Antibacterial Properties of the Mammalian L-Amino Acid Oxidase IL4I1

L-amino acid oxidases (LAAO) are flavoproteins that catalyze the oxidative deamination of L-amino acids to a keto-acid along with the production of H2O2 and ammonia. Interleukin 4 induced gene 1 (IL4I1) is a secreted LAAO expressed by macrophages and dendritic cells stimulated by microbial derived products or interferons, which is endowed with immunoregulatory properties. It is the first LAAO described in mammalian innate immune cells. In this work, we show that this enzyme blocks the in vitro and in vivo growth of Gram negative and Gram positive bacteria. This antibiotic effect is primarily mediated by H2O2 production but is amplified by basification of the medium due to the accumulation of ammonia. The depletion of phenylalanine (the primary amino acid catabolized by IL4I1) may also participate in the in vivo inhibition of staphylococci growth. Thus, IL4I1 plays a distinct role compared to other antibacterial enzymes produced by mononuclear phagocytes.


Introduction
L-amino acid oxidases (LAAO) are homodimeric flavoproteins that catalyse the stereospecific deamination of L-amino acid substrates to a keto-acid along with the production of H 2 O 2 and ammonia. These enzymes are widely expressed in many different organisms from prokaryotes to metazoans, of which snake venom LAAO being the most studied. Their function is still poorly understood. Venom LAAO have been suggested to act as toxins involved in the induction of apoptosis in a variety of different mammalian cell types. In addition, they are associated with the dysfunction of platelet aggregation [1].
Interleukin 4 induced gene 1 (IL4I1) is a secreted mammalian LAAO primarily expressed by activated mononuclear phagocytes, such as macrophages and dendritic cells, under the influence proinflammatory and T helper type 1 (Th1) mediators in vitro. Accordingly, IL4I1 is highly expressed in Th1 but not in Th2 granulomas. This enzyme has also been detected in B lymphocytes following activation by IL-4 and CD40, although at much lower levels [2]. At physiological pH and temperature, IL4I1 degrades the essential amino acid phenylalanine (Phe), producing phenylpyruvate, H 2 O 2 and ammonia. By this activity, IL4I1 depletes the microenvironment of an essential amino acid and induces the accumulation of potentially toxic products. We have demonstrated that IL4I1 is involved in the control of the adaptative immune response via its enzymatic activity [3,4].
Because of H 2 O 2 -dependent cytotoxic effects and the potential toxicity of other resulting catabolites, LAAO family members may play a variety of roles in immune defense in animals. A snake venom LAAO has been shown to present potent antibacterial activities against Gram-positive and Gram-negative bacteria which is related to H 2 O 2 production [5]. An LAAO with protective activity against bacterial mastitis has also been detected in mouse milk [6]. IL4I1 is phylogenetically related to fish LAAO [7], which have been shown to present antibacterial functions [8] and accumulate in ''granuloma-like'' structures induced by the infection with larval nematodes [9]. Recently, a new fish LAAO has been described, which may contribute to the innate immune defense against a variety of bacteria and protozoans [10]. As IL4I1 expression is induced in mononuclear phagocytes by pathogenderived signals, such as Toll-like receptor ligands [2], it may participate in the innate immune defense against pathogens in mammals, in addition to its regulatory effects on the specific immune response.In this work, we thus evaluated the effect of recombinant IL4I1 on Gram-positive and Gram-negative bacteria growth. We show that IL4I1 can act as an antibiotic, which could potentially contribute to the arsenal used by activated macrophages to generate an extracellular environment hostile to bacterial growth.

