Revealing mechanisms underlying variation in malaria virulence: effective propagation and host control of uninfected red blood cell supply

Abstract

Malaria parasite clones with the highest transmission rates to mosquitoes also tend to induce the most severe fitness consequences (or virulence) in mammals. This is in accord with expectations from the virulence-transmission trade-off hypothesis. However, the mechanisms underlying how different clones cause virulence are not well understood. Here, using data from eight murine malaria clones, we apply recently developed statistical methods to infer differences in clone characteristics, including induction of differing host-mediated changes in red blood cell (RBC) supply. Our results indicate that the within-host mechanisms underlying similar levels of virulence are variable and that killing of uninfected RBCs by immune effectors and/or retention of RBCs in the spleen may ultimately reduce virulence. Furthermore, the correlation between clone virulence and the degree of host-induced mortality of uninfected RBCs indicates that hosts increasingly restrict their RBC supply with increasing intrinsic virulence of the clone with which they are infected. Our results demonstrate a role for self-harm in self-defence for hosts and highlight the diversity and modes of virulence of malaria.

Figure 1.

Time series of (a) mean uninfected and (b) infected RBCs per clone (×10⁻² μl^–1); vertical lines show standard deviations across five mice; and (c) relationship between maximum parasitaemia and depth of the RBC trough (×10⁻² μl^–1), suggesting the virulence–transmission trade-off, with dashed lines to indicate quartiles across individuals. Clone colours on the last panel correspond to those on the first two, and range from red for the clone resulting in the deepest RBC trough (AT), through to deep green for the clone resulting in the shallowest RBC trough (CW).

Figure 2.

For eight clones (legend, colours reflect virulence measured as maximum anaemia), across days post infection (x-axis), (a) effective propagation P_e (standard errors shown as vertical lines); (b) changes in RBC density (×10⁻² μl^–1) not due to parasites (range across mice in each clone on each day indicated by vertical bars); values less than 0 (horizontal bar) indicate that destruction by immune effectors or the spleen exceeds replenishment; and (c) effective reproduction number, R_e; values less than 1 (horizontal line) indicate that the parasite population is shrinking; and (d) proportion of reticulocytes at every time point obtained from combined RBC and parasite dynamics (see text); either assuming that normocytes are mostly infected (solid line) or assuming that reticulocytes are mostly infected (dotted line).

Figure 3.

The ratio between minimum b_t and maximum I_t for mice within the eight clones. Values greater than 1 (horizontal line) indicate that more cells were killed by immunity or retention in the spleen in one time-step than by the parasite in one time-step, comparing the maximum of each for each mouse. The degree to which this occurs varies across clones.

Figure 4.

The effect of removing bystander killing or spleen retention on parasite virulence. (a) Simulated uninfected RBCs and (b) corresponding simulated infected RBC numbers (as in figure 1) obtained by taking individual-specific starting RBC densities and individual b_t values, and setting b_t to zero for time-steps when b_t < 0. (c) Observed minimum uninfected RBCs (×10⁻² μl^–1, x-axis) versus simulated minimum uninfected RBC obtained as described earlier (y-axis); vertical and horizontal lines indicate quartiles across individuals; the trough is deeper when no bystander killing is implemented for mice infected by several of the clones; exceptions include many mice in the AD and AS clones.

Figure 5.

Strain differences in patterns of bystander killing and their impact (a) upper quantile of P_e for each clone, versus lower quantile of bystander killing (corresponding to the greatest loss) across individuals for that clone (×10⁻² μl^–1, as in figure 2b), indicating that clones that have highest effective propagation also experience the greatest magnitude of bystander killing (n = 8, y = −2846 + 17 8346 255x, p < 0.05, r² = 0.56); (b) observed (blue circles) and simulated minimum uninfected RBCs (×10⁻² μl^–1, y-axis) for each clone, taking median starting densities across all mice for each clone, and introducing either the median b_t observed across all mice at that time step (black squares), or the maximum between this median value and 0 (black triangles, b_t > 0). Removing bystander killing increases anaemia in all cases, except for the AS clone, with differential magnitude effects across different clones.