Human β-Defensin 2 Induces Extracellular Accumulation of Adenosine in Escherichia coli

Abstract

Human β-defensins are host defense peptides performing antimicrobial as well as immunomodulatory functions. The present study investigated whether treatment of Escherichia coli with human β-defensin 2 could generate extracellular molecules of relevance for immune regulation. Mass spectrometry analysis of bacterial supernatants detected the accumulation of purine nucleosides triggered by β-defensin 2 treatment. Other cationic antimicrobial peptides tested presented variable outcomes with regard to extracellular adenosine accumulation; human β-defensin 2 was the most efficient at inducing this response. Structural and biochemical evidence indicated that a mechanism other than plain lysis was involved in the observed phenomenon. By use of isotope (¹³C) labeling, extracellular adenosine was found to be derived from preexistent RNA, and a direct interaction between the peptide and bacterial nucleic acid was documented for the first time for β-defensin 2. Taken together, the data suggest that defensin activity on a bacterial target may alter local levels of adenosine, a well-known immunomodulator influencing inflammatory processes.

INTRODUCTION

Human β-defensins (hBD) are cationic antimicrobial peptides produced predominantly by epithelial cells (1). Notably, in the intestine, an environment populated by a dense and rich microbial community, epithelial defensins reinforce the host innate defense (2). These peptides are normally expressed at low basal levels, but hBD-2 and -3 can be upregulated by a variety of microbial and inflammatory stimuli (3, 4). In particular, altered hBD-2 expression has been reported to correlate with the development of intestinal inflammation (5, 6). On the other hand, intestinal inflammation has been associated with alterations in the microbiota—for example, with shifts in the relative abundance of Proteobacteria, such as Escherichia coli (7, 8). Elucidating how the interaction between hBD and their bacterial targets impact inflammatory processes can aid our understanding of host-microbe networking in health and disease.

The most-studied mechanism of action of cationic antimicrobial peptides involves membrane permeabilization with eventual cell lysis; however, other mechanisms and cellular targets have been described for many peptides, including human defensins (9–12). Moreover, in addition to their antimicrobial activity, manifold immunomodulatory properties of defensins have been reported, and many of these reports describe direct effects on immune cells and cytokine expression (13–15). In some cases, the interaction of peptides with bacterial components is implicated in their immunomodulatory effects (16, 17). In this context, the present study hypothesized that hBD-2 activity on E. coli could result in the generation of extracellular mediators with relevance for immunomodulation.

MATERIALS AND METHODS

Chemicals.

Lyophilized synthetic human β-defensin 2, human β-defensin 3, and human α-defensin 5 were purchased from the Peptide Institute Inc., sheep myeloid antimicrobial peptide 29 (SMAP-29) from AnaSpec Inc., and magainin I from Sigma-Aldrich. Adenosine (Ado), guanosine (Guo), cytidine (Ctd), and uridine (Urd), as well as their corresponding monophosphates and nitrogen bases, were purchased from Sigma-Aldrich. The trimethylsilylation reagent N,O-bis(trimethylsilyl)trifluoroacetamide (BSTFA) with 1% trimethylchlorosilane (TMCS) was purchased from Supelco Analytical.

Culture conditions.

Escherichia coli strain W (DSM 1116) was cultivated in a mineral medium, referred to below as M4, composed of MgSO₄ (0.02 g/liter), citric acid (0.2 g/liter), K₂HPO₄ (1 g/liter), and NaNH₄HPO₄ (0.32 g/liter) and supplemented with 0.2% glucose. Cultures were inoculated (160 μl; six to eight replicates for each condition in all experiments) in 100-well plates in a Bioscreen C growth analysis system (Oy Growth Curves) and were incubated at 37°C using medium-amplitude shaking. Optical density (OD) measurements were taken every 15 min. At logarithmic phase (4 h), cultures were treated with 40 μl of an aqueous solution of an antimicrobial peptide. Untreated control cultures were handled in parallel in all experiments by the addition of 40 μl of sterile water instead of the peptide. Supernatants were filtered through 0.2-μm-pore-size membranes for further analyses 2 h after treatment, except where indicated otherwise. Three replicate cultures from each condition were monitored for a further 20 to 24 h to generate the corresponding growth curves.

