Biofilm Formation by Pneumocystis spp.

Abstract

Pneumocystis spp. can cause a lethal pneumonia in hosts with debilitated immune systems. The manner in which these fungal infections spread throughout the lung, the life cycles of the organisms, and their strategies used for survival within the mammalian host are largely unknown, due in part to the lack of a continuous cultivation method. Biofilm formation is one strategy used by microbes for protection against environmental assaults, for communication and differentiation, and as foci for dissemination. We posited that the attachment and growth of Pneumocystis within the lung alveoli is akin to biofilm formation. An in vitro system comprised of insert wells suspended in multiwell plates containing supplemented RPMI 1640 medium supported biofilm formation by P. carinii (from rat) and P. murina (from mouse).Dramatic morphological changes accompanied the transition to a biofilm. Cyst and trophic forms became highly refractile and produced branching formations that anastomosed into large macroscopic clusters that spread across the insert. Confocal microscopy revealed stacking of viable organisms enmeshed in concanavalin A-staining extracellular matrix. Biofilms matured over a 3-week time period and could be passaged. These passaged organisms were able to cause infection in immunosuppressed rodents. Biofilm formation was inhibited by farnesol, a quorum-sensing molecule in Candida spp., suggesting that a similar communication system may be operational in the Pneumocystis biofilms. Intense staining with a monoclonal antibody to the major surface glycoproteins and an increase in (1,3)-β-d-glucan content suggest that these components contributed to the refractile properties. Identification of this biofilm process provides a tractable in vitro system that should fundamentally advance the study of Pneumocystis.

Organisms in the fungal genus Pneumocystis cause an often-lethal pneumonia in mammals with a compromised immune status. To date, there have been five formally described species, although all mammalian species examined to date appear to host at least one Pneumocystis species. Those that have been described are P. jirovecii from humans (20), P. carinii (53) and P. wakefieldiae (10, 11) from rats, P. murina (30) from mice, and P. oryctolagi (16) from rabbits.

Limited progress has been made in understanding the life cycle, transmission, and natural history of any Pneumocystis species, due in large part to the absence of a continuous in vitro culture system. Although the incidence of frank pneumonia caused by these organisms has decreased in developed countries such as the United States and European countries, mounting evidence points to new niches being exploited by these fungi. The presence of P. jirovecii in patients with underlying chronic diseases such as chronic obstructive pulmonary disease has been suggested to be a comorbidity factor (42-44). A series of recent case reports have identified P. jirovecii as a significant cause of infection in patients being treated with tumor necrosis factor alpha inhibitors for rheumatoid arthritis and other chronic diseases (22, 25, 28, 54). Other studies in animals and humans suggest that Pneumocystis organisms are also present in the lungs of nonimmunocompromised hosts (24, 60). In contrast, the prevalence of Pneumocystis pneumonia in developing countries, such as those in sub-Saharan Africa, remains high and poorly controlled (37, 46). The strategies used by these organisms to grow and survive in the context of an intact or debilitated host defenses are largely unknown.

Morphological analysis of histologic sections of human and rodent lung tissue, redolent of layers within a microbial biofilm (Fig. 1), as well as recent studies showing the ability of P. carinii to thrive under reduced oxygen conditions in vitro (26) led us to posit that Pneumocystis spp. follow a pathway to biofilm formation within the host lung, similar to that of other fungal pathogens. Since oxygen availability has been shown to influence the metabolic activity in different zones within Pseudomonas aeruginosa biofilms (61), it seemed reasonable to suggest that the masses of Pneumocystis cells that accumulate within an infected mammalian lung lumen also are exposed to a similar oxygen gradient and subsequently have evolved survival mechanisms for hypoxic conditions.

FIG. 1.

P. jirovecii in apparent biofilm structures within an infected human lung. Grocott methenamine silver staining of postmortem lung from a patient with AIDS is shown. (A) P. jirovecii filling the alveolar lumen (arrow) and in an apparent biofilm formation (double arrows). Bar, 100 μm. (B) Higher magnification of P. jirovecii within the alveolar lumen. Note the close apposition of organisms resembling a biofilm structure. Bar, 10 μm. (C) P. jirovecii in the alveolar lumen, illustrating a biofilm-like structure with a base (horizontal, packed organisms) with a protuberance into the lumen. Bar, 10 μm.

To explore our hypothesis, we devised an in vitro system that supported the transformation of two different species of Pneumocystis into biofilm-like structures. The process of formation was well defined and reproducible, resulting in adherent films composed of unique morphological structures. That the biofilms held viable and infective organisms was confirmed by ATP content and fluorescent viability staining and the ability to cause pneumonia in immunosuppressed rodent lungs. The establishment of a biofilm system for Pneumocystis will facilitate studies of life cycle, survival mechanisms, and host cell and parasite interactions and a wide variety of applications heretofore unavailable to the research community.

MATERIALS AND METHODS

Organisms.