Results
We tested the antibacterial effect of conditioned media from IL4I1-expressing THP1 monocytes on four bacterial strains, representative of Gram-negative bacilli and Gram-positive cocci, i.e. Escherichia coli DH5a (E. coli), an Escherichia coli strain auxotrophic for Phe (B2599) and 2 Staphylococcus strains, a methicillin-susceptible Staphylococcus aureus (MSSA) and a coagulase-negative Staphylococcus (CNS). IL4I1 activity in these media was 340.88 pmol/H 2 O 2 /h/ml 666.27 ( Figure 1A). The IL4I1containing medium was then diluted in control-conditioned medium from THP1 cells at increasing ratios ( Figure 1B). We observed that the growth of all four bacterial strains was blocked in an IL4I1 dose-dependent fashion, indicating an antibacterial effect of the enzyme. IL4I1 degrades L-Phe and to a lesser extent L-tryptophan (Trp) to the corresponding a-keto-aminoacid, NH 3 and H 2 O 2 . The activity of IL4I1 is usually measured based on H 2 O 2 production in the presence of Phe ( Figure 1A). Since all the catabolites of the reaction are theoretically produced in equimolar amounts, we expected ammonia accumulation in the presence of IL4I1. Although IL4I1-mediated NH 3 production could not be detected in THP1 and THP1-IL4I1 conditioned media, because of a strong basal level mainly due to glutamine instability, we confirmed the production of ammonia in Phe-containing conditioned PBS ( Figure S1). The 24 h accumulation of ammonia in THP1 cultures (74.0616.2 mM) doubled in the presence of IL4I1 (153.8627.6 mM). Moreover, the presence of IL4I1 induced a 0.5 pH increase in the medium (THP1: pH = 7.5; THP1-IL4I1: pH = 8) that was not modified by bacterial growth. The Phe, Trp and phenylpyruvate content were measured using an HPLC technique. Similar to NH 3 , phenylpyruvate was significantly increased in the presence of IL4I1 ( Figure S2). Phe and Trp content were at the limit of sensitivity of the method. Two samples were transferred to a laboratory specialized in Phe quantification in human serum and urine. No significant variation in Phe content was measured, as expected from the low rate of Phe consumption (less then 10 mM/24 h).
We sought to determine which mechanisms associated with this reaction could be involved in the observed antibacterial effect ( Figure 2). We first explored the role of amino acid consumption in the bacterial microenvironment. For this purpose, bacteria were grown in DMEM/F12 supplemented with glutathione but lacking Phe and Trp. Both amino acids were alternatively added to the medium, alone or in combination ( Figure 2A). E. coli growth was not affected by amino acid deprivation. In contrast, the growth of the three other strains was blocked in Phe-and Trp-deprived media. As expected, the growth of B2599 was restored by Phe, but not by Trp addition. The presence of Trp in Staphylococcus cultures only partially restored their growth. However, MSSA could grow in Trp-depleted medium if Phe was present, while CNS growth was only restored by the presence of both amino acids. Thus, these three strains were variably auxotrophic for Phe and Trp.
We next explored the role of the three catabolites of the enzymatic reaction ( Figure 2B, C, D). Phenylpyruvate was weakly toxic at high concentrations (10 mM). In contrast, H 2 O 2 totally inhibited the growth of the four bacterial strains at 1 mM, despite their endogenous catalase production. This effect was abolished by the addition of glutathione in the culture media. Ammonium (NH4 + ) is considered as a source of nitrogen for bacterial growth. However, addition of the basic compound ammonia (NH 3 ) to the E. coli and staphylococci cultures displayed a toxic effect at doses $ 5 mM, which was reverted by buffering the media with HEPES.
