Wolbachia Promotes Its Own Uptake by Host Cells

ABSTRACT

Wolbachia pipientis is an incredibly widespread bacterial symbiont of insects, present in an estimated 25 to 52% of species worldwide. Wolbachia is faithfully maternally transmitted both in a laboratory setting and in the wild. In an established infection, Wolbachia is primarily intracellular, residing within host-derived vacuoles that are associated with the endoplasmic reticulum. However, Wolbachia also frequently transfers between host species, requiring an extracellular stage to its life cycle. Indeed, Wolbachia has been moved between insect species for the precise goal of controlling populations. The use of Wolbachia in this application requires that we better understand how it initiates and establishes new infections. Here, we designed a novel method for live tracking Wolbachia cells during infection using a combination of stains and microscopy. We show that live Wolbachia cells are taken up by host cells at a much faster rate than dead Wolbachia cells, indicating that Wolbachia bacteria play a role in their own uptake and that Wolbachia colonization is not just a passive process. We also show that the host actin cytoskeleton must be intact for this to occur and that drugs that disrupt the actin cytoskeleton effectively abrogate Wolbachia uptake. The development of this live infection assay will assist in future efforts to characterize Wolbachia factors used during host infection.

INTRODUCTION

Wolbachia are alphaproteobacteria, part of the anciently intracellular Anaplasmataceae, and related to the important human pathogens Anaplasma, Rickettsia, and Ehrlichia (1). However, Wolbachia do not infect mammals but instead are well known for their reproductive manipulations of insect populations, inducing phenotypes such as male killing, feminization, or sperm-egg incompatibility (1). Wolbachia are incredibly successful at establishing infection; the symbiont is both vertically and horizontally transmitted and manipulates host reproduction to persist in populations (2–6). In the last decade, Wolbachia has been shown to provide a benefit to insects where the infection can inhibit RNA virus replication within the host, a phenomenon known as pathogen blocking (7). Because insects are vectors for disease and Wolbachia alter the ability of these vectors to harbor important human pathogens, Wolbachia strains are being used to control the spread of diseases like dengue (8, 9). For Wolbachia to establish itself in an insect population, it must invade host cells, persist during infection, and be transmitted to the next generation. Therefore, a prerequisite for the use of Wolbachia in vector control in insects is the establishment of infection.

Although Wolbachia cells cannot be cultivated ex vivo, they can survive for weeks in cell culture medium when removed from host cells (10). Additionally, Wolbachia cells can be found in host hemolymph, suggesting that during natural infection of whole organisms, they can be found outside host cells (5). However, Wolbachia’s entry into the host cell is still relatively uncharacterized. What little we do know about the infection process comes from elegant transwell assays where Wolbachia cells, after passively exiting host cells and then passing through a 10-mm barrier, were able to colonize uninfected Drosophila cells (11). The infection timeline calculated based on this approach was between 3 h and 6 days, a relatively long time course for most of the Anaplasmataceae (12). In that same work, the authors investigated the importance of clathrin-mediated endocytosis and showed that this host process alone cannot account for all Wolbachia cell entry. In this study, we therefore had two goals: to determine Wolbachia’s contribution to host uptake and to identify a more precise timeline for direct infection of host cells.

Using both infected and uninfected Drosophila melanogaster JW18 cells, we designed a protocol that allows live visualization of Wolbachia establishment. We first asked whether live Wolbachia cells are taken up by host cells more rapidly than Wolbachia cells that have been killed. We then established a timeline and quantified the relative numbers of live versus killed Wolbachia cells entering host cells. Our results suggest that Wolbachia cells indeed do contribute to their own uptake, with live Wolbachia colonizing host cells more quickly and at higher infection rates than their dead counterparts. We then treated uninfected JW18 cells with cytochalasin D, which inhibits the polymerization of filamentous actin, and challenged them with live Wolbachia. We show that pretreatment of host cells with the actin-inhibiting drug cytochalasin D inhibits the uptake of live Wolbachia dramatically, identifying the host actin cytoskeleton as important for the internalization process. We therefore conclude that Wolbachia are not only passively taken up by Drosophila but also actively facilitate the uptake.