Pneumocystis carinii (from rats) and P. murina (from mice) were used immediately after isolation and purification from infected rodent lungs or from cryopreserved aliquots. Infected lungs were removed from rats or mice after approximately 8 weeks of chronic administration of dexamethasone as an immunosuppressant (14). Organisms were released from lung tissue by homogenization followed by washing, hypotonic lysis (to reduce erythrocytes and other host cells), and then centrifugation at 1,000 × g. The organisms were enumerated by tinctorial methods using a rapid Giemsa stain, Hema3 (Fisher Diagnostics, Middletown, VA), and expressed as the number of Pneumocystis nuclei/ml (13). Each nucleus represents one organism. Organisms were cryopreserved in RPMI 1640 containing 7.5% dimethyl sulfoxide (7). Each batch was assayed for ATP content and plated on nutrient agar (36°C), on Sabouraud dextrose agar (25°C), and in RPMI 1640 with a 20% serum supplement (36°C).

Biofilm formation.

RPMI 1640 medium (Gibco Invitrogen, Grand Island, NY) with penicillin (200 U/ml)-streptomycin (200 μg/ml)-amphotericin B (0.5 μg/ml) (Cellgro, Herndon, VA), vancomycin (5 μg/ml) (Fluka, Sigma-Aldrich), 20% calf serum (Atlanta Biologicals, Lawrenceville, GA), and vitamins and amino acids as previously described (14) and containing the desired organism concentration was placed in one of several types of tissue culture insert wells, including polyethylene terephthalate (PET) track-etched membrane cell culture inserts (no. 35-3090; Becton Dickinson Inc., Franklin Lakes, NJ), Transwell (no. 3460; Costar, Corning, Corning, NY), and Millicell-CM hydrophilic polytetrafluoroethylene (PTFE) membranes (PICM-01250 [12 mm] and PICM-03050 [30 mm]) and Millicell-HA insert wells with mixed cellulose ester membranes (PIHA-01250 [12 mm]) (Millipore Corp., Billerica, MA), suspended in each of a well in a 6- to 12-well multiwell polystyrene plate or a 24-well polystyrene plate. The same medium was added to the outside well. Lab Tek II chamber slides (Nunc, Rochester, NY) were used in some cases for microphotography. After 24 h of incubation at 36°C in a water-jacketed incubator with 5% CO₂, the medium in the outside well was aspirated and discarded. The medium within the insert well was carefully removed with a pipette. The biofilm was washed to remove nonadherent organisms by the addition of 500 μl of the medium or phosphate-buffered saline (PBS). The media in the outside well and insert were replenished. This point was considered “day 1” for the biofilm studies. “Day 0” refers to the measurement of the inocula used to seed the inserts. The medium in each outside well was replaced daily to renew nutrients and remove waste products.

Biofilm quantification.

Biofilm formation was measured by a ATP bioluminescence assay used extensively by our lab to assess the viability of Pneumocystis (5, 7-9, 27). The assay is similar in concept to the XTT reduction assay used for fungal biofilm growth, which assesses mitochondrial function, but is a more sensitive and rapid method. The bioluminescence assay is based on the ATP-driven reaction of the substrate luciferin with the enzyme luciferase, which results in the evolution of light as relative light units (RLU) (ATPlite detection kit; Perkin-Elmer). The light evolved is proportional to the amount of ATP in the sample and, correspondingly, to the number of cells from which it was extracted (5). Both unattached organisms found in the supernatant (planktonic) and those attached to the substrates (sessile/biofilm) were measured. Three wells were used for each time point. For sampling of planktonic organisms, supernatants were removed from each insert by gentle pipetting and centrifuged to pellet the cells, followed by addition of 100 μl of ATP lysis solution. The biofilms were then washed with PBS, and 100 μl of ATP lysis solution was added directly onto the insert membrane, mixed, and allowed to incubate at room temperature for 10 min. A 50-μl volume from each phase was removed and placed into wells of a 96-well white plate (Greiner Scientific, Ocala, FL) containing the luciferin-luciferase (substrate-enzyme) reaction buffer. Plates were incubated at room temperature with gentle agitation for 60 min to permit full evolution of light, which is stable for up to 5 h. Luminescence was measured using a POLARStar Optima plate reader (BMG Labtechnologies) with an output of RLU. Quench controls, to control for unanticipated interference of additives with the luciferin-luciferase reaction, were included in each assay.

Phase-contrast and bright-field microscopy.

Biofilms were monitored on a daily basis, and formation was photodocumented using a Nikon Diaphot inverted phase microscope equipped with a SPOT image capture system (Diagnostic Instruments, Inc.). Morphological assessments were performed under phase-contrast, bright-field, and fluorescent microscopy using an Olympus BH2 microscope. For bright-field examination, supernatants were removed and biofilms were washed twice with PBS to remove any residual planktonic cells. The biofilms were harvested by scraping the inserts with a pipette and depositing the material on a microscope slide and then were allowed to air dry. The slides were stained with a rapid Wright-Giemsa stain, Hema3 (Fisher Diagnostics, Middletown, VA), for 1 min in each reagent, a modification of our standard method in which staining is for 10 to 20 s each. Magnification was assigned by calibration with an objective micrometer (Nikon Inc., Melville, NY).

CSLM.

