Micropatterned Macrophage Analysis Reveals Global Cytoskeleton Constraints Induced by Bacillus anthracis Edema Toxin

Abstract

Bacillus anthracis secretes the edema toxin (ET) that disrupts the cellular physiology of endothelial and immune cells, ultimately affecting the adherens junction integrity of blood vessels that in turn leads to edema. The effects of ET on the cytoskeleton, which is critical in cell physiology, have not been described thus far on macrophages. In this study, we have developed different adhesive micropatterned surfaces (L and crossbow) to control the shape of bone marrow-derived macrophages (BMDMs) and primary peritoneal macrophages. We found that macrophage F-actin cytoskeleton adopts a specific polar organization slightly different from classical human HeLa cells on the micropatterns. Moreover, ET induced a major quantitative reorganization of F-actin within 16 h with a collapse at the nonadhesive side of BMDMs along the nucleus. There was an increase in size and deformation into a kidney-like shape, followed by a decrease in size that correlates with a global cellular collapse. The collapse of F-actin was correlated with a release of focal adhesion on the patterns and decreased cell size. Finally, the cell nucleus was affected by actin reorganization. By using this technology, we could describe many previously unknown macrophage cellular dysfunctions induced by ET. This novel tool could be used to analyze more broadly the effects of toxins and other virulence factors that target the cytoskeleton.

INTRODUCTION

Anthrax is a disease caused by Bacillus anthracis, a Gram-positive, rod-shaped, spore-forming bacterium whose spores are considered a serious bioterrorism agent (1). The major known virulence factors of B. anthracis include edema toxin (ET), formed by the association of the protective antigen (PA) component with edema factor (EF). Once in the cytosol, EF acts as a calmodulin (CaM)-dependent adenylate cyclase that increases intracellular cyclic AMP (cAMP) concentrations (2, 3).

At the cellular level, ET effects have been more precisely described over the last decade. PA binds to at least two independent receptors (ANTXR1 or TEM-8 [tumor endothelial marker 8] and ANTXR2 or CMG-2 [capillary morphogenesis protein 2]) on target cells (4). ANTXR2 plays the major role in vivo for toxin entry. A third coreceptor, named low-density lipoprotein (LDL) receptor protein 6 (LRP-6), has also been proposed (5–7). PA subunits associate into heptamers to form a prepore throughout the cell surface via lipid rafts. This PA heptamer enables the binding of EF components to the cell surface in a stoichiometric ratio of 7/3. The toxin-receptor complex is then internalized by clathrin-dependent endocytosis (8, 9). The pH decrease of early endosomes results in the translocation of these factors in multivesicular bodies (MVB), finally merging with the intracellular membrane of late endosomes. This last step is responsible for translocating EF into the cytoplasm, where it stays associated with the membrane of the late endosome (10). Its perinuclear localization generates intracellular cAMP gradients from the cell nucleus to the periphery. In turn, cAMP activates at least the transcription factor CREB in macrophages (11) and EPAC plus Rap-1 in endothelial cells (12).

At the organ level, ET disrupts endothelial homeostasis by playing upon multiple factors, although its role in edema is still being discussed (3, 13). First, ET affects directly the cytoskeleton of endothelial cells by inducing transendothelial macroaperture tunnels (14). Interestingly, the cytoskeleton can sense curvature induced by transendothelial macroaperture through an I-BAR domain protein, missing in metastasis (MIM). A delicate balance exists between the ET-induced macroaperture and resealing by the cytoskeleton through recruitment of Arp2/3 actin polymerization, which induces actin waves that close the macroaperture (14). Second, ET also disrupts Rab11/Sec15 traffic at the exocyst, inducing reduced cadherin expression at the tight junctions (15). Third, ET induces cytoskeletal changes and inhibits chemotaxis through the action of downstream cAMP effectors EPAC and EPAC-related activators Rap-1, Epac, and MR-GEF/rapGEF5 (12).

ET has major disruptive effects on most immune cells (16), including macrophages (11, 17), which represent the first target of the pathogen following inhalation of spores. ET has also profound anti-inflammatory effects on dendritic cells (18) and lymphocytes (19). Finally, the most critical immune effects are mediated through polymorphonuclear cells (PMNs) (20). PMNs were one of the first cellular targets described during anthrax (21) and then abandoned before being put into the limelight again by using myeloid-specific, CMG-2-deficient mice (20).

Taken together, these studies are consistent with what is observed at the animal level, showing in infection models that ET is a critical virulence factor through intranasal infection that facilitates pathogen penetration (22). At the animal level, it has also been elegantly shown in an intoxication model that ET-induced lethality occurs mainly through hepatocytes (13).