We verified if combining all three IL4I1 catabolites could amplify their individual effect. Isomolar amounts of phenylpyruvate, H 2 O 2 and NH 3 were added to a culture medium lacking HEPES but containing minimal amounts of glutathione, Trp and Phe ( Figure 3). Under these conditions, the growth of E.coli and MSSA was sensitive to doses 10 times lower than when H 2 O 2 was used alone. A similar effect was obtained by mixing H 2 O 2 and NH 3 only ( Figure S3). Moreover, the inhibition of bacterial growth was lost in a HEPES-buffered medium containing the three catabolites ( Figure S4), suggesting that NH 3 potentiates the toxic effect of H 2 O 2 via basification of the medium. This was confirmed using NaOH as basificating agent. Indeed, E coli was sensitive to ten times less H 2 O 2 when the pH was increased from 7.5 to 8.5, an increase that can also be obtained with 5 mM NH 3 and that is not toxic by itself ( Figure S5). In summary, both amino acid depletion and the presence of phenylpyruvate, H 2 O 2 or NH 3 affected, to variable degrees, bacterial growth ( Table 1). Moreover, the presence of equimolar quantities of H 2 O 2 and NH 3 from the IL4I1 reaction could have a cumulative effect, indicating that their simultaneous production by the enzyme may amplify its antibacterial properties.
In order to confirm these putative mechanisms of IL4I1 action, we added Phe, Trp, glutathione, and HEPES, either separately or in combination, to bacterial cultures grown in conditioned media containing or not IL4I1. While IL4I1 was still toxic for the bacteria after the addition of a large excess (0.5 mg/ml) of Phe or After 24 hours, bacterial growth was monitored at an OD of 595 nm. Data are given as mean 6 SEM from five independent experiments performed in duplicate. Mean OD in 100% CM THP1 was 0.227 for E. coli, 0.266 for B2599, 0.450 for MSSA and 0.516 CNS, respectively. These OD were considered as 100% bacterial growth and the OD measured in the other culture conditions were calculated as percentages with respect to this control. Statistical analyses were performed using the Mann-Whitney test in comparison to 100% CM THP1. ***p,0.001 for all the bacterial strains at 100% CM THP1-IL4I1; **p,0.01 for E. coli, B2599 and MSSA and # p,0.05 for CNS at 75%; **p,0.01 for E. coli and B2599 at 50%; NS, not significant for MSSA and CNS at 50% and for all the strains at 25%. doi:10.1371/journal.pone.0054589.g001 Trp or both, either H 2 O 2 scavenging with glutathione or NH 3 buffering with HEPES were sufficient to restore full growth of E. coli and staphylococci ( Figure 4). This data suggest that, under our experimental conditions, the antibiotic action of IL4I1 is mainly due to H 2 O 2 and NH 3 production whereas its LAAO activity does not sufficiently deplete the medium of Phe or Trp to inhibit bacterial growth, as indicated by HPLC analysis.
The antibiotic effect of IL4I1 might be due to killing of the bacteria or simply to the blockade of their growth. To analyze this aspect, bacteria grown in conditioned media were reseeded in fresh medium without IL4I1. Whereas bacteria preincubated in THP1 control medium displayed vigorous growth in fresh medium, the growth of IL4I1-preincubated bacteria could not be restored under the same conditions ( Figure 5). This result indicates that IL4I1 is bactericidal.
