Characterization of synergistic antibacterial effect of silver nanoparticles and ebselen

Abstract The emerging and spreading of multi-drug resistant (MDR) bacteria have been becoming one of the most severe threats to human health. Enhancing oxidative stress as mimicking immune system was considered as a potential strategy to fight against infection of MDR bacteria. In this study, we investigated the antibacterial efficiency of such a strategy which combines silver nanoparticles (AgNPs) with ebselen. The results showed that AgNPs and ebselen combination had significant synergistic killing effects both on Escherichia coli (E. coli) and Staphylococcus aureus (S. aureus) in vitro, including model strains of China Veterinary Culture Collection and MDR clinical isolates, which is similar as the combination of silver ion and ebselen. AgNPs exhibited to be a strong inhibitor of bacterial thioredoxin reductase, same as a free silver ion. Ebselen mitigated the cytotoxicity of AgNPs to HeLa cells. However, in a bacteria-cell coexistence condition, the synergistic bactericidal effect was only observed on S. aureus (p<.05), while the temporary synergistic inhibitory effect on E. coli within 4 hours treatment (p<.01). In mice infection model, a combination of AgNPs and ebselen did not increase protection against the challenge of clinical E. coli CQ10 strain. Our data demonstrated that AgNPs and ebselen combination may be a promising strategy to fight against the increasingly MDR bacteria targeting bacterial thiol redox system.


Introduction
In recent decades, the infections of multidrug-resistant (MDR) bacteria have been increasingly horrible threats to human and animal health. It is known that about 700,000 people died from the infection of MDR bacteria worldwide every year and the number will reach up to 10 million by 2050 if the human could not stop them [1,2]. Furthermore, the extensively drug-resistant (XDR) and pan drug-resistant (PDR) bacteria have been frequently isolated from livestock and patients [3]. In particular, the emerging and dissemination of the mcr-1 colistin resistance gene threaten the immune system's last line of defense [4]. The World Health Organization (WHO) recently alerted the coming of a postantibiotic era in which many common bacterial infections will be untreatable [5]. It is urgent to explore novel antibiotics or non-antibiotics antimicrobial strategies on behalf of human health.
Reactive oxygen species (ROS)-based agents would be an alternative weapon in this campaign [3]. ROS are metabolic byproducts of aerobic respiration [3]. Eukaryotes and bacteria have been evolved relatively to comprehensive antioxidant system to maintain their redox homeostasis [6][7][8]. In case of E.coli, the principle oxidative stress regulator OxyR, which induce 30 antioxidant genes, can be quickly oxidized by 0.1 lM H 2 O 2 and subsequently activate antioxidative regulon resulting in a steady-state level of H 2 O 2 about 20 nM [9,10]. However, bacteria usually can not survive in phagosome where the concentration of H 2 O 2 is over 5 lM for the limited antioxidant capacity. A promising non-antibiotic strategy known as antimicrobial photodynamic inactivation (APDI) uses certain wavelength light to activate nontoxic dye (photosensitizer) to generate ROS [11]. However, the outer membrane of Gram-negative bacteria limits the diffusion of dye into the cytoplasm, hence, APDI of Gram-negative bacteria is not so effective as that of Gram-positive bacteria [11].
Silver (Ag) has been used to treat infection over 2000 years and it was recently proved to kill bacteria by inducing ROS, increasing the permeability of the outer membrane and inactivating key respiratory enzymes, etc. However, the silver ion can spontaneously react with the anionic mineral present in the host, such as chloride, phosphate, and protein such as albumin and result in an inactivated complex. Silver nanoparticles (AgNPs) have higher antimicrobial capacity than silver ion and without these disadvantages. It has known that AgNPs have a killing effect on hundreds of bacterial species including MDR bacterial strains and inhibit the formation of various bacterial biofilm [12]. AgNPs also have received extensive studies of antifungal and antiviral effects, including HBV and HIV [13][14][15]. Recently, several studies showed combining AgNPs with general antibiotics could increase the antibacterial effect on MDR stains by induction of ROS [16,17]. However, the potential cytotoxicity and genotoxicity of AgNPs impeded its application for the biosafety concerns [18].
Ebselen is a lipid-soluble organoselenium which previously exhibited neuroprotective, antioxidant and anti-inflammation activities by mimicking the function of glutathione peroxidase (Gpx), and has been in phase 2 clinical trial for the treatment of ischemic stroke and hearing loss. It was recently repurposed to be an antibacterial agent for killing MRSA and vancomycin-resistant S. aureus (VRSA) at a concentration of 20 lM [19,20]. Ebselen also displayed a synergistic antibacterial effect with silver ion against MDR Gramnegative bacteria [21]. The antibacterial mechanism of ebselen was the inhibition of bacterial thioredoxin reductase (TrxR) [22][23][24].
Altogether, the previous studies suggest that combining AgNPs with ebselen could be an alternative non-antibiotic strategy targeting bacterial conserved antioxidant system to combat against MDR, XDR and PDR bacterial infection. In the present study, we have performed a series of synergistic antibacterial assay with AgNPs and ebselen against Gram-positive and Gram-negative bacteria in vitro, cellular condition and mouse infection model on behalf of the biosafety of AgNPs. The data showed that the synergy is a species-dependent model and would rely on the biological situation.