LC-MS.

Bacterial supernatants were analyzed for the presence of defensin-induced compounds by using a 6460 TripleQuad liquid chromatography-mass spectrometry (LC-MS) system (Agilent Technologies, Santa Clara, CA, USA) with electrospray ionization (ESI) coupled to a 1200 series LC. The samples were injected on a reversed-phase C₁₈ column (length, 50 mm; inside diameter [i.d.], 2.1 mm; particle size, 1.8 μm) and were isocratically eluted with 95% water, 5% methanol, and 0.1% formic acid at a flow rate of 0.1 ml/min. Ionization was performed in the positive mode, with the following ion source parameters: N₂ flow, 9 liters/min at a temperature of 300°C; nebulizer pressure, 25 lb/in²; sheath gas (N₂) flow, 7 liters/min at a temperature of 300°C; capillary voltage, 4,000 V; charging voltage, 1,000 V; fragmentor voltage, 135 V. Tandem MS with collision-induced fragmentation was employed to determine the identities of compounds. Quantitative analysis was performed in the multiple-reaction-monitoring (MRM) mode, using commercial compounds as external standards. Specific parameters, expressed as parental ion m/z (collision energy [CE], in volts): quantifier fragment m/z/qualifier fragment m/z, were as follows: for Ado, 268 (30): 136/119; for Guo, 152 (30): 135/110; for Ctd, 244 (30): 112/95; for Urd, 113 (30): 40/70; for AMP, 348 (30): 136/97; for GMP, 364 (30): 152/135; for CMP, 324 (50): 112/95; for UMP, 649 (10): 325/97; for A, 136 (30): 119/92; for G, 152 (30): 135/110; for C, 112 (30): 52/95; and for U, 113 (30): 40/70. In experiments where more than two compounds were quantified simultaneously, chromatographic separation was achieved using a C₁₈ column (length, 250 mm; i.d., 4.61 mm; particle size, 5 μm) in 0.1% formic acid at 1 ml/min with a water-methanol gradient as follows: 0 to 14 min, 100:0 at 20°C; 14 to 28 min, 90:10 at 40°C.

Viability tests and microscopy.

In parallel with the analysis of cell-free supernatants, defensin-treated bacterial cultures were examined for viability and structural changes. For CFU counts, cultures were serially diluted in a solution of 0.9% sterile NaCl and were plated onto LB agar plates. Protein, DNA, and ATP concentrations were determined using the bicinchoninic acid (BCA) protein assay (Pierce Biotechnology), the Quant-iT PicoGreen double-stranded DNA (dsDNA) reagent (Invitrogen/Molecular Probes), and the BacTiter-Glo microbial cell viability assay (Promega), respectively. For fluorescence microscopy, 2 h after hBD-2 treatment, cells were fixed with paraformaldehyde and glutaraldehyde (4%, vol/vol), incubated in 10 mg/ml aqueous 4′,6-diamidino-2-phenylindole (DAPI) for 5 min, washed twice with Tris-EDTA (TE) buffer, and mounted on glass slides for analysis under a Zeiss photomicroscope (excitation at 365 nm; emission at 445 to 450 nm). Alternatively, fixed cells were processed for transmission electron microscopy (TEM), as described previously (18).

Isotopic labeling experiments.

Isotopic labeling and determination of isotopic ratios were performed to elucidate the origin of defensin-induced extracellular compounds. Bacteria were labeled by the addition of 0.5 g/liter [U-¹³C]glucose (99% ¹³C) ([¹³C₆]Glu; Euriso-Top) to the culture medium. Total DNA and RNA were extracted and purified from cultures immediately before defensin treatment by using commercial kits (Qiagen) and were subsequently degraded according to a standard enzymatic protocol (19). Products were lyophilized and were incubated with a trimethylsilylation reagent and pyridine (4:1) for 1 h at 100°C for derivatization. ¹³C/¹²C ratios were determined by gas chromatography-isotopic ratio mass spectrometry (GC-IRMS) in a GC IsoLink system connected via a combustion interface to a MAT 253 spectrometer (Thermo Scientific) as detailed elsewhere (20). Data were normalized by calculations described previously (21).