RESULTS

We were able to successfully isolate live Wolbachia cells from JW18 cells, and using a freeze method, we were able to kill the population of bacteria for our controls. The efficacy of this method was visualized using live/dead staining (Fig. 1). No propidium iodide staining was observed for live, freshly isolated Wolbachia cells, but all cells subjected to freeze treatment internalized the dye, suggesting the cell membranes had become permeable to propidium iodide staining. Each of the samples was used in live microscopy during infection of Drosophila melanogaster cells. Briefly, Wolbachia cells (either live or dead) were stained using SYTOX dye and overlaid onto Wolbachia-free JW18 cells. Although its name implies otherwise, SYTOX is a live-cell dye, chosen because it does not perturb nucleoid morphology or arrest cell growth (13, 14). The process was visualized using ProLong live reagent and fluorescent light microscopy, and pictures were taken of live cells every 2 h for 4 h. Live Wolbachia cells could readily be observed entering host cells by 2 h postinfection, whereas we rarely observed internalized killed Wolbachia cells (Fig. 2). This result confirmed that we could visualize the infection process and that we could quantify Wolbachia’s contribution using a killed control.

FIG 1.

Wolbachia cells are alive after purification from host cells and can be killed with freeze treatment. After isolating Wolbachia cells from Drosophila JW18 cells, we stained them using a combination of SYTO 9 and propidium iodide (PI) (see Materials and Methods). Dead Wolbachia cells are indicated by successful staining using PI. Scale bar = 10 μm.

FIG 2.

Live Wolbachia cells are readily internalized by Drosophila melanogaster. Wolbachia cells (red) stained with SYTOX are visualized infecting JW18 cells (which express Jupiter-green fluorescent protein [GFP], which localizes to microtubules). Representative images during a 4-h time course are shown. Cell perimeters are delineated by Nikon software using GFP signal intensity across z stacks, and Wolbachia cell location within the three-dimensional (3-D) reconstructed cell is identified using the SYTOX stain signal. Wolbachia cells were more readily identified within host cells when live than when killed. Scale bar = 10 μm.

We next proceeded to perform 3 paired experiments, using live and killed counterparts, and counted a total of 454 cells during three 4-h infection time courses (Table 1). Experiments were paired such that the same Wolbachia cells isolated in one experiment were freeze killed and used as a control for that same experiment. No effect of the experiment on the number and volume of Wolbachia cells found within host cells over time was observed (analysis of variance [ANOVA]; df = 2, F = 1.368, P = 0.256, and df = 2, F = 0.311, P = 0.733). However, out of an abundance of caution and to ensure we could directly compare multiplicities of infection (MOIs), we analyzed the data in aggregate as well as in pairwise fashion, comparing each heat-killed control to each live sample. The two measures we took were of Wolbachia counts within cells (individual bacterium-shaped objects) and Wolbachia volumes (the calculated volumes, based on pixels, for each of those objects). Both measures were significantly correlated with each other (Pearson correlation R = 0.92, P < 2.2e−16) (Fig. 3A). We therefore present data from the Wolbachia counts in the main text (see Fig. S1 in the supplemental material for volume-based data).

TABLE 1.

Count of host cells and Wolbachia volume and mean number for each condition over time

Wolbachia cell status	Time (h)	No. of host cells	Mean value (SD) for Wolbachia cells
			Vol (μm3)a	No.
Dead	0	54	0.03 (0.11)	0.5 (1.6)
	2	77	0.24 (1.20)	2.97 (11.5)
	4	78	0.10 (0.24)	2.21 (3.64)
Live	0	81	0.25 (1.42)	3.42 (16.2)
	2	96	0.80 (2.39)	13.5 (38.6)
	4	68	0.52 (0.98)	8.31 (14.8)

^aThe volume of Wolbachia cells shown by red/Texas red/SYTOX staining within JW18 host cells was calculated based on pixels using the NiE Elements GA3 program.