Synthesis and characterization of AgNPs
AgNPs were synthesized by a previously reported method [26]. Briefly, Gibberella sp. (China Type Culture Collection Accession No. M2012524) was cultured in yeast-peptone medium (Sangon, China) for 1 week. Mycelia were collected and its cell-free lysate was used to synthesize AgNPs at 37 C. The synthesized AgNPs were washed twice with ultrapure water and then characterized with scanning electron microscopy (SEM, Hitachi, Japan) and ultraviolet-visible (UV-vis) spectroscopy (Nanodrop 2000, Thermo Scientific, MA, USA). The protein coronae on the surface of AgNPs were characterized by liquid chromatography coupled with tandem mass spectrometry (LC-MS, Thermo Scientific, USA). The AgNPs stocking colloidal solutions were covered with a layer of liquid paraffin and kept at 4 C. For the working solution, AgNPs were washed again and resuspension in fresh distilled water.

Induction of ROS and apoptosis by AgNPs in HeLa cells
The ROS fluorescent probe 2 0 ,7 0 -dichlorodihydrofluorescein diacetate (DCFH-DA, Solarbio Life Sciences, Beijing, China) was added into HeLa cells (4 Â 10 6 cells/mL) with a final concentration of 10 lM. The cells were then cultured at 37 C for 1 h. After washing with PBS three times, the cells were treated with 800 lM Vit. C, 10 lg/mL AgNPs, or 800 lM Vit. C þ 10 lg/mL AgNPs, respectively, at 37 C for 30 min. After another three times washing, the fluorescence intensity at 488 nm (excitation wavelength) and 525 nm (emission wavelength) was measured with flow cytometry (BD, USA) to calculate the production of ROS. RPMI1640 was the control of the treatment. The experiments were repeated three times independently.
HeLa cells (10 6 cells/mL) were treated the same as above. Cells were washed twice with PBS. Binding buffer (195 lL), Annexin V-FITC (5 lL) and propidium iodide (10 lL) were then added into the wells. The cells were gently mixed and incubated in a black box at room temperature for 20 min. The apoptosis analysis was performed with flow cytometry. Each treatment has 6 replicate wells.

Measurement of MBC of AgNPs for E. coli and S. aureus
The overnight cultured E. coli CVCC2081 and S. aureus CVCC1882 were collected by centrifugation and resuspended with RPMI1640 medium, adjusting densities to 4 Â 10 4 , 4 Â 10 5 , 4 Â 10 6 and 4 Â 10 7 CFU/mL, respectively. AgNPs colloidal solutions were diluted with RPMI1640 medium to reach the concentrations of 1.25, 2.5, 5, 10, 20, 40 and 80 lg/mL. The bacterial suspension was mixed with AgNPs at different concentrations with equal volume in a 48-well plate (200 lL/ well) with three replicates each treatment. The mixtures were then incubated at 37 C for 12 h. After neutralizing with 1 mM Na 2 S (Sangon, China), the living bacteria were counted on agar plates after dilution. The experiment was repeated three times independently.
To determine the bactericidal nature of AgNPs, Vit C was used to scavenging ROS induced by AgNPs. 10 7 CFU/mL bacteria were treated with 800 lM Vit. C, 10 lg/mL AgNPs, or 800 lM Vit. C þ 10 lg/mL AgNPs, respectively, at 37 C for 3 h. The live bacteria were counted on agar plates after neutralizing AgNPs with 1 mM Na 2 S.

