There is a tremendous potential for mutation and selection during the asexual proliferation of Plasmodium falciparum. Estimation of mutation rates resulting in drug resistance is important for the development of malaria control strategies. Furthermore, there is significant selection pressure against mutant parasites through competition from wild-type (non-mutant) parasites and from the host immune response. Examination of these components of malaria biology will lead to a greater understanding of Plasmodium dynamics. Thus, the purpose of this thesis was to test the hypothesis that Plasmodium drug resistant mutants arise relatively quickly under drug selection pressure, but in the absence of drug pressure are selected against due to competition with wild-type (non-mutant) parasites and the host's immune response.
The thesis describes new methodology for measuring mutation
rates in the P. falciparum dihydrofolate reductase (DHFR) and cytochrome b (cytb) genes by a combination of mathematical modelling using the Mutation Model and large scale in vitro culturing of parasites. Selection pressure on parasites expressing the mutant DHFR gene by non-mutant (wild-type) parasites was investigated in vitro. The in-host dynamics of parasite survival in the naive host was analysed with a mathematical stochastic simulation model, the Disease Model, to show a close relationship between host immunity and the switching of P. falciparum var genes, which encode variant erythrocyte surface antigens.
The mathematical Mutation Model was developed to simulate P. falciparum growth in vitro and the probability of a mutation occurring during DNA replication. The model was utilized to
determine the number of replicates required to measure mutation rates > 1x10-10 mutational events/gene/replication. Factors which could affect the frequency of mutant positive replicates were also investigated. A reliable in vitro method was established to assay a large number of replicates and detect a small number of viable parasites from a single micro-culture replicate.
The mutation rate in the P. falciparum dihydrofolate reductase (DHFR) gene was calculated from the difference in DHFR mutant frequencies at two separate timepoints. Point mutations were detected in the DHFR gene at codon-46 (TTA to TCA) (a novel mutation) and codon-108 (AGC to AAC), resulting in serine replacing leucine (L46S) and asparagine replacing serine (S108N) respectively in the DHFR enzyme. The L46S and S108N DHFR mutations arose spontaneously in vitro before drug pressure and caused a decrease
in pyrimethamine sensitivity. The P. falciparum L46S mutants were dominated in vitro by the wild-type parasite population. Construction of a Plasmodium DHFR molecular model illustrated the proximity of L46S and S108N to the substrate binding site, the changes in antifolate sensitivities and the possible reason for DHFR wild-type and mutant growth rate differences.
The upper and lower limits on the mutation rates were determined by simulations of the experimental data using the Mutation Model. At a given position in DHFR the mutation rates occurred at <2.5xl0-9 mutations/DHFR gene/replication. This low rate was consistent with frequencies of Plasmodium pyrimethamine resistance in rodent models. A low mutation rate of 5x10-9 events/parasite/replication was also estimated in the P. falciparum cytb
gene. The mutation rates in P. falciparum DHFR and cytb were consistent with mutation rates estimates from other low order eukaryotes.
The mathematical Disease Model has provided the first qualitative framework for the in-host dynamics of a malaria infection which encompasses host variant specific, non-variant specific, non-specific immunity and var gene switching which produces new variants. The upper and lower limits on these parameters, which produced characteristic patterns of chronic parasite recrudescence, were predicted. Values outside these defined limits resulted in non-characteristic patterns of clinical malaria or premature death of the parasite population. Three mechanisms of var gene switching were investigated: ordered, random and uncoupled. Irrespective of the switching mechanism, modelling of fast switching rates (10-2 –
10-3 switching parasites per generation) always resulted in the var gene repertoire being exhausted within 60 days and therefore could not predict chronic patterns of parasitaemia. This showed that a repertoire of fast switching var genes, which had been estimated for P. falciparum in vitro, was not compatible with a chronic infection. The mathematical simulations predicted that not all of var gene switching mechanisms were able to produce the most realistic patterns of parasite recrudescence. However, uncoupled 'on' and 'off’ switching of var genes could predict patterns of parasitaemia similar to chronic parasitaemia described from experimental data and in human infections but only when the var genes switched at medium to slow rates of 3x10-4 to 3x10-7 switching parasites per generation.
This seems the most simplistic mechanism for the control of gene expression. The modelling also predicted a close interaction between the antigenic thresholds governing the immune response and the var gene switching rate. Also a 10 fold increase in the number of parasites which stimulated the non-specific immune response allowed patterns of parasitaemia to be predicted by var genes which switched at slower rates.
The estimation of P. falciparum mutation and switching rates are important results for Plasmodium genetic diversity and antimalarial drug therapy. This study demonstrates that mutants arise not as a consequence of drug pressure, but from spontaneous mutational events. Even with low mutation and switching rates antimalarial resistance will inevitably arise in the global population, due to the very real probabilities of mutant parasites being selected out under drug pressure
and then surviving host immunity until transmission.