Experimental determination of the force of malaria infection reveals a non-linear relationship to mosquito sporozoite loads

Abstract

Plasmodium sporozoites are the infective stage of the malaria parasite. Though this is a bottleneck for the parasite, the quantitative dynamics of transmission, from mosquito inoculation of sporozoites to patent blood-stage infection in the mammalian host, are poorly understood. Here we utilize a rodent model to determine the probability of malaria infection after infectious mosquito bite, and consider the impact of mosquito parasite load, blood-meal acquisition, probe-time, and probe location, on infection probability. We found that infection likelihood correlates with mosquito sporozoite load and, to a lesser degree, the duration of probing, and is not dependent upon the mosquito's ability to find blood. The relationship between sporozoite load and infection probability is non-linear and can be described by a set of models that include a threshold, with mosquitoes harboring over 10,000 salivary gland sporozoites being significantly more likely to initiate a malaria infection. Overall, our data suggest that the small subset of highly infected mosquitoes may contribute disproportionally to malaria transmission in the field and that quantifying mosquito sporozoite loads could aid in predicting the force of infection in different transmission settings.

Fig 1. Acquisition of a blood meal by the biting mosquito is not associated with the successful transmission of Plasmodium.

Single mosquitoes were allowed to probe on individual mice until they obtained a blood meal or lost interest in probing. Following this, mosquito midguts were inspected for the presence of blood, salivary gland sporozoite loads were measured, and mice were followed for blood stage malaria infection. Shown is the percent of mice that became infected after being probed upon by mosquitoes that were or were not successful in obtaining a blood meal. Error bars show 95% confidence intervals calculated using Jefferey’s intervals for binomial distribution [66]. Logistic regression analysis of the probability of infection demonstrates that the acquisition of a blood meal has no impact on the likelihood of malaria infection (p = 0.705). n = total number of mice in each group. Shown are pooled data from 10 independent experiments with 10 to 45 mice per experiment. All experiments in which probe time was not controlled are included in this analysis for a total of 228 mosquito/mouse pairs.

Fig 2. Mosquito salivary gland sporozoite load significantly impacts the likelihood of malaria infection.

Single mosquito feeds were performed and subsequently salivary gland sporozoite loads were measured and mice were followed for blood stage malaria infection. Data are pooled from 20 independent experiments with 7 to 44 mice per experiment. Total n = 412 (A) Log-binning of mosquito salivary gland loads and percent of infected mice in each bin. Mosquitoes are binned using the traditional method of grading salivary gland infections on a log-scale and for each bin the percent of mice positive for malaria infection is plotted. Above each bar is the number of mouse-mosquito pairs in each bin (n). (B) Logistic regression analysis of the probability of malaria infection as it relates to salivary gland sporozoite load. Predicted probability of malaria infection, from logistic regression with a linear spline at 20,000 salivary gland sporozoites, is plotted against mosquito sporozoite load and shown is the smoothed Lowess curve. A statistically significant relationship between salivary gland load and probability of malaria was observed (p<0.001). The x-axis is cut at 100,000 to focus on sporozoite loads more commonly observed in the field. Logistic regression plotted for the full range of sporozoite loads in our dataset is shown in S2 Fig. (C) Mathematical modeling of infection probability as it relates to salivary gland sporozoite load. Three alternative models, (single hit, powerlaw, and threshold), were fit to the entire raw dataset (n = 412) and compared using Akaike Information Criterion to generate Akaike weights (w). The threshold model best describes the probability of malaria infection as a function of salivary gland sporozoite number (highest weight). The data points on the graph are shown for illustrative purposes: Raw data were binned with equal number of mosquitoes in each group (n = 41), except the last group, which has 43 mosquitoes. Error bars on the y-axis show 95% confidence intervals calculated using Jefferey’s intervals for binomial distribution [66] and on the x-axis show 67% confidence intervals calculated using normal distribution. The single hit, powerlaw, and threshold models are described in the Materials and Methods by Eqs 1, 2 and 3, respectively) and fit was determined using the maximum likelihood method (Eq 6 in Materials and Methods).

Fig 3. Mosquito probe time impacts the likelihood of malaria infection.

Single-mosquito feeds were performed with the duration of probing experimentally manipulated at 10 seconds, 1 minute, or 5 minutes. Following this, salivary gland sporozoite loads were measured and mice were followed for blood stage malaria infection. There was a statistically significant association between probe time and probability of malaria infection (Fisher’s exact test p-0.020). Pair-wise comparisons were performed using logistic regression with odds of malaria infection as the outcome, probe time as a categorical predictor and robust variance. The results indicated statistically significant differences between 10 sec vs 5 min (p = 0.021) and 1 min vs 5 min (p = 0.025); while the comparison between the 10 sec vs 1 min groups did not reach statistical significance (p = 0.275). n = total number of mice in each group, pooled from 6 independent experiments with 10 to 15 mice per group per experiment. ns = not significant. See Methods and Materials for details on experimental design and statistical analyses.

Fig 4. The anatomical location of mosquito probing does not impact the likelihood of malaria infection.

Single-mosquito feeds were performed on the ear, abdomen or tail of mice and probe time and blood meal acquisition were recorded. Following this, salivary gland sporozoite loads were measured and mice were followed for blood stage malaria infection. (A) Percent of mice positive for malaria infection when probed-upon at the indicated location. n = total number of mouse-mosquito pairs in each group, pooled from 4 independent experiments with 10 to 15 mouse-mosquito pairs per group per experiment. The risk of infection is not statistically different between the three locations (p = 0.655). (B) Probe time and blood meal acquisition of mosquitoes placed on different anatomical locations. During the location feeds, mosquitoes were allowed to probe until they took a blood meal or became disinterested and flew away. Times to either of these endpoints were measured and were lower for mosquitoes placed on the tail compared to the abdomen and ear, by 45 to 50 percent respectively (p<0.001). Pie Charts: The percent of mosquitoes feeding on the ear, abdomen and tail that succeeded in obtaining a blood meal is shown in red and was 66%, 51%, and 74%, respectively. This difference did not reach statistical significance (p = 0.06, Fisher’s exact test).

Fig 5. The impact of salivary gland sporozoite load on probe time and blood meal acquisition.

Probe time and acquisition of blood meal were recorded for each mosquito feed, and subsequently salivary gland sporozoite loads were measured. (n = 221 mouse-mosquito pairs). (A) Relationship of salivary gland sporozoite load and mosquito probe time. For each mosquito, these parameters were plotted and correlation analysis revealed no significant association (Spearman rank ρ = -0.03, p = 0.66). (B) Binned salivary gland sporozoite loads suggest a relationship to blood meal acquisition. Mosquitoes are binned using the traditional method of grading salivary gland infections on a log-scale and for each bin the percent of mosquitoes obtaining a blood meal is shown. Error bars show 95% confidence intervals calculated using Jefferey’s intervals for binomial distribution [66]. (C) Logistic regression analysis of blood meal acquisition as it relates to salivary gland sporozoite load. Predicted probability of blood meal acquisition as it relates to salivary gland load from logistic regression is plotted against mosquito sporozoite load with 95% confidence intervals shown in gray shading. This analysis indicates that for every increase of 10,000 salivary gland sporozoites, the odds of obtaining a blood meal goes down by 13% (95% CI: from 5% to 21%, p-value = 0.002).