Live In Vivo Imaging of Plasmodium Invasion of the Mosquito Midgut

Sample preparation for live in vivo imaging.Two major obstacles had to be overcome to image Plasmodium ookinete midgut invasion in real time. The first obstacle entailed developing a new strategy to mount blood-fed infected mosquito midguts while maintaining tissue viability for several hours. Our initial attempts to mount dissected midguts ex vivo were unsuccessful due to the loss of tissue integrity. Dissected midguts could not be imaged for more than 20 to 30 min, in agreement with previous reports (7). Levels of ookinete invasion peak at 24-h post-blood meal, a time when the mosquito midgut is under physical stress because of mechanical stretching to accommodate the blood meal and when it can also be damaged by the extensive secretion of digestive proteases. To solve this problem, whole-blood-fed mosquitoes, infected by feeding on mice infected with P. berghei-mCherry parasites (Fig. 1A), were mounted after decapitation and removal of legs and wings to immobilize them (Fig. 1B, top panel). Previous studies have shown that decapitated mosquitoes can survive for long periods of time, even for several days (8). Imaging whole mosquitoes allowed us to maintain an adequate oxygen supply to the gut through the tracheal respiratory system. To minimize any possible injury, mosquitoes were gently knocked down using CO₂ (instead of using air suction followed by ice cooling) and were placed sideways to enable imaging of the midgut through the thinnest and lightest areas of the abdominal cuticle. Mosquitoes were immobilized using malleable adhesive craft putty as a spacer, to confine them to the narrow space between the glass slide and a coverslip (Fig. 1C). Using this mounting setup, we were able to image the blood-fed midgut through the cuticle for more than 4 h using fluorescent confocal microscopy (Fig. 1D and E).

• Open in new tab
• Download powerpoint

FIG 1

Sample preparation for live in vivo imaging of mosquitoes. (A) Experimental design for introduction of fluorescent markers inside mosquitoes. (B) (Top panel) Head, wings, and legs of mosquitoes are removed prior to mounting. (Bottom panel) Side view schematic of a mosquito mounted and imaged on an inverted microscope. (C) Six engorged mosquitoes were aligned and gently pressed between a cover slip and glass slide using adhesive putty. (D) XY image (top) and XZ image (bottom) of the mosquito midgut (green) infected with P. berghei-mCherry (red). Scale bar = 50 μm. (E) Magnified image from box in XY image in panel D showing cells (green) invaded by Plasmodium (red). Bottom panel, XZ image. Scale bar = 10 μm.

The second obstacle was the difficulty of identifying appropriate fluorescent markers to label midgut epithelial cells in P. berghei-infected mosquitoes. These markers must retain strong fluorescence for more than 24 h (time of peak ookinete invasion after blood feeding) at low concentrations, to avoid toxicity. Blood-fed mosquitoes refrain from subsequent feeding until digestion is complete; therefore, dyes were fed in saline solution using a membrane feeder before mosquitoes were fed on P. berghei-infected mice. An alternative method for labeling mosquitoes by systemic injection was also tested for nine potentially suitable fluorescent live-cell tracer dyes (see Table 1). Membrane feeding of CFDA-SE (carboxyfluorescein diacetate succinimidyl ester) in a saline solution worked very well to label all midgut epithelial cells. This is a cell-permeative dye with low toxicity that is commonly used for long-term cell tracing (9). Epithelial caspase-3 activity was detected by feeding NucView to mosquitoes, a substrate that, when cleaved by caspase-3, releases a high-affinity DNA binding dye which produces nuclear fluorescent staining (10). Epithelial F-actin was visualized in vivo by feeding mosquitoes SiR-actin, and this made it possible to follow the reorganization of epithelial actin after ookinete invasion. Systemic injection of the lipophilic membrane dye DiD labeled hemocytes very efficiently, as previously reported (5). These dyes were often used in combination by feeding them together. Labeling multiple components in the same mosquito made it possible to obtain a more comprehensive picture of mosquito-parasite dynamics in vivo.

