ABSTRACT
The recent finding that high numbers of strict anaerobes are present in the respiratory tract of cystic fibrosis (CF) patients has drawn attention to the pathogenic contribution of the CF microbiome to airway disease. In this study, we investigated the specific interactions of the most dominant bacterial CF pathogen, Pseudomonas aeruginosa, with the anaerobic bacterium Veillonella parvula, which has been recovered at comparable cell numbers from the respiratory tract of CF patients. In addition to growth competition experiments, transcriptional profiling, and analyses of biofilm formation by in vitro studies, we used our recently established in vivo murine tumor model to investigate mutual influences of the two pathogens during a biofilm-associated infection process. We found that P. aeruginosa and V. parvula colonized distinct niches within the tumor. Interestingly, significantly higher cell numbers of P. aeruginosa could be recovered from the tumor tissue when mice were coinfected with both bacterial species than when mice were monoinfected with P. aeruginosa. Concordantly, the results of in vivo transcriptional profiling implied that the presence of V. parvula supports P. aeruginosa growth at the site of infection in the host, and the higher P. aeruginosa load correlated with clinical deterioration of the host. Although many challenges must be overcome to dissect the specific interactions of coinfecting bacteria during an infection process, our findings exemplarily demonstrate that the complex interrelations between coinfecting microorganisms and the immune responses determine clinical outcome to a much greater extent than previously anticipated.
INTRODUCTION
Over the past decade, an increasing number of studies on the respiratory tract of cystic fibrosis (CF) patients revealed that, rather than being dominated by a few bacterial pathogens, the CF lung is populated by a much more diverse polymicrobial community, including facultative and obligate anaerobes (1–3). Steep hypoxic gradients within the mucus plugs of the CF airways (4) and inefficient mucociliary clearance create a unique environmental niche and, thus, allow obligate anaerobes to infect and colonize the CF lung (5, 6). Although anaerobes appear to be present in numbers comparable to those of the typical aerobic bacterial pathogens, such as pseudomonads and staphylococci (2), their role in the development and progression of CF lung disease remains unclear (7).
The main obligate anaerobic bacteria in the respiratory tract and the lung mucus of CF patients usually consist of microorganisms that colonize the oral cavity (8). Veillonella parvula, a Gram-negative, strictly anaerobic coccus, is one of the most predominant bacterial species of the oral microbiome (9, 10). Accordingly, Veillonella spp. were repeatedly detected in CF sputum specimens at very high bacterial numbers, together with Pseudomonas aeruginosa (3, 11, 12). Thus, Veillonella species had been reported to be involved in mainly dental infections (13, 14). In addition, severe acute and chronic infections, such as osteomyelitis and discitis (15–19), meningitis (15), prosthetic joint infection (20), pleuropulmonary infections (21), and abscessed orchiepididymitis with sepsis had been observed (22). Endovascular infections by Veillonella species are considered to range from bacteremia to severe endocarditis and fatal cases of sepsis (23–26). Furthermore, Veillonella species are one of the most common anaerobic pathogens in chronic maxillary sinusitis and deep neck infections (27, 28).
Given their pathogenic potential and their relative abundance within the polymicrobial community of the CF lung, anaerobes might well contribute to the course of the disease. However, apart from directly contributing to the pathophysiology in the CF lung, the anaerobes might also display an indirect impact, either by modulating the host immune system or by enhancing the pathogenicity of other bacterial populations within the mixed bacterial CF community. Along this line, recent animal experiments demonstrated that bacterial strains of the oropharyngeal flora, such as Prevotella intermedia, contribute to the pathogenicity of P. aeruginosa (29, 30).
Here, we focused our studies on V. parvula, as the most abundant anaerobe recovered from the respiratory tract of CF patients. V. parvula is well known for its property of coaggregating with other anaerobic and aerobic bacteria (14). We addressed the question of whether and how cocultures of V. parvula and P. aeruginosa influence each other. In addition to in vitro studies, such as growth competition experiments, transcriptional profiling, and analyses on biofilm formation, we used our recently established in vivo murine tumor model (31). In this model, we inject CT26 tumor cells into syngeneic BALB/c mice subcutaneously, and when the tumor has reached a certain size, bacteria are administered intravenously. Most bacteria are able to invade and colonize such tumors. In addition, we could show that Salmonella enterica serovar Typhimurium, as well as P. aeruginosa, forms biofilms in this neoplastic tissue (32). Transcriptional profiling of tumor-colonizing P. aeruginosa revealed that the microenvironment encountered by the bacteria in the tumor closely resembles the microenvironment of the CF lung (33).
