Mosquito age and the stage of Plasmodium falciparum infection affect locomotor activity
Locomotor activity of uninfected female An. gambiae in the absence of human odour significantly reduced with age when assessed in Drosophila activity monitors (χ21 = 5.86, p = 0.02; Fig. 1a-b). This age effect was not observed following infection with P. falciparum (χ21 = 0.83 p = 0.35; Fig. 1b). In the presence of human odour, the age-dependent reduction in locomotor activity was observed in both uninfected (χ21 = 21.12, p <0.001) and P. falciparum carrying mosquitoes (χ21 = 5.73, p = 0.02; the GLMM:lmer model was constrained by age and the random effect of experimental replication; Fig. 1b).
Infection status significantly affected the activity of mosquitoes when comparing the overall locomotion profiles of 7 dpi and 14 dpi individuals with age-matched uninfected females in the presence and absence of human odour (β-estimated = weighted mean value based on all variables plus the random effect of the replication ± SE; presence of human odour: χ21 = 18.06, p < 0.001; absence of human odour: χ21 = 5.73, p < 0.01). In the absence of human odour, 7 dpi females were less active compared to age-matched controls (χ21 = 8.21, p = 0.004), whereas 14 dpi females were as active as their age-matched controls (χ21 = 2.27, p = 0.13; Fig. 1c). The presence of human odour differentially altered the locomotor activity. While there were no significant differences between 7 dpi and age-matched females in the presence of human odour (χ21 = 0.12, p =0.73), the infectious 14 dpi mosquitoes were more active than age-matched uninfected females (14 days control; 14 dc χ21 = 8.80, p = 0.002; Fig. 1c).
Mosquito locomotor activity varies temporally in response to host odour and P. falciparum infection
Diurnal cycles in An. gambiae locomotor and host-seeking activities were evident, with distinct crepuscular activity peaks culminating in the highest activity at the onset of photophase (ZT 23-24), and low to moderate activity during scotophase (ZT 12-23) and low activity during photophase (ZT 0-11)26. To assess the diurnal effect on the locomotor activity of mosquitoes with P. falciparum infection and their respective controls, a mixed model analysis was used (lmer model: constrained by two main explanatory variables, infection status and period of time, as well as the random effect of replication, Supplementary Table 1-2). In the absence of human odour, the diurnal locomotor activity profile of uninfected mosquitoes confirmed the high activity at dawn (ZT 23-24), with the younger cohort being more active during scotophase with a peak at dawn, whereas the older control females displayed an increased activity at dusk that was maintained until dawn (ZT 0-11: χ21 = 27.43, p < 0.001; ZT 11-12: χ21 = 4.58, p = 0.049; ZT 12-23: χ21 = 4.33, p = 0.03; ZT 23-24: χ21 = 32.32, p < 0.001, Fig. 1d, Supplementary Table 2.). In the presence of human odour, the locomotor activity of the younger uninfected cohort increased at dawn, and remained at this activity level until dusk, while the older females only displayed an increase in activity at dusk (ZT 0-11: χ21 = 17.54, p < 0.001; ZT 11-12: χ21 = 7.67, p = 0.005; ZT 12-23: χ21 = 15.21, p < 0.001; ZT 23-24: χ21 = 18.25, p < 0.001, Fig. 1d, Supplementary Table 2.).
The time-of-day, and the stage of P. falciparum sporogony, differentially affected the locomotor activity of the vector (Fig. 1d). In mosquitoes with parasites undergoing sporogony, there was no effect of age on locomotion throughout the 24 h period in the absence of human odour (ZT 11-12: χ21 = 1.71, p = 0.18; ZT 12-23: χ21 = 0.31, p = 0.57; ZT 23-24: χ21 = 3.49, p = 0.06, Supplementary Table 2.), except during photophase, in which infectious (14 dpi) mosquitoes demonstrated lower activity than infected (7 dpi) individuals (ZT 0-11: χ21 = 21.97, p < 0.001). However, in the presence of human odour, the infectious cohort (14 dpi) was more active at dawn (ZT 23-24: χ21 = 15.40, p < 0.001), while being less active than the younger mosquitoes throughout the rest of the diurnal period (ZT 0-11: χ21 = 9.35, p = 0.002; ZT 11-12: χ21 = 20.02, p < 0.001; ZT 12-23: χ21 = 9.73, p = 0.001, Supplementary Table 2.).
Individual locomotor profiles of mosquitoes at 7 dpi and 14 dpi differed significantly from that of age-matched uninfected controls (Fig. 1d). Infection significantly reduced the locomotor activity in 7 dpi females in the absence of human odour when compared with their uninfected counterparts during most time periods (ZT 0-11: χ21 = 8.49, p = 0.03; ZT 11-12: χ21 = 1.008, p = 0.31; ZT 12-23: χ21 = 7.84, p = 0.005; ZT 23-24: χ21 = 13.78, p < 0.001, Supplementary Table 1-2). Mosquitoes carrying the transmissible stage of the parasite (14 dpi) displayed a lower locomotor activity than the uninfected counterparts throughout photophase and scotophase (ZT 0-11: χ21 = 5.92, p = 0.01; ZT 12-23: χ21 = 11.87, p < 0.001, Supplementary Table 1-2), but not at dusk and dawn (ZT 11-12: χ21 = 2.68, p = 0.10; ; ZT 23-24: χ21 = 2.20, p = 0.10). In the presence of human odour, the infected (7 dpi) females showed lower activity than age-matched controls at dawn (ZT 23-24: χ21 = 7.29, p = 0.006), while the inverse activity pattern was observed during photophase (ZT 0-11: χ21 = 12.28, p = 0.001). The infectious mosquitoes (14 dpi) demonstrated a significantly increased locomotor activity at dawn (ZT 23-24: χ21 = 23.78, p < 0.001) and during scotophase (ZT 12-23: χ21 = 25.42, p = 0.001), while there was no difference in activity of the infectious and uninfected females during photophase (ZT 0-11: χ21 = 2.26, p = 0.13), with a significant reduction in dusk (ZT 11-12: χ21 = 16.06, p < 0.001, Supplementary Table 1- 2).