Statistical analysis.

Data obtained in three or more replicates were tested for statistically significant differences between means by one-way analysis of variance (ANOVA), with an α of 0.01, using the Holm-Sidak test for pairwise comparison. The analyses were performed using SigmaPlot, version 11.0 (Systat Software, Inc., Chicago, IL, USA).

RESULTS

Purine nucleosides accumulate extracellularly in E. coli after hBD-2 treatment.

Supernatants sampled from hBD-2-treated E. coli cultures or untreated controls were comparatively analyzed by LC-MS, revealing two peaks present exclusively after hBD-2 treatment (Fig. 1A, peaks 1 and 2). The first was composed of two main ions of m/z 136 and m/z 268, and the second of one major ion of m/z 152 (Fig. 1B). Fragmentation patterns generated by tandem MS were used to identify the compounds in each peak. The ion of m/z 268 fragmented in a single ion of m/z 136. The fragment spectra for the ions of m/z 136 and m/z 152 (Fig. 1C) were then compared to a Web-based data repository (22) and to commercially available standards, and the compounds in peaks 1 and 2 were identified as the purine nucleosides adenosine (Ado) and guanosine (Guo), respectively.

Fig 1.

hBD-2 induces extracellular accumulation of purine nucleosides. Supernatants from E. coli cultures, either left untreated or treated with hBD-2 (20 μg/ml), were analyzed by LC-MS. (A) Ion base peak chromatogram. (B) Characterization of differential peaks (peaks 1 and 2) by mass spectra in scan mode. (C) Fragmentation pattern of target ions by tandem MS at a collision energy of 50 V.

Quantitative characterization (Fig. 2) showed that the treatment of logarithmic-phase E. coli cultures with hBD-2 resulted in the presence of extracellular nucleosides at a low micromolar range, detected exclusively after peptide addition and not in untreated controls. The variation over time after hBD-2 treatment highlighted differences in the temporal dynamics: the accumulation of Ado was accentuated at later time points (Fig. 2A). The dose-response curve (Fig. 2B and C) demonstrated that below 10 μg/ml, hBD-2 was not able to induce accumulation of the compounds; toward higher doses, the concentrations of both compounds increased. Ado was more abundant than Guo at intermediate doses (20 to 30 μg/ml). On the other hand, the response also differed according to cell density: the concentration of Ado was higher than that of Guo when cultures were treated at higher OD values (Fig. 2D and E). The growth curves corresponding to the dose-response and density-response experiments (Fig. 2C and E) corroborated the idea that Ado accumulation correlates with the peptide/cell ratio rather than exclusively with a growth inhibition effect. Moreover, Ado is a well-known signaling molecule with a plethora of biological activities in human cells, including important immunomodulatory properties (23). Therefore, we chose to focus on this compound for further investigation.

Fig 2.

Quantitative characterization of extracellular nucleoside accumulation in response to hBD-2. Samples were analyzed by tandem MS (quantitative MRM mode). Results are means ± standard deviations from three biological replicates. (A) Variations in the extracellular concentrations of Ado and Guo in E. coli cultures over time after treatment with hBD-2 (20 μg/ml). (B) Extracellular concentrations of Ado and Guo in E. coli cultures 2 h after treatment with different doses of hBD-2. (C) Representative bacterial growth curves corresponding to cultures analyzed in the experiment described in the legend to panel B. The broken vertical line indicates the treatment point. (D) Extracellular concentrations of Ado and Guo in E. coli cultures of increasing cell densities, 2 h after treatment with hBD-2 (20 μg/ml). (E) Representative bacterial growth curves corresponding to cultures analyzed in the experiment described in the legend to panel D.

Different antimicrobial peptides have contrasting abilities to induce extracellular adenosine accumulation in E. coli.