Using this model, we set out to investigate species-dependent phenotypic and transcriptomic changes during in vivo coinfections by V. parvula and P. aeruginosa. We found that both V. parvula and P. aeruginosa specifically colonize the neoplastic tissue following intravenous infection of tumor-bearing mice. However, although P. aeruginosa and V. parvula were found in distinct niches within the tumor, in mice that were coinfected with both bacterial species, significantly higher cell numbers of P. aeruginosa were recovered from the tumor tissue and the higher P. aeruginosa load correlated with clinical deterioration.
MATERIALS AND METHODS
Bacterial strains, media, and growth conditions.Anaerobic growth of P. aeruginosa strain PA14 (34) and V. parvula (DSM number 2008) was performed in an anaerobic workstation (Don Whitley Scientific) at 37°C. The growth medium used for all experiments was a modified DSMZ medium number 104 (http://www.dsmz.de/microorganisms/medium/pdf/DSMZ_Medium104.pdf) supplemented with 100 mM KNO3 (PYG-KNO3).
mRNA profiling.RNA extraction, cDNA library preparation, and deep sequencing were performed as previously described (35). In brief, cells were harvested after the addition of RNAprotect buffer (Qiagen), and RNA was isolated from cell pellets using the RNeasy plus kit (Qiagen). mRNA enrichment was performed using the MICROBExpress kit (Ambion). RNA was fragmented and ligated to specific RNA adapters containing a hexameric barcode sequence for multiplexing. The resulting RNA libraries were reverse transcribed and amplified, resulting in cDNA libraries ready for sequencing. All samples were sequenced on an Illumina genome analyzer II-x in the single-end mode with 36 cycles or on a HiSeq 2500 device involving 50 cycles.
Infection of tumor-bearing mice.All animal experiments were carried out in strict accordance with institutional guidelines on procedures involving animals and their care and were fully in compliance with the German Animal Welfare Act (36) and international laws and policies governing the use of animals for scientific purposes (37, 38). The protocol was approved by the appropriate ethical board (Niedersächsisches Landesamt für Verbraucherschutz und Lebensmittelsicherheit [LAVES], Oldenburg, Germany). The permission number is 33.9.42502-04-050/09.
Eight- to 10-week-old female BALB/c mice were subcutaneously (s.c.) inoculated at the abdomen with 5 × 105 CT26 cells. Mice bearing tumors of 4 to 7 mm in diameter were intravenously (i.v.) infected with 5 × 106 CFU of P. aeruginosa PA14 and/or 2 × 108 CFU of V. parvula suspended in phosphate-buffered saline (PBS).
Ex vivo CFU counts of infected tumors.Two days postinfection, mice were sacrificed and tumors were removed and kept in PBS for further analysis. To determine the CFU g−1 in homogenates, each tumor was cut into small pieces and squeezed through a nylon filter with a 70-μm pore size into 2 ml of sterile PBS. After serial dilutions, 10 μl of each dilution step from 104 to 1011 was plated three times on Columbia blood agar plates and incubated at 37°C in an aerobic and anaerobic environment for subsequent CFU counting.
Ex vivo RNA extraction and enrichment.Two days postinfection, mice were sacrificed, tumors were removed, and adherent fatty tissue and fur were cleared. Immediately after tumor removal, the tissue was cut into small pieces and squeezed with a sterile syringe through a nylon filter with 70-μm pore size into a 6-well plate, prefilled with 2.5 ml prechilled RNAprotect bacterial reagent (Qiagen). The resulting mixture was used to rinse the filter again. Afterwards, the mixture was filtered through a nylon filter with a 50-μm pore size. The filtered mixture was collected in a sterile 2-ml reaction tube and centrifuged at 4°C for 2 min at 3,000 × g. The supernatant was transferred to a sterile reaction tube and centrifuged at 4°C for 5 min at 17,000 × g. The pellet was stored at −70°C, and enrichment for bacterial RNA was achieved by using the MicrobEnrich kit (Ambion).
Quantification of gene expression.RNA treatment and comparative analysis of gene expression were performed as previously described (35). Sequence reads were separated according to their barcodes, and barcode sequences were removed. The sequence reads were mapped to a hybrid reference containing the genome sequences of both P. aeruginosa PA14 and V. parvula DSM 2008 using the algorithm stampy (39) with the default options. In this way, nonunique hits corresponding to homologous regions between the two bacterial species were discarded by the mapping algorithm. Furthermore, only high-quality mapped reads (Phred score mapping quality of at least 20) were assigned to the corresponding species and were used to quantify the amount of reads mapping to each gene. The R package DESeq (version 1.10.1) (40) was used for differential gene expression analysis. Briefly, the data for reads per gene were prefiltered to discard rRNA and tRNA genes and then normalized for variation in library size/sequencing depth by using the estimateSizeFactor function of DESeq. Differentially expressed genes were identified using the nbinomTest function based on the negative binomial model.
