A non-destructive sugar-feeding assay for parasite detection and estimating the extrinsic incubation period of Plasmodium falciparum in individual mosquito vectors

Comparing estimates of parasite’s EIP between the classic dissection approach and the non-destructive individual “spit” assay

Destructive approach: mosquito dissection and microscopic observation

A total of 121 mosquito females exposed to parasite isolate A and 114 to isolate B were dissected from 8 to 16 dpbm (between 8 and 20 females/day, median = 14) to assess microscopically the presence and number of oocysts in the midguts and of sporozoites in salivary glands. Salivary gland infections were also confirmed through qPCR. The infection rate was high with 117/121 (96.7%) and 114/114 (100%) of females exposed respectively to isolate A and B harboring parasite oocysts in their midguts (supplementary S44, Fig. S4a). The gametocytemia of isolate B (1208 gam/µl) was higher than that of isolate A (168 gam/µl), resulting in strong difference in the number of developing oocysts between the two isolates (B: 191.65 ± 21, A: 13.86 ± 2, supplementary S4, Fig. S4b, LRT X²₁ = 24.46, P < 0.001). Similar patterns of infection were observed in salivary glands for the sporozoite stages (supplementary S4, Fig. S4c,d).

The few uninfected mosquitoes (N = 4 individuals from isolate A) were excluded from the analysis of the EIP. None of the dissected mosquitoes at 8 and 9 dpbm exhibited ruptured oocysts and the first observations occurred at 10 dpbm for both isolates. As expected, there was a highly significant positive relationship between time post-bloodmeal and the proportion of mosquitoes with ruptured oocysts (LRT X²₁ = 147, P < 0.001, Fig. 1a). The timing of rupturing was similar between the two parasite isolates (LRT X²₁ = 1.39, P = 0.24, Fig. 1a). Using this metric, the estimated EIP₅₀ from the binomial model was 9.98 days for isolate A and 9.49 for isolate B.

The proportion of ruptured oocysts was higher in mosquitoes exposed to isolate A than B (A: 879 ruptured oocysts out of 1503 counted oocysts (58.5%) from 8 to 16 dpbm, B: 3589/19,035 (18.9%) over the same period of time, LRT X²₁ = 1045, P < 0.001, Fig. 1b). This suggests a negative effect of density on parasite growth at the oocyst level, while no such density-dependent effect was observed on the proportion of mosquitoes with ruptured oocysts (Fig. 1a) or with disseminated sporozoites (Fig. 1c,d). To confirm this hypothesis, the relationship between the proportion of ruptured oocysts and the total number of oocysts was explored at the mosquito individual level (n = 190 mosquitoes following the exclusion of the 8 and 9 dpbm time points for which there was no ruptured oocysts). There was a strong negative relationship (LRT X²₁ = 16, P < 0.0001, Fig. S4.2): the smaller the number of developing oocysts in mosquito guts, the greater the fraction of ruptured oocysts from 10 to 16 dpbm was.

Finally, the proportion of mosquitoes with disseminated sporozoites in their salivary glands, the most epidemiologically-relevant metric, did not vary between parasite isolate, be it measured through microscopic observation (LRT X²₁ = 0.96, P = 0.33, Fig. 1c) or qPCR (LRT X²₁ = 0.71, P = 0.40, Fig. 1d). The proportion of sporozoite-infested salivary glands increased with dpbm similarly for both parasite isolate (i.e. no significant dpbm by isolate interaction) and regardless of the method used. Using this metric, the estimated EIP₅₀ from the binomial models were 10.35 (microscopy) and 10.26 days (qPCR) for isolate A and 10.93 (microscopy) and 10.85 days for isolate B.

The estimated time required for the sporozoites to migrate and invade the mosquito salivary glands following the egress from the oocysts was thus 9 h for isolate A (i.e. EIP₅₀ derived from the sporozoites observation in salivary glands (Fig. 1c)—EIP₅₀ derived from the oocyst rupturing data (Fig. 1a): 10.35 days–9.98 days), and 34 h (10.93 days–9.49 days) for isolate B.

Non-destructive approach: the individual “spit” assay

Forty-two (42) females (21 per parasite isolate) were individually placed in tubes for saliva collection from 8 to 16 dpbm. Mosquito survival over the collection period is described in the supplementary S4 (Fig. S4e). At the end of the experiment on 16 dpbm, 37 females (17 in A and 20 in B) were identified as infected using qPCR. Of these, 26 (13 in each isolate) produced cottons containing detectable traces of parasite DNA by qPCR (i.e. positive cottons). The infected females that did not produce any positive cottons from 8 to 16 dpbm (4 females for isolate A and 7 for B) were excluded from the analysis because no EIP values can be derived from these samples. A total of 214 cottons (112 for A and 102 for B) collected from 26 females were thus analyzed.

