Behavioral and postural analyses establish sleep-like states for mosquitoes that can impact host landing and blood feeding

Posture analysis

Basic rest–activity rhythms

The rest–activity rhythms of the three mosquito species were quantified using a Locomotor Activity Monitor 25 (LAM25) system (TriKinetics Inc., Waltham, MA, USA) and the DAMSystem3 Data Collection Software (TriKinetics). Originally, these systems were developed for Drosophila but recently have been utilized to measure the activity levels of several blood-feeding arthropods, including mosquitoes (Lima-Camara et al., 2014; Rosendale et al., 2019; Rund et al., 2012). Individual mosquitoes were placed in 25×150 mm clear glass tubes with access to water and 10% sucrose provided ad libitum. These tubes were placed horizontally in the LAM25 system which allows the simultaneous recording of 32 mosquitoes in an ‘8×4’ horizontal by vertical matrix during a single trial. The entire set-up was held in a light-proof low-temperature incubator supplied with its own lighting system at 24°C, 70–75% RH, under a 11 h:11 h L/D cycle (with 1 h dawn and 1 h dusk transitions). After the acclimation of the mosquitoes for 2 days, activity level was recorded as the number of times (in a minute) a mosquito crosses an infrared beam of the LAM25 in the middle of the locomotor tube. Data collected with the DAMSystem3 for 5 days (with the removal of mosquitoes that were not alive until the end of the assay) were analyzed using the Rethomics platform in R with its associated packages such as behavr, ggetho, damr and sleepr (Geissmann et al., 2019).

Sleep deprivation assay

Following the acclimation of the mosquitoes for 2 days and the establishment of a 24 h baseline day in the LAM25 system, sleep deprivation was conducted in the specific phase of interest. Sleep deprivation was achieved in the mosquitoes through the delivery of vibration stimuli (vibration amplitude=3 G) using a Multi-Tube Vortex Mixer (Ohaus, Parsippany, NJ, USA) attached to the LAM25 system. The vibration amplitude used in our study represents half of the vibration amplitude required to arouse all individuals in an earlier Drosophila-based study (Shaw et al., 2000). In the diurnal Ae. aegypti mosquitoes, three different sleep deprivation protocols were conducted based on modifications from another Drosophila-based study (Kayser et al., 2015): 12 h night-time deprivation (12NTD), 4 h night-time deprivation (4NTD) and 12 h daytime deprivation (DTD). Whereas in the nocturnal An. stephensi mosquitoes, only DTD was conducted. To accomplish 12NTD, a sequence of vibration pulses lasting 1 min, followed by 5 min of rest between pulses was programmed for the entire scotophase subsequent to the baseline day (see Fig. S1A). In 4NTD, vibration pulses lasted for 1 min followed by 1 min of rest between pulses in the first 4 h of the night (Zeitgeber time ZT12–ZT16) following the baseline day. This was done in such a way that the total number of vibration pulses obtainable in 12NTD was delivered in a short time frame (see Fig. S1B). DTD in Ae. aegypti and An. stephensi was conducted similarly to 12NTD, the only difference is that DTD was accomplished during the photophase that succeeds the baseline day (see Fig. S1C,D).

To calculate sleep loss, we used the mean difference of the sleep amounts in the scotophase (12NTD and 4NTD) or photophase (DTD) preceding the deprivation and that of the scotophase (12NTD and 4NTD) or photophase (DTD) during the deprivation. For the calculation of sleep gain, we used the mean difference of the sleep amounts in the photophase (12NTD and 4NTD) or scotophase (DTD) after the deprivation and that of the photophase (12NTD and 4NTD) or scotophase (DTD) before the deprivation.

Host landing and blood feeding assays

Host landing 4 h post sleep deprivation (PSD) was assessed in Ae. aegypti mosquitoes both in laboratory and field conditions. In the lab-based study, sleep deprivation protocol was similar to 12NTD (described under ‘Sleep deprivation assay’). The only difference was that a 17.5×17.5×17.5 cm knitted mesh-nylon cage (BioQuip) housing mosquitoes (10 per replicate) was attached to the Multi-Tube Vortex Mixer to achieve sleep loss, with the entire set-up held in a room isolated from host cues (26±1°C, 75±5% RH and 12 h:12 h L/D cycle).