Biofilm growth assay.Cultivation of static biofilms at the bottom of 96-well microtiter plates was carried out as previously described, with small modifications due to growth under anaerobic conditions (41). In brief, overnight cultures of PA14 wild type and V. parvula growing under anaerobic conditions in PYG-KNO3 medium were adjusted to an optical density at 600 nm (OD600) of 0.02 with fresh medium, and 100-μl amounts were transferred into wells of a 96-well microtiter plate (half-area μClear plate; Greiner Bio-One). The plate was covered with an air-permeable foil and incubated at 37°C in the anaerobe incubator with a humid atmosphere for 24 h. Bacteria were stained by the addition of 60 μl staining solution, including Syto9 and propidium iodide (1:200 dilution of each dye in fresh medium) (Live/Dead BacLight bacterial viability kit; Molecular Probes/Life Technologies). Image stacks of the 24- or 48-h-biofilm-grown bacteria were acquired with a confocal laser scanning microscope (Fluoview 1000; Olympus) equipped with a 60×/1.20 water objective. The z-step size between the image layers was 2 μm. The acquired image stacks were visualized with the software IMARIS (version 6.7.4; Bitplane) and were processed and analyzed with the software tool PHLIP running in Matlab (42). The parameters biovolume (μm3) and colocalization were used for data interpretation.
FISH.Fluorescence in situ hybridization (FISH) was carried out as described previously (43). Briefly, tumor samples were fixed for 24 h at 4°C. Samples were embedded in cold polymerizing resin (methacrylate) and sectioned as described elsewhere (44). The hybridization buffer (20 μl) consisted of 0.9 M NaCl, 20 mM Tris HCl, 0.01% sodium dodecyl sulfate (SDS), 20% formamide, 20 pmol each probe, and 4′,6-diamidino-2-phenylindole (DAPI). Probes were synthesized commercially and 5′ end labeled with a fluorochrome, either Cy3 (indocarbocyanine), Cy5 (indodicarbocyanine), or FITC (fluorescein isothiocyanate) (all from Biomers, Ulm, Germany). Samples were incubated in a dark, humid chamber for 2 h at 50°C. Then, the slides were rinsed with sterile double-distilled water, dried, and mounted with Vectashield mounting medium (Vector Laboratories, Burlingame, CA) without DAPI.
Oligonucleotide probes for FISH.For simultaneous visualization and identification of P. aeruginosa and V. parvula, the specific probes PSM_G (45) and VEPA (46) were used, respectively. The nonsense probe NON-EUB338 (47) was included to test for nonspecific probe binding. For each FISH experiment, positive- and negative-control bacterial strains were hybridized alongside the tumor samples to ensure specificity. Prevotella intermedia, with 2 mismatches at the probe binding site, was a control for PSM_G, and Veillonella dispar, with 1 mismatch at the probe binding site, was a control for VEPA.
Epifluorescence microscopy and image acquisition.For documentation of FISH results, an epifluorescence microscope (Axioplan 2; Carl Zeiss, Jena, Germany) equipped with narrow-band filter sets (AHF Analysentechnik, Tübingen, Germany) was used. Image acquisition was performed with an AxioCam MRm (Zeiss), making use of the AxioVision 4.4 software.
Electron microscopic analysis.P. aeruginosa PA14 and V. parvula mixture-infected subcutaneous tumors from the mouse host were immersion fixed with 4% (wt/vol) paraformaldehyde, 1% (vol/vol) glutardialdehyde–150 mM NaCl–20 mM HEPES, pH 7.5, for 2 h or several days at ambient temperature. The tissue was washed twice in 100 mM sodium acetate, pH 3.0, for 15 min and subsequently labeled with 0.4% (wt/vol) thorium dioxide in 100 mM sodium-acetate, pH 3.0, overnight at 4°C. Labeled cells were washed once for 10 min at ambient temperature in 20 mM HEPES, pH 7.1. Fixed tumor tissue was dehydrated in an ethanol dilution series on ice and embedded in epoxy resin according to the method of Spurr (48). Ultrathin sections (90 nm) were prepared and analyzed by energy-filtered transmission electron microscopy (Libra 120 plus; Zeiss, Oberkochen). Zero-loss energy-filtered micrographs of uranylacetate-poststained sections were registered with a bottom-mount, cooled, 2,048- by 2,048-pixel, charge-coupled device (CCD) camera (Sharp:Eye; Tröndle, Moorenweis, Germany) as described in detail by Lünsdorf et al. (49).
