We previously found that bacterial transfer from infected to uninfected BMDMs requires cell-cell contact (Steele et al., 2016). To resolve how bacteria transfer between cells, we performed transmission electron microscopy (TEM) of bone marrow derived macrophages (BMDMs) during bacterial transfer. For this assay, we mixed BMDMs that had been infected for 20 hr with Francisella tularensis and uninfected BMDMs. The donor and recipient cell were identified based on the cell-cell interaction in the image.

By TEM, the recipient cell appeared to engulf a small protrusion of the donor cell (Figure 1A and B). Notably, the donor cell fragment was contiguous with the cytosol of the host in the initial slices but was surrounded by protrusions from the recipient in sequential slices. These data indicate that BMDMs phagocytose small portions of their neighbours.

Figure 1 with 1 supplement see all

Download asset Open asset

The material that the macrophage acquired appears to include a F. tularensis bacterium based on shape and electron density. F. tularensis is typically identified in TEM images by of the characteristic electron translucent capsule surrounding the bacteria, which this bacterium lacks (Steele et al., 2013) (Example in Figure 5). The fragmentation of the bacterium and lack of capsule suggests that this particular bacterium may be getting degraded during the transfer process or a killed bacterium is being transferred between cells. Cell-cell transfer is a host-mediated process. So killed bacteria, and potentially even bacterial fragments, are fully capable of transferring between macrophages. It is important to note that in the case of F. tularensis, the majority of bacteria are viable following cell-cell transfer (Steele et al., 2016).

The TEM results suggest that cytosolic material transferred with the bacteria (Figure 1B). To test this observation, we labelled infected donor cells with Cell Trace Red, a dye that labels intracellular proteins by binding to free amines (Figure 1—figure supplement 1). We then inverted a coverslip seeded with unstained, uninfected BMDMs onto the dyed, infected cell population. We let the cells incubate like this for 30 min so that cell-cell transfer could occur between the two populations. We then removed the coverslip, which is almost exclusively ‘recipient’ cells (Figure 1F) (0.13 ± 0.23% of cells are infected donors that migrated to the coverslip, three independent experiments, 500 cells per experiment analyzed, mean ± SD).

We stained recipient cells for LAMP-1 and assessed if Cell Trace Red and bacteria were within the same vacuole following bacterial transfer. We found that most Francisella containing vacuoles (FCVs) also contained Cell Trace Red labelled protein from the donor cell cytosol (Figure 1G). From these results, we conclude that both host cytosolic proteins and bacteria are acquired within the same vacuole following bacterial transfer.

Our results indicate that BMDMs phagocytose portions of live cells but does not reveal what happens to the acquired material following transfer. Phagocytosis of extracellular F. tularensis leads to co-localization of bacteria with the early endosomal marker EEA-1. The F. tularensis containing phagosome matures, which results in co-localization with the late endosomal marker LAMP-1 (Craven et al., 2008). The bacteria then rupture and escape the phagosome, entering the cytosol where they replicate. We were interested in whether FCVs follow a similar maturation process after cell-cell transfer.

Using the assay described in Figure 1F with modified co-incubation times, we found that F. tularensis bacteria were typically located in EEA-1⁺ vacuoles at early time points post-transfer (Figure 2A and C). These FCVs matured into LAMP-1⁺ vacuoles over time (Figure 2B and D). Interestingly, the kinetics of LAMP-1 maturation and escape are virtually identical between cell-cell transfer and phagocytosis of extracellular bacteria (Figure 2D). There was a slight delay in EEA-1 maturation following bacterial transfer compared to extracellular bacteria (Figure 2C), but this apparent delay was likely due to much higher variability in the timing of infections through cell-cell transfer, rather than delayed maturation. These data suggest that Francisella interactions with the host are similar regardless of entry route.

