Genetic diversity of Plasmodium falciparum and its relationship to parasite density in an area with different malaria endemicities in West Uganda


Gabriele Peyerl-Hoffmann Department of Infectious Disease and Tropical Medicine, Leopoldstrasse 5, 80802 München, Germany. Fax 00-49-89-336112; E-mail:


Field populations of Plasmodium falciparum can be effectively genotyped by PCR-amplification of selected fragments of the Merozoite Surface Proteins 1 and 2 (MSP1 and MSP2). Genetic diversity of P. falciparum populations in areas with different transmission levels (holo- vs. mesoendemic) was investigated in Kabarole District, West Uganda. 225 samples positive for P. falciparum were analysed by amplification of polymorphic regions and classified according to prevalence of allelic families. A large number of alleles was detected for each locus: 22 for MSP1 block 2 and 24 for MSP2 and, 175 (78%) of MSP1 alleles and 143 (64%) of MSP2 showed multiple infections within a range of 2–8 clones. Significant differences between holoendemic and mesoendemic areas in regards of population structure and number of multiclonal infections of P. falciparum were not apparent. However, a significant correlation between parasite density, selected MSP2 loci and differences between parasite density in monoclonal vs. multiclonal infections occurred. Multiplicity of infection was age-dependent.


Transmission intensity of malaria depends on the density and infectivity of the anopheline vector and on variation in parasite rate in the human host (Molineaux & Gramiccia 1980; Smith et al. 1993). Humans living in malaria-endemic regions acquire semi-immunity to Plasmodium falciparum as a result of natural exposure to multiple infections over many years. Because of differing transmission intensities, semi-immunity at a younger age develops faster in a holo- or hyperendemic area than in an area where transmission is less intense (McGregor 1986; Baird 1995). It is usually assumed that immunity to P. falciparum has two components: an antidisease immunity, which is believed to develop rapidly, and an antiparasite immunity, which is acquired slowly and leads to a marked decrease in parasite densities (Trape et al. 1994; Ntoumi et al. 1995). The risk of an individual falling clinically ill with malaria is determined by many factors (Greenwood et al. 1991). Molecular techniques now offer new possibilities to get a deeper understanding of host–parasite interactions and the biology of the parasite population. The usefulness of the PCR technique together with the minimal requirement of infected samples have led to a multitude of typing studies combining various P. falciparum polymorphic markers such as Merozoite Surface Proteins 1 and 2 (MSP1, MSP2) (Walliker 1994). Typing of P. falciparum in human hosts has been used in work on the diversity of parasite populations (Babiker et al. 1995), the search for markers of parasite virulence (Engelbrecht et al. 1995), in researching the importance of multiplicity of infection (Contamin et al. 1995; Beck et al. 1997) and the geographical distribution of the various alleles of these polymorphic genes of the parasite (Babiker et al. 1997; Zwetyenga et al. 1998; Felger et al. 1999a; Jelinek et al. 1999).

In order to examine genetic diversity and complexity of parasite populations in endemic regions, field isolates were collected in two regions with different endemicity (holo- vs. mesoendemic) in West Uganda. PCR-amplification of polymorphic regions of the two marker genes MSP1 and MSP2 was performed and results of these data were used to describe how genetic diversity and multiplicity of infection vary in accordance to age and parasite density.

Materials and methods

Study area

Three villages in two areas of different transmission (holo- vs. mesoendemic) of Kabarole District in western Uganda were chosen for the survey (Figure 1). The southern, holoendemic part of the region is at an altitude of approximately 900–1300 m. Malaria transmission is perennial with two peaks after the rainy seasons, the first in April and May and the second in October and November. The eastern, mesoendemic part of the region, enclosing the district capital Fort Portal, lies at an altitude between 1500 and 1650 m. Here, malaria transmission is seasonal with a dry period in January/February and July. Kabarole district has approximately 863 000 inhabitants. The predominant malaria parasite is P. falciparum, which accounts for more than 98% of all malaria infections.

Figure 1.

 Study area with holo- and mesoendemic sites and study villages. 1=Kasunga, 2=Burraro, 3=Rwentuha.