In addition to hBD-2, two other human defensins were tested (human β-defensin 3 and human α-defensin 5), and two nonhuman peptides (SMAP and magainin I) were also analyzed for comparison. At the dose of 20 μg/ml, all peptides were equally able to arrest bacterial growth in liquid cultures (data not shown). In contrast, the extracellular concentrations of Ado generated by the bacteria after treatment were remarkably different for each peptide; hBD-2 was the most efficient at eliciting this response (Table 1). In another comparative approach, the accumulation of extracellular Ado in response to hBD-2 was evaluated in a different strain, the probiotic E. coli strain Nissle 1917, using conditions identical to those specified for E. coli strain W. Treatment of the probiotic strain with hBD-2 (20 μg/ml) also resulted in the presence of extracellular Ado (0.632 ± 0.009 μM) 2 h after peptide addition.

Table 1.

Concentrations of adenosine in E. coli supernatants in response to different antimicrobial peptides

Peptide	Description	Extracellular Ado concn (μM)a
Human β-defensin 2	Epithelial β-sheet defensin	0.426 ± 0.013
Human β-defensin 3	Epithelial β-sheet defensin	0.127 ± 0.013
Human α-defensin 5	Paneth cell β-sheet defensin	0.023 ± 0.002
Sheep myeloid antimicrobial peptide 29 (SMAP-29)	Mammalian α-helical cathelicidin	0.265 ± 0.006
Magainin I	Frog skin α-helical defensin	≤0.015

^aMeans ± standard deviations for three biological replicates, analyzed 2 h after peptide addition (20 μg/ml).

Adenosine accumulation does not result from plain cell lysis.

Next, bacterial viability and membrane integrity were investigated in the defensin-treated cultures, in order to evaluate the occurrence of peptide-mediated lysis. Bacterial growth was arrested after treatment with hBD-2 at 20 μg/ml (Fig. 3A); accordingly, metabolic activity was also immediately impaired, and CFU counts showed that the cultures were no longer viable shortly after hBD-2 addition (Fig. 3B). The release of intracellular molecules into the medium was examined as an indication of lysis (Fig. 3C). As with untreated controls, no release of ATP, DNA, or proteins was observed after hBD-2 treatment. Fluorescence microscopy confirmed that intracellular material was kept within discrete cells in hBD-2-treated bacteria, in contrast to cultures subjected to freeze-thaw lysis. Importantly, extracellular Ado levels were not significantly increased following the lysis protocol. These results indicated that although cellular processes were severely affected by hBD-2, intracellular contents were not indiscriminately released into the extracellular medium.

Fig 3.

hBD-2 does not cause indiscriminate lysis. (A) Growth curve of E. coli either left untreated or treated with hBD-2 (20 μg/ml). The broken vertical line indicates the treatment point. Results are means ± standard deviations from three biological replicates. (B) Viability of E. coli cultures after treatment with hBD-2 (20 μg/ml). (Left) ATP-dependent metabolic activity. (Right) Viability on agar plates. Results are means ± standard deviations for four determinations. (C) Assay comparing E. coli cultures that were left untreated, treated with hBD-2 (20 μg/ml), or lysed by a conventional freeze-thaw protocol. (Top) Presence of marker molecules in the supernatants. Samples were collected 2 h after treatment (or immediately for ATP measurements). Results are means ± standard deviations for four determinations. (Bottom) Imaging of intracellular material (DAPI) with reciprocal phase-contrast micrographs. Bar, 2.5 μm. (D) Transmission electron microscopy comparing untreated cultures and cultures treated with hBD-2 (20 μg/ml). Arrows indicate mild plasmolysis (arrow 1), severe plasmolysis with enlarged periplasmic space (arrow 2), and lysed bacteria (arrow 3). Bars, 1 μm (top and center) and 0.5 μm (bottom).

Structural evidence was obtained by transmission electron microscopy (TEM). Untreated E. coli cells (Fig. 3D, left) showed intact membranes and cytoplasm organization, and small areas of plasmolysis were observed in only a few cells (8%). In contrast, in hBD-2-treated cultures (Fig. 3D, right), 82% of cells presented extensive plasmolysis, with large periplasmic spaces. Intracellular contents were retained, except for 8% of visually detected lysed cells. Our observations indicate that plain bacterial lysis was not solely responsible for Ado release after hBD-2 treatment.