Figure 2

Download asset Open asset

Following uptake of extracellular bacteria, F. tularensis requires a type VI secretion system (T6SS) to escape the phagosome (Clemens et al., 2018). Due to the similarities in the vacuole maturation kinetics between extracellular uptake and bacterial transfer, we hypothesized that the T6SS was also required for escape from the recipient cell endosomes after bacterial transfer. To test this hypothesis, we needed a strain that was fully functional in the initially infected cells, but was unable to escape the endosome in the recipient BMDM. To accomplish this goal, we put one of the T6SS structural protein genes (FTL_0119) under anhydrous tetracycline (ATc) induction in a FTL_0119 deletion background. In this strain, the T6SS is only able to form in the presence of ATc.

We infected BMDMs with the T6SS inducible strain after growing the bacteria overnight in broth containing ATc. The inoculation and subsequent infection were in media lacking ATc. Under these conditions, the T6SS strain escaped the initial phagosome and replicated in the cytosol at nearly the same rate as wild-type (Figure 3A, Figure 3—figure supplement 1). These data indicate that the T6SS is largely dispensable for F. tularensis intracellular replication so long as the mutant escapes the initial phagosome. This is crucial because different amounts of replication likely impact the amount of cell-cell transfer.

Figure 3 with 1 supplement see all

Download asset Open asset

Normally, Francisella escapes the FCV over time (Figure 3B). But following cell-cell transfer with the T6SS inducible strain, the FCV remained intact for at least 6 hr (Figure 3B). The phagosomal escape defect is similar between the post-transfer FCV in the T6SS inducible strain and impaired escape in the initial phagosome following inoculation with the uninduced T6SS or a marker-less, in-frame dotU deletion strain. These data indicate that Francisella uses the same machinery to escape both vacuoles regardless of the route of entry.

In addition to showing similarities between phagosomes and post-transfer FCVs, the T6SS inducible strain enables us to accumulate FCVs in recipient BMDMs over time (Figure 3B). This makes it more likely that we can detect bacteria in FCVs post transfer. Additionally, the bacteria serve as a marker to identify vacuoles post-transfer. We will use this strain in later experiments to test FCV formation during cell-cell transfer by TEM and during mouse infections.

F. tularensis is thought to primarily infect macrophages when the cells phagocytose extracellular bacteria. Our previous results suggest that F. tularensis exploits cell-cell transfer to infect cells and that this process may be common in vivo (Steele et al., 2016). It is unclear if macrophages preferentially acquire bacteria through these infection routes, so we tested if BMDMs acquired more bacteria if the bacteria were extracellular compared to acquisition of bacteria from the cytosol of infected cells.

For these experiments, we normalized the amount of bacteria in each sample. Donor BMDMs had approximately 1 million intracellular bacteria, so the control group was infected with 1 million free living bacteria using a synchronized infection where the bacteria were centrifuged onto the BMDMs. To synchronize transfer, we used the same assay as described in Figure 1F. During a brief inoculation, BMDMs acquired 10-fold more bacteria by phagocytosing bacteria from inside of neighbouring, infected BMDMs than uptake of extracellular bacteria (Figure 4A). Approximately 5% of the total bacteria transferred from the infected to uninfected BMDMs. These results suggest that bacteria migrate between cells much better through cell-cell transfer than through phagocytosis of extracellular F. tularensis bacteria.

Figure 4

Download asset Open asset

A small population of infected cells migrates to the recipient population (0.13 ± 0.23%), which will slightly skew our colony forming unit results (three independent experiments, 500 cells examined per experiment, mean ± SD). To account for this, we validated these results by microscopy. We labelled the recipient BMDM plasma membranes with biotin before the two populations were mixed and used the synchronized transfer assay (Figure 1F). After isolating the recipient cells, we stained them with fluorescent streptavidin to ensure they were not contaminating donor cells. Only fully biotin labelled recipient cells were included in the microscopy analysis. We found that recipient BMDMs acquired significantly more bacteria per cell than BMDMs exposed to extracellular F. tularensis (Figure 4B and C). Taken together, cell-cell transfer of bacteria increased the number of bacteria that spread because recipient cells typically acquired multiple bacteria at the same time.