Ookinetes have different fates after they develop in the midgut lumen.Knowing that P. berghei ookinete midgut invasion peaks around 24 h postfeeding (PF), we imaged infected midguts from 22 to 26 h PF to maximize the probability of capturing ookinete dynamics before, during, and after midgut invasion. Z-stack sections were obtained at 1-μm intervals, every 30 s, to capture quick ookinete movements and obtain cross sections that encompassed the full thickness of the mosquito epithelium, from the midgut lumen up to the basal side of epithelial cells. Images were imported into Imaris, a visualization software program, to perform tracking analysis of the ookinetes (see Fig. S1A in the supplemental material).

We tracked 143 individual ookinetes from four mounted midguts and found that the parasites had three possible fates. Although 50% (71/143) of them (see Movie S1 in the supplemental material) invaded the midgut, others (30%; 43/143) came in contact with the midgut surface (Movie S2) but did not invade the cell and 20% (29/143) were motile but remained within the gut lumen (Movie S3) and never made contact with the midgut surface (Fig. 2A to E). Any spherical parasite that lacked the characteristic “banana shape” of ookinetes was presumed to be a zygote and was not included in this analysis. A successful cell invasion event does not guarantee that the parasite will reach the basal lamina and develop into an oocyst. Of the 71 ookinetes that invaded the midgut, 76% (54/71) had already reached the basal lamina and remained there for the duration of the imaging (4 h), 11% (8/71) egressed back into the lumen of the gut and we were able to track 9 ookinetes (13%) as they actively traversed the epithelial cells. We also estimated the time required for an ookinete to traverse the midgut. Because we scanned the tissues every 30 s, we were able to determine that it took these 9 parasites between 60 and 120 s to traverse epithelial cells, with a median of 90 s and an average of 83 s (see Table S1A in the supplemental material).

• Open in new tab
• Download powerpoint

FIG 2

Tracking analysis of ookinetes during live in vivo mosquito imaging. (A to D) Tracked ookinetes (A) were categorized into three possible fates: (B) remaining in lumen (RL), (C) midgut contact without invasion (MC), and (D) midgut invasion (MI). Scale bar = 5 μm. (E) Percentages of ookinetes with outcomes corresponding to each of the fates. (F) Total distance traveled (left) and speed (right) for RL, MC, and MI ookinetes. *, P < 0.05; **, P < 0.01; ****, P < 0.0001.

We then compared the motility of the 54 ookinetes that reached the basal lamina with that of parasites that remained in the gut lumen (n = 29) or that only came in contact with the microvillar surface but did not invade the cell (n = 43). A circular spinning motion was often observed in ookinetes gliding on the gut surface (Movie S1), and those parasites displaced significantly longer distances of traversal (ANOVA, P < 0.0001) (Fig. 2F, left) and moved faster (ANOVA, P < 0.0001) (Fig. 2F, right) than the ookinetes after they invaded the midgut or than those that moved within the blood bolus but did not contact the gut surface.

Caspase activation in invaded midgut cells.We also investigated the kinetics of midgut epithelial cell apoptosis in response to ookinete invasion by tracking caspase activity in cells invaded by ookinetes. NucView was fed to mosquitoes in a saline solution prior to Plasmodium infection. Caspase signal was not detected in midguts of control mosquitoes fed on a healthy mouse, indicating that apoptosis of midgut cells is rare under these conditions, while it was readily observed following ookinete invasion (Fig. S1B).

Caspase activity was detected in a total of 16 single cells or cell clusters from four P. berghei-infected midguts that were imaged for 4 h. Interestingly, the NucView signal was cytoplasmic and not nuclear, indicative of cell damage and DNA release into the cell cytoplasm of invaded cells. A relative level of signal intensity was established for each midgut, where the strongest signal was 100% and the weakest 0%. A minimum threshold was applied, according to which caspase activity with signal intensity that was at least 20% of the highest in the sample was considered to represent a positive result (Fig. 3A and B; see also Table S1B) (Movies S4 and S5). Most of the 16 cells (13/16 = 81%) were already positive for caspase activity when the imaging started (Fig. 3B; see also Table S1B), and 8 cells were still positive when the imaging ended after 240 min (Fig. 3A; see also Table S1B). In 8 cells, caspase activity was detected for close to 4 h (225 to 240 min) (Fig. 3C; see also Table S1B).

