TRANSMISSION OF MIXED PLASMODIUM SPECIES AND PLASMODIUM FALCIPARUM GENOTYPES

Study area and sample collections.

The study was carried out in Antula, a suburb located approximately 5 km north from Bissau, the capital of the Republic of Guinea Bissau. Climate is characterized by a rainy season (June–November) and a dry season (December–May). The country has been considered as mesoendemic-to-holoendemic with intense and seasonal transmission during the rainy season.^23,24 Anopheles gambiae s.s. and Anopheles melas are the main malaria vectors in the area.²⁵

Sampling took place at the end of the rainy season in 2 consecutive years, 1995 (October) and 1996 (November). In 1995, collections were made in 12 households, whereas in 1996, we had access to only 8 of these houses. All houses were mapped and their inhabitants fully identified (including the room and bed net which they slept in). The investigation was approved by the Ministry of Public Health of Guinea Bissau and by the Ethical Committees at Institute of Hygiene and Tropical Medicine of Lisbon and Uppsala University Hospital. Consent also was given by the local village chief. Each person (or parent) was informed of the nature and aims of the study and told that participation was voluntary.

In each household, blood samples were collected by finger-prick during the afternoon from all inhabitants for microscopic detection of gametocytes and for polymerase chain reaction (PCR) analysis. The next morning (7:00–8:00 a.m.), blood-fed resting An. gambiae s.l. females were collected inside the bed nets that had been used by the individuals surveyed the previous afternoon. In the 1996 survey, this sampling scheme was undertaken in 2 consecutive days to assess variations in the P. falciparum population’s composition in blood samples.²⁶

A total of 112 and 71 asymptomatic individuals were sampled in 1995 and 1996 (54 individuals were analyzed in both years, whereas 17 were new inhabitants in the area in the second year). About 45% of the population were <14 years old, age varying between 1 and 78 years old.

Assuming that extrinsic incubation period at an average temperature of 28°C is 9–10 days for P. falciparum, 14–16 days for P. malariae, and 12–14 days for P. ovale,²⁷ mosquitoes were kept in paper cups corresponding to each bed net for 8 days with glucose ad libitum. This period was chosen to allow the development of all species’ oocysts resulting from infections acquired the night before. In 1995, 44 mosquitoes were collected in 6 households (with 60 inhabitants) and held as previously. After this period, dissections were performed, and midguts (Mdg-D8) recovered for PCR analysis. In 1996, a subsample of 100 blood-fed females was dissected the day of collection. From these, 99 midguts (Mdg-D0) and 95 salivary glands (SG1-D0) were recovered. Samples Mdg-D0 were used to determine the parasite composition in blood meals, not being considered for the infection rate estimates. The remaining 192 mosquitoes were held for 8 days, after which they were dissected for midgut (Mdg-D8, 189 recovered midguts) and salivary glands (SG1-D8, 182 recovered salivary glands). We assumed that parasites found in salivary glands, either dissected in D0 or in D8, are the result of previously acquired infections, being considered together through analysis.

Identification of Plasmodium species and Plasmodium falciparum genotyping.

Parasite DNA was obtained from all samples by phenol/phenol-chloroform extraction and ethanol precipitation. Parasite species identification was performed by nested-PCR amplification of the small subunit ribosomal RNA genes.²⁸ P. falciparum genotyping was carried out in all positive samples for this species by nested-PCR amplification of block 2 of merozoite surface protein 1 gene (msp-1), the repeat region of merozoite surface protein 2 gene (msp-2), and the region RII of glutamate-rich protein gene (glurp) as previously described.²⁹ PCR products yielded from blood samples and mosquitoes from the same household were analyzed in the same agarose gel (MetaPhor; FMC Bioproducts, Rockland, USA). A final reading of band sizes was done for each allelic family in all gels to ensure a coherent classification of alleles.

Statistical analysis.

Possible associations between species were determined by (1) pairwise comparisons between observed and expected frequencies of mixed infections under the null hypothesis of species independence (i.e., prevalence of mixed infection is the product of individual species prevalence) and (2) tests of hierarchical log-linear models applied to three-dimensional tables.¹⁶ The most complex model involved a 3-factor interaction (i.e., association between all species). The simplest model was defined by the complete independence of species. Between these 2 extremes were intermediate 2-factor interaction models. The likelihood ratio (G-test) was used to determine the significance of each model. A significance level of 5% was considered. Calculations were made using SPSS Exact Tests 7.0 for Windows (SPSS Inc., Chicago, IL).