Plasmodium falciparum modulates antennal transcript abundance
Paired-end sequencing of each of the libraries constructed from antennal RNA, with a total of 2 400 antennae, generated an average mapping of 26 399 740 million cleaned reads per library. Out of the 13 832 coding genes annotated in the genome of An. gambiae (Agam4.10), a total of 10 115 transcripts were reliably detected above 1 transcript per million (TPM) mapped reads in the antennae, among all experimental and control groups, demonstrating an adequate level of coverage.
A principal component analysis (PCA) of the antennal transcripts was conducted to demonstrate the overall variation among the antennal transcriptomes (infected, infectious, and age-matched uninfected conditions; 4 replicates each; Fig. 2a). The PCA identified that 49.8% of the variation among the libraries was based on the relative infection status in each age group (PC1), while 15.3% of the variance was dependent on age and infection status relative to the controls (PC2; Fig. 2a). All of the biological replicates of the same age and infection status clustered tightly together in the principal component space, except for the infectious samples (14 dpi), demonstrating that variation in the libraries due to handling and processing was successfully minimised. The separation of the four libraries of the 14 dpi samples into two clusters correlates with demonstrated differences in parasite load of the mosquitoes (Fig. 2a; Supplementary Fig. 1).
Antennal transcripts were significantly differentially regulated between the two age-matched cohorts (6 187), of which 3 465 were differentially abundant only between 14 dpi and 14 dc, whereas in 7 dpi and 7 dc, 850 were differentially abundant (Fig. 2b). In total, 4 807 transcripts were significantly differentially regulated between the two age-matched control groups and the two groups carrying P. falciparum parasites, of which 1 071 transcripts were shared between them (Fig. 2c). Within the age-matched control groups, 2 614 transcripts were uniquely regulated, whereas 1 122 were uniquely regulated within the two groups carrying P. falciparum parasites (Fig. 2c; Supplementary Fig. 2). Overall, differentially abundant transcripts were not condition-dependently regulated, except those regulated post-infection during both the infected and infectious stages (1 872), of which more than 80% of the transcripts were down-regulated post-infection (Fig. 2b).
To characterise the functional ontology of the differentially abundant genes in the antennae of infected and infectious mosquitoes with their age-matched controls, a gene ontology (GO) analysis of molecular function (level three) was conducted (Fig. 3). Of the 3 685 differentially abundant transcripts between the two uninfected control groups, the majority (>75%) were functionally classified as structural constituent of cuticle (GO: 0042302) and enzyme inhibitor activity (GO: 0004857; Fig. 3a). Both of these classes were more abundant in older compared to the younger individuals. None of the age-dependent GO terms identified in the uninfected cohort comparison (Fig. 3a) were detected in the pairwise comparisons of the antennal transcriptomes among infected, infectious and their age-matched controls. The three most represented functional classes in the pairwise comparisons between the infected and infectious groups with their age-matched controls, as well as between the infected and infectious antennal transcriptomes, were heterocyclic compound binding (GO: 1901363), organic cyclic compound binding (GO: 0097159) and ion binding (GO: 0043167; Fig. 3b). Two functional classes were regulated differently in the antenna of mosquitoes with a P. falciparum infection, odorant binding (GO: 0005549) and carbohydrate derivative binding (GO:0097367). The number of genes in the functional class odorant binding were regulated in both infected and infectious samples, while those in the carbohydrate derivative binding class were differentially regulated only in infectious samples (Fig. 3b).
A detailed analysis of the major chemosensory gene families associated with the odorant binding functional class, Ors, Irs, Grs, Csps and Obps, was conducted. As the mosquitoes aged, 34% of the chemosensory genes were significantly regulated in the antennae of uninfected females, with chemosensory receptors demonstrating higher abundance in the older females, while the binding proteins were both up- and down-regulated (Fig. 4). Following P. falciparum infection, the abundance of 18 Obps reduced with age, while only one chemosensory gene, Ir41a, demonstrated an increased abundance in the antennae of older females (Fig. 4). The differential abundance of these 18 Obps, along with 3 others, appear to be a result of an increased abundance at 7 dpi compared to the age-matched controls. The other chemosensory gene that was regulated at this age, Ir7u, was down-regulated upon infection.
The age-dependent regulation of chemosensory genes was affected following P. falciparum infection, as 73% of those genes that were up-regulated with age (Fig. 4 column 1), were shown not to increase in abundance post-infection (Fig. 4 columns 2 and 3), resulting in a higher abundance of these transcripts in the controls compared with those at 14 dpi (Fig. 4 column 4). The only exception to this was Ir7u, which was down-regulated at 7 dpi, exacerbating the decreased abundance observed at 14 dpi. Interestingly, when the 7 dpi mosquitoes were compared to 14 dpi, the abundance of the above-mentioned Obps were reversed, returning to pre-infection levels (Fig. 4 column 2).