Similar to microscopic observation, the first positive cottons occurred on day 10 for each isolate. The parasite EIP₅₀ at 27 °C using this assay was 14 days for isolate A and 13 days for isolate B (isolate effect: LRT X²₁ = 0.001, P = 0.97, Fig. 1e).

Individual estimation of EIP in different mosquito species

IRSS

The saliva of 20 An. arabiensis, 20 An. coluzzii, and 20 An. gambiae fed with the blood from a naturally infected gametocyte carrier (parasite isolate C) was collected using the “spit” assay from 8 to 20 dpbm. Of these, 18 An. arabiensis, 17 An. coluzzii, and 20 An. gambiae were confirmed as infected using qPCR on carcasses of dead mosquitoes. Cottons collected from uninfected females (N = 5) were discarded. Two infected females (one An. gambiae and one An. coluzzii) died at 7 dpbm before the collection of saliva has started (full survival results are given in supplementary S5, Fig. S5.2). A total of 368 cotton samples from 53 females (18 An. arabiensis, and 16 An. coluzzii, 19 An. gambiae) were therefore analyzed using qPCR. The proportion of positive cotton balls and the proportion of mosquitoes generating positive cotton samples are given in Table 1 for each species. Over the collection period from 8 to 20 dpbm, 19 individuals (2 An. arabiensis, 10 An. coluzzii and 7 An. gambiae), of the 53 infected females used to collect the saliva, never generated positive cotton samples (Table 1). This was mainly due to early mortality prior to the sporozoites invasion of mosquito salivary glands as this number fell to three (one from each mosquito species) between 13 and 20 dpbm.

EIP, defined as the time between the infectious blood meal and the first day of positive molecular detection by qPCR of P. falciparum from the cotton used to collect saliva, varied among species (LRT X²₂ = 8, P = 0.018, Fig. 2a). The shortest EIP₅₀ was observed in An. gambiae (11 days, min: 9, max: 16), followed by An. coluzzii (11.5 days, min: 9, max: 13) and An. arabiensis (13.5 days, min: 9, max: 18). Following parasite invasion of their salivary glands (i.e. the day when parasite DNA were first detected in cottons), individual females did not systematically generated positive cotton samples (Table 1). The “proportion of positive cottons post-EIP” refers to the sensitivity of our assay; that is, its ability to correctly detect parasite DNA in cotton samples collected on the days following the first day of positive detection. In other words, this is the proportion of cotton samples that tested positive for P. falciparum among those that were used to collect saliva of females that previously generated a positive cotton. Sensitivity was 0.69, 0.27, and 0.71 for An. arabiensis, coluzzii, and gambiae respectively (Table 1). Thus, An. coluzzii tended to deposit detectable quantity of sporozoites on cottons less frequently than the two other mosquito species following the parasite’s EIP.

A major driver of cotton positivity rate following parasite’s invasion of salivary gland was the infection intensity in females used to collect the saliva. First, An. coluzzii mosquitoes, the species with the lowest sensitivity (Table 1), also displayed lower infection intensity than An. arabiensis and An. gambiae (supplementary S5, Fig. S5.1a). Second, there was a positive relationship between the probability to generate Pf-positive cotton samples and infection intensity in individual females (LRT X²₁ = 10, P = 0.002, Fig. S5.1b), regardless of mosquito species (species by sporozoite intensity (Ct) interaction: LRT X²₁ = 1.9, P = 0.4, Fig. S5.1b). However, there was no relationship between the Ct of females used to collect saliva and the Ct of positive cottons (LRT X²₁ = 1.9, P = 0.4).

PSU

Parasite-positive cotton samples were detected in An. stephensi (PSU). Of the 15 mosquitoes sampled, 14/15 An. stephensi individuals generated positive cotton samples at least once during the sampling period. After the dissection of all surviving mosquitoes at the end of the experiments, it was found that 12/12 An. stephensi were infected with salivary glands found harboring sporozoites. EIP50 similarly defined as in 2a was 12 days (Fig. 2b).

The spit assay makes it possible to study the links between different traits at the mosquito individual level. For the 2A IRSS experiment, cotton samples were collected daily up to 20 dpbm but the females used were kept in tubes until their death, thus making it possible to link the EIP with mosquito lifespan at the individual level. There was a significant positive correlation between EIP and mosquito longevity (LRT X²₁ = 21, P = 0.035, Fig. 3) such that short EIPs were also associated with short mosquito lifespan. There was no mosquito longevity by species interaction on EIP (LRT X²₂ = 25, P = 0.07, Fig. 3), although 6 observations in An. coluzzii (very low statistical power) pointed to a negative correlation in this species.