Plasmodium species.

In the human blood sample, infection rates of 46% (52 of 112) in 1995 and 59% (42 of 71) in 1996 were determined by PCR. Age of infected individuals varied (2–73 years old). In the mosquito sample, the infection rate in midguts was 32% (14 of 44) in 1995 and 18% (34 of 189) in 1996. An infection rate in salivary glands of 7% (19 of 277, SGl-D0 + SGl-D8) was determined only in 1996.

Absolute frequency of each Plasmodium species was determined considering the number of cases in which the species is present in either single or mixed infection. When whole populations were analyzed, P. falciparum and P. malariae were more frequent in humans, whereas P. ovale was more frequent in mosquitoes, being totally absent from humans in 1996 (Table 1). At the household level, there was a considerable correlation between Plasmodium species infections in the 2 hosts (Table 2). In 1996, P. ovale–infected and P. malariae–infected mosquitoes were found only in households where P. ovale–infected and P. malariae–infected persons were found in the previous year. The presence of infected persons in households was not indicative of infection in mosquitoes because in 3 households, infected mosquitoes were not found. The inverse—negative inhabitants and infected mosquitoes—did not occur. A thorough individual analysis was precluded because in most of the cases, several persons slept under the same bed net.

Mixed infections were more frequent in 1995 than in 1996 in humans (25% of infected people versus 7%) and mosquitoes (50% of infected mosquitoes versus 4%) (Table 3). Analyzing each household separately, we observed that although single infections occurred in almost every house, mixed infections were more abundant in houses that were not sampled in 1996 (Table 2). Analysis of mixed infections frequency was made only for 1995. The observed number of mixed infections between P. falciparum and P. malariae was significantly higher than expected in the human population (chi-square_{df 1} = 4.22, P = 0.04). In the mosquito population, mixed infections of P. falciparum and P. ovale presented a higher observed number than expected (chi-square_{df 1} = 9.49, P = 0.002). The 2-factor models that best fitted the data were a pair association between P. falciparum and P. malariae in humans (chi-square_{df 3} = 2.327, P = 0.507) and between P.falciparum and P. ovale in mosquitoes (chi-square_{df 3} = 3.617, P = 0.306).

Plasmodium falciparum genotyping.

Allele composition of blood samples taken from the same individual in 2 consecutive days was compared. The value of multiplicity of infection (MOI) (i.e., the number of distinct genotypes detected in an isolate) for each individual was determined combining both or considering the highest. In the human population, mean values of MOI were 2.000 in 1995 ($s_Y^{-}$= 0.208 , minimum = 1, maximum = 4) and 2.489 (s⁻_Y = 0.241, minimum = 1, maximum = 8) in 1996. In mosquitoes, a MOI of 1 was obtained in 1995, whereas in 1996, the mean MOI was 1.712 ($s_Y^{-}$ = 0.127, minimum = 1, maximum = 5). When MOI values in mosquitoes are analyzed, considering the dissected organ and day of dissection, we observed that the mean MOI was 2.050 ($s_Y^{-}$ = 0.185, minimum = 1, maximum = 4) in Mdg-D0 (with blood meal), 1.464 ($s_Y^{-}$ = 0.182, minimum = 1, maximum = 5) in Mdg-D8, and 1.600 ($s_Y^{-}$ = 1.581, minimum = 1, maximum = 2) in salivary glands (SGl-D0 + SGl-D8) (Figure 1).

In the human population, multiple-genotype infections were more abundant than single-genotype infections in both years (single versus multiple infections: 1995, 36% versus 64%; 1996, 32% versus 68% (Figure 1). In mosquitoes, considering all the data, single-genotype infections occurred in 50% of the P. falciparum–infected mosquitoes but when different organs were analyzed separately, multiple-genotype infections predominated only in Mdg-D0 (single versus multiple infections: Mdg-D0, 25% versus 75%; Mdg-D8, 71% versus 29%; SGl, 56% versus 44%) (Figure 1).

Regarding the msp-1, msp-2, and glurp alleles detected in humans and mosquitoes, we observed that mosquitoes harbored less allele diversity (i.e., number of different alleles) than humans in both years. In 1995, only 2 alleles of msp-1-K1 and msp-2-IC type and glurp were detected in mosquitoes (Figure 2). These were also present in the human population, but their prevalence seems to differ. The small amount of data precluded further analysis, however. In 1996, alleles detected in only 1 of the host populations were the less frequent except the msp-2-FC27-300⁻, which was more abundant in the human population but absent from mosquitoes (Mdg-D8 and SGl) (Figure 2). This allele was detected in Mdg-D0 of 5 mosquitoes, however.