We finally measured in vivo the antibacterial properties of IL4I1. A mean of 1.75610 8 CFU of MSSA resuspended in conditioned PBS containing or not the recombinant murine enzyme (specific activity 7.327 nmol/H 2 O 2 /h/ml 61.518) were injected into the peritoneum of C57Bl/6 mice. For this purpose, we used supernatants from HEK293 cells, which produce the murine form of the enzyme, thus avoiding a potential immune reaction to the human form. In addition, these cells are very resistant when cultured in PBS without serum. Twenty-four hours later, mice were sacrificed Figure 2. Susceptibility of bacteria to IL4I1 substrates and catabolites. Bacteria were serially diluted in DMEM/F12, with: (A) no Phe, 35.49 or 500 mg/ml Phe and/or no Trp, 9.02 or 500 mg/ml Trp; (B) ten folds dilutions of phenylpyruvate from 1 mM to 10 mM; (C) ten folds dilutions of H 2 O 2 from 1 mM to 1 mM or 64 mg/ml glutathione with or without 1 mM H 2 O 2 ; (D) ten folds dilutions of NH 3 from 5 mM to 50 mM or 15 mM HEPES with or without 50 mM NH 3 . After 24 hours, bacterial growth was monitored at an OD of 595 nm. Data are given as mean 6 SEM from five independent experiments performed in duplicate. *p,0.05 **p,0.01 ***p,0.001, Mann-Whitney test in comparison to DMEM/F12 without Phe and Trp (A), without phenylpyruvate (B) or according to the bars on the graph (C and D). doi:10.1371/journal.pone.0054589.g002  Figure 6A). This diminution was accompanied by significantly lower levels of plasma IFNc and a trend towards lower levels of the proinflammatory cytokine IL-6 ( Figure 6B). We did not detect significant differences in TNFa and IL-10 concentrations, the level of the latter being identical to those measured in the blood of naïve mice ( Figure S6). As the cytokines were measured 24 h after bacterial injection, this may be due to the kinetics of cytokine production after an acute infection [11]. These results indicate that IL4I1 could protect against bacterial growth in vivo. However, we have previously described IL4I1 as an immunoregulatory enzyme, which inhibits IFNc production by lymphocytes [2]. We thus verified that the diminution of the IFNc levels in the sera of mice receiving IL4I1 was not due to a direct effect of the enzyme on IFNcproducing cells. Mice were thus injected with lipopolysaccharide (LPS) along with IL4I1-PBS or HEK-PBS and cytokines were measured in the plasma at 24 h, while splenocytes were analyzed by flow cytometry for IFNc production. Under these conditions, no difference in cytokine levels was observed between the two groups of mice ( Figure 6C). The IFNc-producing cells were represented by T lymphocytes and NK cells. In all mice that were analyzed, these cells represented 14.665.1% of all T lymphocytes and 34.367.2% of all NK cells, respectively, regardless of the presence of IL4I1 (representative result in Figure S7). Thus, the diminution of plasma IFNc in mice challenged with bacteria and

Discussion
In this paper, we demonstrate that the phenylalanine oxidase IL4I1 is a bactericidal enzyme, which acts primarily through the production of toxic levels of H 2 O 2 and NH 3 . This antibacterial effect was observed on both Gram + and Grambacteria. IL4I1 catalyses the oxidative deamination of Phe and to lesser extent Trp thus producing equimolar amounts of an a-ketoacid (phenylpyruvate), H 2 O 2 and NH 3 [3]. The well-known toxic effect of H 2 O 2 was potentiated by basification of the medium by NH 3 , demonstrating that the IL4I1 antibacterial effect does not simply rely on H 2 O 2 production. Phe or Trp depletion might also participate to growth inhibition in bacterial strains auxotrophic for these amino acids. However, it did not appear to be a major mechanism of action in our in vitro experimental conditions, where no diminution of the Phe content could be evidenced. IL4I1 is produced by mononuclear phagocytes stimulated by bacterial products and pro-inflammatory cytokines, such as type I IFN, IFNc and TNFa [2]. In the context of bacterial infections, IL4I1 could be either secreted at the contact zone between the phagocytic cell and the bacteria, in the recently called ''phagosomal synapse'' [12] or released in the phagolysosome, in both cases contributing to the bactericidal arsenal of the macrophage.
Several amino acid degrading enzymes, produced by myeloid cells in mammals, have been demonstrated to participate in antiinfectious effects together with an immunosuppressive activity directed towards T lymphocytes [13]. These enzymes share a common mechanism of action: amino-acid depletion together with the production of a variety of toxic compounds, constituting a repertoire of weapons against a large spectrum of diverse microbial targets. The redundancy of this system may also suggest its importance for the host response. IL4I1 thus represents a new member of this complex and coordinated antimicrobial system.