Part of the reason cell-cell transfer was so much more efficient at transferring bacteria is that several bacteria entered the same FCV (Figure 5A). Following phagocytosis of extracellular bacteria, the vast majority of cells have a single bacteria per vacuole (Figure 5—figure supplement 1A). In contrast, almost half of recipient cells with bacteria in a LAMP-1 positive vacuole have multiple bacteria within the same vacuole, with about 20% having more than five bacteria in the same vacuole (Figure 5—figure supplement 1A). Recipient cells often take several bites of the same donor cell, so the same cell often has multiple vacuoles containing bacteria (for example, Figure 2A/B and Figure 4C). Thus, recipient BMDMs acquired several F. tularensis bacteria simultaneously.

Figure 5 with 6 supplements see all

Download asset Open asset

More bacteria per vacuole likely means more bacterial effectors were present to modify the vacuole, possibly increasing bacterial escape kinetics. If more bacteria within the same FCV increased phagosomal escape, the percentage of cells with multi-bacterial FCVs should decrease over time. Instead, we found that the percent of LAMP-1⁺ vesicles containing multiple bacteria was consistent across the time points that we examined (Figure 5—figure supplement 1B). These data indicate that having multiple bacteria in a vacuole does not substantially alter escape kinetics. It also suggests that vacuoles with several bacteria are not intrinsically different from vacuoles with a single bacterium.

For these assays, we used the live vaccine strain of F. tularensis. However, we observed identical structures following cell-cell transfer of the highly virulent Schu S4 strain (Figure 5—figure supplement 2). Thus, these results are broadly applicable to virulent F. tularensis.

In Figure 1B, the donor cell plasma membrane is enclosed inside of the plasma membrane of the recipient BMDM. This result suggests that vacuoles formed through phagocytosis of intact cells should have two membranes. We hypothesized that the transferred material would be inside of a structure that resembles an autophagosome, but is made from membranes of two different cells. The plasma membrane of the recipient cell should enclose the structure, just like a typical phagosome. Inside of that should be a portion of the donor cell plasma membrane that was pinched off during transfer. The transferred cytosolic content would be at the core of this structure.

We first tested if bacteria were enclosed in vacuoles containing plasma membrane from the donor cell. To do this, we biotinylated the proteins on the plasma membrane of donor, infected cells. We then synchronized bacterial transfer to uninfected BMDMs and purified the recipient BMDMs. We permeabilized the cell and stained with streptavidin, which will bind to biotinylated proteins that were on the infected donor cell plasma membrane prior to cell-cell transfer. This procedure allows us to track what happens to the donor cell plasma membrane following cell-cell transfer. We found that structures with multiple F. tularensis bacteria were within a vacuole that is at least partially derived from the plasma membrane of the originally infected donor cell (Figure 5A and B, donor cell example in Figure 5—figure supplement 3).

If phagocytosis of the donor cell occurs, then there should be at least two membranes surrounding the transferred bacteria. To test this, we purified recipient cells using the synchronized transfer assay described Figure 1F and examined them by TEM. Since wildtype bacteria will escape the FCV, we primarily used the T6SS inducible strain for our TEM studies. F. tularensis bacteria are the dark, electron dense structures with a small electron translucent clearing around them.

The inducible T6SS strain was only found inside of multi-membraned structures in all of the infected recipients that we identified (Figure 5C and D). Similar to our confocal analysis, multiple bacteria were within the same vacuole and recipient BMDMs often contained several distinct vacuoles with bacteria (Figure 5—figure supplement 4). Notably, we were able to find wildtype bacteria in similar, multi-membranous structures at 40 min post-transfer (Figure 5—figure supplement 4). As expected, only a subset of wildtype bacteria were found inside of vacuoles due to phagosomal escape. From these results, we conclude that the FCV is a double-membraned structure composed of membrane from two different cells.