FIG 3.

Live Wolbachia cells are internalized more quickly and at higher numbers. Numbers calculated based on live imaging of SYTOX-stained Wolbachia cells over a 4-h time course. (A) Calculated Wolbachia numbers and volumes per host cell were highly correlated. (B) Aggregate internalized Wolbachia cell counts over the time course were significantly higher when Wolbachia were live than when they were dead (χ² = 5.6407, df = 1, P = 0.0175). (C) Across three independent experiments (A, B, and C), we consistently observed higher Wolbachia cell counts per cell when Wolbachia cells were live than with heat-killed (dead) matched samples (zero-inflation GLM coefficient estimate −0.4736, Z = −2.464, P = 0.137). The numbers of cells per experiment were as follows: zero hours, dead, 15, 35, 10, and live, 24, 23, 34; 2 h, dead, 27, 35, 15, and live, 29, 32, 35; 4 h, dead, 20, 43, 15, and live, 20, 28, 20.

Live Wolbachia cells are internalized more quickly and at higher numbers than their dead counterparts.

Over the course of the infection, we saw increases in the total numbers and volumes of Wolbachia cells internalized per cell (with median peaks at around 13 and 3 in the live versus dead Wolbachia overlays, respectively) (Table 1). In aggregate, across all the sampled cells, the number of total Wolbachia cells internalized over the time course of infection was greatest for the treatment with live Wolbachia cells (χ² = 5.6407, df = 1, P = 0.0175) (Fig. 3B). Indeed, a logistic regression significantly predicts infection status based on both time (estimate = 0.507, standard error of the regression [stderr] = 0.070, Z = 7.256, P = 3.89e−13) and whether the assay involved live Wolbachia cells (estimate = 0.723, stderr = 0.211, Z = 3.429, P = 0.0006). For two of the three experiments, the largest number and volume of internalized Wolbachia cells was observed in the live overlay after 2 h (23.1 and 12.4 Wolbachia cells, on average, observed within a host cell) (Fig. 3C). The numbers stabilized at an average of 8.3 Wolbachia cells per host cell at 4 h (Table 1).

Because our data are necessarily zero inflated (due to the large number of uninfected cells in the dead overlay and at early time points), we used a zero-inflated generalized linear model (GLM) to compare the numbers of Wolbachia cells internalized across time for each condition. We observed a significant effect of both time (estimate = −0.079, stderr = 0.014, Z = −5.707, P = 1.15e−08) and whether the bacteria were live or dead (estimate = −0.474, stderr = 0.192, Z = −2.464, P = 0.0137) on the number of Wolbachia cells internalized, corroborating our observations.

Cytochalasin D treatment reduces Wolbachia uptake.

Previously, Wolbachia has been shown to use the host cytoskeleton for localization within oocytes and for maternal transmission (2, 4). Therefore, we sought to confirm the fraction of uptake of Wolbachia cells we observed that was based on the host cell cytoskeleton. We treated uninfected JW18-tet cells with cytochalasin D prior to infection with Wolbachia and, as expected, observed a dramatic decrease in internalization (Fig. 4A; Fig. S3). The median counts observed were generally 1 or zero for each time point (mean numbers of Wolbachia cells within host cells per time, μ = 2.33 at 0 h, μ = 0.167 at 2 h, and μ = 1.62 at 4 h). Indeed, compared to overlays using dead Wolbachia cells, the cytochalasin D-treated samples were statistically indistinguishable (estimate = −2.055, stderr = 1,712.3, Z = −0.001, P = 0.950) (Fig. 4B).

FIG 4.

