Differential responses of human dendritic cells to metabolites from the oral/airway microbiome

Summary

Small molecule metabolites that are produced or altered by host-associated microbial communities are emerging as significant immune response modifiers. However, there is a key gap in our knowledge of how oral microbial metabolites affect the immune response. Here, we examined the effects of metabolites from five bacterial strains found commonly in the oral/airway microbial communities of humans. The five strains, each isolated from cystic fibrosis patient sputum, were Pseudomonas aeruginosa FLR01 non-mucoid (P1) and FLR02 mucoid (P2) forms, Streptococcus pneumoniae (Sp), S. salivarius (Ss) and Rothia mucilaginosa (Rm). The effect of bacterial metabolites on dendritic cell (DC) activation, T cell priming and cytokine secretion was determined by exposing DCs to bacterial supernatants and individual metabolites of interest. Supernatants from P1 and P2 induced high levels of tumour necrosis factor (TNF)-α, interleukin (IL)−12 and IL-6 from DCs and primed T cells to secrete interferon (IFN)-γ, IL-22 compared to supernatants from Sp, Ss and Rm. Investigations into the composition of supernatants using gas chromatography–mass spectroscopy (GC-MS) revealed signature metabolites for each of the strains. Supernatants from P1 and P2 contained high levels of putrescine and glucose, while Sp and Ss contained high levels of 2,3-butanediol. The individual metabolites replicated the results of whole supernatants, although the magnitudes of their effects were reduced significantly. Altogether, our data demonstrate for the first time that the signature metabolites produced by different bacteria have different effects on DC functions. The identification of signature metabolites and their effects on the host immune system can provide mechanistic insights into diseases and may also be developed as biomarkers.

Graphical Abstract

Small molecule metabolites produced by human-associated microbes are thought to have a big impact on the immune system. We exposed dendritic cells (DCs) harvested from healthy human adults to microbial supernatents from five human airway isolates (two Pseudomonas spp., Rothia mucilaginosa and two Streptococcus spp.) and five molecules we found to be abundant in those cultures. The microbial cultures and metabolites had differential effect on DCs; glucose and putrescine were more abundant in Pseudomonas cultures and induced inflammatory responses and T cell priming, while 2,3-butanediol was more abundant in Streptococcus cultures and exerted no significant effect on DCs at the concentrations we tested.

butanediol, butanedione, dendritic cells, glucose, oral microbiome, putrescine

Introduction

Mucosal surfaces are bombarded constantly with innocuous antigens in the form of inhaled antigens, food and so forth. The mucosal immune system exists in a state of immunosuppression to prevent reactivity to the harmless antigens. Dendritic cells (DCs), found beneath the epithelial cell lining in the mucosa, are considered key players for both immunity and tolerance [1–3]. Mucosal DCs can either sense infections directly via projection of their dendrites into the lumen or indirectly via factors secreted by epithelial cells [4]. Epithelial cell DC cross-talk is also essential for maintaining tolerance in the mucosa [3,5]. Recent studies have highlighted the role of retinoic acid (RA) and transforming growth factor (TGF)-β in mucosal tolerance. RA and TGF-β produced by epithelial cells have been demonstrated to act on DCs to render them tolerogenic and induce T regulatory cells [6,7]. Thus, the interaction between immune and non-immune cells is thought to play a major role in immunity and tolerance in the mucosal environment. Emerging evidence suggests that a third factor, the resident microbial community, has an equally important role in shaping the immune response in the mucosa [8].

Humans evolved in a world first inhabited and dominated by microbes. We each contain a unique resident microbial community (the microbiome) that becomes established during the first few years of life, and which encodes metabolic and genetic diversity greater than the set of human genes themselves [9]. Estimates suggest that at least half of the detectable metabolites in a human are produced or altered by microbes [10–12]. Microbial metabolites impact immune development and influence cells throughout the body, including stem cells and immune cells, and also include precursors to neurotransmitters which could affect behaviour [13,14]. Microbiome composition impacts health, and has been shown to be correlated with a number of conditions, including allergies, autoimmune diseases (i.e. inflammatory bowel disease and type 1 diabetes), cancer, obesity, autism and infection susceptibility. In addition, the gut microbiota respond to changes in nutrition by modulating specific bacterial-derived bioactive metabolites; for instance, the production of butyrate and other short chain fatty acids from fibre. These metabolites can affect the integrity of tight junctions in the gut epithelia and can influence whether molecules normally retained in the gut can cross over more easily into the blood [11]. Butyrate also has an anti-inflammatory effect on DCs [15].

