterized to be 100% Fy antigen-positive (more than 85% are Fy[a+b−]) (4, 17), and as high as 90% α⁺-thalassemic (18). If, as previously suggested (14, 15), α⁺-thalassemia predisposes children to P. vivax infection, examination of similar host populations would present a unique opportunity to observe the influence of this malaria species on FY sequence polymorphism. To this end, we have studied individuals from two different bioclimatic regions of P. vivax-endemic East Sepik Province, PNG. The inclusion of these two populations enabled the study of communities where environmental factors are likely to vary the exposure to malaria-infected mosquitoes, and was predicted to increase the genetic diversity of the study because of the relative isolation of the resident Abelam and Urat linguistic groups (16, 19).

Methods

Study Area and Population.

The study was conducted in villages in East Sepik, PNG. Details describing South Wosera (n ≈ 6000) and villages near the government administrative unit in Dreikikir (n ≈ 4000–5000) have been provided earlier (16, 19). Study subjects from these areas are members of the Abelam (n = 531) and Urat (n = 381) linguistic groups, respectively. Sample collection was organized through the Papua New Guinea Institute of Medical Research, Goroka. Blood samples from West African individuals (n = 200) from 17 villages distributed between Senegal and Ghana were collected as part of the studies conducted in conjunction with the Onchocerciasis Control Programme, Ouagadougou, Burkina Faso (20). North American random blood donors, whose race or ethnicity was self-identified at the time of sample collection (n = 400), were obtained from the National Histocompatibility Laboratory, American Red Cross/University of Maryland Medical System, Baltimore, MD. Blood samples from all individuals participating in this study were collected under clinical protocols approved by the corresponding institutional review boards.

Malaria Diagnosis.

Blood films were prepared and examined as described previously (16). Briefly, thick and thin films were stained with 4% Giemsa and examined under oil immersion objective (100×) for 100 fields. Parasite species were identified by using both thick and thin film preparations. Parasite densities were recorded as the number of parasites per 200 white blood cells (WBC). Previous studies have determined the average WBC count for individuals living in this area to be approximately 8000 WBC/μl.

DNA Extraction.

Whole blood was collected in K+EDTA vacutainers and stored at −20°C until DNA extraction could be performed. DNA extraction was performed by using the QIAamp blood extraction protocol (Qiagen, Valencia, CA) or by protocols described previously (21).

PCR-Restriction Fragment Length Polymorphism (RFLP) Genotyping.

PCR amplification was performed in a solution (25 μl) containing 2.5 pmol of the appropriate positive-strand and negative-strand primer; 67 mM Tris⋅HCl (pH 8.8); 6.7 mM MgSO₄; 16.6 mM (NH₄)₂SO₄; 10 mM 2-mercaptoethanol; 100 μM dATP, dGTP, dCTP, and dTTP; 2.5 units of thermostable DNA polymerase; and 10–50 ng of purified human genomic DNA. Strategies used to characterize FY DNA sequence polymorphisms (Fig 1A; refs. 6–8) include fragment A (primers −FYPup 5′-GTAAAATCTCTACTTGCTGGAAG-3′ and FYPdn 5′-CCATGGCACCGTTTGGTTCAGG-3′), fragment B (primers FYup 5′-GACTCTTCCGGTGTAACTCTGATG-3′ and FY851[−] 5′GGCCAAGACGGGCACCACAATG-3′), fragment C (primers FY2014[+] 5′-ACTCAGTTCAGGGAACATATC-3′ and FY2224[−] 5′-TGCTGACAGAGATGCAAGCT-3′), and fragment D (primers FYPup and FY851[−]). The thermocycling program used for each of the above amplifications was 30 seconds at 94°C, 30 seconds at 65°C, and 30–60 seconds at 72°C for 40 cycles. RFLP genotyping was performed following supplier-recommended protocols (New England Biolabs) as follows, fragment B/StyI, fragment C separate assays by using BanI, AciI, and MwoI. Restriction fragments were visualized after electrophoresis on either 4% (5:1, GTG:NuSieve:LE; fragment A/StyI fragments) or 2% (fragment B/BanI, AciI and MwoI fragments) agarose gels (FMC) stained with a 1:10,000 dilution of SYBR Gold (Molecular Probes) and detected with a Storm 860 (Molecular Dynamics). Genotyping of α-globin was directed by a combination of primers described previously (22).