Pneumocystis biofilm structure and viability were evaluated at various time points after staining with 20 μM FUN-1 and 10 μM concanavalin A (ConA)-Alexafluor 488 (Molecular Probes, Invitrogen Detection Technologies, Carlsbad, CA) following procedures established for evaluation of Candida biofilms (4), The dyes were added to the culture media, and after 30 min at 37°C, the medium was removed, cells were washed twice with PBS, and the biofilm was embedded in Fluromount-G (Southern Biotech). The FUN-1 dye is processed during the incubation period into strikingly red fluorescent inclusion bodies. This conversion of the dye is accompanied by a reduction in the diffuse green cytoplasmic fluorescence. In combination with the ConA-Alexafluor 488 staining, live cells will exhibit discrete sequestered red fluorescence that will appear orange to yellow in combination with the green fluorescence of the ConA counterstain. Confocal scanning laser microscopy (CSLM) was performed with an LSM 510 laser scanning microscope (Zeiss, Germany). ConA staining was observed using the 488-nm laser line and a 505- to 550-nm band-pass emission filter. FUN-1 staining of live cells was visualized using the 543-nm laser line with a 560-nm long-pass filter. The gain was set to unstained organisms as “black” to control for any autofluorescence. To determine the three-dimensional biofilm structure, a series of horizontal (x-y) optical sections with a thickness of 0.18 μm at 0.50-μm intervals was taken throughout the biofilms. Eight-bit confocal images of green (ConA) and red (FUN-1) fluorescence were obtained using multitracking, which reduced the likelihood of bleedthrough. The images were captured and processed using Zeiss LSM Image Browser software.

Surface glycoprotein staining.

All Pneumocystis species contain a family of glycoproteins that are a major antigenic component on the surface of all developmental forms of these organisms, the major surface glycoproteins (MSGs) (57). Biofilms were evaluated for the presence of MSGs by staining with a monoclonal antibody (RA-E7) directed to an epitope of the protein component of these molecules (35). Biofilms were recovered from the insert surface by scraping with a pipette tip, air dried, and fixed with acetone. The biofilms were probed for the expression of MSG by an indirect immunofluoresence assay as previously described (34). Briefly, the slides were blocked with 1% bovine serum albumin (BSA) in PBS (PBS-BSA) for 60 min and then incubated in a 1/500 dilution of RA-E7 in PBS-BSA for 60 min at 37C. The slides were washed in PBS and then reacted with an Alexafluor 594-labeled F(ab′)₂ fragment of goat anti-mouse immunoglobulin G (heavy and light chains) (Molecular Probes, InVitrogen, Eugene, OR) diluted 1/1,000 in PBS-BSA for 60 min at 37C. After washing, the slides were mounted with fluorescent mounting medium and examined under oil immersion with a Nikon E600 epifluorescence microscope. Controls included addition of the Alexafluor 594-labeled F(ab′)₂ fragment of goat anti-mouse immunoglobulin G conjugate directly on the biofilm preparations, monoclonal antibody RA-E7 alone, and PBS without additional reagents. No red autofluorescence was observed.

Effects of farnesol on the viability and formation of P. carinii biofilms.

Farnesol (Acros Organics, Thermo Scientific Fisher, NJ), a quorum-sensing molecule elucidated by Candida albicans, was added at the time of inoculation of P. carinii into PET inserts at a concentration of 100 μM (48) and then assayed for ATP content of the supernatant and biofilm cells over a 1-week period. Two separate experiments were performed with two batches of P. carinii. Results were expressed as the percent ATP versus that for untreated P. carinii inoculated in the same insert system. Statistics were evaluated using GraphPad Prism v. 4 (GraphPad Software, Inc., San Diego, CA).

Measurement of β-1,3-Glucan.

Measurement of the (1,3)-β-d-glucan content in P. carinii biofilms over time was conducted using the Glucatell kinetic assay, following the vendor's instructions (Associates of Cape Cod, Inc., East Falmouth, MA). The reaction is based on activation of factor G, a serine protease zymogen, which in turn activates the proclottin enzyme that cleaves pNA from the chromogenic peptide substrate, creating a chromophore that absorbs at 405 nm. The rate of optical density increase is measured against standards. At each time point, supernatants from duplicate wells were removed and inserts were washed twice with PBS. The membranes were then scraped with a pipette tip and the contents combined and air dried within a sterile biosafety cabinet. The dried samples were resuspended in 1 ml of 1 M sodium hydroxide and shaken at room temperature for 1 h. Samples were prepared as 10-fold dilutions in pyrogen-free water in test tubes certified to be free of glucan (Associates of Cape Cod). Glucan standards were prepared in the same manner according to the vendor's instructions. Samples were read at 2-min intervals for 1.2 h (40 cycles). The (1,3)-β-d-glucan was measured as slope/minute, raw data − blank, and calculated using a linear regression curve of standards from 25 to 200 pg/ml. The correlation coefficient for the studies presented was 0.992, which is within the acceptable range suggested by the vendor (>0.980). Glucan content was expressed as pg/ml. Statistical analysis was not performed since the duplicate wells were pooled to provide sufficient material for assay.

Passage of biofilms.

The ability to passage biofilms was evaluated using small (12-mm, Corning Transwell) and large (30-mm, Falcon, BD) inserts (0.4-μm pore size) that could both be contained in a six-well polystyrene plate. P. carinii was inoculated onto the small inserts and washed after 24 h, and the medium was replaced. Each small insert was then placed inside the larger insert well, to which 1 ml of the RPMI+ medium was added. The smaller insert hung within the larger insert, permitting communication between the supernatants but not the insert membranes. Both inserts were placed within a well of a six-well polystyrene plate. The outer well contained 2 ml of the RPMI+, which was replenished daily. The inserts were sampled for ATP content every 3 to 4 days and examined macroscopically for evidence of organism transfer and biofilm formation, by detection of the formation of a white film. The larger insert that was then populated with the first-passage organisms was removed and placed in another of the six-well polystyrene wells with fresh medium. An uninoculated small insert was then hung within the confines of the large insert that contained the organisms. A representation of this process is in Fig. S1 in the supplemental material.