The gut microbiome has been the major focus of recent studies on the effect of bacterial metabolites on the immune system. The airway microbiome has been less studied, and there is a scarcity of information regarding the metabolites secreted by microbes in the airways and their effects on immune cells. Although the gut and the lungs are mucosal tissues, their physiology and anatomy is quite different. In lung mucosa, the surface of the epithelia is oxygen-rich and is lined with lipid-rich surfactants. Airway fluid experiences steep oxygen gradients, and while microbes are aspirated into the lungs from the dense microbial community of the oral cavity, the airways have reduced bacterial loads (usually < 10⁴ microbes per ml in healthy lungs; detected microbes may be transient) due to vigorous mucociliary clearance mechanisms [16–18]. In contrast, the gut mucosa is almost anaerobic and nutrient-dense, with ∼10¹² microbes per ml [19]. These differences result in very different microbial community composition and abundance in the lung and gut mucosa. Emerging evidence suggests that the airway microbiome resembles the oral microbiome closely and is quite different from the nasal microbiome [16,20]. Interestingly, as people age, microbial communities in the oropharynx and anterior nares become more similar to each other, suggesting that ageing disrupts the ability to maintain distinct microbial communities at different sites [21]. Although the density of bacteria in the airways is significantly less compared to the gut, airway microbiome dynamics and microbe-to-microbe interactions are recognized increasingly as critical factors in lung diseases [18,22]. One recent study found that enrichment of oral taxa in lower airway bronchoalveolar lavage samples led to a T helper type 17 (Th17) immune response in a culture-independent study of samples from 49 people [23].

Given the importance of the microbiome in influencing local immunity, in this study we examined the effect of oral microbial metabolites on human DC functions.

Materials and methods

Blood donors

Peripheral blood samples were obtained from healthy young volunteers. This study was approved by the Institutional Review Board of the University of California (Irvine, CA, USA).

Generation of supernatants from oral bacteria and identification of metabolites in bacterial culture supernatant by gas chromatography–mass spectroscopy (GC-MS)

Clinical isolates collected from cystic fibrosis (CF) patients at the UCSD adult CF clinic were used.

The five cultures (Table 1) were started from a single colony and grown statically in 2 ml Todd–Hewitt (TH) medium at 37°C in ambient conditions for 4 days. Optical density 600 nm measurements show that the Pseudomonas aeruginosa and Streptococcus spp. strains grew to saturation, while Rothia mucilaginosa did not grow as robustly in these conditions (Table 1). Culture cells were pelleted in a microfuge and supernatants were collected for cell-culture assays and metabolomics analysis. Metabolomics analysis with GC profiling that captures ∼500 metabolites was performed at the UC Davis West Coast Metabolomics core facility. Twenty microlitres of supernatant was extracted using the plasma extraction protocol, which includes an extraction step in acetonitrile : isopropyl alcohol : water before derivatization and gas chromatography–time-of-flight (GC-TOF) analysis [24].

Table 1

Description of bacterial isolates

Abbreviation	Description	*OD 600 nm
P1	Non-mucoid Pseudomonas aeruginosa, PaFLR01	1.147
P2	Mucoid Pseudomonas aeruginosa, PaFLR02	1.527
Rm	Rothia mucilaginosa	0.671
Sp	Streptococcus pneumonia	1.242
Ss	Streptococcus salivarius	1.262

*

Cultures were started from a single colony and grown statically in 2 ml Todd–Hewitt (TH) medium at 37°C in ambient conditions for 4 days. OD = optical density.