Figure 1.

(A) FY gene locus (human chromosome 1q22-23) and PCR-RFLP genotyping strategies for FY promoter and ORF (6–8). The scale at the top of the figure provides orientation to the FY gene sequence (−/+ coordinates relative to the transcription start site). Positions of previously characterized single nucleotide polymorphisms identified along the linear representation of the FY gene [by using nucleotide (amino acid) nomenclature consistent with erythroid gene product (8)] include T-46C, G625A (G44D), C765T (R91C), G798A (A102T). Promoter-specific (−304 to +25, 328 bp; fragment A) and ORF-specific (+474 to +851, 387 bp; fragment B) amplicons were subject to StyI and BanI restriction endonuclease digestion, respectively. StyI RFLP analysis reveals polymorphism where the 77-bp fragment corresponds with −46T (active GATA-1 site) and the 65-bp fragment corresponds with −46C (inactive GATA-1 site). BanI RFLP analysis reveals polymorphism where the 151- and 212-bp fragments correspond to the Fy^a-specific sequence 625G (encoding G) and the 363-bp fragment corresponds with the Fy^b-specific sequence 625A (encoding D). Additional polymorphisms include C765T (R91C), eliminating an AciI restriction endonuclease cleavage specific for the FY*B^weak allele (14), and G798A (A102T) eliminating an MwoI restriction endonuclease cleavage site in the wild-type FY allele. PCR-based cloning and DNA sequence analysis were applied to the 3′ untranslated region microsatellite (fragment C, arrow) and contiguous promoter-ORF (fragment D) sequences. Allele-specific analysis of fragments C and D enabled precise determination of the linear order of DNA sequence polymorphism potentially obscured by heterozygosity in direct DNA sequencing approaches. (B and C) PCR-RFLP analysis of individual study subjects, lane 1; West African, lane 2; African-American, lanes 3–5; Papua New Guineans, lanes 6 and 7; cloned amplicons (fragment D) derived from the individual in lane 5. (B) Results from StyI digestion of the promoter-specific amplicon-producing restriction fragments of 138, 108, 77, and/or 65 bp. The 12-bp product was not visualized after staining of agarose gels. (C) Results from BanI digestion of the ORF-specific amplicon-producing restriction fragments of 363 and/or 212 and 151 bp. The figure shows the reverse SYBR Gold-stained image scanned with the Storm 860.

Cloning and DNA Sequence Analysis.

After amplification of fragments C and D (Fig. 1A) from human genomic DNA templates, PCR products were cloned into pCR2.1TOPO (Invitrogen). Multiple clones from subjects of interest were sequenced by using fluorescence-based sequencing protocols on an ABI377 automated sequencer (Perkin–Elmer).

Microsatellite Genotyping.

Microsatellite primer pairs identified in Fig. 3 were selected from the Version 9 Marker Screening Set listed by the Marshfield Center for Medical Genetics (http://www.marshmed.org/genetics) and GDB6.3 (http://www.gdb.org/gdb/gdbtop.html). One primer of each microsatellite primer pair was 5′ end-labeled with Cy5 at the time of synthesis (Research Genetics, Huntsville, AL). Microsatellite amplification was performed by using the same buffer conditions as described above. The thermocycling program used for amplification of all microsatellites was 2 minutes at 94°C; 5 cycles of 30 seconds at 94°C, 30 seconds at 65°C, and 60 seconds at 72°C; 25 cycles of 30 seconds at 94°C, 30 seconds at 60°C, and 60 seconds at 72°C; 5 cycles of 30 seconds at 94°C, 60 seconds at 55°C, and 2 minutes at 72°C; and a final extension of 7 minutes at 72°C. After PCR amplification of human genomic DNA samples, Cy 5-labeled microsatellite amplicons were visualized after electrophoresis on 6% denaturing polyacrylamide gels (7.7 M urea/32% formamide). For each microsatellite, allelic determination was normalized by inclusion of reference samples on each gel.

