Physical Sequestration of Bacillus anthracis in the Pulmonary Capillaries in Terminal Infection Free

Abstract

Anthrax is an acute, toxin-associated infection caused by the bacterium Bacillus anthracis [1]. The cause of death still remains elusive, although different hypotheses have been put forward and tested. The lethal effects of the toxins, and especially the lethal toxin, have been implicated [2]. Septic shock has been suggested, with death resulting from an excessive host systemic inflammatory response originating from the effects of the toxins, from the growing bacteria, or both sources [3].

Among animal models of anthrax [4], the mouse is exquisitely sensitive to infection with encapsulated B. anthracis strains, even in absence of toxin production [4]. Thus, the polyglutamate capsule plays a major role in the high virulence of B. anthracis in this model. However, its exact contribution remains poorly understood.

Using bioluminescent encapsulated nontoxinogenic B. anthracis, we have previously shown that spores germinate and multiply at the site of entry (cutaneous inoculation site, nasal associated lymphoid tissue, and Peyer patches for cutaneous, inhalational, and digestive infections, respectively) [5]. After rapid migration to the draining lymph nodes, bacteria disseminate into the blood and enter the spleen, where they multiply, before reaching the lung by the hematogenous route [5]. Death then occurs in a few hours. Interestingly, irrespective of the route of infection—inhalational, cutaneous, or digestive—the lung is the main terminal infected organ [5].

To address the events occurring in the lung at the terminal stage of infection, we characterized pulmonary lesions after a cutaneous inoculation (ie, in the absence of primary infection in the respiratory sphere). In this way, we could focus on common pathophysiological mechanisms that occur in any form of anthrax. Infection was performed with an encapsulated nontoxinogenic B. anthracis to explore the mechanisms independent of toxin effects. In this study, we provide evidence that the high pathogenicity of B. anthracis in the absence of toxins is most probably linked to lung tissue injury resulting from a size-dependent blockade of alveolar capillaries by the fast-growing encapsulated bacilli and its pathophysiological consequences.

METHODS

B. anthracis Strains and Mice Infection

OF1 mice (Charles River, L′Arbresle, France) were used for all infection experiments. Flk1^GFP/+ mice were used for capillary morphometric analysis. B. anthracis strains used were the bioluminescent encapsulated 9602P(ΔpagA)-lux strain (50% lethal dose, < 25 spores subcutaneously) [5] and the nonencapsulated 9602R-lux strain (pXO1⁺, pXO2⁻) [6]. All animal experimental protocols were approved by the Institut Pasteur Safety Committee and Animal Experimentation Ethics Committee (CETEA-2013-0088/MESR-01168.01).

Cutaneous infection was performed by injecting 10 µL of spore suspension (mean level [±SD], 4.03 ± 0.16 log₁₀ colony-forming units) into the ear dermis of 14 mice as previously described [5]. Bioluminescence images were acquired as previously described [5].

Histological and Ultrastructural Analyses

Histological, immunohistochemical, and ultrastructural analyses were performed as described previously [7, 8]. A complete description can be found in the Supplementary Data.

Hypoxia Detection

In situ tissue hypoxia was detected using Hypoxyprobe-1 kit (Hypoxyprobe INC, Burlington, Massachusetts) as indicated by the manufacturer. A complete description can be found in the Supplementary Data. Blood pH was determined using the Vetstat electrolyte and blood gas analyzer (Idexx Laboratories, Westbrook, Maine).

Two-Photon Imaging and Morphometric Analysis of Lung and Brain Capillaries

Two-photon excitation fluorescence imaging of control noninfected Flk1^GFP/+ mice was performed, and 3-dimensional capillary diameter analysis was carried out in the lung and brain, using IMARIS (Imaris Bitplane, Zurich, Switzerland) software. A complete description can be found in the Supplementary Data.

Statistical Analysis

Statistical significance was determined by the Student t test.

RESULTS

Critical Lung Lesions at the Terminal Phase of B. anthracis Cutaneous Infection

At the late stage of a cutaneous infection with spores of an encapsulated nontoxinogenic B. anthracis bioluminescent strain, when mice display the first clinical signs of illness and bioluminescent signal was high in the lung [5], the lung displayed critical lesions (Figure 1A), characterized by neutrophils and hemorrhages in alveolar spaces, and encapsulated bacteria invading alveolar walls and spaces. In alveolar spaces, bacteria were preferentially located in close apposition to the alveolar surface. In bronchi and bronchioles, bacteria were closely apposed to the luminal surface, grouped in small colonies. Bacteria were also observed in large pulmonary blood vessels, either along the vascular endothelium, or they formed microcolonies included in an amorphous material that was stained with the polyclonal anticapsule antibody.

Figure 1.