Sample collection

At the beginning of a 1-year prospective study on the effectiveness of insecticide-treated curtains in co-operation with the Gesellschaft für Technische Zusammenarbeit (GTZ) and the Ministry of Health of Uganda, inhabitants of the villages were informed about this investigation. Informed consent was obtained from adults or parents of children before they were enrolled in the study. We collected 406 blood samples (holoendemic, 136, mesoendemic, 270) from residents during the transmission season from July 1997 to March 1998. Rwentuha village yielded 136 samples, Kasunga village 206, and Burraro village 64. Ten microlitres of periphal blood were collected by lancet prick from a fingertip for direct microscopic diagnosis, and the same amount of blood obtained by lancet prick was spotted directly onto filter paper (Whatman 3 MM chromatography paper). Blood spots on filter paper were allowed to air-dry, then placed individually in plastic bags and stored at room temperature. Filter papers were transported to the laboratories of the Department of Infectious Diseases and Tropical Medicine of the University of Munich for further analysis.

Microscopic examination

Thin and thick blood smears were air-dried in the field, transported to the central laboratory at Buhinga Hospital, Fort Portal and stained for 45 min with 3% Giemsa stain using standard procedures (Gilles & Warrel 1993). The slides were read by experienced technicians at the health centre examining at least 100 oil-immersion fields before a slide was considered negative. For those who were found to be infected with malaria parasites, the approximate number of parasites per μl of blood was determined (Dowling & Shute 1966). For quality control, 10% of the slides were retested.

Preparation of the DNA template

The DNA template for the nested PCR assay was prepared from the whole blood spot using the Chelex (Bio-Rad Laboratories, Hercules, CA, USA) boiling method (Kain & Lanar 1991; Jelinek et al. 1996). Each spot was boiled in 180 μl of ChelexTM solution and approximately 100 μl of DNA template was obtained and, 1–2 μl of this was used in the nested PCR assay.

Primers and PCR amplification

All 406 blood samples collected on filter paper were initially analysed by using P. falciparum genus and species-specific primers in a nested PCR approach (rPLUS1 and rPLUS5; rFAL1 and rFAL2). Details of this method have been described elsewhere (Singh et al. 1999). PCR genotyping was performed as described by Snounou et al. (2000), using repetitive regions found in two polymorphic genetic markers, namely block 2 of MSP1 and block 3 of MSP2. Allelic variants of MSP1 (MAD20, K1 and RO33) and MSP2 (IC and FC27) were detected by allelic family specific nested PCR. PCR conditions and the primer sequences have been described in detail by Snounou et al. (2000). All PCR amplifications (UNO-Thermoblock, BioMetra, Göttingen, Germany) contained a positive control (standardized laboratory strains e.p. K1, 3D7) and a negative control (containing no target DNA).

Product analysis

After the addition of 5 μl loading puffer to the amplified product, 10 μl were analysed by agarose gel electrophoresis. The concentration of agarose gel was 2%, a 3 : 1 mixture of NuSieve® agarose and SeaKem® agarose (BMA Products, Rochland, ME, USA). DNA was visualized under ultraviolet light after being stained with ethidium bromide and results were photographed. Band sizes for successful amplification of the DNA were 205 bp for rFAL1 and rFAL2, 150–300 bp for MAD20, 150–300 bp for K1, 160 and 240 bp for RO33, 400–750 bp for IC, and 250–500 bp for FC27.

Data processing and analysis

Individual data points and laboratory data were independently entered into a database. The technician who performed the amplification of DNA by PCR was blinded to the results of microscopy and vice versa. Before analysing the data they were cross-checked and then transformed into SPSS, Version 9. With this software all necessary descriptive statistics as well as parametric (t-test, one-way analysis of variance with post hoc multiple comparisons of mean according to Scheffè and Bonferroni) and non-parametric methods (Spearman’s correlation coefficient, χ2) were calculated. Two population measures of parasitaemia were used: the parasite rate and the geometric mean of parasite density. The prevalence of each allelic family was estimated by calculating the percentage of fragments assigned to one family by PCR with family specific primers among the overall number of fragments detected for that locus in the group considered. The total number of multiple infections and the number of infections belonging to the allelic families (MAD20, K1, RO33; IC and FC27) were analysed with respect to age, endemic area and parasite density. The complexity of infection was calculated for each typing reaction (MSP1, MSP2) independently as the average number of distinct fragments per PCR-positive sample. Spearman’s correlation coefficients were calculated to assess associations between multiplicity of infection and parasite densities.


Of 406 blood samples collected on filter paper in the villages, 225 samples were confirmed as P. falciparum positive by nested PCR. In contrast to the number of 140 microscopically positive samples, 85 (21%) additional P. falciparum-positive cases were detected by PCR. Fifty-six of those samples came from the holoendemic and 29 from the mesoendemic area. All 225 positive samples (holoendemic, 56; mesoendemic, 169) were further genotyped to define and differentiate parasite populations. The geometric mean of parasite density was 63.8 parasites per μl of blood (range 4–31 960) with a confidence interval of 43.6–93.4. Samples that were positive by PCR but negative by microscopy (n=85) were assigned a density of 4 parasites/μl of blood. The geometric mean of the parasite density decreased with age; children up to 2 years had the highest rate with 375 parasites/μl; the geometric mean parasite density of adults older than 25 years was 25/μl of blood.