Cytochalasin D treatment reduces the number of internalized Wolbachia cells. The numbers were calculated based on live imaging of SYTOX-stained Wolbachia cells over a 4-h time course after treatment of host cells with 1.5 μg/mL of cytochalasin D, a toxin that inhibits actin polymerization. (A) Wolbachia counts were rarely higher than zero across all time points for host cells treated with cytochalasin D and did not vary based on time point. (B) The numbers of internalized Wolbachia cells in cytochalasin D-treated host cells compared to the numbers using dead Wolbachia cells in three independent experiments (zero-inflation GLM coefficient estimate −0.4736, z = −2.464, P = 0.137). The numbers of cells counted for the cytochalasin D-treated cells across time were 30 (0 h), 36 (2 h), and 26 (4 h).

DISCUSSION

All intracellular bacteria must manipulate the host to enter, persist, replicate, and be transmitted. For those that move between hosts, they must be able to survive extracellularly and reinfect host cells. Orientia tsutsugamushi, which causes scrub typhus, uses clathrin-mediated endocytosis to enter a host cell, escapes the endosome and utilizes dynein and microtubules to transport itself to the nucleus for replication, and then exits the host cell enveloped in host membrane to continue its life cycle (15). For the endosymbiont Wolbachia, little is known about its ex vivo lifestyle, although more than a decade ago, Wolbachia pipientis strain wAlbB was observed to survive outside host cells for 1 month and to be infectious at least 7 days after extraction from the host (10). Additionally, Wolbachia cells can be purified from Drosophila host hemolymph and used to inject fruit fly abdomens, after which they establish the canonical infection of the ovaries, suggesting that the microbe has a free-living form conducive to horizontal transfer (16). Understanding how Wolbachia moves between hosts and establishes infection is a primary research goal for the use of Wolbachia in vector control, but additionally, as Wolbachia bacteria clearly move between host species frequently (17, 18), understanding how they may accomplish this is important to the natural history of the microbe.

Here, we used cultured Drosophila cells to identify a timeline of Wolbachia entry into host cells and to characterize Wolbachia’s role in its own uptake. We show that a population of live Wolbachia cells can directly infect established cultured JW18 host cells in as little as 2 h (Fig. 2 and 3). We show that this population of live Wolbachia cells enters host cells much more rapidly and in greater numbers than do killed controls (Fig. 2 and 3 and Table 1). This strongly implies that Wolbachia bacteria play a role in their own uptake; if this was purely a passive process, the live and dead populations would enter host cells at comparable rates and in similar numbers. This work confirms a previous study that has shown that heat-killed Wolbachia cells directly placed into uninfected cell culture are not detectable by either fluorescence in-situ hybridization (FISH) or PCR after incubation for 5 days, while live Wolbachia cells are easily detectable by both methods after this period of time (10). Our results are the first to show a direct comparison between the rates of initial colonization of live and dead Wolbachia cells by direct infection in cell culture in less than 24 h.

Work from the Sullivan laboratory clearly established that Wolbachia cells use dynamin- and clathrin-mediated endocytosis to enter host cells, echoing results from Orientia, but inhibition of these pathways did not prevent entry completely (11). In that same work, the Sullivan laboratory established that cell-to-cell contact is not needed for Wolbachia’s intercellular transmission. If dynamin- and clathrin-dependent endocytosis was inhibited using drugs, fewer Wolbachia cells were internalized. These results indicate both that the host plays an important role in taking up Wolbachia (perhaps why we observe dead Wolbachia cells being internalized) and that Wolbachia may utilize redundant routes of entry. Major differences between that work and ours include the fact that White et al. did not isolate Wolbachia from host cells and perform a timed infection, instead allowing Wolbachia to exocytose from JW18 host cells and float over to neighboring, uninfected S2 cells (11). Additionally, White et al. looked at transfer between two different cell backgrounds, while we used uninfected counterparts of the same background (JW18). It is likely that for these reasons, the timing we observed for internalization is quite different.