• Open in new tab
• Download powerpoint

FIG 3

Caspase activity is dynamic in infected midguts. (A) Snapshots of a cell increasing in caspase activity (green) (midgut in red, ookinetes in blue). Time stamps in hours and minutes from the moment the signal could first be detect until it reached maximum signal. Scale bar = 20 μm. XY view on top panels and XZ view on bottom. (B) Snapshots of cells decreasing in caspase activity (green). Timestamps in panels A and B represent the amount of time (hours:minutes) that elapsed from the moment that the cell reached maximum signal until it was no longer detectable. Scale bar = 20 μm. (C) Distribution of the durations of caspase signal in minutes. Several cells were positive for caspase for most of the 4-h imaging session. The vertical red line indicates the median.

Actin reorganization in response to ookinete invasion.Previous confocal studies with fixed tissues showed that the actin cytoskeleton undergoes dramatic rearrangements as cells damaged by Plasmodium invasion are extruded from the midgut epithelial layer, by forming an actin ring at the base of damaged cells that is gradually constricted (1). Because traditional actin markers, such as phalloidin, are highly toxic (11) and impermeable to live cells, we decided to use the F-actin-specific SiR-actin fluorophore (12).

Midgut cells invaded by Plasmodium ookinetes exhibited strong F-actin staining. These damaged cells could be seen protruding into the lumen of the midgut as they were extruded from the epithelial sheet (Fig. 4A; see also Movie S6). Five midguts were imaged for 4 h starting at 24 h PF. A total of 34 single cells or cell clusters with F-actin signal were observed, and a 20% minimal signal threshold was also applied. The median duration of positive F-actin signal was 217 min for 26 cells that were imaged for a total of 240 min (Fig. 4B; see also Table S1C). We also established how long it took for a cell to achieve maximum F-actin levels and for it to decrease to background levels. We captured the appearance of F-actin in 12 invasion events, and F-actin levels increased over a median period of 135 min (Fig. 4C; see also Table S1D). Once maximum levels were reached, it took about 20 min longer for F-actin to decrease to background levels, with a median of 155 min (n = 17) (Fig. 4D; see also Table S1E). We also found that some ookinetes that remained intracellular when F-actin levels increased were still able to egress from the cell (Movie S7A), while others are completely surrounded by F-actin and trapped inside the cell. Parasites that failed to egress were eliminated when the cell budded from the midgut epithelium (Movie S7B).

• Open in new tab
• Download powerpoint

FIG 4

Actin undergoes rearrangement in invaded midgut cells. (A) Snapshots of two cell clusters invaded by ookinetes (red) and undergoing actin rearrangement (cyan) as seen in XY (top) and XZ (bottom) views. The cell cluster on the left had decreasing F-actin levels, while the cells on the right had increasing F-actin levels (midgut in green). Timestamps represent passage of time in hours after imaging start. Scale bar = 30 μm. (B) Duration of F-actin signal in single cells or cell clusters with sustained activity. (C) Time it took for F-actin to reach maximum levels. (D) Time it took for F-actin to decrease back to basal levels. The horizontal black lines indicate the median.

Hemocytes patrol the basal surface of the midgut.We previously showed that the mosquito midgut releases prostaglandin E2 (PGE2) when epithelial cells come in contact with bacteria from the gut microbiota or with the immune elicitors that they release (13). Ookinete invasion triggers PGE2 synthesis, and this is a strong hemocyte chemoattractant (13). We investigated hemocyte dynamics following ookinete invasion in live mosquitoes. Hemocytes were fluorescently labeled by systemic injection of DiD, a lipophilic dye that specifically labels this cell population (5). The next day, mosquitoes were infected by feeding on a P. berghei-mCherry-infected mouse. We found that hemocytes were actively patrolling the basal surface of infected midguts (Movie S8A) and that a few hemocytes appeared to be tracking specific ookinetes (Movie S8B), but the specificity and functional relevance of these transient interactions remain to be determined.