Polymorphism of MSP1 and MSP2

Alleles of MSP1 and MSP2 were classified according to the size of their PCR-amplified fragments. Both genetic markers and the corresponding allelic families were very diverse. Amplification of the allelic family MAD20 was positive in 93 samples (41.3%) and yielded eight different fragments (160–300 bp). Allelic family K1 was positive in 182 samples (81.1%) and produced up to 10 different bands (160–380 bp). Allelic family RO33, positive in 80 samples (35.5%), led to amplification of four different fragments with a predominance of 160 bp length (91.3%). The additional bands of RO33 probably are because of unspecific bands or to new RO33 variants that contain deletion and were not further tested in this study. 150 samples (66.7%) of the allelic family IC and 125 samples (55.6%) of FC27 were positive and produced up to 12 fragment sizes (IC: 400–759 bp, FC27: 200–550 bp). We saw 1080 fragments in 46 band lengths. Comparing prevalence and polymorphisms between the allelic families of MSP1 and MSP2 between holoendemic and mesoendemic areas, we found no significant difference, indicating that very similar parasite populations were circulating then.

Complexity of the infection

175 (78%) samples of MSP1 and 143 (64%) of MSP2 contained multiclonal infections with at least two clones. We found up to eight different bands per allelic family. The estimated mean number of clones was 2.6 with MSP1 and 2.2 with MSP2. The influence of age on the complexity of P. falciparum infection showed that children in the age group of 3–5 years had the highest average number of different clones in their samples: 3.2 with MSP1 and 2.3 with MSP2. The average number of MSP1 and MSP2 bands per isolate was slightly lower in the older age groups (Table 1), but this difference was not significant. Further stratification into the allelic families showed no significant influence of age on multiplicity of infection. There was little variation concerning multiplicity of infection within the two different endemic areas. There were no significant differences in mean multiplicity of infection between the holoendemic (MSP1 2.9, MSP2 2.0) and the mesoendemic area (MSP1 2.5, MSP2 2.3).

Table 1.   Geometric mean of density (parasite/μl) and multiplicity of infection in MSP1 and MSP2 stratified by age group (n = 224) Thumbnail image of

Parasite density and multiplicity of infection

The relationship between geometric mean parasite density and multiplicity of infection was analysed particularly for the mesoendemic area (n=169, Table 2) because of a lack of PCR results from microscopic positive samples in the holoendemic area. A clear trend of increasing parasite density with increased multiplicity was visible here for genetic markers on both genes, MSP1 and MSP2. In MSP1, 137 samples showed multiclonal infection (two or more bands) with a geometric mean parasite density of 182 parasites/μl compared with 32 samples with monoclonal infection and a geometric mean density of 90 parasites/μl (P > 0.005). In MSP2, the geometric mean of parasite density of 111 samples with multiclonal infections was 310 parasites/μl and differed, but not significantly, from the geometric mean among the 58 monoclonal infections (45 parasites/μl). For allelic families, significant differences were detected for IC only (P < 0.004). The association between multiplicity of the target genes and parasite density was calculated by Spearman correlation. Here, MSP1 (Spearman’s r=0.208; P < 0.001) and MSP2 (Spearman’s r=0.367, P < 0.001) showed a significant positive correlation. Figure 2 shows this correlation referring to the different age groups.

Table 2.   Geometric mean parasite density of P. falciparum according to multiplicity of infection in the mesoendemic area (n = 169, number of parasite clones per patient) Thumbnail image of
Figure 2.

 Correlation between multiplicity of infection and parasite density in the mesoendemic area (n=169).