Figure 3.

Test for association between microsatellite alleles and FY*A^null. Association between microsatellite and FY*A and FY*A^null alleles was performed on 47 FY*A/FY*A and 15 FY*A/FY*A^null unrelated individuals from South Wosera following the approach described by Weir (24). P values were adjusted (Bonferroni correction) for the number of alleles observed at each individual locus. Linear organization of microsatellites in relation to the FY locus was performed by using the “Build your own map” function, supplied by the Marshfield Center for Medical Genetics (http://www.marshmed.org/genetics). Genetic distances between FY-specific amplicons and microsatellite (MS) markers were determined following radiation hybrid mapping by using the G3 RH panel (Research Genetics) and the Stanford Human Genome Center Radiation Hybrid Mapping E-mail Server (http://www-shgc.stanford.edu/RH/rhserverformnew.html). D1S2635 maps to within 0.3–0.6 cM of FY. ns, not significant.

Flow Cytometry.

Five-microliter blood samples were suspended in PBS (50 μl) and then pelleted by microcentrifugation (4.5–6 × 10³ × g). This washing procedure was repeated twice. After the second washing, the red blood cell pellet was resuspended in 50 μl of PBS/Fy6 mouse monoclonal antibody (23) (Fy6 was kindly provided by J. Barnwell, Centers for Disease Control and Prevention, Atlanta, GA) at a 1:50 ratio, then incubated for 15 minutes at 37°C. The samples were then washed twice in PBS, resuspended in 15 μl of PBS/goat anti-mouse Ig labeled with phycoerythrin at a 1:3 ratio, and incubated for 15 minutes at 37°C. Red blood cells were then washed twice in PBS and resuspended in 1 ml of PBS. Flow cytometry was performed with a Coulter Elite ESP flow cytometer; forward light scatter, side light scatter, and fluorescence data were analyzed.

Statistical Analysis.

Odds ratios and confidence intervals were calculated by using standard methodologies and computed with the programs available through Simple Interactive Statistical Analysis (SISA, http://home.clara.net/sisa/two2hlp.htm). The association between microsatellites and Fy locus-specific markers was analyzed by using procedures described by Weir (24).

Results and Discussion

Genotyping Studies.

PCR-based genotyping of α-globin polymorphism used strategies developed by Oron-Karni et al. (22) to amplify a 298-bp fragment downstream from the α2 gene and a 446-bp fragment downstream from the α1 gene. Consistent with other studies, the frequency of α⁺-thalassemia (−α/−α or −α/αα) was ≥98% in both linguistic groups.

PCR-RFLP genotyping (Fig. 1A) of FY promoter (Fig. 1B) and open reading frame (ORF) (Fig. 1C) polymorphisms was performed for a West African (lane 1), an African-American (lane 2), and three Abelam speakers (lanes 3–5). The West African was homozygous for the inactive-GATA-1 site (65-bp fragment) and FY*B (363-bp fragment) RFLPs (FY*B^null/FY*B^null); the African-American was heterozygous for the GATA-1 site (65- and 77-bp fragments) and ORF (151-, 212-, and 363-bp fragments) RFLPs (FY*A/FY*B^null); the three Abelam speakers were heterozygous for the GATA-1 (65- and 77-bp fragments) and homozygous for FY*A (151- and 212-bp fragments) RFLPs (FY*A/FY*A^null). PCR-RFLP and DNA sequence analyses of cloned allelic amplicons confirmed that the 65-bp restriction fragment (lane 7) was due to the same T-46C transition observed in Fy(a−b−) Africans, and that this mutation was on the ORF allele encoding the Fy^a antigen (625 G). Within the PNG study population, 23 Abelam speakers were heterozygous, FY*A/FY*A^null, for this new allele (Table 1). This report of the FY GATA-1 mutation in PNG and of the association between the inactive GATA-1 promoter site and FY*A suggests the identification of an FY*A^null allele. The latter contrasts with the association between the GATA-1 mutation with FY*B observed in Africans (8) and African-Americans (FY*B^null; Tables 1 and 2), suggesting that this new allele is independent of African origin. Results on 400 North Americans (Table 2) illustrate that FY*A^null is not observed in a more ethnically diverse study population where recombination between FY*A and FY*B^null would be possible. Identification of FY*B^null alleles in non-African populations (Table 2) is likely to be the result of admixture with the African-American genetic background. Earlier studies have observed African-based genetic polymorphisms in the coding region of ICAM-1 (25) and the TNF-α promoter (TNFAp-4; ref. 26) in Caucasian-, Hispanic-, and Asian-American populations studied here.