The morphology of these FCVs is unusual. The T6SS inducible bacteria were found within complex, membranous structures that most closely resemble residual bodies (Novikoff and Shin, 1978). Residual bodies are the undigested remnants of phagosolysosomes or autophagolysosomes (De Duve and Wattiaux, 1966). Since the mutant cannot escape, the vacuole likely matures down the normal degradative pathway. In comparison, TEM of wildtype F. tularensis indicate that the bacteria are in a more electron dense, less membranous compartment. This is likely due to modifications of the vacuole by the T6SS, which is essential for F. tularensis to escape the vacuole.

A possible alternative explanation for the formation of FCVs containing multiple bacteria is phagocytosis of extracellular vesicles that contain several bacteria. In theory, this structure would also form a double membraned compartment after phagocytosis that could enclose several bacteria. To assess the likelihood that extracellular vesicles explain our phenotype rather than contact dependent cell-cell transfer, we analyzed how many cells had FCVs containing several bacteria when the infected and uninfected populations were physically separated.

We only observed FCVs containing several bacteria under conditions where cell-cell contact was possible (assay from Figure 1F). When infected and uninfected BMDMs were physically separated by 2–3 millimetres using a Transwell membrane, very few recipient BMDMs became infected and we did not find a single recipient cell with multiple bacteria in the same LAMP-1⁺ vacuole in any of the experiments (Figure 5—figure supplement 5). In this setup, bacteria and extracellular vesicles can pass through the membrane, but whole cells from the different populations could not touch. In contrast, uninfected cells that were able to physically touch infected cells had readily observable multi-bacterial FCVs. Thus, cell-cell transfer needs to occur for recipient cells to acquire multiple bacteria within the same FCV.

To test if multi-bacterial FCVs form in vivo, we intranasally inoculated mice with the inducible T6SS strain. The lungs were harvested 24 hr post inoculation and a single cell suspension of cells was allowed to adhere to a coverslip for 1 hr. The cells were then washed, fixed with paraformaldehyde and stained for LAMP-1. We then examined the adherent cells ex vivo by microscopy.

We readily found multi-bacterial clusters in LAMP-1⁺ vacuoles in lung cells ex vivo, similar to our results using cultured BMDMs (Figure 6). Based on the similarities with our in vitro results, cell-cell transfer of bacteria occurs during F. tularensis infections in the lung and is not a cell culture phenomenon.

Figure 6 with 1 supplement see all

Download asset Open asset

It is difficult to determine how bacteria infect cells in vivo because there are several different potential routes that likely all occur. To estimate how much cell-cell transfer may be occurring in vivo during F. tularensis infections, we quantified the number of cells with T6SS mutant bacteria clustered within the same LAMP-1⁺ vacuole. We excluded cells where bacterial growth occurred because these were most likely the initially infected cells (example in Figure 6—figure supplement 1). We found that 19.02 ± 2.93% of the recipient cells had three or more bacteria within the same LAMP-1 positive vacuole (mean ± SD, from 4 experiments with two mice pooled per sample, 45 putative recipient cells identified).

Importantly, this estimate does not include any transfer events where only one or two bacteria transferred at a time. Additionally, only about 70% of the T6SS mutant bacteria were found in LAMP-1 positive vacuoles in vitro, even though the mutant was trapped within the FCV (Figure 3B). Based on these conservative criteria, a minimum of about 1 in five infected recipient cells become infected via cell-cell transfer in the lung during F. tularensis infection.

The EEA-1 antibody (N-19, sc-6415) and LAMP-1 antibody (1D4B) were acquired from Santa Cruz Biotechnologies (Dallas, Texas) and the Developmental Studies Hybridoma Bank (University of Iowa, Iowa City, Iowa) (Hughes and August, 1981) respectively. The secondary antibodies used were anti-Rat (cat# 26-4826-82) and anti-goat (cat# PA1-28662) from ThermoFisher Scientific (Waltham, Massachusetts).

Bone Marrow derived Macrophages were differentiated and used as previously described (Steele et al., 2016). BMDMs were seeded the night before infection at 200,000 cells per well in a 12 well plate. For all microscopy studies, the imaged BMDMs were seeded onto an acid treated coverslip.

