Plasmodium falciparum: population genetic analysis by multilocus enzyme electrophoresis and other molecular markers.

we have studied 12 gene loci coding for enzymes. Seven rele-loci were analyzed in all populations, namely: glucose phosphate iso- merase (GPI, E.C. 5.3.1.9); glutamate dehydrogenase (GDH, E.C.1.4.1.2); hexokinase (HK, E.C.2.7.1.1); lactate dehydrogenase leucine amino peptidase pepti- dase 1, substrate L -leucyl-leucine-leucine loci, subtelomeric developed for testing deviations from random in of The

Abderrazak, S. B., Oury, B, Lal, A. A., Bosseno, M.-F., Force-Barge, least two local populations, the Congo population and one population from Burkina Faso, as well as in the cosmopolitan sample, and less P., Dujardin, J.-P., Fandeur, T., Molez, J.-F., Kjellberg, F., Ayala, F. J., and Tibayrenc, M. 1999. Plasmodium falciparum: Population genetic definitely in the other Burkina Faso population. However, no linkage disequilibrium is detected in the Congo population with the molecular analysis by multilocus enzyme electrophoresis and other molecular markers. Experimental Parasitology 92, 232-238. The population assays. We failed to detect any nonrandom association between the different kinds of genetic markers; that is, MLEE with RAPD or RFLP, structure of Plasmodium falciparum, the agent of malignant malaria, is uncertain. We have analyzed multilocus enzyme electrophoresis RAPD with RFLP, and so on. Although isoenzyme data show statistical departures from panmictic expectations, these results suggest that in (MLEE) polymorphisms at 7-12 gene loci in each of four populations (two populations in Burkina Faso, one in Sudan, one in Congo), plus the areas under survey, P. falciparum populations do not undergo predominant clonal evolution and show no clear-cut subdivisions, un-one "cosmopolitan" sample consisting of parasite cultures from 15 distant localities in four different continents. We have also performed like Trypanosoma cruzi, Leishmania sp., and other major parasitic species. We discuss the epidemiological and taxonomical significance random amplified polymorphic DNA analysis (RAPD) and restriction fragment length polymorphism (RFLP) and characterized gene varia-of these results. it is often assumed that it approximates a "potentially pan-MATERIALS AND METHODS mictic" model (Walliker 1985). We have earlier proposed that published data suggest nonrandom association between Parasite stocks. We have studied five samples of P. falciparum gene loci in certain natural populations of this parasite (Table 1). Four are local samples, Bobo 1 and Bobo 2 (Burkina Faso), (Tibayrenc, Kjellberg, and Ayala 1990; Sudan, and Congo. The fifth is a "cosmopolitan" sample consisting of 1991). This proposition has been controversial (Dye, Davis, 29 stocks from a total of 15 locations in different continents; namely and Line 1990; Walliker et al., and Luzzatte 1990;Tibayrenc, Africa (total: 22 stocks;Burkina Faso: 1 stock;Cameroon: 5;Gambia: Kjellberg, and Ayala 1991) and evidence has been presented 4; Ghana: 1, Guinea: 1; Kenya: 1; Mali: 1; Uganda: 1; Zambia: 1; undetermined: 5), Asia (total: 4 stocks; China: 1; Thailand: 3), Latin against the occurrence of linkage disequilibrium in P. falci-America (Brazil and Honduras: 1 stock each), and Papua New Guinea parum (Babiker et al. 1991;Conway and McBride 1991). 20 10-mer primers (kit A; obtained from Operon Technologies, Ala-The pC4.H32 insert contains a 0.5-kb imperfectly repeated sequence found in subtelomeric regions of multiple chromosomes. Restriction meda, CA, U.S.A.). The following six primers were used in this study: OPA 02 (5Ј-TGCCGAGCTG-3Ј), OPA 07 (5Ј-GAAACGGGTG-3Ј), site variations both within and outside of the 0.5-kb repeat contribute to the fingerprint polymorphisms. OPA 08 (5Ј-GTGACGTAGG-3Ј), OPA 09 (5Ј-GGGTAACGCC-3Ј), OPA 10 (5Ј-GTGATCGCAG-3Ј), and OPA 18 (5Ј-AGGTGACCGT-Data analysis. We have earlier developed methods suitable for testing deviations from random mating in populations of parasitic 3Ј). Each reaction was carried out in 15 l of reaction mixture containing 1ϫ reaction buffer II (100 mM Tris-HCl, pH 8.3; 500 mM protozoa (Tibayrenc, Kjellberg, and Ayala 1990;Tibayrenc et al. 1991). The segregation tests used in earlier studies are not applicable to P. KCl, Perkin-Elmer-Cetus, Norwalk, CT, U.S.A.), 1.0 mM MgCl 2 , 0.2 m primer, 0.2 mM each dNTP, 2.5 UI of Taq DNA polymerase (Gibco falciparum, because the parasite forms we have sampled (i.e., asexual forms) are haploid. However, the recombination tests previously pro-BRL, Grand Island, NY, U.S.A.). This reaction mixture was added to 10 l of template