Determination of serum TNF-α levels by ELISA.In order to obtain serum, mice were retroorbitally bled 1 h, 6 h, and 48 h after mono- or coinfection or uninfected mice were bled at the same time points, and the blood was transferred into a Microvette 500 serum gel (Sarstedt, Germany) and centrifuged for 5 min at 10,000 rpm. The serum was transferred into a reaction tube and frozen at −20°C. In order to determine serum tumor necrosis factor alpha (TNF-α) levels, the Mouse TNF-α ELISA Max standard kit (Biolegend, USA) was used according to the manufacturer's instructions. Therefore, serum samples were thawed, and 1 to 25 μl of each sample was used in the enzyme-linked immunosorbent assay (ELISA). The values obtained were calculated back to the amount of TNF-α per ml of blood for graphic depiction of the results.
Microarray data accession number.Raw and processed RNA data have been submitted to the Gene Expression Omnibus (GEO) under the accession number GSE58388.
RESULTS
Coculture of P. aeruginosa PA14 and V. parvula in vitro.Both bacterial species, P. aeruginosa and V. parvula, are recovered on a regular basis and at high bacterial cell numbers from respiratory tract material of CF patients. To investigate whether one species has an effect on the growth and survival of the other, we cocultivated P. aeruginosa PA14 and V. parvula DSM number 2008 in vitro under defined anaerobe growth conditions using a modified DSMZ medium (number 104) enriched with KNO3 as an alternative terminal electron acceptor (PYG-KNO3) and with only gentle agitation. The CFU counts of both bacterial species were determined over time. As shown by the results in Fig. 1, the growth of P. aeruginosa PA14 was not significantly influenced by the presence of V. parvula. However, the growth of V. parvula was drastically impaired by P. aeruginosa PA14. At 36 h postinoculation, no vital cocci could be recovered from cocultures. Interestingly, incubation of V. parvula with 25% or 50% (vol/vol) culture supernatant of stationary-phase P. aeruginosa PA14 (OD600 of 2) diluted in PBS did not lead to significant alterations in CFU counts. Similar cultures, where supernatants of V. parvula cultures with the same optical density were added to the V. parvula as controls, also did not show any growth differences (data not shown). These results suggest that direct contact with PA14 is required for impaired V. parvula growth and survival. Alternatively, the corresponding soluble effector molecule could be very short lived.
Growth of P. aeruginosa PA14 and V. parvula under mono- and coculture conditions, shown by CFU measurements of P. aeruginosa PA14 alone, P. aeruginosa PA14 in coculture, V. parvula alone, and V. parvula in coculture under anaerobic growth conditions over time. The PYG growth medium was supplemented with 100 mM KNO3. The mean results ± standard deviations of three independent replicates are presented.
To analyze the mutual influence of both bacteria in more detail, we monitored the transcriptional profiles of both species after 20 h of in vitro growth in mono- or cocultures (Table 1). Interestingly, the expression profile of P. aeruginosa PA14 was nearly unaffected by the presence of V. parvula, and only three P. aeruginosa PA14 genes were significantly downregulated (>2.5-fold) in the presence of V. parvula. In contrast, the expression of 27 genes (9 of them being hypothetical genes) of the anaerobic cocci were significantly upregulated in the presence of P. aeruginosa.