As ET effects are critical on myeloid cells such as macrophages during an infection and as ET also impairs cytoskeleton organization (23), we wanted to dissect cellular effects of anthrax ET upon the macrophage cytoskeleton. Indeed, previous reports already describe some effects of ET on macrophage functions such as chemotaxis (12) or phagocytosis (24), through cytoskeleton disruption (11).

Classical cell cultures grown on a flat adhesive microenvironment already provide some clues to ET's disruptive effects upon the cytoskeleton (11). However, this loose-constraint environment is far from reproducing in vivo the highly structured microenvironment and induces a high variability in size, shape, and morphology that impairs any quantitative analysis of the actin cytoskeleton, leaving many questions unanswered. Cell micropatterning methods have been recently developed, based on the use of a culture substrate with microscopic features that imposes a defined cell adhesion pattern (25). Cell micropatterning imposes a stable cellular form, which induces through geometrical constraint a reproducible internal organization of the actin and microtubule cytoskeleton. Cell micropatterning has recently proven to be useful in analyzing subtle changes in cell morphology triggered by drug treatment or Salmonella infection (26). Cell micropatterning has been developed using a highly adhesive transformed cell line, such as HeLa cells.

Here, we report the adaptation of cell micropatterning to both loosely adhesive bone marrow-derived macrophages (BMDMs) and primary peritoneal macrophages. We show that ET induces a dynamic disruption of the actin cytoskeleton and subsequent concentration around the nucleus, which then cause shrinking of the cell. Finally, the cell architecture is completely disrupted, as the cell nucleus is displaced and modified in shape. Cell micropatterning reveals a potent model for studying the multiple implications of cytoskeletal disruption on immune cell functions.

MATERIALS AND METHODS

Mice and ethics statement.

Female C57BL/6 mice, aged 6 to 12 weeks, were obtained from Charles River and were used for the isolation of peritoneal macrophages and the generation of bone marrow-derived macrophages (BMDMs). Mouse handling and the experimental procedures were performed in accordance with the French Government guidelines for the care and use of laboratory animals and were approved by the Institut de Recherche Biomédicale des Armées ethics committee (approval number 2012/34.0).

Recombinant proteins and reagents.

Protective antigen (PA) and edema factor (EF) were purchased from List Biological. A selective inhibitor of phosphodiesterase-4, Rolipram, was obtained from Enzo Life Sciences. Forskolin (catalog no. F6886) was purchased from Sigma-Aldrich.

Cell culture of HeLa cells and peritoneal macrophages and generation of BMDMs.

A human cervical adenocarcinoma line (HeLa) was cultured in RPMI 1640 (Pan Biotech GmbH) supplemented with 10% heat-inactivated fetal calf serum (Gibco Life Technologies)–100 U/ml penicillin–0.1 mg/ml streptomycin–2 mM l-glutamine (Gibco Life Technologies) and maintained at 37°C under a humidified atmosphere with 5% CO₂.

Primary peritoneal macrophages were collected by peritoneal washing with 5 ml RPMI 1640. The cells were pelleted by centrifugation (1,500 rpm, 5 min, 4°C) and then suspended at 2.5 × 10⁶ cells/ml in RPMI 1640 supplemented with 10% heat-inactivated fetal calf serum–100 U/ml penicillin–0.1 mg/ml streptomycin–2 mM l-glutamine in 6-well plates. After 2 h at 37°C in a 5% CO₂ atmosphere, cells were washed 3 times with RPMI 1640 to remove nonadherent cells and adherent cells (peritoneal macrophages) were subsequently cultured in RPMI 1640–10% heat-inactivated fetal calf serum for 1 day. At day 2, peritoneal macrophages were collected, washed, and resuspended in fresh medium at 5 × 10⁵ cells/ml for further use.

BMDMs were generated from proliferating mouse bone marrow progenitors. Briefly, bone marrow cells were isolated by flushing the bone marrow shafts of femurs and tibias from mice. Red blood cells were lysed with ACK lysis buffer (Gibco Life Technologies). After washing, bone marrow cells were seeded in 6-well flat-bottom plates (Costar-Corning) at 10⁶ cells/ml in RPMI 1640 supplemented with 10% heat-inactivated fetal calf serum–100 U/ml penicillin–0.1 mg/ml streptomycin–2 mM l-glutamine (Gibco Life Technologies), henceforth referred to as complete medium (CM). Macrophage colony-stimulating factor (M-CSF; Peprotech) was added to each well at 20 ng/ml. Cells were maintained at 37°C in a humidified atmosphere with 5% CO₂. On days 3 and 5, 70% of the medium was replaced by fresh CM containing M-CSF at 10 ng/ml. At day 7, the nonadherent cells were purged and the macrophages (adherent cells) were collected, washed, and resuspended in fresh medium at 5 × 10⁵ cells/ml for further use.