P. falciparum genotypes present in both host populations were compared at the household level. In 1995, this comparison was possible in only 1 of the houses because of unsuccessful parasite genotyping of P. falciparum–infected mosquitoes collected in other houses. Of the 7 infected mosquitoes, 5 were infected with the same single genotype that was not detected in the inhabitants of the house. In 1996, occasional correlations could be established between P. falciparum–infected inhabitants and mosquitoes. Where a clear correlation was observed, mosquitoes exhibited single-genotype or dual-genotype infections, whereas inhabitants exhibited complex infections (Table 4). When parasite genotypes found in blood meals (Mdg-D0) were compared with the parasite genotypes of the blood samples, a correlation was established in only 5 mosquitoes.

Presence of gametocytes.

Prevalence of gametocytes was 9% (10 of 112; 70% of the gametocyte carriers were <14 years old) in 1995 and 15% (11 of 71; 36% of the gametocyte carriers were <14 years old) in 1996. In 1995, gametocytemias ranged from 16 to 128 gametocytes/μl for P. falciparum, from 16 to 96 gametocytes/μl for P. malariae, and from 16 to 32 gam/μl for P. ovale. In 1996, gametocytemias ranged from 16 to 96 gametocytes/μl for P. falciparum and 16 gametocytes/μl for P. malariae.

Gametocyte patency was not indicative of mosquito infections. Most correlations of infections in individuals and mosquitoes were established with individuals without visible gametocytes in the peripheral blood. Regarding Plasmodium species, from the 6 gametocyte carriers who seemed to have transmitted their infections to mosquitoes, 5 were <14 years old in 1995. Regarding P. falciparum genotypes, we verified that from the 11 persons with whom it was possible to establish a correlation with infections in mosquitoes, 8 were less than 14 years old. From these, only 4, aged 7 and 10 years old, showed a patent gametocytemia.

Plasmodium species.

Although differences in Plasmodium species frequency in the 2 host populations were observed when overall results are considered, the distribution of the parasite species in the mosquitoes collected in each house correlated with that found in the blood of the inhabitants. This correlation suggests that the infections found in midguts dissected after 8 days of maintenance were acquired in the last blood meal immediately before collection. Infections found in salivary glands also correlated with infections found in humans. Further, no infected mosquitoes were found in households where the inhabitants were negative, which might have indicated that infections had been acquired elsewhere. Therefore, each household is likely to work as an individual unit of transmission. All species found in mosquitoes also were found in inhabitants except for P. ovale in 1996, which was not detected in the human population. P. ovale–infected mosquitoes were found, however, only in the houses where this species was detected in their inhabitants the year before. Because P. ovale causes chronic infections and is characterized by short patency periods, it is possible that the same persons were still infected by P. ovale, transmitting this parasite to the mosquitoes.

Regarding mixed infections, the positive association between P. falciparum and P. malariae in the human population agrees with previous studies.^12–14 The result obtained in the mosquito population is different, however, because a positive association between P. falciparum and P. ovale better describes data. This result could be biased by the small sample size and by the fact that most P. ovale–infected mosquitoes were collected in 1 house where a P. ovale–gametocyte carrier lived. It also can be due to a higher “visibility” of this species in mosquitoes when compared with humans, owing to its already mentioned short patency periods. Alternatively, it can reveal a different pattern of distribution of Plasmodium species in the 2 hosts as suggested by McKenzie and Bossert.³⁰ P. ovale is normally suppressed by P. falciparum,⁴ but it can be more transmissible to mosquitoes if the suppression leads to enhancement of transmissibility.³¹ Some studies have reported alterations in gametocyte production when a Plasmodium mixed-species infection occurs.^32,33

Plasmodium falciparum genotypes.