Differential regulation of genes in P. aeruginosa and V. parvula in coculture versus monoculture as identified by in vitro RNA transcriptome sequencing
Biofilm formation of P. aeruginosa PA14 is influenced by the presence of V. parvula.V. parvula has been shown to enhance the aggregation properties of adjacent bacterial species, such as Actinomyces and Streptococcus sanguis (14). Here, we tested whether the presence of V. parvula alters the capability of P. aeruginosa to enter a biofilm state of growth. We cultivated P. aeruginosa PA14 and V. parvula separately and in cocultures without agitation in PYG-KNO3 medium in 96-well plates. Twenty-four- and 48-h-old biofilms that had developed at the bottom of the plate were analyzed by counting CFU (Fig. 2A). We also used a confocal laser scanning microscope to determine the overall structure of the biofilms (Fig. 2B) and their total biovolume, as well as the distribution of live and dead cells according to a cell wall permeability stain assay (Fig. 2C). Under the experimental conditions used, both strains formed mature biofilms. CFU measurements revealed that P. aeruginosa reached high bacterial cell numbers already after 24 h of growth and was not affected by the presence of V. parvula. In contrast, the cell numbers of V. parvula increased significantly with prolonged incubation, but only in the absence of P. aeruginosa PA14. At both time points, the biofilm formed by V. parvula was homogeneous, without the formation of bacterial aggregates (Fig. 2B), and the fraction of dead bacteria increased to approximately 50% after 48 h of biofilm growth (Fig. 2C). In comparison to that of V. parvula, the biofilm of P. aeruginosa was more heterogeneous, and we observed a higher fraction of bacterial aggregates, including a high proportion of dead bacteria already after 24 h of incubation (Fig. 2B). After 48 h, more cells contributed to the total biovolume of the P. aeruginosa biofilm, exhibiting a balanced ratio between dead and live cells (Fig. 2C). Interestingly, the biofilm of P. aeruginosa in coculture with V. parvula presented some significant differences from its biofilm under monoculture conditions. The flat biofilm formed after 24 h by V. parvula was absent in the presence of P. aeruginosa. In cocultures, we observed a higher number of bacterial clusters. After 48 h, these aggregates displayed an increased size and were interspersed with vital bacteria that did not contribute to the total biomass of the biofilm (Fig. 2C). The increased size and presence of cluster formation may indicate that V. parvula promotes the development of robust P. aeruginosa aggregates and mature biofilm structures.
Biofilm formation of P. aeruginosa PA14 and V. parvula under mono- and coculture conditions. (A) CFU counts of P. aeruginosa PA14 and V. parvula under monoculture conditions and in mixed cultures demonstrate the eradication of veillonellae in cocultures after 48 h. The mean results ± standard deviations of three independent replicates are presented. (B) The corresponding biofilm development at 24 h and 48 h postinoculation demonstrates the different architecture and pattern. (C) Analysis of strain-dependent biofilm biovolumes, including live and dead cell distribution as determined by cell wall permeability.
V. parvula enhances P. aeruginosa colonization of solid subcutaneous tumors in mice.It has previously been shown that facultative anaerobic bacteria, such as Salmonella Typhimurium and Escherichia coli, are able to colonize solid subcutaneous tumors of the colon carcinoma cell line CT26 in a mouse tumor model (50). We have further demonstrated that P. aeruginosa, although rapidly cleared from the blood and other organs, is also able to gather and multiply specifically in the murine tumor tissue following intravenous administration (31, 32). Of note, whereas facultative anaerobic bacteria like Salmonella colonize the whole tumor tissue, including the anaerobic center, the nonfermenter P. aeruginosa colonized mainly the outer rim of the tumor tissue. Genetic profiling of P. aeruginosa isolated from such tumors revealed that the growth conditions encountered by the bacteria in the murine tumors and the CF lung must be very similar.
To understand more about the interplay between P. aeruginosa PA14 and V. parvula in vivo, we administered the two bacteria alone or together to tumor-bearing mice.
We intravenously infected tumor-bearing mice with V. parvula and determined the CFU counts within the tumor and various organs 48 h postinfection. As observed before for the facultative anaerobes (31) and P. aeruginosa (30, 32), V. parvula accumulated in the tumor tissue. However, accumulation of V. parvula was only detected when sufficient numbers of bacteria were intravenously administered (2 × 108 CFU). Lower infection doses frequently resulted in complete clearance of the bacteria. The bacterial titers in all peripheral organs tested, including lung, liver, and spleen, were always significantly lower (<104 CFU/g tissue, data not shown) than those in the corresponding mouse tumor (>1010 CFU/g) (Fig. 3). Interestingly, the V. parvula CFU/g tumor tissue showed no significant changes upon coinfection with P. aeruginosa. In contrast, the massive presence of intratumoral V. parvula increased the P. aeruginosa PA14 CFU/g tumor tissue significantly (Fig. 3). In addition, the extent of necrotic tissue was greater in most of the coinfected tumors (data not shown). Furthermore, considerably enhanced clinical symptoms (such as disorientation, scruffy fur, and lethargy) were observed already 24 h postinfection for mice that were coinfected with PA14 and V. parvula. As shown by the results in Fig. 4, the survival rate of mice at 48 h postinfection was drastically decreased under coinfection conditions.
Colonization of CT26 tumors by P. aeruginosa PA14 and V. parvula in a mouse model. Tumor-bearing mice were infected i.v. with P. aeruginosa PA14 (5 × 106 CFU), V. parvula (2 × 108 CFU), or both species (same CFU counts as for single infections). At 48 h p.i., four tumors were homogenized and plated to determine the CFU per gram of tissue. The experiment was repeated three times. The results of one representative experiment with P. aeruginosa PA14 versus P. aeruginosa PA14 in coculture with V. parvula are presented (n ≥ 4). Statistical significance was assessed with the Mann-Whitney test; ns, not significant.