Table 1.

Summary of FY alleles in PNG and African study groups

	South Wosera†	Dreikikir‡	West Africa
FY*A	1039	762	0
FY*Anull	23	0	0
FY*B	0	0	0
FY*Bnull	0	0	400

^†Abelam speakers. 

^‡Urat speakers. 

Table 2.

Summary of FY alleles in North American study groups^†

	Caucasian-American	African-American	Asian-American	Hispanic-American
FY*A	91	24	159	103
FY*Anull	0	0	0	0
FY*B	109	31	33	90
FY*Bnull	0	145	8	7

^†Self-identified ethnicity. 

Flow Cytometric Analysis of Fy Expression.

Evidence correlating FY*A^null with the Fy antigen-negative phenotype was developed by flow cytometry, comparing Fy6-specific (23) phycoerythrin labeling of erythrocytes from six PNG FY*A/FY*A and six PNG FY*A/FY*A^null individuals (Fig. 2). Results show a significant difference in Fy6-specific mean fluorescence intensity for the homozygous wild-type [187.36 (±33.29)] vs. the heterozygous [93.33 (±12.81)] genotypes (Mann–Whitney U, P = 0.004). These findings indicate a gene-dosage effect, where individuals with two erythroid-functional alleles express approximately twice the amount of Fy on erythrocytes compared with individuals with one erythroid-functional allele. Furthermore, these results illustrate that T-46C affects Fy expression on the FY*A and FY*B genetic backgrounds in a similar manner (Fig. 2 legend).

Figure 2.

Flow cytometric analysis of phycoerythrin-labeled Fy6-specific fluorescence intensity comparing erythrocytes from 12 PNG study subjects. Mean fluorescence intensity (MFI) is indicated on the y axis, and FY genotype is indicated along the x axis. MFI readings were compared for six PNG FY*A/FY*A^null individuals (shaded circles), 93.33 (SEM = 12.81), and six PNG FY*A/FY*A individuals (solid circles), 187.36 (SEM = 33.29), (Mann–Whitney U = 36, P = 0.004). The open circle represents the average MFI for experiments omitting primary (Fy6) and secondary phycoerythrin-labeled goat anti-mouse IgG incubation with erythrocytes from an individual with an FY*A/FY*A genotype (MFI = 5.98). In a separate experiment, the average MFI for individuals with the FY*B/FY*B genotype (n = 2) was 2.72 times greater than that for individuals with the FY*B/FY*B^null genotype (n = 2).

Association Between FY Genotype and P. vivax Infection.