Cell seeding and incubation with toxins.

For cell patterning, HeLa cells, peritoneal macrophages, and BMDMs were deposited on adhesive micropatterned coverslips coated with fibronectin FN650 (CytooChips; Cytoo Inc.), according to the manufacturer's instructions, with few modifications for peritoneal macrophages and BMDMs. Briefly, peritoneal macrophages and BMDMs were plated at a density of 50,000 cells/ml. After 60 min, nonadherent cells were removed by gentle flushing with fresh medium and were incubated for full spreading for an additional 4 to 6 h at 37°C in a humidified atmosphere of 5% CO₂ prior to treatment. For nonpatterned cells, HeLa cells and BMDMs were seeded on 8-well BD BioCoat Fibronectin Cellware plates (BD Biosciences) 1 day before treatment in the same way as micropatterned BMDMs.

For toxin treatment, patterned and nonpatterned peritoneal macrophages and BMDMs were incubated with PA at 1 μg/ml and EF at 100 ng/ml for 6 or 16 h at 37°C in a 5% CO₂ atmosphere. For some experiments, cells were incubated with PA alone at 1 μg/ml.

Immunostaining and image acquisition.

After treatment, cells were washed with phosphate-buffered saline (PBS), fixed with 4% formaldehyde (Sigma-Aldrich) in cytoskeletal buffer (CB) [10 mM 2-(N-morpholino)-ethanesulfonic acid sodium salt (MES), 3 mM MgCl₂, 138 mM KCl, 2 mM EGTA, and 0.32 M sucrose] for 30 min, and permeabilized in 0.25% Triton X-100 (Sigma-Aldrich) for 5 min. Cells were blocked with 2% bovine serum albumin (BSA) in CB for 30 min. For actin staining, cells were incubated with Alexa Fluor 488-phalloidin (Life Technologies) in staining solution for 30 min. Cells were incubated in 0.2 μg/ml Hoechst 33248 (Sigma-Aldrich) for 5 min in PBS for nucleus staining before mounting in Prolong Gold antifade reagent (Life Technologies).

Microscopic image acquisitions were performed with an Olympus IX-81 inverted microscope equipped with an MT20 illumination system (Olympus). Images were captured using an Orca-ER charge-coupled device (CCD) camera (Hamamatsu Photonics), controlled by Cell software (Olympus) using a 40× objective (UPlan FLN; numerical aperture [NA], 0.75).

Image processing.

All images were processed using ImageJ (National Institutes of Health). For analysis, all images from single cells per condition were assembled into one stack and aligned using the stained fibronectin micropattern and a strategy developed earlier (25). The median spatial distribution of actin or nuclei was measured by calculating the median intensity of each pixel over the stack. The median of F-actin and nuclei intensity distribution was named the “reference cell,” as coined by Théry et al. (25), with a difference being that the calculation was based not on the mean intensity but on the median intensity of each pixel. The reference cells were represented with an intensity gradient of a fire lookup table (LUT) to facilitate the interpretation of actin distribution and with a median intensity projection (MIP). The reference cells resulted from the assembly of Z-projected images of actin (green) and nuclei (blue).

Plot profile intensity and morphological parameters of shaped cells were measured by ImageJ on the stack of aligned cells. For morphological analysis, 3 parameters were calculated for each cell: the area; the circularity index (4π × area/perimeter²), which assumes values between 1 for circular shape and 0 for elongated morphology; and the aspect ratio (length of major axis/length of minor axis).

Density map analysis of F-actin structures and P value calculations were performed by Cytoo Cell Architects (Cytoo Inc.), according to the protocol developed by Schauer et al. (27).

cAMP and EF measurement.

BMDMs were seeded on 96-well plates at 200,000 cells per well and cultivated for 24 h before intoxication. After subjecting cells to experimental conditions (PA, ET, forskolin, Rolipram, and ET plus Rolipram), total proteins were extracted by removing BMDM culture medium (as supernatant) and then adding lysis buffer chilled to 4°C (radioimmunoprecipitation assay [RIPA] buffer; Cell Signaling Technology), supplemented with a protease-phosphatase inhibitor cocktail (Cell Signaling Technology). The cells were incubated in lysis buffer on ice for 10 min and centrifuged at 13,000 rpm for 10 min, and cell lysates were collected for protein dosage (Bradford assay) and cAMP/EF measurement.