In the field-based experiment, adult female mosquitoes were released into similar cages described earlier (10 mosquitoes per replicate), which were then placed into 47.5×47.5×47.5 cm knitted mesh-nylon cages (BioQuip). To achieve bulk sleep deprivation, the entire set-up was situated in a city environment, where high activity occurs and located near an air conditioning unit. This air conditioning unit was operating periodically over 24 h, but mainly in the scotophase, likely providing disturbance both through vibration and sound to prevent sleep. The control set-up was placed in a similar environment to the sleep-deprived counterpart, but located in a secluded area of the property where mosquitoes experienced reduced disturbances. Both locations were covered to allow the mosquitoes to remain in shaded areas. This experiment was conducted independently three times, which yielded similar differences. The presence of potential hosts within range of detection was likely much higher in the location with increased likelihood of disturbance. To determine the number of mosquitoes that landed on a host mimic, we used techniques adapted from previous studies (Barnard et al., 2011; Hagan et al., 2018). A host mimic (Hemotek feeder) filled with a mixture of water and 100 µl artificial eccrine perspiration (Pickering Laboratories, Mountain View, CA, USA) heated to 37°C was covered three times with parafilm and placed on top of the experimental cage. Incidental contact was distinguished from foraging contact by using mosquitoes that landed and remained for at least 5 s on the feeder. By recording using a video camera (7 White, GoPro, San Mateo, CA, USA), the number of mosquitoes that made foraging contact was counted in the lab experiment at 10, 20, 40 and 60 min after the artificial host was turned on. In the field experiment, this was counted only after 5, 10 and 15 min. Results were expressed as a proportion of the total mosquitoes that remained alive at the end of the experiment and compared with the control group.

To assess the influence of sleep deprivation on blood-feeding propensity in mosquitoes, Adult female Ae. aegypti were exposed to the legs of a volunteer human host for 5 min after 4 h PSD (approved by the University of Cincinnati IRB 2021-0971). The set-up and sleep deprivation protocol in this experiment were similar to the lab-based host landing assay described above. The number of mosquitoes that successfully blood fed (shown by engorged abdomen) in the sleep-deprived group was compared with that of control (non-sleep deprived group) and expressed as a proportion of the total mosquitoes that stayed alive throughout the assay.

Reduction in host responsiveness

To determine whether prolonged sleep-like states reduced the response of mosquitoes, we performed basic host cue response studies on two species (Ae. aegypti and Cx. pipiens). Mosquitoes were observed through video and after 0, 30, 60, 120 and 240 min of inactivity, the experimenter entered the room and exhaled on the cage to provide a host cue. The number of mosquitoes that took flight within 30 s following exposure to experimenter breath was used as a proxy for host response. Each time point was conducted on 8–14 mosquitoes for each species.

Circadian timing and sleep-like period differ among multiple mosquito species

One important hallmark of sleep is that organisms (studied so far) experience reversible prolonged periods of immobility/inactivity during a particular phase of the circadian day (Hendricks et al., 2000; Prober et al., 2006; Raizen et al., 2008; Shaw et al., 2000; Yokogawa et al., 2007). To determine periods of putative sleep (lack of activity) in mosquitoes, we quantified the rest–activity rhythm of all three mosquito species using an infrared-based activity monitoring system during a 24 h circadian day. In Drosophila-based studies, sleep is usually defined as a period of inactivity lasting for at least 5 min and the occurrence of rest (putative sleep) is inversely related to the number of activity counts (beam breaks) recorded per a given time (Hendricks et al., 2000; Huber et al., 2004; Shaw et al., 2000). This short period of 5 min is not appropriate for mosquitoes because this threshold would overestimate sleep duration in mosquitoes since they are less active than Drosophila melanogaster in the absence of host cues. Rather, we quantified the sleep profile for mosquitoes using a period of inactivity lasting 120 min based on the time required for 50% of mosquitoes to enter a sleep-like posture (Fig. 1C), which also represents a threshold beyond which there was a significant reduction in the arousal of mosquitoes to host cues (Table 1).