F. tularensis live vaccine strain and Schu S4 were grown on chocolate agar supplemented with isovitalex for 3 days. The day before infection, the cells were seeded into Chamberlain’s defined media. The culture was shaken at 37°C overnight, typically 16–18 hr.

All bacterial infections where uptake of extracellular bacteria was tested were performed by synchronizing the infection. Briefly, cells were chilled on ice and had cold media containing bacteria at either a multiplicity of infection of 5 (experiments comparing bacterial uptake to cell-cell transfer) or 100 (all other assays). The cells were placed into a chilled centrifuge and spun for 5 min at 300x g. The cells were then heat shocked in a 37° water bath and incubated for 30 min. Following incubation, the media was exchanged with media containing 10 ug/ml gentamicin.

BMDMs were seeded directly into a 12 well plate or placed on coverslips. 20 hr post inoculation, the media in both wells was replenished with fresh media containing 10 ug/ml gentamicin and the uninfected BMDMs on the coverslip were inverted onto the infected BMDMs. After a 30 min incubation, the coverslip was removed and placed back in the original well. The cells were then harvested at the indicated time. This procedure had a 0.13 ± 0.23% frequency of infected donor cells attaching to the coverslip (three independent experiments, 500 cells examined per experiment, mean ± SD).

The tetracycline inducible promoter containing plasmid pEDL17 was modified to constitutively express GFP and express either kanamycin resistance (empty vector controls) or DTL_0019 (dotU) (LoVullo et al., 2012). Genes under the control of the tetracycline inducible promoter were induced by 250 ng/ml of anhydrous tetracycline (ATc) (Cayman Chemicals, Ann Arbor, Michigan) in Chamberlains Defined Media for 18 hr prior to inoculating cells. In all induction experiments, ATc was added to the overnight growth media for all samples except the uninduced control. ATc was removed prior to inoculation of host cells unless otherwise indicated. pEDL17-GFP empty vector had a slight but consistent delay in phagosomal escape compared to the pkk214-GFP plasmid used in most of our studies.

BMDMs were synchronously infected with either infected BMDMs in the presence of gentamicin or extracellular bacteria. Extracellular bacteria were added at a multiplicity of infection of 5. This MOI was chosen because the donor BMDMs averaged five bacteria per cell based on bacterial colony forming units at 20 hr post inoculation and the initial number of BMDMs seeded.

After a 30 min incubation, the slide with infected cells in the cell-cell transfer sample was returned to its original well and 10 ug/ml of gentamicin in fresh media was added to all samples. At 2 hr after the start of transfer, the cells were scraped from the plate, lysed by vortexing, serially diluted and quantified.

BMDMs were infected for 19 hr with GFP-expressing F. tularensis. 10 ug/ml gentamicin added 2 hr post inoculation. The infected cells were labelled with biotin using an EZ-link sulfo-NHS- biotin labelling kit following the manufacturer’s protocol (ThermoFisher Scientific). The cells were washed with complete media and incubated for 30 min after biotin labelling. A coverslip with uninfected, unlabelled cells were placed onto the infected cells for 1 hr. The cells were then lifted off, fixed with 4% paraformaldehyde, washed with PBS followed by 50 mM ammonium chloride, stained with wheat germ agglutinin (cat# W32466 ThermoFisher Scientific) and imaged by confocal microscopy.

Microscopy results were acquired using a Leica DM4000 upright fluorescence microscope. For a bacterium or bacterial cluster to be counted as within an EEA-1 or LAMP-1 positive vacuole, the fluorescent marker had to completely surround the bacteria, as in Figure 2A and B. Bacteria that co-localized with these markers but where the markers did not form a coherent vacuole were not counted as being within a vacuole.

Occasionally, BMDMs will have bacteria in multiple vacuoles or some in vacuoles and some not within a vacuole. For these cells, if any of the bacteria were contained in a LAMP-1 or EEA-1 positive vacuole, they were counted as having bacteria inside of a vacuole. Likewise, for determining the percentage of cells with multiple bacteria within individual LAMP-1 positive vacuole, any cell with multiple bacteria in the same vacuole was counted, even if bacteria were not in a vacuole elsewhere in the cell.