The adenylyl cyclase EF activity measured by assessing the production of cAMP used a competitive enzyme immunoassay (EIA) as described previously (28). Briefly, the assay consisted of three steps. (i) The enzymatic reaction assay for EF was carried out as follows: each sample was diluted in 20 mM HEPES buffer (pH 7.2) with 0.5 mg/ml bovine serum albumin, 10 mM MgCl₂, 10 μM CaCl₂, 1 mM ATP, and 10 μM CaM. The standard curve of EF was performed with serial dilutions of EF in untreated cell lysates. The adenylyl cyclase activities were then assayed for 30 min at 30°C. (ii) The second step was elimination of ATP interference and increase of cAMP sensitivity by chemical transformations (production of acetylated cAMP). (iii) The third step was the EIA: acetylated cAMP was placed in competition with a cAMP enzymatic tracer, cAMP coupled to acetylcholinesterase (AChE), for binding to anti-cAMP polyclonal antibodies, for 90 min at room temperature. The amount of tracer bound to the anti-cAMP antibody site was then measured by incubation with AChE substrate (Ellman's reagent) at room temperature overnight. The absorbance was recorded at 405 nm using a spectrophotometer microplate reader (Multiskan Ascent; Labsystems). EF activity was expressed as picograms per milligram of protein for cell lysate and as picograms per milliliter for supernatant.

To detect cAMP, samples were treated without dilution or with dilution in EIA buffer (0.1 M phosphate buffer [pH 7.4], 0.15 M NaCl, 0.1% BSA, 10 μM CaCl₂, 1 mM EDTA, 0.01% sodium azide). A standard curve of cAMP was performed with serial dilutions of 7,500 pmol/ml of cAMP in untreated cell lysates, diluted in agreement with the dilution of the sample. To increase the sensitivity of the detection, the assay consisted of steps 2 and 3 described above. cAMP content was expressed as picomoles per milligram of protein.

Statistical analysis.

Statistical analysis was performed using GraphPad Prism software (GraphPad Software, Inc.). Comparisons between groups were performed using a Mann-Whitney test to determine statistical significance (P < 0.001).

RESULTS

Micropatterning model for BMDMs.

In a structured tissue, cells integrate constant information from their microenvironment as previously shown by micropatterned cell strategy (29). Micropatterning has been developed by using mainly highly adherent cells, such as transformed human HeLa cells. However, those cells are transformed and do not reproduce the phenotype of differentiated immune cells such as macrophages. We have thus established an adaptation of the general micropatterning protocol using commercial crossbow and L micropatterns to culture BMDMs, as these shapes induce “polar” organization of the cell in contrast to nonconstrained cells (Fig. 1A). The nucleus-centrosome-Golgi axis that forms the global orientation of cell polarity is oriented from regions that lack extracellular matrix toward those that are abundant (25). In accordance with previous reports, we have observed a polar organization in BMDMs similar to what has been previously described with HeLa cells, compared to nonconstrained cells (Fig. 1A). Interestingly, the global aspects of the cells on the pattern and organization of actin fibers were very different in highly adherent HeLa cells versus loosely adherent BMDMs. Such results suggest that cell adhesion is due to cytoskeletal organization, cellular tension, and contraction that may provide spatial cues for the establishment of cell shape and actin cytoskeleton organization. With the tightly adherent HeLa cells, the cytoskeleton was more spread out, filling up the concavity of the patterns (Fig. 1B). On the L patterns, HeLa cells formed triangles filling up the nonadhesive side, and on the crossbow, the bowstrings were tensed by stress fibers. In sharp contrast, we did not see any significant bundles of actin stress fibers on the bowstrings or cortical actin along the “extrados” (26), suggesting a specific cytoskeleton architecture in these highly adherent cells (Fig. 1B). The median intensity of F-actin staining from >60 cells, by projecting the median intensity projection (MIP) of each aligned cell into a median intensity map for the reference cell, allowed quantitative measurements of actin rearrangements induced in micropatterned cells (Fig. 1B). The F-actin distribution is represented with a heat map for the “median cell.” BMDM global shapes on crossbow and L micropatterns were significantly different from what was previously described with HeLa cells. For the rest of the experiments, we focused on BMDMs, which represent a better cellular model for anthrax infection.

FIG 1.

Micropatterned BMDMs as a model for studying host-pathogen interactions. HeLa cells and BMDMs were grown on classical fibronectin-coated coverslips or on micropatterned coverslips. Alexa Fluor 488-phalloidin-stained F-actin is depicted in green, Hoechst 33248-stained nucleus is shown in blue, and Alexa Fluor 650-stained fibronectin is shown in red. (A) Nonconstrained HeLa cells and BMDMs are shown in the left panels, while HeLa cells and BMDMs patterned on the L are shown in the middle panels and those patterned on the crossbow are shown in the right panels. For L- and crossbow-patterned cells, the design of the pattern is depicted in the white square, while one significant cell (red square) is enlarged in the large red square on the right. (B) Empty patterns are shown in the left panels. Reference cells (the median intensity of fluorescence of n cells) of HeLa cells (middle panels) and BMDMs (right panels) are depicted with the median intensity projection (MIP) representation or fire lookup table (LUT) representation. The number of cells used is shown at the bottom of each panel. On L patterns, the hypotenuse of the triangle is referred to as the “nonadhesive” side, and on the crossbow pattern, the adhesive side is the “extrados,” while the nonadhesive sides are called the “bowstrings.” Bars, 10 μm.