Inoculation of passaged biofilms into P. carinii-naïve rats.

The ability of the passed organisms to induce infection in immunosupressed rats was evaluated. Biofilms grown on Millicell-CM 12-mm inserts were passed on day 15 of culture by transfer of the small insert to a large insert. After 1 week, the small inserts were removed and the entire contents of each of the large insert wells (passaged organisms) were harvested, pooled, and cryopreserved. One half of the pooled organisms were inoculated into each of three immunosuppressed Sprague-Dawley rats (3.35 × 10⁵ [log 5.52] nuclei each) as previously described (12). Uninoculated immunosuppressed rats served as controls. Immunosuppression continued for an additional 75 days. The rats were then sacrificed and the lungs processed for microscopic enumeration of P. carinii (7).

RESULTS

Insert composition and ATP content.

Several insert types from different manufacturers were assayed for their ability to support biofilms over a 2-week period (Table 1.). The data are expressed as the total ATP as RLU within a well, which included measurements of both the planktonic and biofilm cells and as a ratio of the ATP content of the adherent cells/total ATP. Inspection of the RLU for each insert type revealed various levels of decreases for the different inserts from day 1 to day 14, although by day 10, all systems experienced notable drops in ATP. Inserts composed of hydrophilized PTFE (Biopore-CM from Millipore) and Millicell-HA cellulose rapidly formed tightly adherent biofilms that were able to maintain ATP levels higher than those of P. carinii in other insert types over 10 days. The kinetics of biofilm formation among each insert type were generally similar in that the percent ATP content associated with the adherent populations (adherent/total ATP) generally increased or remained stable over the 2- to 3-week time period while the planktonic ATP levels decreased, with the exception of the PET inserts, where little biofilm formation was observed except at day 10. Of the two matrices that best supported biofilm formation, the PTFE inserts are suitable for addition of coatings such as collagen or vitronectin, are easily prepared for electron microscopy, and are compatible with cellular and/or fluorescent stains. The cellulose inserts are opaque and cannot be used for other applications, and they were not used in further experiments.

TABLE 1.

Effect of support matrices on the formation of P. carinii biofilms^a

Insert typeb	Day 1		Day 3		Day 7		Day 10		Day 14
	Total ATP (RLU, 103)	Adherent/total ATP	Total ATP (RLU, 103)	Adherent/total ATP	Total ATP (RLU, 103)	Adherent/total ATP	Total ATP (RLU, 103)	Adherent/total ATP	Total ATP (RLU, 103)	Adherent/total ATP
PET (Becton-Dickinson)	3,089.91	4.2	1,127.06	7.6	33.55	14.1	16.91	48.7	11.26	17.1
Polyester Transwell (Corning)	2,669.85	38.9	779.37	17.3	150.90	22.7	24.66	42.0	7.69	49.5
Millicell-CM, hydrophilic PTFE membranes	3,474.58	75.2	2,831.48	84.1	2,635.54	91.2	613.64	82.3	30.89	44.0
Millicell-HA cellulose insert	3,444.34	97.2	2,173.53	95.7	1,554.86	96.1	61.78	92.4	13.45	69.6
Lab Tek II chamber slide (Nunc)	3,568.70	89.2	1,127.06	7.6	491.80	23.6	87.94	52.8	20.31	68.8
Cell culture cluster (Corning)	2,667.23	81.5	779.37	17.3	54.15	19.6	6.58	4.3	5.90	73.3

^aA characterized batch (as described in Materials and Methods) of P. carinii from a single infected rat (06-2-23) was used as inocula for these experiments at an initial density of 5 × 10⁷/ml.

^bDetails on vendors and catalog numbers are in Materials and Methods.

The Pneumocystis biofilms could be observed macroscopically.

Titration of P. carinii inoculum numbers at 1 × 10⁶ to 1 × 10⁸ using the Millicell-CM PTFE inserts showed little differences in ability to form biofilms over 2 weeks, with the ATP levels increasing in the biofilm compartment over time with each inoculum (data not shown). In each case, the biofilms began as small adherent clusters, which grew larger over time regardless of inoculum size. The P. carinii biofilms could be observed macroscopically as thick white films with visible clumps, as illustrated when inoculated into Lab Tek II chambers (Fig. 2A) or the Millicell-CM PTFE inserts (Fig. 2B).

FIG. 2.

Macroscopic biofilm formation by P. carinii. (A) Duplicate wells showing biofilm formation after 1, 5, and 10 days on a Lab Tek II chamber slide. (B) PTFE inserts (12 mm) without inocula (left) and after 14 days of biofilm formation (right). Photographs were taken with a 35-mm SLR digital camera on the bench top. Bar, 1 cm.

Phase-contrast monitoring of the biofilms.