DNA. Two DNA concentrations were used for each posed, namely d1, d2, e, and f (Tibayrenc, Kjellberg, and Ayala 1990) are usable whatever the ploidy level of the organism under study. These sample: 12 and 4 ng per reaction. The RAPD PCR analysis for each primer was repeated at least once. Negative controls for each primer recombination tests are based on the null hypothesis of random genetic exchange and appraise different consequences of linkage disequilibrium contained all of the above components except for 10 l of distilled water in place of P. falciparum DNA. Thermocycling was performed between loci. Test d1 relies upon a combinatorial analysis and gives the probability with a GeneAmp PCR system 9600 thermocycler (Perkin-Elmer-Cetus), using the following thermal profile: 94ЊC for 5 min, then 45 of sampling the most frequent genotype as many as or more times than actually observed in a population. Test d2 measures the probability cycles of 94ЊC for 1 min, 36ЊC for 1 min, 72ЊC for 2 min, followed by a final 72ЊC extension for 7 min. Amplification products were of observing as many as or more individuals of any genotype than actually observed of the most common genotype. Test e gives the electrophoresed on 1.5% agarose gels and visualized by ethidium bromide staining.
probability of observing as few as or fewer different genotypes than found in the sample. Test f estimates the probability of observing a Polymerase chain reaction (PCR) amplification of antigen gene. We used published sequences of genes that have conserved 5Ј and 3Ј linkage disequilibrium as high or higher than actually found. The d2, e, and f tests are based on Monte Carlo simulations, with 10 3 iterations ends but contain a region with blocks of repeated sequences which vary in size and DNA sequence from strain to strain (Kemp et al. in the present study. 1987). We selected highly conserved regions and chose primers that spanned the repeats in four surface antigens: ring-infected erythrocyte surface antigen (RESA, Favalaro et al. 1986), the precursor of the major merozoite surface antigen-1 (MSA-1; Machay et al. 1985), the RESULTS AND DISCUSSION major surface antigen-2 (MSA-2; Fenton et al. 1991;Smythe et al. 1991), and circumsporozoite surface protein (CSP; Dame et al. 1984).
0.142 to 0.516 in a given population.
A total of 1 to 5 g of DNA was amplified in a final volume of  Note. N, number of genetic loci tested; ND, not done; NS, nonsignificant; MLEE, multilocus enzyme electrophoresis; RAPD, random amplified polymorphic DNA; RFLP, restriction fragment length polymorphism; Ag, antigene genes. particular parasite isolate exhibits two electromorphs at one for the 'g' test (Tibayrenc, Kjellberg, and Ayala 1990;. In the present case, lack of correlation or more loci. Since the parasite forms sampled are haploid, we interpret these two-allele instances as mixtures and have between different genetic markers is illustrated in Fig. 1, which shows for the Congo sample that the unweighted pair-removed them for calculating the statistics for nonrandom association, given that the multilocus associations are ambig-group method with arithmetic averages (UPGMA) dendrograms (Sneath and Sokal 1973) based on isoenzymes are uous whenever two alleles are observed at two or more loci. We have repeated the statistical tests for nonrandom not congruent with those based on RAPD. A totally different picture is obtained with Trypanosoma cruzi, the agent of associations using the allelic frequencies obtained when the mixed-culture data are included; the results are not materially Chagas' disease, in which dendrograms based on different sorts of markers are congruent with one another (Tibayrenc different from those shown. For the d1 test, the observed and expected (assuming et al. 1993). In summary, several cases of significant linkage disequi-random association between loci) frequencies of the most common genotype are shown in Table 1, in the two columns librium have been found in different populations of this sample for the isoenzyme data. But there is no evidence of preceding d1. The test is statistically significant for the Bobo 1, Bobo 2, and Congo populations and also for the cosmopol-linkage disequilibrium in the Congo population, on the basis of the other molecular data, whether the different sets of itan sample and the total data set. All other tests are statistically significant for the cosmopolitan sample and the com-markers are considered separately or in combination. It is therefore apparent that in the Congo population of bined total; the f test is significant also for the Bobo 1 and the Congo populations.