ET triggers a global actin cytoskeleton rearrangement in BMDMs.

Micropatterned BMDMs were intoxicated with 100 ng/ml of ET and further analyzed at different time points. We observed a progressive actin rearrangement with time after ET exposure. ET induced a persistent actin meshwork collapse into a dense F-actin centroid starting at 6 h for L patterns (Fig. 2A). BMDMs incubated with PA only did not show any modification of actin organization versus control BMDMs (see Fig. S1A in the supplemental material). Quantitative analysis of the cell area, circularity index (4π × area/perimeter²), and aspect ratio (length of major axis/length of minor axis) showed a significant decrease of each parameter (Fig. 2B). On crossbow patterns, we also observed a progressive actin cytoskeleton concentration around the nucleus at 6 and 16 h (Fig. 2C). At 16 h, cells were shrunk with numerous holes in the actin staining on the fibronectin patterns (Fig. 2C). In accordance with the L pattern, quantitative measurements of cell shape were significantly affected. Cell area decreased very significantly along time, as well as the circularity index, while the aspect ratio increased after 16 h (Fig. 2D). The aspect ratio is a ratio between the longest and largest dimension of the cell, and focal adhesion loss seems to affect the bow more than the stalk in the pattern.

FIG 2.

B. anthracis ET induces a global actin cytoskeleton rearrangement. BMDMs were treated with 100 ng/ml of ET (e.g., PA at 1 μg/ml and EF at 100 ng/ml) for 6 or 16 h. F-actin is stained in green with Alexa Fluor 488-phalloidin. (A) The effects of ET on BMDM actin are shown on nonconstrained cells versus L-patterned cells 6 h after intoxication. Individual patterned cells are shown in the middle panel (enlarged for one representative in the red square), and the reference cells are shown in the right panels. Bars, 10 μm. (B) The effects of ET on L-patterned cells are shown for cell areas in the left panel, for circularity in the middle panel, and for aspect ratio in the right panel. Each dot represents one cell. (C) The effects of ET on BMDM actin are shown on nonconstrained cells versus crossbow-patterned cells 6 and 16 h after intoxication. Individual patterned cells are shown in the middle panel (enlarged for one representative in the red square), and the reference cells are shown in the right panels. Bars, 10 μm. (D) The effects of ET on crossbow-patterned cells are shown for cell area in the left panel, for circularity in the middle panel, and for aspect ratio in the right panel. Each dot represents one cell. (E) The density maps of F-actin are shown for control cells and after 6 and 16 h of intoxication. The colors represent the smallest regions that contain the percentages of analyzed structures from 10% (red) to 90% (yellow). ***, P < 0.001; ns, nonsignificant.

Next, we tried to statistically compare the actin cytoskeleton rearrangements induced by ET. The average intensity maps were transformed into density maps by using software developed by K. Schauer's laboratory at the Institut Curie and licensed by Cytoo Cell Architects (27, 30). Fluorescence signal is transformed into a cloud of coordinate points by segmentation. Coordinates were replaced by Gaussian functions (kernels) that were summed, revealing the underlying density of F-actin throughout the cell. Red colors represent areas with highest probabilities of F-actin, while yellow colors represent areas with less F-actin (Fig. 2E). Density maps allowed statistical comparisons between actin organizations (Table 1). These data clearly confirmed the high statistical significance (P < 0.001) of the actin organization by comparing the control BMDMs with those cells treated with ET for 6 h and 16 h, or the 6-h representations were compared with the 16-h-treatment group.

TABLE 1.

Statistical significance of the density map of F-actin analysis

Sample 1	Sample 2	P value	Significance
ET 6 h	Control	>3.4E−38	***
ET 16 h	Control	>3.4E−38	***
ET 16 h	ET 6 h	6.46880757627423E−33	***

We also illustrated how ET induced different contractions of actin depending on the cell axis and pattern shape. On the crossbow, contraction of the actin cytoskeleton was more pronounced on the x than on the y axis, with a concentration of intense pixels in the center (Fig. 3A). On the L pattern, we saw mainly a loss of adhesion at the cell edge (on x and y axis) and contraction on the diagonal at the nonadhesive side (Fig. 3B).

FIG 3.