The biofilms of P. carinii and P. murina both began as microscopic clusters, which grew over time and coalesced with other clusters, often forming connecting branches that produced three-dimensional structures. Figure 3 illustrates this process for P. murina, but it should be noted that P. carinii biofilms followed the same progression. As shown in Fig. 3B and C, small microclusters of P. murina were observed during the first day and week of culture on insert membranes by inverted phase microscopy. These clusters continued to increase in size up until the end of the experiment, on day 10. Evident in Fig. 3D are the very large, expansive clusters that formed on the insert as a biofilm but also extended up into the supernatant.

FIG. 3.

Temporal development of P. carinii biofilms on inserts. Organisms were inoculated onto cell culture insert membranes at a density of 2 × 10⁷/insert, placed in multiwell plates, allowed to adhere for 24 h, washed with PBS, and placed back in the well with fresh medium. (A) Uninoculated membrane of insert. (B) Small microclusters (arrows) forming 24 h after washing; the inset shows a magnified microcluster. (C) Seven-day biofilm. Note larger branching clusters attached to membrane (arrows); the inset shows a magnified cluster. (D) Ten-day biofilm, showing an example of a large cluster (note lower magnification). Bars, 100 μm (A to C) and 200 μm (D).

The large clusters produced by either species were highly refractile and contained hundreds to thousands of organisms enmeshed within a matrix. Examples of clusters formed by P. murina are shown in Fig. 4A and B, and examples of those formed by P. carinii are shown in Fig. 4C. Many of the clusters formed elongated extensions, as seen in Fig. 4A and C. Smaller extensions could be seen on the ends of some branches, as seen in Fig. 4B. The more robust biofilms were formed by P. murina organisms inoculated as fresh isolates (as in Fig. 4A and B). The macroscopic films were quite impressive, spreading across the membrane in thick branching clusters often spanning several low-power microscopic fields. These clusters were autofluorescent when excited at 488 nm (Fig. 5).

FIG. 4.

Biofilm formation by P. murina and P. carinii with highly refractile large clusters. (A and B) Large clusters of P. murina from 10-day-old biofilms viewed unstained under phase-contrast microscopy. Note that the clusters spanned well beyond the microscopic field (×10). (C) Cluster of P. carinii at day 14 showing elongated extensions. White arrows indicate elongated morphologies in both P. murina and P. carinii biofilm structures. The white arrowhead in panel B indicates a small bud-like extension. Bars, 10 μm.

FIG. 5.

Autofluorescence of a P. carinii biofilm cluster. (A) Differential interference contrast image of a cluster with anastomosed extension from a 7-day-old biofilm; (B) same cluster showing autofluorescence at 488-nm excitation. Bar, 10 μm.

The process of biofilm formation.

After 24 to 48 h in the insert system, the organisms retained the density and staining characteristics that we and others have observed in standard tissue culture samples (see Fig. 7A). However, in the biofilm system, after 3 to 4 days a noticeable change began to occur (Fig. 6 A). The clusters, especially the cysts, took on a highly refractile property around the periphery of the cysts, with some focal points of refractility within the cysts that resembled spores (Fig. 6A). Over time (10 to 14 days), the refractile property encompassed the entire cluster (Fig. 6 B). After 2 to 3 weeks, elongated and refractile branches could be observed (Fig. 6C). The refractility coincided with poor uptake of the stain, as can be observed in the images shown in Fig. 7B to H.

FIG. 7.

The morphology of Pneumocystis changes dramatically during biofilm formation. (A) P. carinii from the supernatant of a 3-day-old standard short-term culture stained with Hema3, illustrating the differences in morphology from the biofilm structures. (B to H) Images were taken from 16-day-old biofilms inoculated with P. murina (obtained as a fresh isolate). The images were obtained from films on inserts that were scraped with a pipette tip, aspirated, air dried, and stained with Hema3, a rapid Wright-Giemsa stain, as described in Materials and Methods. Images were viewed under oil immersion. Bars, 10 μm. (B) P. murina cluster showing a cyst-like structure with a stalk (arrow). (C) A cyst in a chain with intracystic forms. (D) Refractile cluster beginning to form extensions (arrow). (E) Two large clusters forming a linkage or bridge structure (arrow). (F) A series of cyst-like structures forming a chain (arrows). (G) Cyst containing a structure which appears to be trophic forms undergoing binary fission or conjugation (arrow). (H) Cyst-like forms containing intracellular spore-like morphologies (arrows). (I to L) Images were taken from P. carinii biofilms at day 9 of the first passage. (I) Cluster mass seemingly comprised of three clusters that have joined in the center (arrow). (J) Cyst-like structures in a cluster with obvious intracellular contents resembling spores (arrows). (K and L) Chain-like forms. Bars, 10 μm.

FIG. 6.

Process of biofilm formation monitored by phase microscopy. Phase microscopy of wet mounts (unstained) of P. carinii biofilms is shown. (A) Cluster of P. carinii after 4 days in the Millicell-CM insert system. Arrows indicate focal points of refractility. (B) P. carinii after 10 days in the insert system. Arrows indicate peripheral refractility. (C) P. carinii after 14 days in the insert system. The refractile property encompassed the entire cluster, with elongated extensions beginning to occur (arrows). Bar, 10 μm.

The morphology of Pneumocystis changes dramatically during biofilm formation.