Many microbes alter the host actin cytoskeleton to either inhibit or promote uptake, with mechanisms ranging from the use of bacterial effectors that induce membrane ruffles to the nucleation of host actin filaments to promote intercellular spread (19, 20). Previous work has established that Wolbachia interacts with the host actin cytoskeleton, both requiring it for efficient host colonization and transmission and using a secreted effector that bundles host actin to facilitate transmission (4, 21). To build on and expand the known Wolbachia routes of entry, we endeavored to show that the host cell actin cytoskeleton is needed for this initial entry process to occur. We treated uninfected JW18 cells with the drug cytochalasin D and then exposed these treated cells to live Wolbachia cells as before. Importantly, live Wolbachia cells were unable to enter the treated cells (Fig. 4), and indeed, we observed numbers even more reduced than in the dead Wolbachia controls. This result suggests that the host actin cytoskeleton must be functional for the initial colonization step to occur (Fig. S3).

Our experiment used freezing to kill Wolbachia cells for comparison to freshly isolated bacteria from JW18 cells. Importantly, the freeze-thaw process will alter the bacterium in many ways, including perturbing the membrane or completely lysing the cells. It is possible that some membrane-bound receptor on the surface of Wolbachia cells was abrogated by the freeze-thaw process, limiting the ability of the previously frozen microbes to be internalized by Drosophila cells. An “invasin” of this sort has not been identified for Wolbachia, but many outer membrane proteins of the bacterium are highly expressed, including the Wolbachia surface protein (Wsp). This protein homolog (WspB) was recently implicated in Wolbachia’s polar localization in Drosophila ovaries (22). Additionally, when Wolbachia finds itself outside host cells, it may alter its protein expression, likely to allow it to survive extracellularly. Understanding the changes in Wolbachia gene expression during ex vivo growth and during host infection will help identify Wolbachia loci that are important for host internalization. Many bacterial pathogens, for example, Coxiella, use acidification of the vacuole to detect host uptake (23), and it may be that Wolbachia similarly responds to the acidification of its host-derived vacuole after uptake, deploying a repertoire of effectors to modulate host cell biology.

In conclusion, this work shows that Wolbachia promotes its own uptake by host cells, which helps elucidate how this bacterial symbiont has become such a globally successful infection. We also show that the host actin cytoskeleton is necessary for Wolbachia uptake, which further expands the knowledge necessary to use the microbe in vector control efforts.

MATERIALS AND METHODS

Insect cell culture.

Drosophila melanogaster JW18 cells were treated with tetracycline to remove Wolbachia infection. The infection status was confirmed by PCR after four passages in the presence of antibiotic. Cells with and without Wolbachia bacteria were maintained in the dark, in T25 unvented cell culture flasks at room temperature. Schneider’s insect medium was used and supplemented with 10% heat-treated fetal bovine serum and 1% penicillin/streptomycin solution.

Wolbachia isolation and overlay for rate-of-uptake screening.