Lung parenchyma lesions and intracapillary sequestration of encapsulated Bacillus anthracis at the terminal phase of anthrax. A, Lung tissue samples were obtained at the terminal stage of infection (22 hours after cutaneous inoculation into the ear pinna), when mice began to harbor clinical signs of illness and the bioluminescence signal was high in the lungs. At low magnification (a), no histological lesion was detected. At higher magnification (b), round-shaped clear spaces (diameter, <5 µm) were identified in the alveolar walls (arrowheads, also visible in c); hemorrhages (black stars) and rare neutrophils (arrows) were detected in alveolar spaces. The bronchial/bronchiolar epithelial cells (c) displayed cytoplasmic microvacuolation, leading in some cells to formation of giant vacuoles displacing the nucleus at the cell periphery and to cell degeneration. Rare neutrophils were also detected in bronchial/bronchiolar epithelium (arrows). Immunohistochemical analysis with an anticapsule polyclonal antibody (d–h) revealed the presence of many bacteria in close apposition to alveolar cells (e; Δ, accumulation of capsular material in the alveolar space), bronchiolar epithelial cells (f; arrowhead, rare erythrocytes could be entrapped in bacterial microcolonies), and endothelial cells from large arteries and veins (g and h; arrowheads, bacteria along endothelial cells). In these large vessels, bacteria can form microcolonies attached to the endothelial lining that contain entrapped erythrocytes and other blood cells (h). B, Ultrastructural analysis was performed when infected mice displayed lung bioluminescence but no clinical signs. Bacteria (arrowheads) are inside the alveolar capillaries surrounded by a large clear halo (thickness, 0.5–2.5 µm) representing bacterial capsule as described in spleen [9]. The mean diameter (±standard error of the mean) of encapsulated bacilli, measured using ImageJ software, on 50 bacteria randomly selected in the different images was 2.73 ± 0.40 µm. This is most probably an underestimation, owing to the dehydration steps during histological analysis. The intraluminal space available for erythrocyte (white star) transit was limited. All experiments were performed at least twice on groups of at least 3 mice. Abbreviations: Al, alveoli; En, endothelial cells.

Open in new tabDownload slide

Ultrastructural analysis at an earlier stage of the pulmonary colonization process, when the lung bioluminescent signal was high but the mice did not yet display any clinical signs (Figure 1B), showed encapsulated bacteria inside the lumen of alveolar capillaries, narrowing or obstructing the vascular space available for erythrocyte transit. Bacteria were round to spindle shaped, depending on the section plan, and were surrounded by a large clear halo representing the bacterial capsule.

Size-Related Trapping of Blood-Circulating Encapsulated B. anthracis in Alveolar Capillaries

We hypothesized that blood-circulating encapsulated bacilli were trapped in the lung capillaries at this stage of infection because of their physical size. Bacteria at the bacteremia stage of infection (18–20 hours [7]) were in chains of 2–8 bacterial cells per chain (mean length [±SEM], 12.2 ± 0.7 µm; n = 48), with a mean (± SEM) of 4.2 ± 0.2 bacterial cells (n = 48; Supplementary Data).

We mimicked this bacteremic stage of infection by inoculating intravenously bioluminescent encapsulated nontoxinogenic B. anthracis bacilli of a similar size (11.8 µm; Figure 2A). In <1 minute, a strong bioluminescence signal was detected in the lung, increasing in intensity at 5 minutes with appearance of bioluminescence in the spleen. Gram staining showed many bacteria in lung capillaries without any lesion. Bioluminescent nonencapsulated bioluminescent bacilli of a similar size (Supplementary Data) and encapsulated bacilli of smaller size (mean length, [±SEM] 3.6 ± 0.1 µm, with a mean [±SEM] of 1.2 ± 0.1 bacilli/chain; n = 40) were not detected in the lung in the same period (Figure 2A).

Figure 2.

A, Sequestration in alveolar capillaries is dependent on bacterial size. Encapsulated bioluminescent Bacillus anthracis bacilli (mean length, 11.8 or 3.6 µm) were injected intravenously and the bioluminescence signal acquired at the indicated time points. The injected bioluminescent bacilli produced in vitro were in (1) mean chains (±standard error of the mean [SEM]) of 3.9 ± 0.1 bacilli (n = 65), with a mean length (±SEM) of 11.8 ± 0.3 µm (n = 65; range, 5–16 µm) and a mean diameter (±SEM) of 3.3 ± 0.02 µm (n = 204), or (2) mean chains (±SEM) of 3.6 ± 0.1 µm, with a mean (±SEM) of 1.2 ± 0.1 bacilli/chain (n = 40). Lung tissues were processed for histological analysis 5 minutes after inoculation, for hematoxylin-eosin (HE) and Gram staining (arrowheads, bacteria revealed by Gram stain). Data are representative of at least 2 experiments with groups of 3 mice. Many 11.8-µm bacteria were detected in the lung capillaries without any lesion, whereas few 3.6-µm bacteria were found and, in each case, as small aggregates.