Genetic diversity appears to play a major role in the natural acquisition of immunity to malaria infections. For hyper-and holoendemic areas the number of malaria infections and parasite densities of P. falciparum in adults is lower than in children, which can be characterized as a function of antiparasite immunity (Day & Marsh 1991; Baird 1995). The geometric mean of parasite densities of 225 PCR positive samples from Kabarole District of West Uganda was highest in the age group of children up to 2 years (375/μl) and decreased continuously with age (> 25 years: 25/μl). Similar age trends have been observed in Dielmo, a highly endemic area in Senegal (Trape et al. 1994). As the average parasite densities decreases with age, there are many samples with low densities close to the detection limit of micrsocopy among older individuals. Direct comparison of age-stratified parasite densities in the holo- and mesoendemic areas was not possible because all PCR-positive samples from the holoendemic area (n=56) were microscopically negative. Yet, all 56 (41%) samples (n=136, holoendemic) were positive by species-specific nested PCR. Their parasite density as well as that of the other 29 samples from the mesoendemic area was therefore assumed at four parasites per μl blood. While microscopic examination may fail to detect low-grade parasitaemia, nested PCR can be a useful tool for the detection of asymptomatic carriers in endemic areas.

Although the allelic families of the investigated isolates were highly diverse, a similar pattern of distribution of MSP1 and MSP2 allelic families was detected in both endemic areas of Kabarole District. One reason for this lack of significant differences in strain distribution between the holo- and mesoendemic areas could be the relatively close proximity (100 km) and frequent travel between study sites that may allow for exchange of parasite populations. The number of different MSP1 and MSP2 alleles observed among the 225 isolates was large: 22 variations of allelic families were detected in MSP1, and a further 24 in MSP2. This is less than in a holoendemic area in Dielmo (MSP1 33; MSP2 47) (Konate et al. 1999) but more than in a mesoendemic area in Sudan (MSP1, MSP2 13) (Babiker et al. 1997). Altogether, these results are in line with previous studies (Beck et al. 1997; Kyes et al. 1997; Irion et al. 1998; Felger et al. 1999b). However, PCR methods underestimate allelic polymorphisms as there may be daily individual variations; alleles of identical size but with point mutations may not be differentiated and alleles present at very low density in a complex blood infection are likely to remain undetected (Contamin et al. 1996; Daubersies et al. 1996). As a result of the study design as a cross-sectional survey, unfortunately it was not possible to show the daily individual variations of strains.

The number of possible combinations between the amplification products of the MSP1 and MSP2 genes provides a highly distinctive method for the identification of different clones in the investigated samples (Babiker et al. 1997; Haddad et al. 1999). Among the isolates that were investigated in this study, multiplicity of infection was highest in the age group 3–5 years with allelic families of MSP1 (average 3.2 clones) and in the age group 6–10 years with MSP2 (average 2.5 clones) (Table 1). Multiplicity of infection appeared to decrease with age but unlike in other studies (Felger et al. 1999; Konate et al. 1999; Smith et al. 1999), this trend was statistically not significant. When comparing isolates from holo- and mesoendemic areas, 78 and 64% of the samples, respectively, contained multiple MSP1 and MSP2 genotypes, comparable with studies in Zimbabwe (Creasy et al. 1990), Senegal (Ntoumi et al. 1995) and Tanzania (Babiker et al. 1997). However, the differences between the holo- (MSP1: average 2.9 clones; MSP2: average two clones) and mesoendemic area (MSP1: average 2.5 clones; MSP2: average 2.2 clones) were not significant. Work by Konate et al. (1999) showed similar results from the holoendemic village in Dielmo (Senegal) with a multiplicity average of 2.8 clones with MSP1 and three clones with MSP2, while multiplicity of infection in the mesoendemic area of Ndiop (Senegal) with 1.5 clones (MSP1 and MSP2) was clearly lower than the mesoendemic area investigated in this study in Uganda. This might be because of increased and prolonged rainfalls during October–December 1997 in the Kabarole District of western Unganda, which in turn caused increased reporting of malaria cases in local health centres in January and February 1998 (Kilian et al. 1999). The higher number of multiple P. falciparum infection in the mesoendemic area could be a result of this rain pattern.

Multiplicity of infection of MSP1 and particulary of MSP2 was positively correlated to parasite density in most age groups (Fig. 2). Studies from Tanzania reported a similar correlation between multiplicity of infection and parasite density across age groups (Beck et al. 1997; Felger et al. 1999; Smith et al. 1999).

Parasite genotyping is an established tool in molecular epidemiology and enables researchers to gain more understanding of premunition in P. falciparum. But we need more studies to elucidate the dynamics of transmission of multiple infections and about the potential of genetic recombination and the distribution of genetic traits such as drug resistance.


This work was supported in part by Friedrich Baur Stiftung, Förderprogramm für Forschung und Lehre der Medizinischen Fakultät der Universität München, and the Gesellschaft für Technische Zusammenarbeit (GTZ). Gabriele Peyerl-Hoffmann is supported by the Democh Maurmeier Stiftung. We thank all villagers and village health workers for their assistance.