To visualize the rate of uptake of Wolbachia cells, ~2,000 uninfected JW18 host cells were seeded into each well of a Cellvis 96-well glass bottom plate with fresh Schneider’s medium and incubated in the dark at room temperature overnight to allow them to settle. The following day, confluent Wolbachia-positive JW18 cells grown in a T25 flask were first counted and then transferred into a 15-mL conical tube containing 500 μL sterile glass beads and SYTOX orange dye (250 nM) (ThermoFisher). To release Wolbachia cells from JW18 host cells, the conical tubes containing glass beads were then vortexed for 4 pulses of 5 s each. The lysates were divided into 500-μL aliquots, and each aliquot was placed onto an Ultrafree-MC-SV Durapore polyvinylidene difluoride (PVDF) 5.0-μm centrifugal filter column (Millipore). The columns were spun at 10,000 × g at 10°C for 10 min to purify Wolbachia cells from host cells (10). The resulting pellets were then transferred into a new 1.5-mL centrifuge tube containing fresh Schneider’s medium and resuspended. Extractions from the same T25 flask were divided such that half would be used immediately (live) and the other half frozen at −20°C for at least 3 days (killed) to serve as a control for the infection. To prevent bleaching, 50 μL of ProLong live reagent (ThermoFisher) solution was added to each tube (250-μL volume total) prior to imaging and the tubes incubated away from light for 15 min according to the manufacturer’s instructions. JW18 cells are not all infected at the same density (10, 24), and because live Wolbachia cells cannot be quantified by quantitative PCR (qPCR), we controlled for the multiplicity of infection by counting the number of infected JW18 host cells that we lysed before isolating Wolbachia cells and by using the matched extracted Wolbachia cell sample as a dead control for each experiment. The Wolbachia cells were then added to each seeded well at an MOI of 20:1 host cell equivalents. That is, for each seeded JW18 cell, we overlaid the Wolbachia contents from 20 infected JW18 cells. The plate was then gently spun down for 5 min at 500 rpm to ensure that Wolbachia and uninfected JW18 host cells were proximal. For cytochalasin D treatments, the same infection protocol was followed, but preceding overlay, host cells were treated with 1.5 μg/mL cytochalasin D in Schneider’s medium for 30 min. Cytochalasin D at this concentration has been used previously by other researchers using Drosophila as a model for infections by Chlamydia, Acinetobacter, and Serratia, as well as studying basic endocytic processes (25–28).

Live/dead staining of isolated Wolbachia cells.

Wolbachia cells isolated from infected JW18 cells were stained with a 1:1 mixture of SYTO 9 (ThermoFisher) and propidium iodide (ThermoFisher) either immediately following isolation (live) or after storage at −20°C for at least 3 days prior to imaging (dead). Live Wolbachia cells were used straight away, without any long incubations, and kept away from light prior to imaging using the same magnification, intensities, and exposure times.

Microscopy and analysis of images.

All images were obtained using a Nikon Ti2 Eclipse inverted microscope at ×100 magnification with a Nikon DS-Qi2 camera attachment. As noted above, cells were seeded at a density no greater than 2,000 cells per well; this allowed us to more clearly delineate individual cells and avoid cell clumps during the infection and made the image processing smoother. Intensities and exposure times were as follows: green/fluorescein isothiocyanate (FITC) for 200 ms with SOLA pad at 4% and red/Texas red for 400 ms with SOLA pad at 8%. All z stacks were taken with a step size of 0.5 μm with between 100 and 300 contiguous planes imaged per field, with identical settings applied across all fields, regardless of treatment and condition. All cells in a field were analyzed, regardless of Wolbachia infection status. Only cells where we could not clearly delineate cellular boundaries were excluded. All image processing was done within the Nikon NiE Elements software package.

To process the images and count internalized Wolbachia cells, Nikon NiE Elements software was used to first deconvolute all z stacks (using the automatic 3D deconvolution setting). This deconvolved set of stacks was then used as the input to the NiE Elements GA3 program. Each individual host cell boundary was identified using the look-up tables for green/FITC to delineate the edge of the cell. GA3 was then used to collect internalized Wolbachia data for each image, using the threshold parameters specified in Table S1. The GA3 program then used the entire stack for that cell to identify what was “in” versus “out”—any red/Texas red/SYTOX staining seen within the cell was counted as interior Wolbachia cells and their volume measured by GA3. For each live-versus-dead condition, three experiments were performed (labeled A, B, and C).

Counts and volume data from Nikon NiE Elements were tested for normalcy and overdispersion in R, and a zero-inflated GLM with a Poisson distribution (zeroinfl in the PSCL library) was used to test for significant differences in the numbers and volumes of total Wolbachia cells inside host cells over time under each condition. A logistic regression (glm with family = binomial) was used to predict infection status of cells based on Wolbachia cell volume and number.