P. falciparum there is no evidence of predominant clonal evolution, nor of clear-cut subdivisions ('discrete typing In summary, when isoenzyme data are considered, there is evidence of significant linkage disequilibrium in two units' or DTUs; Tibayrenc 1998), such as are observed in other parasitic species, like Trypanosoma cruzi or Leish-(Bobo 1 and Congo) of the four local populations studied; and d1 indicates nonrandom multilocus associations also in mania (Tibayrenc, Kjellberg, and Ayala 1990;. Indeed, the evidence of linkage disequilibrium ob-Bobo 2. Other molecular assays were performed, but only in the tained in the present study of P. falciparum is far weaker than that for these other parasites. This result is not likely Congo population. All tests for linkage disequilibrium based on these assays are negative (see Table 1). No statistically to be due to lack of resolution of the tests, owing to low levels of genetic variability that would lead to statistical significant correlation is found, either between different pairs of markers, that is, between isoenzymes and RAPD, or be-type II errors . All the samples included in the present study show notable levels of genetic diversity tween RAPD and RFLP, or any other pair combination. Correlation between independent sets of genetic markers is (Table 1).
In the UPGMA dendrogram elaborated for the Congo strong evidence for linkage disequilibrium and is the basis Two UPGMA dendrograms (Sneath and Sokal 1973) showing the genetic relatedness revealed by MLEE (right) and RAPD (left) analyses for 31 stocks of Plasmodium falciparum from the Congo. Lack of agreement between the two dendrograms is taken as an indication that linkage disequilibrium is not strong within this sample. population with isoenzyme data (Fig. 1), two clusters are clustering pattern is not confirmed by the RAPD dendrogram (Fig. 1). apparent that are defined mainly by three loci: Gdh, Gsr, and Ldh. Three alleles occur at Ldh, and two at each of Gdh The fact remains that, when isoenzyme data are considered, there is highly significant linkage disequilibrium in and Gsr. The top cluster includes the 8 stocks with genotype 2/2/2 (allele 2 at each locus). The bottom cluster includes several populations. Several possible explanations that are not exclusive of one another can be explored. First, culture all 18 stocks with genotype 1/1/1, plus 5 stocks representing four additional genotypes (six other genotypes, with an ex-bias selection, which would have eliminated a large portion of the possible genotypes, could play a role. This explanation pected joint frequency of 0.264, are not included in the dendrogram). The two complementary genotypes 1/1/1 and is not acceptable in the Bobo 1 and Bobo 2 populations, which derive from placenta samples and were not cultured. 2/2/2 are both present in much greater frequencies than expected (Table 2) and account for most of the genetic Second, geographical separation could lead, through genetic drift, to different allelic frequencies among populations and disequilibrium in the Congo population. Nevertheless, this tently pooled (Wahlund effect). But this explanation will not work either, because the Bobo and Congo populations have been collected in relatively limited geographical areas (within less than a 20-km-diameter area, an area within REFERENCES which people readily move). Moreover, when a microorganism is found to be panmictic, such as Neisseria gonorrhoeae, Ayala, F. J. 1998 may very well be the case that some kind of uniparental Ben Abderrazak, S., Guerrini, F., Mathieu-Daude, F., Truc, P., propagation (Tibayrenc, Kjellberg, and Ayala 1990) obtains Neubauer, K., Lewicka, K., Barnabe, C., and Tibayrenc, M. 1993. in these populations of P. falciparum. This might occur as Isozyme electrophoresis for parasite characterization. In "Protocols a consequence of high rates of self-fertilization, a situation Congo population of P. falciparum, the only one we have surveyed with molecular markers other than isozymes. By