Isolation and culture of human monocyte-derived DCs

DCs were prepared essentially as described previously [25,26]. Peripheral blood mononuclear cells were separated over Ficoll density gradient centrifugation. Monocytes were selected positively using CD14 coupled magnetic beads (Stemcell Technologies, Vancouver, Canada). The resulting monocytes were cultured under a humidified atmosphere of 5% CO₂ at 37°C in RPMI-1640 supplemented with 10% fetal bovine serum (FBS), 1 mM glutamine, 100 U/ml penicillin, 100 µg/ml streptomycin, 50 ng/ml human recombinant granulocyte–macrophage colony-stimulating factor (rGM-CSF) (Peprotech, Rocky Hill, NJ, USA) and 10 ng/ml human recombinant IL-4 (rIL-4) (Peprotech). Half the medium was replaced every 2 days with fresh medium and monocyte-derived DCs were collected after 6 days. The cells were CD11c⁺, CD14^–, human leucocyte antigen D-related (HLA-DR⁺) at this stage, indicating successful differentiation to DCs. DCs were cultured with bacterial supernatants as indicated. Control DCs were cultured in Todd–Hewitt medium. After 24 h, the cells were collected for flow cytometry and supernatants were stored for cytokine determination. For these experiments, bacterial supernatants were collected once. DCs from six different individuals were used to assay the response to the supernatants.

For experiments with individual metabolites, immature DCs were obtained as above and then cultured with 2,3-butanediol (20 mM), 2,3-butanedione (0·2 mM), putrescine (1 mM) or glucose (15 mM) for 24 h. The metabolites were obtained from Sigma-Aldrich (St Louis, MO, USA).

Cytokine production by DCs

Cytokines, tumour necrosis factor (TNF)-α, interleukin (IL)−6, IL-12p70, IL-10, monocyte chemoattractant protein 1 (CCL-2) and CXC chemokine ligand (CXCL)−10 in the supernatants for each experiment were measured by specific enzyme-linked immunosorbent assay (ELISA) kits (BD Pharmingen, San Jose, CA), as per the manufacturer's protocol.

DC–CD4 T cell cultures

Immature DCs were stimulated with bacterial supernatants or metabolites as described above. After being cultured for 24 h, cells were collected and washed before 2 × 10⁴ DCs were cultured with 1 × 10⁵ purified, allogeneic CD4 T cells (negative selection kit from Stemcell Technologies). Six days later, the supernatant was collected and the secretion of IL-22 (R&D Systems, Minneapolis, MN, USA), IL-10 and interferon (IFN)-γ (BD Pharmingen) was assessed using ELISA. This procedure was repeated six times.

Statistical analyses

Statistical analysis was performed using GraphPad Prism (GraphPad Inc., San Diego, CA, USA). Differences between unstimulated and stimulated conditions were tested using paired t-tests. A P-value of < 0⋅05 was considered statistically significant. Metabolite intensity analysis and visualization was carried out with Metaboanalyst version 3.0 (http://www.metaboanalyst.ca/).

Results

Differential response of DCs to metabolites from different bacteria in the oral microbiome

In order to determine the effect of metabolites from airway microbial community members on DCs, we exposed DCs to culture supernatants from five bacterial strains found commonly in oral and respiratory tract microbial communities. These strains were isolated from CF patient sputum samples (Table 1), and are known to be associated with CF pathology and also pneumonia. P. aeruginosa is one of the most common pathogens causing nosocomial infection in patients suffering from respiratory diseases, chemotherapy cancer patients, immunocompromised hosts and young adults with CF [27,28]. Pseudomonas spp. are more common in very young or elderly patients [21,29]. Both the non-mucoid (P1) and mucoid (P2) are associated with respiratory diseases; however, the mucoid form develops most often in chronic respiratory infections, especially CF [30]. R. mucilaginosa (Rm) is a Gram-positive coccus member of the family Micrococcaceae, which forms part of the normal microflora of the human mouth and the upper respiratory tract. Although this organism is believed to be of low virulence, it is recognized increasingly as an opportunistic pathogen causing pneumonia in immunocompromised hosts [31], and can produce toxic molecules that could be carcinogenic, such as acetaldehyde [32]. Streptococci are Gram-positive bacteria and the association of S. pneumoniae with pneumonia is well established. Evidence suggests that Sp also plays a significant role in the pathogenesis of the CF lung, where it is reported to form biofilms readily and have increased antibiotic resistance [33,34]. S. salivarius (Ss) is part of the normal human oral microbiome and is reported to be one of the most prevalent streptococci in the CF lung, where it may produce fermentation products, including 2,3-butanedione, that modulate the physiology of opportunistic pathogens such as P. aeruginosa [35,36].