Addition of IL4I1 to bacteria injected in mice diminished the bacterial load in the spleen, concomitantly reducing the inflammatory response, independently of the previously demonstrated immunomodulatory effect of IL4I1. In humans, we previously reported a high level of IL4I1 production in macrophages associated with tuberculosis granuloma [2], suggesting a role for IL4I1 in both the containment of the bacterial dissemination and modulation of the Th1 cell response, in order to preserve the organ from the consequences of uncontrolled inflammation. Such a mechanism has also been proposed for the indoleamine 2,3 dioxygenase enzyme in the context of listeria granuloma [14].
IL4I1 is phylogenetically derived from bony fishes LAAO [7], some of which have been shown to use the enzyme to limit the growth of parasite larvae in structures resembling granuloma [9]. Thus, as it has been described for other aminoacid-catabolising enzymes [13], IL4I1 may have evolved from ancestral innate antimicrobial functions to acquire a regulatory effect on the adaptive immune system.

Cell Culture, Media and Reagents
Monocytic THP1 and Human embryonic kidney 293 (HEK) cell lines were cultivated respectively in RPMI 1640 and DMEM Figure 5. IL4I1 bactericidal effect. Bacteria were grown 24 hours in serial dilutions of conditioned medium of THP1 or THP1-IL4I1 cells and OD measured. The lowest OD for both conditions was selected (t = 0). New DMEM/F12 containing 1% FCS was added, then bacterial re-growth was measured after 7 and 24 hours and compared between bacteria pre-cultured in THP1 and THP1-IL4I1 conditioned medium. Data are given as mean 6 SEM from five independent experiments performed in duplicate. *p,0.05 and **p,0.01, Mann-Whitney test. doi:10.1371/journal.pone.0054589.g005 (Life technologies, Saint Aubin, France) containing 10% FCS (PAA, les Mureaux, France), 2 mM L-glutamine, 100 UI/ml penicillin, 100 mg/ml streptomycin, and for THP1 1 mM sodium pyruvate (all from Life Technologies). Stable THP1 cell lines expressing human IL4I1 (THP1-IL4I1) were obtained as described in Marquet et al [2]. Stable HEK cell lines expressing murine IL4I1 (HEK-mIL4I1) were obtained as described in Boulland et al [3]. Other media used were DMEM/F12 (1:1) without Phe and Trp (Clinisciences, Montrouge, France). Phe and Trp from Clinisciences were used at concentrations starting from those found in RPMI 1640. Glutathione, HEPES, hydrogen peroxide (H 2 O 2 ) and phenylpyruvate were from Sigma Aldrich (Lyon, France), and ammonia (NH 3 ) from Honeywell Riedel-de Haën (Seelze, Germany). PBS, Amplex Ultra Red and LB agar were obtained from Life Technologies.

Preparation of Conditioned Medium
THP1 and THP1-IL4I1 were cultured in DMEM/F12 containing 1% FCS (0.5610 6 cells/ml) for 2 days and the supernatants used as culture medium for bacteria, in the presence or absence of additional Phe (0.5 mg/ml) and Trp (0.5 mg/ml), glutathione (64 mg/ml) and HEPES (15 mM). HEK-mIL4I1 and HEK transfected with the empty vector were cultured in PBS with calcium and magnesium for 24 hours and the resulting supernatants were used for mouse injections (conditioned PBS from HEK cells defined HEK-PBS and conditioned PBS from HEK-mIL4I1 cells, IL4I1-PBS).

Enzymatic Activity Measurement
IL4I1 activity was measured in conditioned medium from THP1, THP1-IL4I1, HEK and HEK-mIL4I1. The enzymatic assay was performed as described in Carbonnelle-Puscian et al. [15] using Amplex Ultra Red and expressed as pmol H 2 O 2 produced in the presence of Phe per hour per ml of conditioned medium.