Based on historical observations of field-based mosquito feeding behavior, we hypothesized that Ae. aegypti – a diurnal mosquito ‘day biter’ (Rund et al., 2020; Tuchinda et al., 1969) – would have increased activity during the photophase (day time) and that rest (putative sleep) would be well consolidated in the scotophase (night time). Laboratory measurements in Ae. aegypti showed that activity increased from mid-day till the onset of light off, but activity reduced significantly throughout the night after light off (Fig. 2A). Putative sleep for Ae. aegypti decreased from mid-day till the end of the photophase, but putative sleep was well consolidated in the scotophase (Fig. 2D). As expected, the sleep bouts reported in the circadian day were the exact inverse to the flight activity profile.

Open in a separate window

Fig. 2.

Timing and amount of sleep differ among multiple mosquito species. Basic activity rhythm of (A) Aedes aegypti, (B) Culex pipiens and (C) Anopheles stephensi over a 24 h period. The y-axis represents the mean beam crosses in an activity monitor made by all the mosquitoes. Sleep profile of (D) Ae. aegypti, (E) Cx. pipiens and (F) An. stephensi averaged into a single 24 h period. The y-axis shows the proportion of time spent sleeping (defined as inactive periods of 120 min), averaged for each mosquito within a 30 min time window. The x-axis for all the plots represents the Zeitgeber time (ZT0–ZT24). The solid lines and the shaded areas display means and their 95% bootstrap confidence interval, respectively. White and black horizontal bars represent the photophase and scotophase, respectively. Comparison of (G) total sleep and (H) daytime and night-time sleep among the three mosquito species. Error bars denote s.e.m. sleep amount. Different letters indicate significant differences between treatment groups (Kruskal–Wallis test with Dunn's multiple comparison post hoc, P<0.05). In all the analyses, n=60 for both Ae. aegypti and Cx. pipiens and n=34 for An. stephensi.

Comparative analysis conducted in Cx. pipiens – a crepuscular–dark active species (Veronesi et al., 2012) – showed that activity was consistently low during the day but increased in anticipation of light off (dusk) (Fig. 2B) and putative sleep was reduced significantly from dusk into the first half of the night (Fig. 2E). For the nocturnal mosquito ‘night biter’ (Rund et al., 2016), An. stephensi increased activity from early night into the mid-night, with activity reducing as day approached (Fig. 2C). Putative sleep occurred throughout the day for An. stephensi (Fig. 2F).

Sleep amount in minutes was quantified for our laboratory strains of mosquitoes, with comparisons made among the three species. Statistical analysis shows there was a significant difference in the mean total sleep (Kruskal–Wallis test: χ²=7.221, d.f.=2, P=0.027), where the difference existed only between Ae. aegypti and Cx. pipiens (Dunn's multiple comparison: P=0.027; Fig. 2G). As expected, length of daytime and night-time sleep differed among the species (Kruskal–Wallis test: daytime, χ²=36.831, d.f.=2, P<0.001; night-time, χ²=65.519, d.f.=2, P<0.001; Fig. 2H); however, there was no difference between Cx. pipiens and An. stephensi for either day or night. Together, these results reveal the marked differences in timing and amount of sleep-like periods in different mosquito species.

Sleep deprivation induces sleep rebound in Aedes aegypti and Anopheles stephensi depending on the phase of perturbation

Sleep deprivation in mosquitoes was assessed for subsequent sleep rebound when individuals are normally active. In Ae. aegypti, sleep deprivation by mechanical disturbance was conducted for 12 h during the night, 4 h during the night, and 12 h during the day. However, in An. stephensi, sleep deprivation was only done for 12 h during the day for comparative observations with the day-active Ae. aegypti.