Open in new tabDownload slide

B, Hypoxic zones are prominent in lung endothelial cells and brain parenchyma at the terminal phase of anthrax. Lung and brain tissue specimens were collected when clinical signs appeared; the hypoxic probe pimonidazole was revealed by immunofluorescence associated with spectral deconvolution microscopy. Hypoxia was detected in the lung and, more specifically, in the endothelial cell lining (white arrowheads) of small to large blood vessels (BV). Multiple foci of hypoxia were also detected in the cerebral cortex, hippocampus, and cerebellar cortex, with hypoxia detected in neurons and peripheral neuropil. Few bacteria were detected in the brain and were randomly distributed in the vascular network, with no obvious colocalization with hypoxic zones. The central panel shows comparison, at low magnification, between hypoxia labeling and Gram staining of serial sections of cerebral cortex (arrowheads, bacteria revealed by Gram stain). C, Proposed events occurring in the lung at the terminal phase of anthrax. During the bacteremic phase (early terminal phase), encapsulated B. anthracis with a mean length of 12 µm circulate in the venous bloodstream and are sequestered in the lumen of lung alveolar blood capillaries, consistent with the lung acting as a filter. Because of their size, they cannot transit through the capillary network. The bacilli colonize and multiply locally, producing CO₂, consuming O₂, and altering the circulation of erythrocytes; these phenomena interfere with the physiological function of gas exchange in the lung and lead to acidosis (a link between hypoxia and acidosis is a well-known phenomenon in intensive care units [10]). At the late terminal phase, local hypoxia at the alveolar endothelial lining results in cell injury, tissue lesions, exit of bacteria, and hemorrhage in the alveolar spaces; this leads to further aggravation of the perturbation of lung function and to formation of numerous hypoxic zones in the brain parenchyma, leading to death. Thus, in the absence of toxins, the high pathogenicity of B. anthracis in mice is linked to sequestration of encapsulated bacilli in the lumen of the alveolar capillaries and its consequences. In the case of anthrax, the toxins produced by the encapsulated B. anthracis will add their deleterious effects to this toxin-independent lung-targeted pathogenicity [11]. This sequestration phenomenon might be present in infectious processes with other pathogens.

Open in new tabDownload slide

Thus, bioluminescent chains of encapsulated bacilli of a similar size to those circulating during a natural infection were immediately trapped in the lung capillaries, while smaller-sized encapsulated bacilli or nonencapsulated bacilli flowed freely through.

To provide a complete view of the parameters involved in this host-pathogen interaction, we performed a morphometric analysis of the green fluorescent protein–expressing alveolar capillaries by 2-photon microscopy. The alveolar capillary diameter was estimated to be 2–11 μm (n = 300). These results suggest that the small size of normal alveolar capillaries may contribute to in situ bacterial trapping.

Hypoxia in Lung Endothelium and Brain

Physical sequestration of encapsulated B. anthracis within alveolar capillaries coupled with lung lesions raised the hypothesis of perturbations of pulmonary gas exchanges. We thus mapped the hypoxic areas in the lung and brain (Figure 2B). Hypoxic areas in the lung were restricted to endothelial cells lining alveolar capillaries and larger blood vessels. In the brain, multiple foci of hypoxia were detected in the cerebral cortex, hippocampus, and cerebellar cortex. Few random bacteria were detected, and no colocalization of bacteria with hypoxic zones was observed. Mean blood pH (±SEM) was decreased in the B. anthracis–infected animals, relative to that in noninfected animals (7.27 ± 0.05 [n = 10] vs 7.42 ± 0.03 [n = 4]; P = .01).

DISCUSSION

Irrespective of the route of a lethal infection (ie, cutaneous, inhalational, and gastrointestinal), the lung is the final main target of encapsulated B. anthracis before death, with or without the presence of toxins [6, 8]. Our study shows that bacterial seeding of the lung vascular space at the terminal stage of infection occurs through physical sequestration of large, blood-circulating encapsulated bacteria within the lung alveolar capillaries.

The respective size of blood circulating bacteria (13 µm) and the alveolar capillary diameter were compatible with this phenomenon. Indeed, encapsulated bacilli of a similar size, when inoculated intravenously, were immediately trapped within the pulmonary vascular bed. In contrast, encapsulated bacteria of shorter length were able to transit through the lung vasculature. Wiggs et al, studying cell transit through rabbit alveolar capillaries, reported that 86% of 3.08-µm beads passed through the lungs, whereas only 15% of 5.85-µm beads were able to transit [12].