Given this reduction in Fy expression on erythrocytes of heterozygous individuals, it was next determined whether there was an association between FY*A/FY*A^null and reduced prevalence of P. vivax blood-stage infection. Although the prevalence of P. vivax infection was approximately 2-fold greater in FY*A/FY*A (83/508 = 0.163) compared with FY*A/FY*A^null (2/23 = 0.087) individuals, this difference was not statistically significant (odds ratio = 2.05, 95% confidence interval = 0.47–8.91). Moreover, consistent with in vitro parasite invasion studies (10), these results show that heterozygous individuals are susceptible to infection by P. vivax. For other malaria species, prevalence of P. falciparum (FY*A/FY*A, 158/508 = 0.311; FY*A/FY*A^null, 5/23 = 0.217; odds ratio = 1.63, 95% confidence interval = 0.59–4.45) and P. malariae (FY*A/FY*A, 65/508 = 0.128; FY*A/FY*A^null, 3/23 = 0.130; odds ratio = 0.98, 95% confidence interval = 0.28–3.38) infection was similar based on FY promoter genotype. It is important to acknowledge that these findings provide only a preliminary assessment of differential susceptibility to P. vivax infection based on this new FY promoter genotype. More definitive examination of this issue requires a longitudinal, age-matched, case-control study in which individuals are treated to clear existing Plasmodium infection and then followed prospectively to compare not only incidence, but also time to infection in FY*A/FY*A and FY*A/FY*A^null individuals.

In contrast to FY*B^null fixation in the absence of P. vivax in Africa, which precludes further study of how this parasite acts as a selective agent for the T-46C promoter mutation, the apparent recent emergence of FY*A^null in PNG provides the opportunity to study the significance of this mutation for susceptibility to malaria infection and clinical morbidity. Support for a recent emergence of FY*A^null, relative to FY*A, is presented in Fig. 3. These findings identify specific microsatellite alleles associated with FY*A^null at D1S2635 (adjusted P = 0.0012) and ATA73A08 (adjusted P = 0.025) that map to within 3 cM of the FY gene locus on human chromosome 1. No significant associations were observed between microsatellite alleles and FY*A, indicating that longer time, enabling more genetic recombination, has obscured haplotype associations with the wild-type allele. Beyond this comparison of haplotype associations with the FY*A and FY*A^null alleles in the Abelam-speaking population, it would be premature to estimate when FY*A^null arose in this setting. To assess properly the origin of this new allele would require broader sampling and assessment of population structure within Abelam-speakers (estimated population size, 40,000; ref. 27) and among neighboring communities.

Conclusion

The emergence of FY*A^null raises important questions regarding the complex relationships between malaria parasites and the human host. Whereas Duffy-negativity clearly explains resistance to P. vivax merozoites in individuals and host populations, little is known regarding the factors that contributed to fixation of the FY*B^null allele throughout ethnically diverse African populations. To complicate this discussion, although early studies have reported that P. vivax was responsible for significant disease and death in Britain (28), it has become consistently regarded as a “benign” or “non-lethal” malaria parasite. Therefore, the mechanisms through which it might act as a selective agent are not clear. The evolution of the FY*A^null allele and its emergence in a P. vivax-endemic region associate this very specific mutation with vivax malaria for a second time. This association may suggest that infection by this parasite cannot be benign and that it is influencing polymorphism in the human genome through mechanisms not previously considered.

Additionally, identification of FY*A^null in PNG must also be considered as an important variable with the potential to influence severe pathogenesis resulting from P. falciparum infection. Recent studies have proposed that α⁺-thalassemia increases susceptibility to P. vivax infection to serve later as a natural vaccine against severe falciparum malaria. As the frequency of α⁺-thalassemia was found to be ≥98% in the East Sepik Province groups studied here, this hypothesized heterologous immune phenomenon should be at work within these host populations. If this complex hypothesis is correct, an increase in the frequency of FY*A^null leading to Duffy-negativity and resistance to P. vivax blood-stage infection may be suppressed to maintain the proposed protective effect of P. vivax infection, because severe falciparum malaria kills millions of children each year. Alternatively, it may be necessary to reevaluate the proposed heterologous immunological relationship if the frequency of FY*A^null rises in this P. vivax-endemic region of PNG.

Abbreviation

PNG

Papua New Guinea

RFLP

restriction fragment length polymorphism

MFI

median frequency intensity