Representative confocal images were acquired using a Leica SP-8 microscope and the same samples were used for representative images and data analysis where applicable.

Mice were anesthetized with avertin and intranasally inoculated with 10,000 colony forming units of the inducible T6SS strain. The bacteria were grown in chamberlains defined media containing 250 ng/ml ATc overnight and 250 ng/ml of ATc was added to the PBS containing bacteria that the mice were inoculated with.

At 24 hr post inoculation, the lungs were harvested. Single cell suspensions were made by finely chopping the lung with scissors and incubating the lung pieces in five units per ml of dispase (Stem Cell Technologies, Cambridge Massachusetts) for 30 min at room temperature. The lung homogenate was put through a 40 μm cell strainer to isolate individual cells. The sample was centrifuged at 250x g for 5 min and the supernatant was removed. RPMI with 10% FBS and supplemented with non-essential amino acids, Glutamax, and sodium pyruvate was added to the cells for the remainder of the incubation.

The lung cell samples were placed onto chamber slides and incubated at 37° for 1 hr to allow cells to adhere. The sample was then washed with PBS and fixed with 4% paraformaldehyde. The samples were then stained for LAMP-1.

All animals were handled according to approved institutional animal care and use committee (IACUC) protocol #4946 at Washington State University.

BMDMs were inoculated with wildtype bacteria or ΔdotU ATc inducible dotU strain. 10 μg/ml of gentamicin was added to the BMDMs at 2 hr post inoculation. The cells were then incubated for a total of 20 hr. At 20 hr post inoculation, BMDMs seeded on a Nunc Thermomax plastic coverslip (ThermoFisher Scientific) were inverted onto the infected cells for synchronized transfer. The cells were incubated with recipient cells for 1 hr, purified and fixed for 1 hr with 2% paraformaldehyde and 2% glutaraldehyde in 0.1M cacodylate buffer (pH 7.2).

The BMDMs were microwave fixed using a Pelco Biowave Pro 36500 Laboratory Microwave System for 1 min at 350 Watts with a temperature restriction set to 38°C. The cells were rinsed 0.1M Cacodylate buffer 2X, 10 min each and 1X in distilled water for 10 min. The cells were incubated in 2% OsO₄ and 1.5% potassium ferrocyanide, 2 mM CaCl₂ in 0.05M cacodylate buffer (pH 7.2). They were rinsed 3X in distilled water, 10 min each and dehydrated in progressive graded ethanol series (10 min each step). The ethanol was replaced with acetone and the cells were infiltrated in 1:1 acetone and Spurr’s resin overnight on a rotator. The acetone was evaporated over 6 hr. Fresh 100% resin was added and left to infiltrate overnight on a rotator, this step was repeated two more times. The cells were embedded and cured for 24 hr at 70°C. Sectioning, staining and imaging were done as described in Froelich et al. (2011).

The statistical test for each experiment is listed in the figure legend. All details about the number of replicates, experiments, and cells analyzed are included in the figure legends.

The only cells excluded from our microscopy analyses were cells that had clearly undergone efforocytosis (large vacuole with DAPI stained nucleus in vacuole) and highly infected cells that were in a different focal plane on top of adherent cells because these are almost certainly donor cells that migrated to the coverslip.

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

We would like to thank Daniel Mullendore at the Franceschi Microscopy and Imaging Core at Washington State University for preparing the samples for transmission electron microscopy and assistance in acquiring the images. We would like to thank Amanda Foreman for creating the graphical summary Figure 7.

Animal experimentation: All animals were handled according to approved institutional animal care and use committee (IACUC) protocol #4946 at Washington State University.

1. Received: January 25, 2019
2. Accepted: April 18, 2019
3. Accepted Manuscript published: April 24, 2019 (version 1)
4. Version of Record published: May 3, 2019 (version 2)

© 2019, Steele et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.