Alveolar capillary diameter is remarkably similar in different species, with mean values (±SEM) of 6.3 ± 1.0 µm in the rat, 6.02 ± 1.95 µm in the rabbit, and 6.03 ± 1.38 µm in the dog [13, 14]. Considering the combined length of the bacterial chain and the diameter of encapsulated B. anthracis, alveolar capillary diameter will not favor transit of bacteria and red blood cells through the alveolar capillary network. This phenomenon has been suggested for megakaryocyte trapping in the lung microvessels [15]. Restriction of leukocyte circulation through pulmonary capillaries has also been suggested to be related to their respective diameter and deformability and to account for the marginated leukocyte pool [14].

The capsule surrounding B. anthracis played a role in this sequestration phenomenon, as its absence decreased the amount of trapped bacteria. Close interactions have been described between encapsulated B. anthracis and the liver sinusoidal endothelium [16]. Nevertheless, sequestration in the lung microvasculature was of a different mechanism, as encapsulated bacilli of a small size were still able to transit through the lung vascular bed.

The presence of bacteria and erythrocytes in the alveolar and bronchiolar spaces strongly suggests that the integrity of the blood-air barrier (alveolar epithelium and endothelial lining) is altered by the encapsulated nontoxinogenic B. anthracis. At an earlier stage, encapsulated B. anthracis and capsular material were present within the alveolar capillaries, almost completely blocking the vascular lumen; erythrocytes showed an altered morphology, with cell deformation around the encapsulated bacilli, as previously described in the liver sinusoids [16]. It was tempting to hypothesize that such intravascular filling with encapsulated bacteria and capsular material leading to potential decrease in erythrocyte transit would alter the functioning of the blood-air alveolar barrier, thus limiting gas exchanges and resulting in poor blood oxygenation and pH decrease. Indeed, we detected blood pH decrease and multifocal hypoxia in pulmonary endothelial cells. Furthermore, one may speculate that local bacterial metabolism will further decrease oxygen availability for the host cells, through oxygen bacterial consumption, and lead to increased production of CO₂ by the multiplying bacteria. The link between hypoxia and acidosis is a well-known phenomenon in intensive care units.

Local consequences of these phenomena would be alteration of the integrity of the alveolar barrier, thus furthering the intensity of hypoxia to devastating effect, as we showed in parenchyma of the brain, the organ most sensitive to hypoxia, displaying numerous foci of hypoxia. This is in accordance with our observations that mice displayed acute neurological symptoms at the terminal stage of infection, just before death.

As the diameter of alveolar capillaries in human is similar (7.5 µm at maximal dilation [14]), one may consider B. anthracis trapping within the lung vascular bed to be also relevant to human anthrax. Indeed, aggregates of bacilli were obstructing lung capillaries and venules in victims of the Sverdlovsk 1979 outbreak, with a gradient of bacteria from blood to air space, suggesting a hematogenous origin rather than an endobronchial origin [17]. This intracapillary sequestration and local gas modifications may be of further importance during anthrax, as an increased local CO₂ concentration may also lead to increased capsulation and toxin secretion [1] and, hence, to increased pathophysiological effects.

In conclusion, we propose that, in the absence of toxins, the high pathogenicity of B. anthracis in mice is linked to sequestration of encapsulated bacilli in the lumen of alveolar capillaries (Figure 2C), with the lung acting as a filter. Other virulence factors, such as anthrolysin O and proteases, may also play a role in the local lung pathogenesis. In the case of anthrax, the toxins produced by the encapsulated B. anthracis will add their deleterious effects to this toxin-independent lung-targeted pathogenicity [2]. This has therapeutic implications: by decreasing the lesions of the lung endothelium during the bacteremic phase of anthrax, the chance of survival among infected patients might increase. Interestingly, a correlation between the lung bacilli load and the duration of antibiotic treatment was observed in the victims of the US anthrax attack in 2001—early initiation of antibiotic treatment probably prevented intravascular bacterial growth [18]. This sequestration phenomenon might be a more general process during infection with other pathogens.

Notes

Acknowledgments. We thank Franck Verdonk, for scientific discussions on hypoxia pathophysiology; Pierre Rocheteau, for his technical help to detect tissue hypoxia; Alejandro Piris Gimenez, who was involved in the initial, preliminary experiments; and Michèle Mock, for openness to and availability for fruitful discussions during the development of this study and for her great contribution to reviviscence of research on B. anthracis and anthrax, a truly Pasteurian activity, at the Institut Pasteur, which led to the constitution of a high-quality world-wide scientific network with far-reaching consequences, both at the fundamental and applied scientific levels and because of its societal usefulness in health protection and defense preparedness.

Disclaimer. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Financial support. This work was supported by the Agence nationale de la recherche-Accompagnement spécifique des travaux de recherches et d'innovation Défense (ANR-11-ASTR-022-01 to P. L. G., G. J., and J.-N. T.).

Potential conflicts of interest. All authors: No reported conflicts. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.

References

Author notes