Aedes aegypti mosquitoes subjected to vibration pulses (1 min of vibration followed by 5 min of rest) throughout the night recorded a significant sleep loss of about 558 min when we compared with sleep during the preceding night (Wilcoxon signed rank test: V=1122, P<0.001; Fig. S1E). This sleep loss promoted a significant rebound in the subsequent photophase, with a gain of approximately 76 min of sleep (paired t-test: t=3.463, P=0.001; Fig. 3A). A significant sleep loss of nearly 159 min occurred in Ae. aegypti mosquitoes that experienced vibration pulses (1 min of vibration followed by 1 min of rest) in the first 4 h of the night (Wilcoxon signed rank test: V=1672.500, P<0.001; Fig. S1F). Even this short amount of lost sleep early in the night was adequate to induce sleep rebound in the following day; a significant sleep gain of nearly 1 h was reported (paired t-test: t=3.846, P<0.001; Fig. 3B).

Open in a separate window

Fig. 3.

Sleep deprivation induces sleep rebound in both Aedes aegypti and Anopheles stephensi. Comparison of sleep amounts in the phase of interest before and after sleep deprivation in (A) 12 h night-time sleep deprivation (12NTD) experiment in Ae. aegypti (n=48), (B) 4 h night-time sleep deprivation (4NTD) experiment in Ae. aegypti (n=59), (C) 12 h daytime sleep deprivation (DTD) experiment in Ae. aegypti (n=64) and (D) 12 h daytime sleep deprivation (DTD) experiment in An. stephensi (n=36). Comparison of average bout durations in the phase of interest before and after sleep deprivation in (E) 12 h nighttime sleep deprivation (12NTD) experiment in Ae. aegypti (n=48), (F) 4 h night-time sleep deprivation (4NTD) experiment in Ae. aegypti (n=59), (G) 12 h daytime sleep deprivation (DTD) experiment in Ae. aegypti (n=48, individuals with zero values excluded) and (H) 12 h daytime sleep deprivation (DTD) experiment in An. stephensi (n=13, individuals with zero values excluded). Test of significant difference between groups was carried out using paired t-test or Wilcoxon signed rank test where applicable (n.s.=not significant, **P<0.01, ***P<0.001).

In the Ae. aegypti mosquitoes subjected to vibration pulses (1 min of vibration followed by 5 min of rest) during the photophase, the amount of sleep lost by comparing with sleep amount in the baseline day was approximately 436 min (Wilcoxon signed rank test: V=2013, P<0.001; Fig. S1G). However, this failed to yield a significant sleep gain in the subsequent night, indicating that sleep deprivation during a normally active period does not generate a rebound (paired t-test: t=0.378, P=0.707; Fig. 3C). This was not the case for An. stephensi mosquitoes, as daytime sleep deprivation in this species mirrored that of night-time sleep deprivation in Ae. aegypti, which was expected as this species is active at night. A significant sleep loss of about 594 min was reported (Wilcoxon signed rank test: V=666, P<0.001; Fig. S1H), which induced a significant sleep recovery in the subsequent scotophase (Wilcoxon signed rank test: sleep gain=196 min; V=73, P<0.001; Fig. 3D).

The influence of sleep deprivation in mosquitoes was also examined in relation to another sleep architecture, i.e. sleep bout duration. Results showed that sleep deprivation promoted a significantly increased sleep bout duration in the subsequent light phase for both the 12 h night-time (Wilcoxon signed rank test: V=994, P<0.001; Fig. 3E) and 4 h night-time deprivations in Ae. aegypti (Wilcoxon signed rank test: V=1364, P<0.001; Fig. 3F). As expected, daytime sleep deprivation in Ae. aegypti did not significantly impact sleep bout duration in the subsequent night (paired t-test: t=0.481, P=0.633; Fig. 3G and Fig. S1I). Although, daytime sleep deprivation in An. stephensi significantly promoted sleep gain in the subsequent night, sleep bout duration was not significantly affected (Wilcoxon signed rank test: V=64, P=0.216; Fig. 3H and Fig. S1J). From our results, mosquitoes deprived of sleep during the normal periods of low activity, experience sleep rebound in the subsequent phase, but there was no sleep recovery if sleep deprivation occurs during their normally active period.

We thank Diane Eilerts for fruitful discussions about the project, and Zhijian Jake Tu for providing females of the Indian strain of Anopheles stephensi, a representative of the type form. The following reagent was obtained through BEI Resources, NIAID, NIH: Aedes aegypti, Strain LVP-IB12, Eggs, MRA-735, contributed by David W. Severson.