Determination of Bacterial Growth
For the analysis of bacterial growth in conditioned medium, Escherichia coli (E. coli) and E. coli B2599 (Phe auxotroph bacteria) from Life Technologies and ASAP (A Systematic Annotation Package for community analysis of genomes, University of Wisconsin), respectively, were used. Methicillin-susceptible Staphylococcus aureus (MSSA) and Coagulase negative Staphylococcus (CNS) were obtained from routine diagnostic specimens. Bacteria were grown on LB agar. For use in experiments, 24 hours old bacterial colonies were picked and resuspended in 100 ml of PBS. Bacteria were serially diluted 1/20 in conditioned medium (10 ml in 200 ml final, in 96-well flat bottom plates). After incubation for 24 hours, bacterial growth was monitored using a microplate spectrometer (Optima Fluostar, BMG Labtech, Champigny, France) by measuring the optical density (OD) at 595 nm [16]. All the experiments were analysed on these serial dilutions, allowing the selection of the dilution still containing at 24 h bacteria in the growth phase for the control (OD595 usually from 0.3 to 0.5). The corresponding bacterial content in CFU/ml was measured by serial dilution plating on LB agar and colony counting. For an OD of 0.3, the CFU content per ml was 9.9610 7 for E. coli, 14.0610 7 for B2599, 69.5610 7 for MSSA, 26.0610 7 for CNS. For the analysis of the bactericidal effect, following 24 hours of bacteria incubation in 150 ml conditioned media the OD was measured (time = 0), then 150 ml of fresh DMEM/F12 containing 1% FCS was added to the dilution that allowed obtaining the lowest bacterial content in both IL4I1 and control conditions. These cultures were used for subsequent OD measures at 7 and 24 hours.

Mouse Injections
Naïve female C57BL/6 mice (6 to 10 weeks old) were purchased from Charles River. Groups of 3 mice were injected intraperitoneally with a mean of 1.74610 8 CFU of MSSA or 20 mg lipopolysaccharide resuspended in 500 mL conditioned PBS from HEK or HEK-mIL4I1. After 24 hours, the mice were sacrificed and the spleens and blood collected. All animal experiments were performed in accordance with local institutional (ComEth) guidelines and with appropriate personal and institutional authorizations. The animals were kept under strict observation during the first hours of the experimental procedure to identify signs of distress. No sign of distress were observed. After 24 h, blood (100 ml) was taken via the retrorbital sinus and mice were sacrificed by cervical dislocation (without anaesthesia) in order to have access to internal organs (spleens) and evaluate bacterial charge.
For the in vivo bacterial count assay, dissociated spleen sample supernatants were serially diluted and then inoculated on LB agar plates. Bacterial counts were performed after overnight incubation at 37uC.

ELISA
Cytokines interferon c (IFNc), tumor necrosis factor a (TNF), interleukin 6 (IL-6) and interleukin 10 (IL-10) were detected in mouse plasma samples diluted four fold in sample diluent by ELISA according to the manufacturer's instructions (BD Biosciences, Le Pont de Claix, France).