Survival rates of mice bearing mono- and coinfected CT26 tumors. Tumor-bearing mice were infected i.v. with V. parvula, P. aeruginosa PA14, or both species. Mouse survival was measured at 48 h postinfection, before tumors and organs were dissected for further experiments. The average survival rates ± standard deviations of the means during 4 experiments, each performed with 5 mice, are reported. Statistical significance was assessed with the Mann-Whitney test. ***, P ≤ 0.001.
Spatial separation of P. aeruginosa PA14 enables unaffected colonization of V. parvula.To determine the specific niches within the tumor that are colonized by P. aeruginosa PA14 and V. parvula, respectively, we coinfected mouse tumors and detected both bacterial strains in thin sections by the use of ribosome-based fluorescence in situ hybridization (FISH). The analysis revealed no significant differences in colonization patterns or bacterial distributions between single infections or mixed infections (Fig. 5). In mixed infections, both species were found to be clearly separated from each other. Whereas the anaerobic cocci formed large clusters in the inner core of the tumor and developed very distinct nest-shaped (globular/spherical) colonies (Fig. 5), P. aeruginosa PA14 cells were mainly found in the outer rim of the tumor tissue. Only a few areas showed direct contact between the two populations. This distribution pattern was seen in six independent tumor samples. The closest proximity between V. parvula and single isolated pseudomonads as measured by electron microscopy was about 2 μm (Fig. 6). Within the colonies formed by V. parvula in deeper tissue, we were never able to detect P. aeruginosa.
Visualization of P. aeruginosa and V. parvula in tumor tissue, with overview (A) and insets at higher magnification (B, C, D). All tumor tissue samples were hybridized simultaneously with species-specific probes for P. aeruginosa (PSM_GFITC) (green) and V. parvula (VEPACy3) (yellow-orange). Nonsense binding of the probes was excluded by using NON-EUB338Cy5 (magenta). Host cell nuclei were stained using the nonspecific nucleic acid stain DAPI (blue). The outer part of the tissue is colonized by P. aeruginosa (B), whereas V. parvula was found exclusively in deeper parts of the tumor (C, D). Only a few areas showed close contact between the two populations.
Ultrastructural analysis of coinfected CT26 tumors from mouse host. (A) Larger view, showing a cluster/microcolony of Veillonella parvula (VP) within the tumor tissue. Pseudomonas aeruginosa cells (PA) are found individually distributed within the host tissue. (B) Detailed view of a Veillonella cluster. Bacterial cells of various sizes and obviously different physiological states are shown, surrounded by a high density of outer membrane vesicles (omv). pg, peptidoglycan; chr, bacterial chromosome; cp, cytoplasm; om, outer membrane.
Prolonged TNF-α response during P. aeruginosa PA14 and V. parvula coinfection.Depletion of TNF-α has previously been demonstrated to retard blood influx and to delay bacterial colonization of the tumor tissue (51), clearly suggesting that the induction of this cytokine significantly influences the bacterial invasion and colonization process. In order to investigate whether the increased P. aeruginosa cell counts in coinfected mice could be correlated to increased TNF-α levels, we measured the cytokine levels over time in monoinfected versus coinfected animals. As shown by the results in Fig. 7, TNF-α is readily detectable in the blood of all infected mice 1 h after administration and is detected at somewhat lower levels at 6 h. Importantly, TNF-α becomes undetectable in monoinfected mice at later time points. Only under conditions of coinfection of P. aeruginosa and V. parvula can significant amounts of this cytokine still be detected 48 h after i.v. infection. This prolonged production of TNF-α might contribute to an enhanced influx of P. aeruginosa cells into the tumor tissue and might at least partly explain the high bacterial cell counts observed in coinfected tumor tissue. Thus, whereas the high TNF-α titers might be the result of increased P. aeruginosa cell counts, they just as well might as be their cause, since TNF-α enhances bacterial influx into the tumor tissue. In addition, the prolonged high concentration of TNF-α in the blood of coinfected animals might explain the deterioration of the general health status and the lethal outcome for some of the mice.
TNF-α release into the blood of mice bearing CT26 tumors mono- or coinfected with P. aeruginosa PA14 and V. parvula. Tumor-bearing mice were infected i.v. with V. parvula, P. aeruginosa PA14, or both species. TNF-α concentrations in the blood of CT26 tumor-bearing BALB/c mice at different times p.i. were determined by ELISA analysis. The average results ± standard deviations for 4 biological replicates are reported. Statistical significance was assessed with the Mann-Whitney test. *, P ≤ 0.05.