The images in Fig. 7 (except for Fig. 7A) were obtained from films on inserts that were scraped with a pipette tip, aspirated, air dried, overstained with Hema3 (a rapid Wright-Giemsa stain), and viewed under oil immersion (13). Biofilm images from both P. murina and P. carinii and primary and passaged biofilms are shown to illustrate the similar morphologies between the two species and between the primary and passaged structures. The images in Fig. 7B through H were taken from a 16-day P. murina biofilm. Those in Fig. 7I through L were samples from a passaged P. carinii biofilm after 9 days on the inserts. Immediately apparent from the images is the difference in staining patterns compared to organisms that were obtained from aspiration of standard supernatant cultures (Fig. 7A) versus the biofilms (Fig. 7B to L). Whereas most of the supernatant-derived organisms stained readily, showing blue cytoplasm with reddish-purple nuclei, the structures from biofilms excluded much of the stain, were highly refractile, and formed chains and elongated structures (Fig. 7F, K, and L). A striking feature was the linkages formed between organism clusters, which appeared to facilitate the anastamosis of the separate biofilm clusters into larger, more expansive mats (Fig. 7E and I). This morphology was common among the more mature biofilms (>7 days). Many of the sphere-shaped forms contained apparent spore-like structures that were highly refractile (Fig. 7C, G, H, and J). The structure shown in Fig. 7G contained what appeared to be organisms undergoing binary fission (or conjugation) within an enclosed sphere. Many of the organism chains had elongated tips (Fig. 7D). In other cases, clusters contained cyst-like structures that were recognizable but that were larger and more aberrant than typically observed in standard suspension cultures (Fig. 7B, C, G, and H). We propose adoption of the term “allomorphs” to describe these oddly shaped structures, as suggested for alternative forms of Histoplasma capsulatum (18).

CSLM.

FUN-1 is routinely used to assess the integrity of fungal biofilms (1, 2, 4). All cells, live or dead, take up the dye, resulting in a diffuse green fluorescence. Metabolically active cells rapidly process the dye, which causes a shift from green to bright red fluorescence that is sequestered in intracellular inclusion bodies. ConA-Alexafluor is used to highlight the mannose-containing extracellular matrix. Biofilms stained in this manner can be used to assess viability and for measurement of the biofilm thickness.

After 7 days of culture in the Lab Tek II chambers, the biofilms consisted of mostly metabolically active cells showing the red inclusion bodies (Fig. 8) or a yellow green fluorescence indicating dual staining with the ConA-Alexafluor stain. The green- or yellow-staining areas indicate mannose or glucose components of extracellular material or surface carbohydrates. The images also show tunnels throughout the structure (top and side), indicating that there is spatial heterogeneity in the three-dimensional structure. The thickness of the biofilm at day 7 was 14.59 μm, somewhat less than that of a typical fungal biofilm (e.g., 25 to 40 μm for a 72-h biofilm of Trichosporon asahii) (17), but this was not unexpected since Pneumocystis organisms do not appear to form the typical hyphal masses produced by other fungi in biofilms.

FIG. 8.

Merged orthogonal images of P. carinii biofilms with FUN-1 and ConA-Alexafluor. After 7 days of culture in Lab Tek II chambers, FUN-1 and ConA-Alexafluor 488 were added to the P. carinii culture medium and cells were incubated, washed, and mounted in Fluromount-G. The images were merged to produce this representation. The center panel shows viable organisms (red, arrows) overlaid with ConA-staining mannose residues (arrowheads) to produce the yellow color. The side panels show the structure and thickness of the biofilm, approximately 14.59 μm. Bar, 10 μm.

Effects of farnesol on the viability of P. carinii biofilms.

Exposure of C. albicans to 300 μM farnesol prior to establishment of a biofilm rendered the fungus incapable of making a biofilm (48). At a much lower concentration, 100 μM, farnesol was able to dramatically decrease ATP levels in P. carinii in both the supernatant and adherent compartments of the biofilm system to less than 2% of that in untreated controls after 7 days of exposure (Fig. 9).

FIG. 9.

ATP content of P. carinii biofilms exposed to farnesol. Farnesol (100 μM) was added to the inserts with the inocula. Biofilms were fed with medium containing the farnesol on a daily basis. Bars reflect the percent ATP content compared to untreated control biofilms over a 7-day period. Error bars indicate standard deviations.

Measurement of (1,3)-β-d-glucan.

The increasing refractility of the Pneumocystis biofilms was a prominent feature of biofilm formation. Since P. carinii is known to possess a glucan synthase gene, that for Gsc-1, and contains (1,3)-β-d-glucan in its cell wall (19, 32), we hypothesized that this compound could be a key component in the morphological changes that the organisms underwent during the process. The (1,3)-β-d-glucan content in duplicate P. carinii biofilms was measured over time using the Glucatell kinetic assay. Recently, the assay has been used for the diagnosis of P. jirovecii infection with apparent strong sensitivity and specificity (29, 39, 55). Figure 10 shows that the amount of glucan (pg/ml) steadily increased over time in the P. carinii biofilms, resulting in a 61% increase. The (1,3)-β-d-glucan in a single P. murina biofilm assay increased by 75% but was more erratic in the kinetics than for P. carinii (data not shown).

FIG. 10.