Supernatants collected following bacterial culture for 4 days in TH medium contain metabolites produced by the bacteria during that time. Monocyte-derived DCs were cultured with bacterial supernatants at various ratios of AIM V (serum-free medium for DCs) to bacterial supernatant ranging from 1 : 1 to 1 : 10. The ratio of 1 : 5 (AIM V to supernatant) was found to be optimal, as ratios 1 : 1 and 1 : 2 were toxic to DCs and a ratio of 1 : 10 was too dilute to have an effect on DCs (data not shown). DCs cultured in a 1 : 5 ratio of AIM V to uninoculated TH medium were used as controls for the experiment.

Investigations into DC cytokine secretion after exposure to the bacterial metabolites revealed that P1 and P2 supernatants were able to activate DCs and induce the secretion of significant levels of proinflammatory cytokines, TNF-α, IL-6 and IL-12p70 (Fig. 1). Significant induction of the anti-inflammatory cytokine, IL-10, was also observed with these supernatants. P1 and P2, however, differed in chemokine secretion with P2 displaying reduced chemokine secretion (Fig. 1). Stimulation with P2 supernatants induced levels of CXCL-10 that were reduced significantly compared to the controls. Secretion of CCL-2 was also reduced in P2, although the reduction was not significant. In contrast to P1 and P2, supernatants from Sp and Ss were poor inducers of DC cytokine secretion. Compared to the controls, the levels of TNF-α, IL-6, IL-12p70, IL-10 and CCL-2 were similar while CXCL-10 were reduced significantly (Fig. 1). Rm displayed intermediate activity, in that it induced the secretion of IL-6 and IL-10 from DCs to levels lower than P1 and P2. Secretion of other mediators was comparable to controls. These data suggest that different bacteria from the oral microbiome have differential effects on DC activation.

Fig. 1.

Differential response of dendritic cells (DCs) to metabolites from different bacteria in the oral microbiome. DCs were cultured overnight with supernatants from bacterial culture. Cytokine secretion was quantified by enzyme-linked immunosorbent assay (ELISA). Bar graphs depict the level of each cytokine [tumour necrosis factor (TNF)-α, interleukin (IL)−12p70, IL-6, IL-10] and chemokine (CCL-2 and CXCL-10). Results are mean ± standard error (s.e.) of six experiments. DC indicates controls where DCs were exposed to uninoculated Todd–Hewitt (TH) medium. P1: Pseudomonas aeruginosa FLR01 non-mucoid (P1); FLR02 mucoid (P2); Streptococcus pneumoniae (Sp); S. salivarius (Ss); and Rothia mucilaginosa (Rm).

DCs activated with different bacterial supernatants induce distinct T cell cytokines

DCs activated with different bacterial supernatants were cultured with purified allogenic T cells to determine the T cell responses. In keeping with DC cytokine secretion, DCs exposed to P1 and P2 supernatants induced the secretion of significant levels of IFN-γ, IL-10 and IL-22 from T cells (Fig. 2). The secretion of IL-10 was higher in P1 compared to P2. Secretion of all three cytokines was similar in co-cultures of DCs and T cells stimulated by supernatants from Rm, Ss, Sp and the controls (Fig. 2). These data confirm that among the bacterial supernatants tested, P1, P2 were most efficient in activating DCs.

Fig. 2.

Dendritic cells (DCs) activated with different bacterial supernatants induce distinct T cell cytokines. DCs exposed to bacterial supernatants were cultured with T cells for 6 days and the secretion of T cell cytokines was assayed by enzyme-linked immunosorbent assay (ELISA). Bar graphs depict the levels of the cytokines interferon (IFN)-γ, interleukin (IL)−10 and IL-22. Results are mean ± standard error (s.e.) of six experiments. DC + T: dendritic cells + CD4 T cells; P1: Pseudomonas aeruginosa FLR01 non-mucoid (P1); FLR02 mucoid (P2); Streptococcus pneumoniae (Sp); S. salivarius (Ss); and Rothia mucilaginosa (Rm); T: CD4 T cells.