Extracellular adenosine is derived from preexistent bacterial RNA.

Isotopic labeling experiments were performed to investigate whether extracellular Ado in hBD-treated E. coli cultures originated from preexistent nucleic acids. Cultures grown in a medium containing either [¹²C]Glu or [¹³C]Glu were treated with hBD-2 (20 μg/ml). The isotopic ratio of extracellular Ado was compared to that of nucleic acids extracted immediately before treatment and was found to be in the same range as the preexistent RNA in labeled cultures (Fig. 4A). Targeted tandem MS (Fig. 4B) demonstrated that ribonucleotides and pyrimidine nucleosides were also present in the supernatants of treated cultures, supporting the conclusion that extracellular Ado is derived from bacterial RNA. In contrast, deoxyribose compounds were not detected. Interestingly, the addition of total RNA purified from E. coli increased the concentration of Ado detected after hBD-2 treatment (Fig. 4C). This was observed exclusively in cell-free supernatants from hBD-treated cultures; the increase was negligible in the corresponding harvested cells, indicating the involvement of an extracellular bacterial factor. Electrophoretic analysis suggested a direct interaction between the peptide and RNA; however, this was not sufficient to cause Ado accumulation in sterile medium alone.

Fig 4.

Extracellular Ado is derived from preexistent RNA. (A) Isotopic (¹³C) enrichment of adenine nucleosides in total DNA and RNA and of extracellular Ado 2 h after hBD-2 treatment. E. coli was grown either in unlabeled M4 medium or in [¹³C]Glu-containing M4 medium and was treated with hBD-2 (20 μg/ml). Means ± standard deviations for three replicates are shown. SN, supernatant. *, P < 0.001; n.s., not significant. (B) Extracellular concentrations of different ribonucleosides and related compounds released by E. coli cultures either left untreated or treated with hBD-2 (20 μg/ml). Compounds were quantified by tandem MS 2 h after treatment. Means ± standard deviations for triplicate measurements from three biological replicates are shown. (C) Influence of extracellularly added RNA on Ado accumulation. Cells and supernatants from E. coli cultures treated with hBD-2 (20 μg/ml) were separated immediately after peptide addition. Harvested cells resuspended in fresh medium, cell-free supernatants, and sterile medium were further incubated at 37°C for 2 h in the absence or presence of total RNA purified from E. coli (1 μg/ml). (Top) Bar chart showing the relative concentration of extracellular Ado generated under each condition. Values are expressed relative to extracellular Ado levels from E. coli cultures after the same hBD-2 treatment. Means ± standard deviations for five biological replicates are shown. (Bottom) Electrophoretic profile of samples in a 2% agarose gel stained with ethidium bromide.

DISCUSSION

The key question in this study—whether human defensin triggers an immunomodulatory response from enteric bacteria—was approached by searching bacterial metabolites released from E. coli into the extracellular medium after hBD-2 treatment. Our results demonstrated that hBD-2 induces selective accumulation of purine nucleosides, including Ado, which is an important and well-studied immunomodulator, validating our initial hypothesis. The experimental setup was designed by considering the effective concentrations reported for hBD-2, ranging from 10 to 50 μg/ml (24–26). In a physiologic context, the precise local concentrations of peptide at the site of defensin action are not known, and it is reported that most of the antimicrobial activity secreted by the intestinal mucosa is retained in the mucus layer, contributing to uneven distribution and variable local concentrations (27). In human fecal samples, concentrations of >50 ng/ml were found; these were estimated to be a dilution reflecting even higher concentrations present in the mucosa (28).

The possible effects of Ado release in the context of the human body, particularly in the context of the intestinal mucosa, are manifold, depending on which of the four known Ado receptors are being stimulated in the host (29). The micromolar concentrations found in our in vitro studies are in the range of basal levels of Ado in human tissues (<1 μM) and consistent with 50% effective concentrations (EC₅₀s) reported for the A₁ and A₃ receptors (0.3 μM) and the A_2A receptor (0.7 μM) (30). The A_2A receptor has been shown to be implicated in the anti-inflammatory activity of Ado (31) On the other hand, the predominant receptor type expressed in intestinal epithelial cells, A_2B, has a lower potency (EC₅₀, 24 μM) and can trigger proinflammatory signals (30, 32, 33). The final in vivo outcome of the phenomenon reported here is difficult to predict. Nevertheless, exploring localized production of Ado at the site of contact with the microbiota represents a valuable asset for the therapeutic manipulation of Ado signaling, which is a promising approach to the treatment of inflammatory disorders, though often limited by the risk of side effects due to the broadness of Ado activities (23).

