Does the drug sensitivity of malaria parasites depend on their virulence?

Here, the first experimental tests of how genetically related parasites, differing in virulence, perform under a range of drug treatment regimes are presented. It is well known that serial passage increases parasite virulence [30, 31]. This makes it possible to derive parasites, which differ in virulence but come from the same clonal lineage. Thus, the effect of chemotherapy on asexual- and transmission-stage densities can be tested on genetically similar parasites differing in virulence. To mimic the possible causes of sub-optimal treatment that can occur in the field (reviewed in the Introduction), parasites were challenged with sub-optimal doses and sub-optimal duration of drug treatment when symptoms first appeared.

The original Plasmodium chabaudi isolate was obtained from a thicket rat from the Central African Republic [32]. Genotype CW_avir (CW175) was obtained after cloning of the wild isolate and 4 passages in mice. CW_vir (CW202) was derived from CW_avir by 11 serial passages in mice [30] and became more virulent in the process [33]. Samples of CW_avir and CW_vir were passaged a further three times in C57Bl/6J mice before the experiments described here. None of the parasite lines involved, nor any of their ancestors had been exposed to antimalarial drugs prior to these experiments.

Drug sensitivity to the antimalarial drug pyrimethamine was tested in vivo using 6–10 week old C57Bl/6J female mice (Harlan), with access to food (41B maintenance diet, Harlan) and drinking water supplemented with 0.05% para-amino benzoic acid [34]ad libitum. Mice were infected with 10⁵ parasites of either CW_vir or CW_avir by intraperitoneal injection and given treatment with the antimalarial drug pyrimethamine from day 5 after infection, when first symptoms of disease occurred. The experimental setup is shown in Table 1. Pyrimethamine was administered during four consecutive days in experiment 1 (five mice per group) and experiment 2 (four mice per group) at doses of 0, 1, 3 and 5 mg/kg or, in experiment 3, on a single day at doses of 0, 4, 8, 12 or 20 mg/kg (four mice per group). In experiment 1, four mice infected with CW_vir were excluded from analysis, three of these were severely anaemic and had to be euthanized early in the experiment (dose 0, n = 1; dose 1, n = 2) and parasites were not detected at time of treatment in another mouse (dose 5; Table 1). In experiment 2, all four mice in the untreated CW_vir group had to be euthanized a few days after the peak of infection on day 11 (n = 1) or 12 (n = 3), and available data have been used in the analysis unless indicated otherwise. Pyrimethamine was dissolved in dimethyl sulphoxide at the required concentration and injected intraperitoneally with a maximum total volume of 50 μL. To monitor treatment efficacy, parasite dynamics and virulence (weight loss, anaemia) were followed until three weeks post infection, when mice no longer show disease symptoms. Asexual parasites were counted per 1000 RBCs by examination of Giemsa-stained thin smears from tail bleeds. Red blood cell density was determined by flow cytometry (Beckman Coulter) and used to calculate parasite density/mL of blood. Blood samples for RNA (10 μL) and DNA (5 μL) extraction were stored to quantify gametocytes by quantitative reverse-transcriptase PCR and for sequencing to detect pyrimethamine resistance mutations.

Ten μL blood samples were added to 30 μL nucleic acid purification lysis solution (Applied Biosystems) and 15 μL PBS, gently mixed and stored at -80°C. RNA was extracted using the ABI Prism 6100^® [35] and cDNA was obtained by reverse transcriptase PCR (high capacity cDNA archive kit, Applied Biosystems) according to manufacturer's protocols. Gametocytes were counted using quantitative real-time PCR with primers based on the gametocyte-specifically expressed gene PC302249.00.0 [36].

During qPCR analysis for gametocyte quantification, three different dilution series of positive control samples, initiated from different mice with P. chabaudi gametocytes, were tested in concentrations ranging from 7 to 7*10⁶ gametocytes/mL. The average threshold cycle (Ct) values ± 1 standard deviation are shown in Figure 2. Gametocyte numbers have a high correlation to Ct values (R² = 0.99; p < 0.0001). The absolute detection limit for the assay was determined to be at least 700 gametocytes/mL of blood. Lower gametocyte densities were detected in 88% (221 gametocytes/mL), 67% (71/mL), 43% (22/mL) and 12% (7/mL) of the tested control samples. This decline in probability of detection is most likely related to the probability of target nucleic acid being present in the 10 μL blood sample.

Pyrimethamine is an antifolate drug that inhibits the enzyme dihydrofolate reductase (DHFR) that is essential in the folic acid pathway, ultimately leading to pyrimidine synthesis. Pyrimethamine inhibits parasite multiplication but does not directly kill gametocytes. As with other malaria parasite species, resistance to pyrimethamine in P. chabaudi is conferred by point mutations in the dhfr gene, resulting in conformational changes of the enzyme and decreased binding capacity of the drug [37, 38]. Samples from the day of infection and day 21 were analysed to detect resistance mutations in the dhfr gene, using the following protocol. DNA was extracted using InstaGene Matrix (BioRad) according to the standard protocol for DNA preparation from whole blood. The extracted DNA, forward primer pcdhfr-10 5'-GCTATTTCTTTCTACATTTGC-3' and reverse primer pcdhfr-11 5'-TTTAAAATGATGAGCATGCTC-3' were used to amplify the region of the dhfr gene that contains possible mutations involved in pyrimethamine resistance. The PCR reaction included 13.75 μL water, 10 μL 5× PCR buffer (Promega), 3 μL 25 mM MgCl₂, 1 μL dNTPs 10 mM, 0.25 μL Taq 5 u/μL, 1 μL of each primer 10 μM and 20 μL DNA sample. The temperature programme used for amplification was 95°C for 60 seconds, then 30 cycles of each 95°C for 60 seconds, 52°C for 60 seconds, 65°C for 60 seconds and ending with 65°C for 600 seconds and storage at 4°C. PCR products were purified using Qiaquick PCR purification kit (Qiagen Ltd. UK) according to manufacturer's protocol and sequenced.