The gene expression of both species is altered during cocolonization of solid tumors.An alternative cause of the high mortality of coinfected mice might be enhanced virulence of the opportunistic pathogen P. aeruginosa as a result of specific bacteria-to-bacteria interactions in vivo. In order to elucidate whether and how the bacteria influence each other during coinfection in the tumor tissue, we recorded the ex vivo transcriptional profiles of the murine tumors infected with pure cultures of PA14 and V. parvula, as well as the profiles of tumors coinfected with both bacterial pathogens. We performed ex vivo RNA sequence analysis as recently described by Dötsch et al. (35), with some modifications, and also compared the in vivo transcriptomes with those of the in vitro mono- and cocultures.
Independent from mono- or coinfection/coculture of V. parvula with P. aeruginosa, we found 158 genes that were differentially regulated in V. parvula under in vivo and in vitro conditions (76 genes were upregulated and 82 were downregulated in vivo; see Table S1 in the supplemental material). Many of the genes that were highly expressed in vivo are involved in central bacterial metabolism, indicating that the in vivo conditions promote the growth of V. parvula.
Table 2 presents the data for the set of genes that were differentially regulated in V. parvula during monoinfection of the mouse tumor and coinfection with P. aeruginosa. Twenty-four V. parvula genes were downregulated in vivo in the presence of P. aeruginosa, and only 1 was upregulated more than 2.5-fold (at a P value of ≤0.05). A strong downregulation of genes involved in biotin metabolism was found. Furthermore, many transcriptional changes of V. parvula in vivo in the presence of P. aeruginosa involved hypothetical genes putatively encoding transport and transmembrane signaling proteins, such as ABC transporters or periplasmic-binding proteins.
V. parvula genes that were differentially regulated in the presence of P. aeruginosa as identified by ex vivo RNA transcriptome sequencing
On the other hand, independent of whether or not P. aeruginosa was coinfected/cocultivated with V. parvula, we found 680 genes that were differentially regulated in PA14 under in vivo as opposed to in vitro conditions (see Table S2 in the supplemental material). As expected, P. aeruginosa genes that were highly expressed in vivo were genes of the type III secretion system and genes involved in iron acquisition and amino acid biosynthesis, as well as in energy metabolism. Thus, similar to the results observed for V. parvula, our in vivo conditions appear to promote the growth of P. aeruginosa.
We next determined which genes were differentially regulated in P. aeruginosa PA14 during coinfection with V. parvula and monoinfection of the mouse tumor (Table 3). Only 10 P. aeruginosa genes were overexpressed in vivo in the presence of V. parvula (most of them were annotated as hypothetical, unclassified, or unknown), whereas 78 genes were downregulated more than 2.5-fold (at a P value of ≤0.05). As observed for V. parvula, many of the P. aeruginosa genes that were downregulated during coinfection with V. parvula encode hypothetical proteins or are involved in transport of small molecules or adaptation/protection processes (Table 3). Among the P. aeruginosa genes most strongly downregulated in the presence of V. parvula were the PA14_52880, PA14_52900, and PA14_52910 genes. Those genes are located in a single operon encoding proteins involved in lipid metabolism and transport.
P. aeruginosa genes that were differentially regulated in the presence of V. parvula as identified by ex vivo RNA transcriptome sequencing
DISCUSSION
Although it has been suggested that CF patients may benefit from antibiotic treatment that includes treatment of anaerobes (52, 53), obligate anaerobes are not yet a routine target in the treatment of CF lung infections. Nevertheless, it is evident that anaerobic bacteria are prevalent in the lungs of CF patients (3). However, their contribution to CF lung pathophysiology is far from being understood. More information on how anaerobes interact with other CF pathogens and the general lung microbiome, as well as the host immune system, is needed in order to develop and apply more effective regimens to treat chronic lung disease in CF.
In this study, we analyzed how the dominant CF pathogen P. aeruginosa interacts specifically with V. parvula, an anaerobic bacterium that is routinely recovered from the respiratory tract material of CF patients at cell numbers comparable to or even higher than those of P. aeruginosa (3). We found that neither the growth of P. aeruginosa PA14 nor its transcriptional profile is significantly changed in the presence of V. parvula in vitro, although PA14 rapidly kills V. parvula, either by direct cell-cell contact or via a short-lived molecule, as supernatants of PA14 had no influence on the growth of V. parvula.