Our study demonstrated that the ability to induce Ado accumulation in E. coli is not a general feature of different antimicrobial peptides, nor is it exclusive to β-sheet molecules. The secondary structure of antimicrobial peptides is important for membrane-targeted effects, but interactions with other cellular targets and alternative mechanisms of action are found across different structural classes (34). The peptide most efficient at inducing the accumulation of Ado was hBD-2. This peptide is produced by epithelial cells in the intestinal mucosa and is upregulated in inflamed tissue (5). Furthermore, the probiotic E. coli strain Nissle 1917, tested under conditions identical to those used for E. coli strain W, responded to hBD-2 by accumulating extracellular Ado at even higher levels. Taken together, these data are consistent with a possible therapeutic relevance for this interaction in the inflamed gut. Further studies to investigate the species specificity of the response, including other Gram-negative bacteria and also Gram-positive bacteria, would be interesting. Nevertheless, our observations with E. coli Nissle 1917 are noteworthy, since this probiotic strain has been employed successfully in the treatment of intestinal inflammatory disorders, and interestingly, its ability to induce hBD-2 production by intestinal epithelial cells has been reported as a potential beneficial effect for patients (35–37).

With regard to the possible mechanisms leading to increased concentrations of extracellular Ado, our results implicate a distinctive nonlytic kind of membrane damage, resulting in extensive plasmolysis. To our knowledge, no such effect had been reported yet for hBD-2, but retraction of the cytoplasmic membrane after sublethal treatment with porcine β-defensin has been described for Salmonella enterica serovar Typhimurium (38). Studies with model vesicles revealed that hBD-2 caused the release of a small marker, but not of molecules of 3,000 Da or more, and that this peptide is unlikely to be arranged in channel-forming oligomers (39). This model supports the conclusion that small metabolites could be released from E. coli following hBD-2 treatment independently of a lysis event. Moreover, stress-induced plasmolysis is known to cause the selective release of periplasmic enzymes from E. coli, including a nucleotidase activity that is able to convert AMP into Ado (40). In this sense, a periplasmic nucleotidase could be the bacterial factor contributing to Ado accumulation in defensin-treated E. coli, and the response would be influenced by membrane architecture and the subcellular distribution of the enzyme.

In addition to membrane-targeted effects, it is possible that other hBD-2 activities are involved in the generation of extracellular Ado. Our data indicated that Ado accumulation was favored at lower peptide/cell ratios, and we speculate that these conditions would promote plasmolysis and avoid general lysis. It has been reported that sublethal concentrations of antimicrobial peptides can elicit selective responses from bacteria, an effect not necessarily related to the microbicidal activity (41), and the use of peptide concentrations close to the lethal dose may favor the identification of alternative mechanisms and targets (42). Such mechanisms, already described for some cationic peptides, include direct binding and/or degradation of target nucleic acids (11, 43, 44), which is in agreement with our observations. While the nature of hBD-2–RNA interaction is not yet clear, this work represents, to our knowledge, the first suggestion of a nucleic acid-targeted activity for hBD-2.

Conclusion.

The central finding of this study—the identification of adenosine as an extracellular bacterial metabolite following defensin treatment—represents a thus far undisclosed link between known players in intestinal inflammation: epithelial antimicrobials and Ado signaling. While the immunomodulatory properties of Ado have been well studied in diverse inflammatory conditions, the view of indigenous bacteria as a direct source of Ado has only recently begun to be appreciated. In conclusion, besides contributing to the understanding of host-microbe interactions, this work offers a new perspective on the mechanism of action of human β-defensins, to be considered beyond bactericidal effects, as a component of the interkingdom signaling network.