(1,3)-β-d-glucan content in P. carinii biofilms. Duplicate Millicell-CM inserts were harvested over a 3-week period and the (1,3)-β-d-glucan content measured using the Glucatell assay kit as described in Materials and Methods. Day 0 refers to the (1,3)-β-d-glucan content of the inoculum.

Biofilms produce an abundance of MSGs.

Biofilms reacted with a monoclonal antibody specific for Pneumocystis spp. and targeted to a common epitope within the P. carinii MSG family of proteins to produce a robust fluorescence (Fig. 11), suggesting that these surface glycoproteins may play a prominent role in the morphogenetic conversion to adherent biofilms. Indeed, these mannose- and glucose-containing proteins could be a major component of the extracellular matrix or could contribute to the refractile properties of the biofilms.

FIG. 11.

MSGs are abundant on the surface of P. carinii biofilms. Staining reactivity of monoclonal antibody RA-E7, directed to a protein epitope of the MSG family of antigens, with a cluster from a 7-day P. carinii biofilm is shown. Bar, 10 μm.

Inoculation of passaged biofilms into P. carinii-naïve rats.

The ability of the passed organisms to induce infection in immunosupressed rats was evaluated by inoculation of passed organisms into immunosuppressed rats (Table 2), the standard animal model protocol that our laboratory uses for cryopreserved organisms (12). The cyst-to-troph ratio of slightly more than 1:1 was lower than those reported for natural infections, which usually range from 1:5 to 1:20, but the levels of organism burdens were within the range for moderate infections (59), showing that the passaged fungi were viable and remained infective (59). The lower organism burdens may have been due to the lower numbers in the inocula used in the present study than in the standard model (3.35 × 10⁵ versus 1 × 10⁷).

TABLE 2.

Organism burdens of immunosuppressed rats inoculated with passaged P. carinii biofilms

Rat no.	Cysts per lung (log10)	Nuclei per lung (log10)
22	6.28	7.15
23	6.06	7.62
24	5.72	6.54

DISCUSSION

Biofilms are organized microbial communities enmeshed within a matrix of exopolymeric substances and attached to biotic or abiotic matrices. Fungal and bacterial pathogens exist predominantly within biofilms, rather than as free-floating, planktonic organisms. There are a myriad of advantages of biofilm formation for microbes, including the exchange of plasmids within the closely apposed community, reduced susceptibility to antimicrobial agents, resistance to host immune clearance, and dispersion of populations from aggregates in infection foci. A significant proportion of microbial pathogens causing human disease have been shown to use a biofilm strategy during the infective process (47). While bacteria have been the focus of most biofilm studies, increased appreciation of the widespread nature of these microbial communities have lead to several studies reporting this phenomenon in pathogenic fungi, including Cryptococcus neoformans (38), Aspergillus fumigatus (45), Trichosporon asahii (17), Candida spp. (21, 31, 36, 47, 49-52), and Blastoschizomyces capitatus (15).

Within the mammalian lung, Pneumocystis trophic forms adhere to type I pneumocytes and spread along these cells, which comprise over 90% of the alveolar surface. Pneumocystis cells grow in three-dimensional clusters that fill the alveolar lumen space (Fig. 1.). There, different developmental forms are in close apposition with one another and form layer upon layer of cells within a dense matrix that could be construed as a biofilm. We postulated that like other fungi, Pneumocystis spp. form biofilms, and this process could be established in an in vitro system.

In the present study, biofilms were formed by two species of Pneumocystis on inert matrices of insert wells suspended in supplemented RPMI 1640 medium. There were specific requirements for the composition of the matrix, with a hydrophilized PTFE membrane providing the most robust growth. Two species of Pneumocystis, P. murina and P. carinii, were evaluated since differences in ability to form biofilms have been reported to exist among Candida species (47). One difference between the two species was the observed ability of P. murina to form more extensive biofilms, especially from organisms that were freshly isolated from the immunosuppressed mouse lung, than their P. carinii counterparts, but the process of biofilm formation and the morphological structures formed were strikingly similar.

The structure of these biofilms resembled those of other fungi when analyzed by fluorescence labeling and CLSM, with three-dimensional structures that indicated channels for nutritional exchange with spatial heterogeneity. The biofilms reached an average thickness of about 15 μm, which was less than that for other fungi. However, it is clear that Pneumocystis spp. do not form the typical hyphal elements that contribute to the fungal biomass, but they did produce morphologies unique to the biofilm setting. A novel finding in this study was the observation that both species of Pneumocystis were able to undergo morphological transformations that resulted in similar but unusual and oddly shaped forms, which we termed “allomorphs” after the name given to alternative phenotypic forms of Histoplasma. capsulatum (18). Formation of these allomorphs could be in response to changes in pH, oxygen availability, or other environmental stimuli unique to this environment.

The formation of Pneumocystis biofilms followed a reproducible process that could be observed at the light microscopic level. Clusters of organisms gained refractile properties, formed elongated extensions as they spread across the insert membrane, and ultimately produced large three-dimensional structures that were visible at the macroscopic level. A moderate (61%) increase in (1,3)-β-d-glucan content over time suggested that this polymer may play a role in the biofilm process and could contribute to the increased refractility. The signature family of surface glycoproteins of Pneumocystis spp. were produced in the biofilms. These mannose/glucose-rich glycoproteins have often been observed to enmesh the organisms within the lung alveoli and could function as a component of the extracellular matrix, an important component of fungal biofilms.