Distinctive composition of secreted metabolites from different bacteria

We next determined whether the differential response of DCs to the bacterial supernatants is due to strain-specific differences in supernatant composition detected using GC-MS. The heat map shows an overview of the different metabolites secreted by different bacteria (Fig. 3). Many metabolites were found in all bacterial supernatants; however, there were a few unique metabolites associated with each strain's supernatant. Sp and Ss supernatants had very high levels of 2,3-butanediol (Bdl), a product of fermentation that is increased in the breath of CF patients [36]. In addition to Bdl, Ss (but not Sp) supernatant also had a high level of polyunsaturated fatty acids: elaidic acid, linoleic acid and oleic acid. Ss supernatant also had a high level of glutamic acid. In contrast to Sp and Ss supernatants, Bdl was not detected in the supernatants from P1, P2 and Rm. Instead, high levels of putrescine and glucose were detected in P1 and P2 supernatants. Putrescine or tetramethylenediamine is a polyamine produced by mammalian cells, and bacteria such as P. aeruginosa, for protection of the cell wall against antibiotic and oxidative damage [37]. Glucose is not a preferred carbon source for P. aeruginosa, which may account for the high levels of glucose observed in the supernatant. Supernatants from Rm had high levels of glutamine and a few other amino acids, along with elevated citric acid. However, the presence of citric acid was also detected in the TH medium control, suggesting a lack of consumption rather than production. These results indicate that although bacteria in the oral microbiome secrete many similar metabolites, each strain has its own distinct metabolic signature.

Fig. 3.

Distinctive composition of secreted metabolites from different bacteria. The metabolites in the supernatants of bacterial cultures were determined by gas chromatography–mass spectroscopy (GC-MS) profiling. The heat map depicts the intensity of ions from metabolites found in bacterial supernatants; the bacterial strains are identified at the top of the relevant column: red: TH media (control); green: P1 (Pseudomonas aeruginosa FLR01 non-mucoid); blue: P2 (Pseudomonas aeruginosa FLR02 mucoid; cyan: Rm (Rothia mucilaginosa); magenta: Sp (Streptococcus pneumonia); yellow: Ss (S. salivarius). TH = Todd–Hewitt media.

Response of DCs to unique metabolites present in the bacterial supernatants

Next, we examined whether the differential response of DCs to various bacterial supernatants is a consequence of the unique metabolites present in the supernatants. To discern this, DCs were exposed to Bdl, putrescine, glucose or a combination of putrescine and glucose at the concentrations mentioned in the Methods. In addition, we also used 2,3-butanedione (Bdn), which is a volatile compound produced in the same acetoin fermentation pathway as Bdl. Similar to Bdl, Bdn was also detected in the breaths of individuals with CF [36]. Bdn is more volatile than Bdl, which may be why only Bdl was detected by GC-MS in the bacterial supernatants. The effect of elaidic acid, linoleic acid and oleic acid was not studied because polyunsaturated fatty acids have already been reported to inhibit DC activation [38].

The optimal test concentration for each metabolite was identified by exposing DCs to various concentrations. Concentrations higher than those used here were found to be toxic to the cells, as determined by 7-aminoactinomycin D (7-AAD) staining (Supporting information, Fig. S1). Exposure of DCs to Bdn and Bdl (signature metabolite of Ss and Sp that were found in the breath of CF patients) did not result in the induction of TNF-α, IL-6, IL-10 or IL-12p70 compared to controls (Fig. 4). The effect was thus similar to what was observed with Ss and Sp bacterial supernatants. In contrast, the exposure of DCs to putrescine, one of the signature metabolites of P1 and P2, resulted in significant secretion of IL-12p70 compared to controls. The secretion of TNF-α, IL-6 and IL-10 was also slightly higher than controls in the putrescine group, but was not significant. DCs exposed to glucose, the other signature metabolite of P1 and P2 supernatants, displayed significant secretion of TNF-α, IL-6, IL-10 and IL-12p70 versus the control group (Fig. 4). More importantly, exposure of DCs to a combination of glucose and putrescine together resulted in significantly enhanced levels of all cytokines compared to each metabolite alone (Fig. 4). Putrescine and glucose thus seem to work in synergy. Altogether, these data suggest that DCs activated with the signature metabolite of the bacterial supernatants respond as they did to the supernatants themselves. The magnitude of cytokine induction was higher when DCs were exposed to the total supernatants than with a single metabolite, suggesting that other common metabolites also influence the DC response.