The maximum and cumulative measures of parasite and gametocyte density achieved by each infection during the 21-day experiments were used as response variables in the analysis. Plasmodium chabaudi has a cell-cycle of approximately one day for asexual parasites, therefore, cumulative parasite density is a measure of the total number of parasites present during a defined period of the infection. Gametocyte density correlates positively with infectivity to mosquitoes [21, 39] and, when summed over the course of an infection, provides a measure of transmission potential. Cumulative parasite and gametocyte counts were log₁₀-transformed to meet assumptions of normality and homogeneity of variance. Analyses were performed using general linear models in SPSS version 14.0 (SPSS Inc., Chicago, USA). Full models for analysis of parasite data included the main effects of clone (CW_avir or CW_vir; virulence classification) and dose, their 2-way interaction, experiment (1–3) as a random factor, and the covariates weight and anaemia at the day of infection as well as parasite density at the start of treatment. The clone × dose interaction indicates that CW_vir and CW_avir were differentially affected by drug dose, the hypothesis under test (Figure 1). Models were minimized using step-wise deletion.

As found previously [33], CW_vir was indeed more virulent in the absence of chemotherapy, generating greater weight loss and anaemia than CW_avir (Figure 3; mean difference ± s.e. of maximum weight loss: 15 ± 2% of starting body mass; F_1,21 = 68.6, p < 0.001; mean difference ± s.e. 12 ± 2% of starting RBC density; F_1,21 = 47.9, p < 0.001).

In the absence of drugs, CW_vir achieved higher maximum asexual parasite densities than CW_avir (mean difference ± s.e. 9.9 ± 2.5 × 10⁸, F_1,21 = 21.5, p < 0.001). Drug treatment reduced maximum asexual parasite density and delayed peak parasite density in a dose-dependent manner (Figure 4). At all doses, cumulative asexual parasite density was reduced by drug treatment (dose effect, F_7,90 = 58.9, p < 0.0001), but remained significantly higher in CW_vir than CW_avir infections (Figure 5; clone effect, F_1,90 = 153.4, p < 0.0001). The relative reduction in parasites densities induced by the different doses differed between the CW_vir and CW_avir (clone × dose interaction F_7,90 = 3.6, p = 0.002), with the impact disproportionately greater on CW_avir. Thus, sub-optimal chemotherapy enhanced the relative difference in cumulative parasite densities between the two clonal lineages (Figure 5).

In the absence of drug treatment, CW_vir infections had higher peak gametocyte densities (mean difference ± s.e. 34.4 ± 12.8 × 10⁶, F_1,19 = 7.3, p = 0.014) as well as higher total gametocyte production (Figure 6; F_1,15 = 30.9, p < 0.001) than CW_avir infections.

Total gametocyte numbers were significantly reduced by treatment (dose effect, F_7,82 = 16.0, p < 0.001) and the higher transmission potential for CW_vir is generally maintained under conditions of drug treatment (clone effect F_1,82 = 51.7, p < 0.001).

Overall the experiments, the effect of drug treatment on the two parasite lines was marginally different (clone × dose interaction F_7,82 = 2.1, p = 0.06). However, breaking these data down by experiment showed that there was no general pattern (Figure 6). Gametocyte densities in the third experiment, which involved one day of treatment show similar patterns (Figure 6C) to those seen in the asexual densities (Figure 5), but the disproportionate reductions in transmission potential of CW_avir at the higher doses was not statistically significant (clone × dose F_4,17 = 1.62, p = 0.196). Due to loss of data for the untreated CW_vir group in experiment 2 (see Methods), it is not possible to determine the differences in transmission potential between the two lines in the absence of treatment (Figure 6B). Among the remaining (treated) groups, the different doses had similar effects on both CW_vir and CW_avir (clone × dose F_2,20 = 0.11, p = 0.901). In experiment 1, the clone differences did vary with treatment (clone × dose F_3,24 = 9.45, p < 0.001), but inspection of Figure 6A shows that chemotherapy is causing similar proportionate reductions in gametocyte numbers for both lines except at an intermediate dose, which was disproportionately effective against CW_vir (or ineffective against CW_avir). The cause for this effect is unknown and it was not observed in the repeat experiment (Figure 6B).

Mutations known to confer resistance to the drug pyrimethamine in P. chabaudi [37, 38] in parasites of either clonal lineages were not observed, either at the start of the infections, or among parasites surviving treatment and present on day 21 post-infection, the last day of the experiment.