Determination of mosquito sugar feeding rate

IRSS

Assay 1

The overall mosquito survival rate during the 8 days-assay was 89%, with An. gambiae dying at a faster rate than both An. arabiensis and An. coluzzii (8/30 vs 0/30 and 3/30, X²₂ = 11, P = 0.004). Dead mosquitoes were excluded from the analysis. A meal was scored as having been taken when filter papers at the bottom of the cups contained either blue, yellow or green fecal dots. The papers were changed daily and the presence of colored fecal dots was scored daily. Of the 79 mosquito survivors, 31 individuals generated colored filter papers all 8 collection days. The average number of positive days was 6.89 ± 0.14 (min: 2 days, median: 7 days, max: 8 days). The overall detected sugar feeding frequency over this period was 544 meals/632 feeding opportunities = 86%. The sugar feeding rate of An. arabiensis and An. coluzzii females was higher than that of An. gambiae (89%, 91% and 76% respectively, GLMM with mosquito identity considered as a random effect: X²₂ = 13.9, P < 0.001).

Assay 2

The 16 females used to assess the presence of colored fecal dots from 14 to 24 dpbm were all infected (mean Ct = 22.2 ± 1.37). A meal was scored as positive when filter papers at the bottom of the cups contained either blue, yellow or green fecal dots. A total of 91 colored cottons soaked in 10% glucose were retrieved from these infected females over the collection period, of which 63 were positive to P. falciparum (i.e. a 69% sensitivity) (Table 2, supplementary S6). The color of the cotton (yellow or blue) had no incidence on the probability to detect P. falciparum (X²₁ = 0.008, P = 0.93). Colored dots were observed on 61 filter papers of a total of 91 observations; that is, a daily feeding frequency of 67%.

Mosquitoes expectorating sporozoites of P. falciparum on cotton pads during the night (as evidenced by Pf-positive cotton samples, n = 63) egested colored droppings on 40 occasions (63.4%). Contrary to our prediction, the absence of P. falciparum in cottons was not necessarily associated to the absence of colored dots: there were 21 occurrences of parasite-negative cotton samples but positive sugar meals, and only 7 occurrences of parasite-negative cotton samples and negative sugar meals (X²₁ = 0.69, P = 0.40, Table 2). Starting from dpbm 15, there was a high proportion of green dots or combinations of green + yellow dots or green + blue dots. This shows that colored dots observed at 16:00 on a given day did not necessarily come from the digestion of a sugar meal taken during the same night, but could also result from previous sugar meals, meaning that mosquito sugar digestion can last > 24 h. These results suggest that measuring colored fecal dots does not reliably provide evidence of either sugar feeding by night, or of transmission of parasites to cotton pads.

PSU

The 15 An. stephensi mosquitoes maintained at 27 °C for 1 week produced sugar fecal dots most of days, suggesting a regular feeding. Due to some mortality, there were only 79 sample-days, but on 66 of these days mosquitoes produced at least 4 or more new dots on filter paper, suggesting they were likely feeding that day on the dyed sugar meal about 83.5% of days. No dots were observed on 8 sample-days, or 10.1% of the time, with some variation of days with fewer than 4 dots observed making up the difference, where it was unclear if they were producing new dots or still digesting the previous day’s meal.

Color fecal dot production was monitored for An. stephensi (8 days) or G3 strain mosquitoes (7 days) housed at either 20 °C, 27 °C, and 32 °C. As might be expected, G3 strain mosquitoes produced fewer fecal dots at lower temperatures and greater numbers of dots at high temperatures. An. stephensi showed a similar pattern, although mortality at 32 °C was very high in this test which could have affected average dot production due to some individuals dying before digestion was completed (Table 3).

Generally, these results suggest that using the sugar-feeding assay to detect parasites will be more effective at warmer temperatures, and more false negatives are predicted at lower temperatures if mosquitoes are feeding less often. Around typical temperatures used in many infection studies (27 °C) mosquitoes are estimated to have fed about 80–90% of days sampled.

Infection duration

IRSS

From a total of 34 cotton balls collected from 25 to 53 dpbm, 17 were positive. We could detect P. falciparum sporozoites in old cotton samples from 23, 26, 33, 36 and 39 dpbm.

PSU

For samples at PSU tested in groups of 1, 4, 5, or 100+ in cups, surviving An. stephensi were carried through until day 40 dpbm, although only a few of these samples were analyzed. From this, there was evidence of salivary gland infection and transmission to cotton substrate as late as day 40 post infection from a group of 3 (those remaining from a set of 5), and also the larger group of 100, which by this point had about 20 mosquitoes remaining. Dissected mosquitoes were all found to have sporozoites in their salivary glands at 41 days post infection. This is evidence that infection can persist at least until day 40 post infection, and possibly until mosquito death.

Source: Ecology - nature.com