Fig. 4.

Response of dendritic cells (DCs) to unique metabolites present in the bacterial supernatants. DCs were exposed to signature metabolites of the bacteria and cytokine secretion was determined. Bar graphs depict the levels of the cytokines tumour necrosis factor (TNF)-α, interleukin (IL)−12p70, IL-6 and IL-10. Results are mean ± standard error (s.e.) of four experiments. Bdn-2,3: butanedione; Bdl-2,3: butanediol; Puet: putrescine; Glu: glucose; Glu + Puet: glucose + putrescine.

Cytokine secretion by T cells primed with DCs exposed to the signature metabolites

We investigated whether the DCs activated with signature metabolites induced T cell cytokine responses similar to those observed with the total supernatants. Secretion of IFN-γ and IL-22 was increased significantly over the control group when the T cells primed with DCs were exposed to glucose and putrescine together (Fig. 5). T cells primed with all other DCs induced levels of IFN-γ and IL-22 that were similar to the controls. IL-10 secretion by T cells was not significantly different in any of the groups (Fig. 5). These data confirm that the unique metabolites present in the bacterial supernatants, and total bacterial supernatants, influence DC functions similarly. The nature of the DC responses was similar, but the magnitude of the effect of total supernatant was higher than the single metabolite.

Fig. 5.

Cytokine secretion by T cells primed with dendritic cells (DCs) exposed to the signature metabolites. DCs exposed to signature metabolites were cultured with T cells before measuring T cell cytokine secretion. Bar graphs depict the levels of the cytokines interferon (IFN)-γ, interleukin (IL)−10 and IL-22. Results are mean ± standard error (s.e.) of four experiments. DC + T: dendritic cells + CD4 T cells; Bdn-2,3: butanedione; Bdl-2,3: butanediol; Puet: putrescine; Glu: glucose; Glu + Puet: glucose + putrescine; T: CD4 T cells.

Discussion

The microbiome and the metabolites produced by the microbes are emerging as critical mediators of immunity and tolerance in mucosal tissues. Here, we demonstrated that the metabolites secreted by various airway microbes are distinct and have different effects on the DC responses. P. aeruginosa produce high levels of putrescine and do not metabolize glucose, compared to streptococci. Metabolites produced by P. aeruginosa strains P1 and P2 are also able to induce DCs to secrete proinflammatory cytokines TNF-α, IL-6 and IL-12. Conversely, Streptococcus spp. (Sp, Ss) produce high levels of Bdl, which does not induce DC activation and cytokine secretion. R. mucilaginosa did not produce any detectable signature metabolite, despite Rm supernatants having an intermediate effect on DC activation. These results shed light on the important roles of microbial metabolites in regulating host immune responses.

Metabolites produced by the microbiome enable communication between the host and microbial community. Recent reports, particularly from the gut, highlight the role of metabolites in host physiology and pathophysiology where they control a large range of metabolic, inflammatory and even behavioural processes [39,40,41,42]. Information regarding the metabolites produced by microbes in the oral/airway microbiome is scarce, but is of great importance because of the role these metabolites probably play in various lung pathologies. Previous studies by Whiteson [36] highlighted the potential for Bdn or Bdl to serve as biomarkers for microbial activity in the breath of CF patients. Bdn is known to be toxic to humans; it is also known as diacetyl and is the main flavouring for microwave popcorn, where factory workers have developed bronchiolitis obliterans [43,44]. In this study, we confirm that Bdl and Bdn are produced mainly by the Streptococcus spp. and do not activate the DCs independently (Fig. 4). The production of pH neutral fermentation products, such as Bdl and Bdn by Sp and Ss, may therefore be a mechanism used by the bacteria to survive as undetected commensals in the anaerobic conditions of the oral cavity. However, during chronic respiratory infection with P. aeruginosa Bdn and Bdl may influence the ability of this opportunistic pathogen to induce inflammation. A recent study demonstrated that P. aeruginosa encapsulated in agar beads in the presence of Bdl persisted longer in the murine respiratory tract, induced enhanced TNF-α and IL-6 responses and resulted in increased colonization in the lung tissue by environmental microbes [45].