Analysis of actin distribution along different axes on patterned cells. (A) The reference cells (MIP) for F-actin on crossbow patterns have been overlaid for control cells (green) and for ET at 6 h (red) and 16 h (blue). The MIPs for pixels on each axis (x axis and y axis) are represented in the histograms with an overlay of control cell (green line), ET at 6 h (red line), and ET at 16 h (blue line). Bars, 10 μm. (B) The reference cells (MIP) for F-actin on L patterns have been overlaid for control cells (green) and ET at 6 h (red). The MIPs for pixels on each axis (x axis, y axis, and diagonal axis) are represented in the histograms with an overlay of control cell (green line) and ET at 6 h (red line). Bars, 10 μm.

ET also alters actin organization of primary peritoneal macrophages.

Our first results, although interesting, were limited to BMDMs that may harbor specific features. To demonstrate that the actin global cytoskeleton rearrangement induced by ET could also affect primary macrophages, we performed experiments on primary peritoneal macrophages (Fig. 4). We observed a similar pattern of actin retraction on crossbow with ET at 6 h. Peritoneal macrophages incubated with PA only did not show any modification of actin organization versus control peritoneal macrophages (see Fig. S1B in the supplemental material).

FIG 4.

B. anthracis ET induces an actin cytoskeleton rearrangement on micropatterned peritoneal macrophages. Peritoneal macrophages were grown on crossbow-micropatterned coverslips. Alexa Fluor 488-phalloidin-stained F-actin is depicted in green, and Alexa Fluor 650-stained fibronectin is shown in red. (A) Peritoneal macrophages patterned on a crossbow shape are shown in the left panels. The design of the pattern is depicted in the white square, while one significant cell (red square) is enlarged in the large red square on the right. Reference cells of peritoneal macrophages (right panels) are depicted with the median intensity projection (MIP) representation or fire lookup table (LUT) representation. The number of cells used is shown at the bottom of each panel. Bars, 10 μm. (B) Peritoneal macrophages were treated with 100 ng/ml of ET (e.g., 1 μg/ml PA and 100 ng/ml EF) for 6 h. F-actin is stained in green with Alexa Fluor 488-phalloidin. Individual patterned cells are shown in the left panel (enlarged for one representative in the red square), and the reference cells are shown in the right panels. The number of cells used is shown at the bottom of each panel. Bars, 10 μm.

ET enters the cells and modulates intracellular cAMP levels.

Next, we correlated the effects of ET incubation with intracellular cAMP levels, as ET is known to be a calcium and calmodulin-dependent adenylate cyclase. We did not observe a significant increase of cAMP in the BMDMs at 6 and 16 h after incubation with ET, in contrast to the high increase induced by forskolin (Fig. 5A). This could be due to the kinetics of cAMP production that decreases over time as previously shown on alveolar macrophages and a very high phosphodiesterase activity in macrophages, as suggested by the fact that the addition of Rolipram, a specific phosphodiesterase-4 inhibitor (31), very significantly increased the cAMP levels. Another option would be that a very low threshold of cAMP would trigger actin remodeling. To rule out a potential biochemical inhibition of EF activity after its entry in the cell, or a blockage of its entry, we also measured adenylate cyclase activity of EF in the supernatant and in the cell (Fig. 5B and C). We observed a significant calcium-calmodulin-dependent adenylate cyclase activity in the cells, demonstrating that EF had entered the cells after 6 and 16 h at a very significant level and was biochemically active.

FIG 5.

Effects of ET on intracellular cAMP. BMDMs were treated with PA (1 μg/ml), ET (1 μg/ml PA plus 100 ng/ml EF), forskolin (100 μM), Rolipram (10 μM), and ET plus Rolipram. After 6 or 16 h of incubation, supernatant and cell lysates were collected to measure cAMP concentration in cell lysate and EF adenylate cyclase activity in cell lysate and supernatant. The data are the most representative of 3 experiments. (A) Quantification of intracellular cAMP under indicated conditions and periods of time following intoxication. Data are presented as picomoles of cAMP per milligram of protein lysate. (B) Quantification of the EF adenylate cyclase activity in cell lysates under indicated conditions and periods of time following intoxication. Data are presented as picograms of EF per milligram of protein lysate. (C) Quantification of the EF adenylate cyclase activity in supernatant under indicated conditions and periods of time following intoxication. Data are presented as picograms of EF per milliliter of supernatant.

ET induces a nuclear shape deformation and a nucleus displacement.

We also analyzed the shape of the BMDM nucleus, as the nuclear shape can be affected by cytosolic constraints, and its shape and size are associated with differentiation and several diseases (32). On the crossbow, the nucleus increased in size with a loss of circularity and a kidney-like shape at 6 h, before returning to a normal size that was still deformed and elongated at 16 h (Fig. 6A and B). We saw the same for BMDM nucleus on the L pattern at 6 h (Fig. 6C and D).

FIG 6.