Like for biofilm formation by various Candida species, metabolic activity was highest within the first 24 h after inoculation, followed by a plateau thereafter to 48 h (3, 49). Our studies were carried out much further, to 21 days in some cases, and showed significant decreases in ATP content over these extended times. A decrease in metabolic activity in fungal biofilms is a common characteristic. However, visual inspection of the biofilms at these later time points showed horizontal and vertical expansion of the biofilm mass and increasing complexity of the biofilm morphology, suggesting that the decrease in metabolic activity did not halt the biofilm formation.

Another striking finding of these studies was the ability to pass the biofilms and that the passaged P. carinii retained the ability to produce infections in the immunosuppressed rat model. The method of passage was not an active one, where the biofilms were scraped and used to inoculate naïve inserts. Rather, inserts with biofilms were placed in naïve wells that subsequently became populated with organisms, both in the supernatant and on the membranes. Detachment of cells from bacterial biofilms has been shown to play an important role in dissemination of infection (56), and such a mechanism may play a role in the dissemination of the organisms in this system, as the supernatants of the two inserts were in communication. Alternatively, the possibility of specialized cells that may disseminate the infection cannot be ruled out. In C. albicans biofilms treated with antifungal regimens, a subpopulation of perister cells arise that are antifungal tolerant and able to produce new biofilms (33), providing a precedent for such phenotypic variation.

Quorum sensing is a chemical means of cell-to-cell communication within prokaryotic and eukaryotic biofilms. The process has been implicated in preventing overpopulation, controlling competition for nutrients, and mediating dissemination, an especially important process in Candida biofilms (47). Two quorum-sensing molecules, farnesol and tyrosol, have been biochemically and functionally defined in Candida biofilms (6, 23). At high concentrations, farnesol blocks the morphological transition from yeast to filaments, while tyrosol accelerates germ tube formation, suggesting that the fungus can react both positively and negatively in response to environmental stimuli. The close apposition of the Pneumocystis organisms in the biolayers creates an ideal environment for exchange of chemical signals that could mediate mating or changes in cell cycle. Sensitivity to relatively low concentrations of farnesol supports the contention that a similar communication system may be operational in Pneumocystis biofilms. Alternatively, quorum-sensing molecules play a role in staving off other microbial predators, as in the case of the inhibitory effects of farnesol on the swarming motility in Pseudomonas aeruginosa strains isolated from cystic fibrosis patients (40). Farnesol could also be acting in this manner by reducing the viability of the P. carinii.

In contrast to a previous study that reported to have established a long-term culture method for P. carinii using culture inserts that required daily supplementation with S-adenosylmethionine (41), we found that this supplement was not required and, if added, caused a dramatic fall in viability to undetectable levels by 24 h (data not shown). In addition, the morphological changes documented herein were not reported in the other study, and it is doubtful that these two systems are comparable.

We propose that biofilm formation, documented here in an in vitro setting, occurs in the mammalian lung and confers several advantages for the survival of Pneumocystis. The multilayered structure and morphological changes offer protection from host immune responses, such as oxygen free radicals, by the nature of the extracellular matrix and refractile layer, through the creation of a barrier between the organism in this matrix and the alveolar lumen, where host immune cells reside. Phagocytosis would be physically hindered by the large and extensive structures that could not be engulfed by macrophages. Glucan in the cell wall of P. carinii has been shown to mediate inflammatory cytokine production through NF-κB-dependent mechanisms (19), and the increased elucidation of (1,3)-β-d-glucan by Pneumocystis biofilms could further exacerbate this process, thereby increasing the pathology of the immune responses associated with the pneumonia (58). Like for other fungal biofilms, it is anticipated that the biofilm structures will decrease the efficacy of anti-Pneumocystis drugs. Preliminary data in our laboratory suggest that this is the case.

The ability of Pneumocystis to survive in an environment with reduced availability of oxygen (26) is in keeping with the microenvironments produced by the layers of the biofilms and would be an additional survival mechanism that coincides with the production of these structures. Finally, little is known about the way in which Pneumocystis organisms disseminate from one alveolus to the next to produce pneumonia or new areas of colonization. It is reasonable to speculate that the large extensive mats of the biofilm facilitate the process of emigration to other alveoli using a process similar to that of other fungi which use hyphal or pseudohyphal extension or by detachment of portions of the biofilm to seed uninfected alveoli. Alternatively, specialized cells could migrate from the established biofilms to colonize other areas of the lung.

The significance of the novel in vitro system that supports Pneumocystis biofilms described here is severalfold. It provides investigators with organisms that remain viable in vitro for weeks, permitting approaches such as the establishment of a transformation system as a genetic tool, long-term assessment of anti-Pneumocystis drugs with the ability to passage the treated organisms to evaluate whether the agents are static or cidal, and an opportunity to perform life cycle studies to better understand the metabolic capacities of this family of fungal organisms. Moreover, the concept of biofilm formation by Pneumocystis provides new insights into the mode of replication and dissemination in the mammalian lung and offers a potential survival strategy for these organisms in the hostile milieu of this environment by affording protection from the immune response and other defenses by the host or a shelter from therapeutic agents.