We also found high levels of putrescine in the supernatants of P1 and P2. Studies have demonstrated that P. aeruginosa relies upon the arginine decarboxylase and agmatine deiminase pathways to convert arginine into putrescine, which protects it from antibiotic treatment and possibly from the host immune response. Polyamines such as putrescine also aid in biofilm formation by P. aeruginosa [37]. High levels of all polyamines, including putrescine, were reported recently to be present in the bronchoalveolar lavage fluid (BALF) of children with CF and correlated positively with neutrophil infiltrate [46]. Recent studies also suggest a role of aberrant polyamine biosynthesis in carcinogenesis and tumour immunity. Our results suggest that putrescine is a weak activator of DCs and induces low levels of TNF-α, IL-6, IL-12 and IL-10 in DCs (Fig. 4). The effect of putrescine on DCs has not been investigated though other polyamines, such as spermine, have been reported to suppress inflammatory cytokine secretion in human PBMCs and M1 macrophages responding to LPS [47].

In addition to putrescine, glucose was also elevated in P1 and P2 supernatants (Fig. 3). P. aeruginosa are capable of using a variety of carbon sources. Succinate is often a preferred carbon source, and while many strains can grow or survive in minimal media with glucose as the carbon source, glucose is not used when another carbon source is available. Glucose or elevated glucose present in human tissues due to diabetes mellitus would be less likely to be consumed by P. aeruginosa, with consequences for immune cell activation. Interestingly, there are only a few studies delineating the effect of glucose on DCs. The study by Lu et al. [48] found that high glucose induced a proinflammatory cytokine profile in human DCs leading to DC maturation, consistent with our results. In contrast to the activating effect of high glucose on differentiated DCs, when DCs undergoing differentiation were exposed to high glucose-containing medium or hyperglycaemic sera, ROS production was increased and DC differentiation and maturation was impaired [49]. This suggests that the effect of glucose may vary with the stage of DC differentiation. Diabetic patients, however, display a greater susceptibility to respiratory infections. Our results also suggest that putrescine and glucose present in P1 and P2 supernatants synergize to enhance proinflammatory response of DCs (Fig. 4). Thus, the metabolites may work in tandem to enhance or dampen the effect of each other.

Finally, R. mucilaginosa is a widespread member of oral microbial communities which causes disease only in immunocompromised individuals. Thus, it may be a mechanism of its survival to go undetected by the immune system.

Altogether, our data demonstrate for the first time that different bacteria from the oral microbiome produce signature metabolites that have different effects on DC functions. The identification of signature metabolites and their effects on the host immune system can provide mechanistic insights into the diseases and may also be developed as biomarkers as metabolite detection could be fast and inexpensive. Furthermore, some signature metabolites may be detected before the onset of major diseases. In addition, the metabolites may also be used as novel therapeutic candidates and immune modulators for treatment of various airway diseases.

Acknowledgements

NIH grants AG045216 (to A. A.) and UL1 TR000153 to National Center for Research Resources and the National Center for Advancing Translational Sciences. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH. K. W. is supported by a Gilead CF Research Scholars award (app_00b072) and a UC Davis West Coast Metabolomics Center pilot grant (DK097154), where metabolomics data was collected with help from Megan Showalter and Professor Oliver Fiehn among many others. The authors would like to thank Dr Heather Maughan for feedback concerning the manuscript.

Disclosure

The authors declare no financial conflicts of interest.

Author contributions

K. W. provided bacterial supernatants, performed GC-MS analysis and helped in writing the manuscript. S. A. determined the effect of supernatants and metabolites on DCs. A. A. analysed the data, prepared the figures and wrote the manuscript.

Author notes

These authors contributed equally to this study.