We also found a trend toward an earlier and stronger aggregation during the biofilm formation process in cocultures of the two bacterial species in vitro. This indicates that the presence of V. parvula triggers aggregation of P. aeruginosa PA14, as has been described before for cocultures with other bacterial species (14). This might be of clinical relevance, as P. aeruginosa forms bacterial aggregates in the chronically infected CF lung and it has previously been demonstrated that in oral cavities, V. parvula plays an essential role within multispecies biofilms. where the presence of V. parvula leads to a higher resistance of cocci, such as Streptococcus mutans, against chemical antiseptics (54, 55).
We further exploited our recently established mouse tumor model to study the interaction of the two bacterial species in vivo during an infection process. V. parvula is frequently isolated from solid skin and lung tumors of cancer patients (56, 57). Thus, it was reasonable to expect that V. parvula colonizes the mouse tumor tissue following intravenous infection. As it turned out, the colonization was very effective. Predictably, the obligate anaerobic bacterium colonized the necrotic hypoxic center of the tumor, whereas P. aeruginosa was mainly found in the well-oxygenated outer rim. If it occurred at all, direct contact of the two bacterial species in vivo was rare. This is in agreement with the finding that P. aeruginosa effectively kills V. parvula if cocultured in vitro. Nevertheless, despite the absence of direct contact of the bacteria and constant and very high V. parvula CFU counts, we found significantly increased cell numbers of P. aeruginosa in the tumor tissue coinfected with V. parvula. More importantly, the increased bacterial cell counts correlated with clinical deterioration.
The increased bacterial cell numbers per se may cause clinical deterioration and might explain the observations of an increased and sustained inflammatory response. At the same time, elevated TNF-α titers might directly contribute to the aggravation of the clinical symptoms. On the other hand, it has been suggested previously that the production of TNF-α significantly influences bacterial cell counts in the tumor tissue due to its positive effect on blood influx into the tumor (51). Our observation of prolonged production of TNF-α during coinfection experiments thus suggests an enhanced influx of P. aeruginosa cells into the tumor tissue and also, possibly, a lower capability of the host immune system to contain the P. aeruginosa cells within the tumor.
Furthermore, V. parvula might specifically interact with P. aeruginosa in vivo and have an effect on P. aeruginosa virulence, thus resulting in clinical deterioration and increased CFU counts in the infected tumor. To address this, we performed ex vivo transcriptional profiling and isolated bacterial RNA directly from mouse tumors that had been infected with P. aeruginosa and V. parvula individually or during coinfection experiments. Interestingly, whereas during in vitro cocultures, P. aeruginosa did not exhibit significant changes in its transcriptional profile in the presence of V. parvula, we found 80 P. aeruginosa genes that were differentially regulated in vivo only in the presence of V. parvula. Apart from hypothetical genes, genes involved in central intermediary metabolism, transport of small molecules, and adaptation/protection processes were among the downregulated genes. However, no genes directly involved in P. aeruginosa virulence were found. This in vivo-specific transcriptional profile suggests that the presence of V. parvula improves the growth conditions for P. aeruginosa. This might also be concluded from the upregulation of genes involved in lipid metabolism, since the bacteria should be able to scavenge on the increased number of dead cells. As a possible consequence, P. aeruginosa might be recovered from the infected host site at significantly higher cell numbers. Therefore, favorable nutritional conditions in vivo might contribute to increased P. aeruginosa cell counts in the tumor tissue and, thus, clinical deterioration.
In conclusion, a major reason for the largely unexplored mutational impact of coinfecting bacteria is that many of the specific bacterial interactions only become apparent under in vivo infection conditions. However, as demonstrated in this study, the specific in vivo interactions might affect clinical outcome to a much greater extent than previously anticipated.
ACKNOWLEDGMENTS
This work was supported by an ERC starter grant (RESISTOME 260276), the Federal Ministry of Education and Research (project number 1616038C), and the President's Initiative and Networking Fund of the Helmholtz Association of German Research Centers (HGF) under contract number VH-GS-202. V.P. has been supported by the Helmholtz International Graduate School for Infection Research under contract number VH-GS-202. A.L. was funded by the International Research Training Group 1273 supported by the German Research Foundation (DFG).
TWINCORE is a joint venture of the Hannover Medical School, Hannover, Germany, and the Helmholtz Centre for Infection Research, Braunschweig, Germany.
FOOTNOTES
- Received 23 June 2014.
- Returned for modification 5 August 2014.
- Accepted 5 November 2014.
- Accepted manuscript posted online 10 November 2014.
Supplemental material for this article may be found at http://dx.doi.org/10.1128/IAI.02234-14.
- Copyright © 2015, American Society for Microbiology. All Rights Reserved.