Analysis of nuclear shape deformation. BMDMs were treated with 100 ng/ml of ET (e.g., 1 μg/ml PA and 100 ng/ml EF) for 6 or 16 h. The nuclei are stained in blue with Hoechst 33248, and the fibronectin patterns are shown in red with Alexa Fluor 650. (A) The effects of ET on BMDM median intensity projection (MIP) of nucleus are shown on crossbow-patterned cells 6 and 16 h after intoxication. Empty patterns are on the left, MIPs of nuclei are in the middle panels, and the merged images are in the right panels. Bars, 10 μm. (B) The effects of ET on nuclei of crossbow-patterned cells are plotted for nucleus area (top panel), circularity (middle panel), and aspect ratio (bottom panel) for control cells and at 6 and 16 h. Each dot represents one cell. (C) The effects of ET on BMDM MIP of nucleus are shown on L-patterned cells 6 h after intoxication. Empty patterns are on the left, MIPs of nuclei are in the middle panels, and merged images are in the right panels. Bars, 10 μm. (D) The effects of ET on nuclei of L-patterned cells are plotted for nucleus area (top panel), circularity (middle panel) and aspect ratio (bottom panel) for control cells and at 6 h. Each dot represents one cell. ***, P < 0.001; ns, nonsignificant.

We further calculated whether the nucleus was also displaced in the cell. We calculated the centroid of the nucleus coordinate on the x and y axes, as shown in Fig. 7A. We observed that the nucleus was significantly displaced, as the distance between the axis origin and the centroid was significantly shorter in the x and y axis on the L pattern (Fig. 7B and C). In accordance with our observation on the L pattern, we also used the density map to analyze nuclear displacement on the crossbow (Fig. 7D). We effectively observed a displacement of the nucleus that is significantly going toward the extrados of the crossbow (Fig. 7E and Table 2). In both models (L and crossbow), the nucleus tended to be displaced toward the more adhesive edge of the cell.

FIG 7.

Nucleus displacement is correlated with actin reorganization. (A) The diagram shows how the coordinates of the nucleus centroid (in red) are calculated along the axes (x and y) drawn along the pattern (in blue). (B) The picture highlights displacement of the nucleus induced by ET by the overlay of the nucleus of a control (green) and cells treated with ET for 6 h (red). Bar, 10 μm. (C) Coordinates of the nucleus centroid on the x (left panel) and y (right panel) axes are plotted for control and cells treated with ET for 6 h. Each dot represents one cell. ***, P < 0.001. (D) The density maps of the nucleus are shown for control cells and after 6 and 16 h of treatment with ET. Colors represent the smallest regions that contain the percentages of analyzed structures from 10% (red) to 90% (yellow). (E) The median intensity value (±standard error of the mean) of each pixel (blue, nucleus; green, actin) is calculated on a diagonal starting at the bottom left corner of the L micropattern and going up to the nonadhesive edge of the cell. Bars, 10 μm.

TABLE 2.

Statistical significance of the density map of Hoechst 33248 staining analysis

Sample 1	Sample 2	P value	Significance
ET 6 h	Control	>3.4E−38	***
ET 16 h	Control	>3.4E−38	***
ET 16 h	ET 6 h	2.62398991424106E−11	***

To correlate the nuclear displacement with reorganization of the actin cytoskeleton, we compared the median intensity value of each pixel on an axis starting from the bottom left corner of the L micropattern and extending to the external edge of the cell on the concavity side (Fig. 7E). The maximum of the median intensity of each pixel for the nucleus was at 109 pixels on the diagonal from the edge of the cell in controls, whereas it moved to 62 pixels after 6 h of ET incubation. In the meantime, for actin the maximum of the median intensity moved from 151 pixels to 89 pixels from the edge of the pattern. This clearly shows that the nucleus displacement is correlated with a displacement of actin at the center of the cell. Interestingly, the value of maximum of the median intensity increased for the nucleus staining (190.70 ± 6.3 [arbitrary units {AU}] for control versus 235.0 ± 6.4 for ET treatment for 6 h) and for the actin staining (122.2 ± 7.4 [AU] for control versus 220.2 ± 8.9 for ET treatment for 6 h), indicating that both structures were squeezed. This observation indicates that the nucleus was progressively displaced toward the adhesive edge in a close correlation with a concentration of actin on its nonadhesive edge.

DISCUSSION

For the first time, we have set up a protocol for patterning loosely adherent immune cells and used it to analyze the effect of a bacterial toxin on the actin cytoskeleton. We have quantified unexpected disruptive effects of B. anthracis ET on the actin cytoskeleton of BMDMs.