Statistical Analysis
All experiments were performed in duplicate (in vitro bacterial count assay) or triplicate (in vivo experiments) and were performed 3 to 5 times as stated in the figure legends. Data are given as mean 6 SEM. For statistical analysis, the nonparametric Mann-Whitney test was performed using the GraphPad Prism software. p values less than 0.05 were considered statistically significant. Figure S1 Quantitative determination of ammonia/ammonium. NH 3 and NH 4 + were measured in 24 hours Phe-containing conditioned PBS from THP1 and THP1-IL4I1, using an enzymebased assay. Results from 6 and 5 independent samples, respectively, with mean 6 SEM, are shown. *p = 0.03, Mann-Whitney test. (TIF) Figure S2 HPLC analysis of Phe, Trp and phenylpyruvate content in THP1 and THP1-IL4I1 conditioned media. Twenty ml of DMEM/F12 media were separated by a mixed mode ion exchange and reverse phase HPLC technique. (A) 2 mM of Phe (left), phenylpyruvate (center) or Trp (right) were added to DMEM/F12 to identify the retention times (black arrows). Red arrows indicate the dimethylaminobenzoic acid internal standard (2 mM). (B) Representative chromatograms of 24 hours conditioned media: THP1, left; THP1-IL4I1, right. Lower panels are enlargements of the circled areas corresponding to the phenylpyr-uvate peaks. The first peak corresponds to phenylpyruvate and the second peak to a degradation product. The red dotted line is placed at 100 units on the absorbance scale to allow a better visualization of the difference in peak areas. (C) Results expressed as peak area ratio (for phenylpyruvate, the sum of both peaks was considered) are shown for five independent samples, which were analysed in two tests. Left: phenylpyruvate (*p = 0.016, Mann-Whitney test), medium: Phe (not significant), right: Trp (not significant). Note that for phenylpyruvate, the concentrations corresponding to the ratios shown are less than 600 mM. (TIF) Figure S3 Cooperative effect of H 2 O 2 and NH 3 on bacterial growth inhibition. Bacteria (MSSA) were serially diluted in RPMI 1640 with or without 1 mM to 1 mM isomolar combinations of two of the following products: phenylpyruvate, H 2 O 2 and NH 3 . After 24 hours, bacterial growth was monitored at an OD of 595 nm. Data are given as mean 6 SEM from five independent experiments performed in duplicate. **p,0.01 and *p,0.05, Mann-Whitney test in comparison to RPMI 1640. (TIF) Figure S4 Susceptibility of bacteria to catabolite mix in HEPES containing medium. Bacteria were serially diluted in DMEM/F12 with or without isomolar addition of phenylpyruvate, H 2 O 2 and NH 3 from 1 mM up to 1 mM. After 24 hours, bacterial growth was monitored at an OD of 595 nm. Data are given as mean 6 SEM from five independent experiments performed in duplicate. **p,0.01, Mann-Whitney test in comparison to DMEM/F12. (TIF) Figure S5 Potentiation of H 2 O 2 antibacterial effect via basification of the medium. (A) E. coli growth is preserved in pH ranging from 6.5 to 9.5. RPMI has a pH of 7.5. The table displays the various additions made to RPMI to obtain pH ranging from 6.5 to 13. E. coli was cultured 24 hours in the different RPMI-based media and bacterial growth was monitored at an OD of 595 nm. Data are given as mean 6 SEM from 2 to 8 independent experiments performed in duplicate (B) E. coli becomes sensitive to a non-toxic dose of H 2 O 2 in basic medium (pH equal or superior to 8.5). E. coli was cultured 24 hours in RPMI containing or not NaOH (2.5 or 5 mM) and/or H 2 O 2 (100 mM). The pH of the media are indicated in the figure. Bacterial growth (OD 595 nm) are given as mean 6 SEM from four independent experiments performed in duplicate *p,0.05 and **p,0.01, Mann-Whitney test, according to the bars on the graph. (TIF) Figure S6 Cytokines in naïve mice plasma. Interferon-c (IFNc), tumor necrosis factor a (TNF), interleukin-6 (IL-6) and interleukin-10 (IL-10) were measured by ELISA in diluted plasma samples from naïve mice. Two to three mice were analyzed. The mean result is indicated by the horizontal bar. (TIF) Figure S7 IFNc production in T cells and NK cells from mice injected with IL4I1 and LPS. Mice were injected i.p. with LPS resuspended in HEK-PBS (n = 3) or in IL4I1-PBS (n = 3). Splenocytes were collected at 24 h and restimulated in vitro with PMA and ionomycin. Intracellular IFNc was measured by flow cytometry in the NK1.1 and the CD3 positive lymphocyte populations. No significant difference was observed in the splenocytes from mice receiving or not IL4I1. The dot-plots show representative results in NK (right) and T cells (left) from one mouse. (TIF) Methods S1

Supporting Information
(DOC)