The impact of innate immunity on malaria parasite infection dynamics in rodent models

Abstract

Decades of research have probed the molecular and cellular mechanisms that control the immune response to malaria. Yet many studies offer conflicting results on the functional impact of innate immunity for controlling parasite replication early in infection. We conduct a meta-analysis to seek consensus on the effect of innate immunity on parasite replication, examining three different species of rodent malaria parasite. Screening published studies that span four decades of research we collate, curate, and statistically analyze infection dynamics in immune-deficient or -augmented mice to identify and quantify general trends and reveal sources of disagreement among studies. Additionally, we estimate whether host factors or experimental methodology shape the impact of immune perturbations on parasite burden. First, we detected meta-analytic mean effect sizes (absolute Cohen's h) for the difference in parasite burden between treatment and control groups ranging from 0.1475 to 0.2321 across parasite species. This range is considered a small effect size and translates to a modest change in parasitaemia of roughly 7-12% on average at the peak of infection. Second, we reveal that variation across studies using P. chabaudi or P. yoelii is best explained by stochasticity (due to small sample sizes) rather than by host factors or experimental design. Third, we find that for P. berghei the impact of immune perturbation is increased when young or female mice are used and is greatest when effector molecules (as opposed to upstream signalling molecules) are disrupted (up to an 18% difference in peak parasitaemia). Finally, we find little evidence of publication bias suggesting that our results are robust. The small effect sizes we observe, across three parasite species, following experimental perturbations of the innate immune system may be explained by redundancy in a complex biological system or by incomplete (or inappropriate) data reporting for meta-analysis. Alternatively, our findings might indicate a need to re-evaluate the efficiency with which innate immunity controls parasite replication early in infection. Testing these hypotheses is necessary to translate understanding from model systems to human malaria.

Keywords: innate immunity; meta-analysis; plasmodium; rodent malaria; within-host dynamics.

Figure 1

PRISMA flowchart (22) of the literature search and data extraction. Our complete dataset included 74 articles for analysis of innate immune perturbations (see Supplementary References ). Box colours indicate numbers of articles for the literature search (green), screened (pink and purple), included (blue) and excluded studies (yellow), with the number of articles that were removed (red arrows). The following abbreviations refer to: n, number of articles; Syst, Keyword search; Bac, Backwards; For, Forwards.

Figure 2

The predicted mean impact of innate immune perturbations from the meta-analytic model. The impact is expressed as the absolute difference in peak parasitaemia in a treatment relative to control group. Note that these plots do not indicate directionality, and so do not indicate that treatment groups have a higher parasitaemia than controls (or vice versa). Instead, they simply illustrate the size of the difference between treatment and control groups. Each panel refers to a different model system: (A) P. chabaudi, (B) P. yoelii non-lethal, (C) P. berghei and (D) P. yoelii lethal. Each data point represents experimental sampling points (the size of each dot indicates sample size) and the regression lines and 95% confidence intervals are shown.

Figure 3

Predicted impact of host (A) age in weeks and (B) sex on the absolute Cohen’s h effect size. (A) The regression line is shown in red with a 95% confidence band, and the points represent studies, with point size reflecting the sample size. (B) Mean absolute effect ± SEs in three sex categories (female only N = 7, male only N = 1 and mixed/unreported N = 23). The grey shading denotes the reference for Cohen’s effect size bands; small (0.2) and medium (0.5) (23).

Figure 4

The cartoon (left, A) illustrates the relative position in an immune signalling network of a target of experimental perturbation: input (purple) refers to factors with an upstream signalling position (e.g., pattern recognition receptors (PRR) and transcription factors) whereas output (orange) refers to downstream effectors (e.g., cytokines, nitric oxide and proteases). The plot (right, B) shows the mean and ± SE effect size for input and output interventions in the P. berghei model. The grey shading denotes the reference for Cohen’s effect size bands; small (0.2), medium (0.5) and large (0.8) (23). The cartoon was created using BioRender.com.

Figure 5

Mean ± SE effect size for the method of immune perturbation in the P. berghei model. GM = genetic modification (N = 13), Drug = agonist or antagonist chemical intervention (N = 13) and Mixed = any combination of two or more techniques from genetic modification, drug administration, surgery and adoptive cell transfer (N = 3). The grey shading denotes the reference for Cohen’s effect size bands: small (0.2) and medium (0.5) (23).