The host cytoskeleton is a common target of many bacterial pathogens through exotoxin secretion. C2 toxin of Clostridium botulinum was the first to be described to ADP-ribosylate actin (33). ADP-ribosylating activity has been extended for the iota family, which includes Clostridium perfringens iota toxin, Clostridium spiroforme CST, and Clostridium difficile CDT, and for Bacillus cereus vegetative insecticidal protein (VIP) (review in references 34 and 35). C3 of C. botulinum was the first toxin to be shown to act as a Rho GTPase ADP-ribosylating toxin (36), but numerous other protein toxins produced by bacteria have been since shown to deregulate Rho family GTPases through either mimicking or covalently affecting GTPases, mainly by ADP-ribosylation (37, 38). Those alterations of the Rho GTPases induce a massive disruption of actin that is used by enteric pathogens to cross the epithelial barrier. Interestingly, Rho GTPase alterations are so common that they constitute a danger signal sensed by NOD1 activation (39). EF has never been shown to have ADP-ribosylating activity (for a review, see reference 23), and has never been suggested to have any direct effect on actin, although a close evolutionary path between Clostridium and Bacillus toxin families is suggested by sequence analysis (34). At the amino acid sequence level, Clostridium and Bacillus toxins share 80 to 85% identity within the iota family (which includes iota toxin, CST, and CDT), 31 to 40% identity between C2 and iota family toxins, and 26 to 30% identity between PA and clostridial binding components (34). Furthermore, these bacteria are Gram-positive, spore-forming bacilli commonly found in soil.

ET induces significant morphological and cytoskeletal changes in macrophages (11, 24), primary human microvascular endothelial cells (HMVEC) (12), and neutrophils (40). Some morphological differences have been observed between cells after ET treatment, such as formation of filopodial protrusions in mammalian cells (41) versus reduction in macrophages (24), or cell rounding (24, 41). Flattened cell morphology was observed in HMVEC after incubation with ET (12). In any case, every cell type had reduced spread morphology, a lowered F-actin content, and actin redistribution to the cell margin in a time-dependent manner (11, 24, 40, 41).

It is also interesting that EF also affects RAB11, another GTPase that controls intracellular vesicular trafficking, and disrupts Notch/DLL4 signaling at the adherens junctions (15). EF acts on RAB11 through the increase of cAMP and protein kinase A (PKA) and EPAC–RAP-1 activation. It is also interesting that the cholera toxin ADP-ribosylates the stimulatory G protein (G_sα), which causes a pathological intracellular increase of cAMP, and RAB11 disruption through PKA and EPAC–RAP-1 (42). How cAMP increase affects RAB11 regulation is still unknown, but this may explain some effects observed on the cytoskeleton organization (43).

So far, the slight but consistent effects of ET on various cytoskeleton cell types could not be quantified, because no technology was developed to address this question. We have used the cell micropatterning technique developed in the field of cell biology to quantify the effects of ET. The idea of exploring cytoskeleton-pathogen interactions by micropatterning is not completely new, as it has been previously used to address host cell-Salmonella interactions (26). However, these studies were performed on HeLa cells, which are human transformed cells. Even if HeLa cells are easy to pattern and have some value in the field of cell biology, they are far from being relevant for studying host-pathogen interactions in anthrax. Moreover, transformed cell lines behave differently from BMDMs for amoeboid migration, suggesting that HeLa and BMDM cytoskeletons bear significant differences in terms of organization (44). We have then extended the technique of patterns on BMDMs by adapting a protocol on commercial micropatterns.

Here, we show that ET induces a collapse of F-actin, packed in the center of the BMDMs. This actin disruption induces a loss of adherence, decrease of cell surface, and blebs in the cell center (see Fig. S2 in the supplemental material).

It is not clear if the changes of nuclear shape are due only to cytosolic actin reorganization and the disruption of F-actin cables, followed by the pressure of actin in the cell center, or if they also involve the nucleoskeleton. It has already been shown that cytosolic constraints affect nuclear shape and finally chromatin organization and cell cycle (45). More interestingly, cell geometry and nuclear shape can modulate gene expression (46). The effects of ET on BMDM nuclear shape may also participate in the broad transcription alterations induced by ET (11).

A close look at the effects that we show here may explain numerous alterations previously described in phagocytosis (24) or cytokine production after intoxication by ET (18). Actin disorganization may also affect other general cellular functions, such as protein synthesis and sorting, compartment organization, and cell cycling. Macrophages play a crucial role in the immune system of scavenging pathogens and pollutants, presenting antigen to lymphocytes, and regulating the immune system by producing cytokines.

From a therapeutic perspective, it would also be of great interest to decipher how ET alters cytoskeleton organization.

Eventually, our technique can prove powerfully effective to analyze and quantify slight cytoskeleton alterations induced by bacterial toxins or other virulence factors. As micropatterned BMDMs can be affected by many different virulence factors, they represent a valuable new tool for quantitative analysis of host-pathogen interactions.