In mosquitoes, the proportion of single-genotype infections was higher than multiple-genotype infections, whereas the opposite occurred in the humans. In midguts obtained after 8 days of mosquito maintenance and in salivary glands, single-genotype infections predominated. Conversely, in midguts obtained the day of collection, multiple-genotype infections prevailed, which reflected the presence of asexual parasites in the blood meal. Most of the mosquitoes dissected in D8 presented single-genotype or dual-genotype infections, whereas the humans who exhibited the same parasite genotypes carried more complex infections. Mosquitoes seem to acquire only 1 or 2 genotypes from an infective person. This may be due to the asynchrony in gametocyte production of different genotypes³⁴ or different ability between genotypes to produce gametocytes.³⁵ A lower number of msp-1, msp-2, and glurp alleles was detected in mosquitoes in both years, and the proportion of certain allelic families differed between humans and mosquitoes. Similar results were obtained in studies undertaken in Papua New Guinea¹⁹ and Tanzania.²⁰ In our study, although MAD20 and FC27 alleles were more abundant in humans, these appeared to be less transmissible to mosquitoes. A loss of sensitivity in PCR analysis of parasite DNA derived from mosquitoes is unlikely to justify these differences. Parasite DNA was prepared from dissected organs in which no PCR inhibition occurred.³⁶

Comparison of P. falciparum genotypes present in humans and mosquitoes in each household revealed that despite the correlation between Plasmodium species, the composition of P. falciparum populations seems to differ. These results could be explained by the characteristics of P. falciparum gametocyte production—the maturation period. P. falciparum gametocytes appear in peripheral blood approximately 10 days after schizogonic stages, implying that the detected genotypes in humans still are not being transmitted to the mosquitoes. The sensitivity threshold of PCR also could contribute to this result. Detection of low-density genotypes in blood samples can be hampered in multigenotype infections.³⁷ Nevertheless, low-density genotypes can be more transmissible to mosquitoes, as observed by Taylor et al²² with two clones of P. chabaudi. The differences found in allelic family frequencies also corroborate this. Parasites characterized by K1 and RO33 msp-1 alleles and IC msp-2 alleles seem to be transmitted more efficiently than msp-1-MAD20 and msp-2-FC27. In midguts dissected in D0, a correlation between genotypes present in midguts and in the peripheral blood was possible only in a few mosquitoes. Mosquitoes must have bitten people who slept in the room or house where they were captured because all were fully engorged, and An. gambiae s.s. is markedly endophagic and endophilic in this region, with a biting peak at 12 a.m. and 3–5 a.m. (K. Pålsson, unpublished data). We cannot exclude fully, however, the possibility of having fed outdoors or in other households. A more reasonable explanation lies on the asynchrony of different P. falciparum genotypes in the peripheral blood combined with the different moments in which blood collections and mosquito feeds were undertaken.

This study, in which the genetic diversity of oocysts and sporozoites was analyzed, and previous studies, in which heterozygote deficits were found in individual oocysts,^17,19 seem to indicate the limited genetic diversity of a single inoculum. Druilhe et al³⁸ stated that diversity of a single inoculum is composed of genetically related but diverse parasites not distinguishable by the PCR markers used. Considering that the same genetic markers were used to analyze P. falciparum diversity in both host populations, the diversity was still considerably higher in the human population exposed to these inocula. The higher genetic diversity found in humans is likely to result from superinfection. A limited genetic diversity of the inocula would benefit the parasite because a protective immunity would take longer to develop, as suggested by Mendis et al.³⁹

Several ongoing control programs may alter the proportions of parasite populations prevalent in a given region, which could have consequences on the acquisition of protective immunity or malaria pathology, if interactions occur between them.⁴⁰ Further studies on human malaria parasites other than P. falciparum and on interactions of Plasmodium parasites within vectors in suitable host/parasite models or in natural infections in endemic areas should be performed.

Our study indicated that transmission dynamics of Plasmodium species and genotypes is complex and can be affected by parasite interactions, requiring analysis at a microepidemiologic level. Further evaluations on parasite transmissibility or infectivity based on gametocytemia or gametocyte carriage assessed by optical microscopy can yield a misleading description of malaria transmission dynamics. Gametocyte patency was not an indicator of mosquito infections per se because most correlations of infections in individuals and mosquitoes were established with individuals without visible gametocytes in the peripheral blood. Regarding age of gametocyte carriers, gametocytes were detected in all age groups, but data suggest that younger individuals are more infective, being the most important infective reservoir in endemic areas, as stated by several authors.^41–43

A thorough analysis in which individual humans and resulting mosquito infections could be analyzed should be envisaged to clarify these epidemiologic questions further. The methodology used in this work could be applied to persons sleeping alone under a bed net. In addition, newly developed methods to identify which P. falciparum clones are producing gametocytes^44,45 should be applied in such studies. The development of similar molecular tools directed at other human Plasmodium species is also desirable.