Inhibition of recombinant E.coli TrxR by AgNPs
The experiment was performed with the method previously described [21]. The dosage of AgNPs was adjusted to the same concentration of silver ion with reference [21], in detail, 0.54 lg/mL AgNPs is equal to 1 lM silver ion, and then 2-fold serial dilution. Pure recombinant 100 nM E.coli TrxR, in the presence of 200 lM NADPH, was incubated with a serial concentration of AgNPs which is equal to 1, 2, 3, 4, 5 lM silver ion in mass for 10 min, respectively, then the assay buffer containing 5 lM E. coli Trx and 2 mM DTNB was added and TrxR activities were detected with a DTNB reduction assay.

Fractional inhibitory concentration index of AgNPs and ebselen
The synergism between AgNPs and ebselen against E. coli and S. aureus were determined by two-dimensional microdilution assay. Briefly, bacteria were cultured in salt-free Luria/Millerbouillon growth medium. Minimum inhibitory concentration (MIC) of AgNPs and ebselen against E. coli or S. aureus (4 Â 10 5 CFU/mL) were measured separately. Subsequently, AgNPs and ebselen were combined with a two-dimensional model in 96well microtiter plate and the final concentration was 2 MIC, 1 MIC, 1/2 MIC, 1/4 MIC, 1/8 MIC, 1/16 MIC, and 1/32 MIC, respectively. 4 Â 10 5 CFU/mL E. coli or S. aureus was then added into each well. The plates were incubated at 37 C for 12 h. The fractional inhibitory concentration (FIC) was measured by the automatic growth-curve tester (Biosceen, Turku, Finland). The FIC index was calculated as the formula:

FIC index ¼ MIC of AgNPs in the combination
=MIC of AgNPs alone þ MIC of Ebselen in the combination =MIC of Ebselen alone: The combinational effect was judged by the following standard: (i) antagonistic effects, FIC index > 2.0 indicate, (ii) additive effects, 0.5 < FIC index < 2.0 indicate, and (iii) synergistic effects FIC index < 0.5.

Synergistic antibacterial effect of AgNPs-ebselen combination in vitro
The synergistic antibacterial effect of ebselen and AgNPs were analyzed at the concentration of 10 7 CFU/mL. The culture of E. coli CVCC2081 and S. aureus CVCC1882 were added into a 100-well plate (100 lL/well). Three treatment groups were set as AgNPs group (0.625, 1.25, 2.5 and 5 lg/mL), ebselen group (5 and 20 lM) and AgNPs þ ebselen group. Three wells were set for each treatment and incubated at 37 C. OD 600 was detected every hour by an automatic growthcurve tester (Biosceen, Finland). The experiment was repeated three times. The live bacteria in each treatment were checked on the agar plate after 9 h treatment.

Cytotoxicity of AgNP-ebselen combination on HeLa cells
The AgNPs-ebselen combination was formulated with orthogonal design, in which the final concentrations of ebselen were 0, 5, 10, and 20 lM and AgNPs were 2.5, 5, 10, 20 and 40 lg/mL, respectively. The mixture compound was added into 60 h grown HeLa cells and incubated at 37 C for 12 h. After washing twice with PBS, the intracellular ATP was detected as previously described. Three wells were set in each group and the experiment was repeated three times.

Antibacterial effect of AgNPs-ebselen combination in the presence of HeLa cells
The 2 Â 10 7 CFU/mL E. coli CVCC2081 and S. aureus CVCC1882 were added to HeLa cells after it is grown for 60 h, respectively and treated with 10 lg/mL AgNPs, 20 lM ebselen, or 10 lg/mL AgNPs þ 20 lM ebselen at 37 C for 12 h. The live bacteria were counted at 4, 8 and 12 h after treatment. Three replicate wells were set for each treatment. The experiment was repeated three times.
Anti-MDR clinical E. coli strain in vitro and in vivo STEC CQ10 was isolated from piglet and resistance to 14 antibiotics with mcr-1 gene [25]. The antibacterial experiment in vitro was carried out just as previously described for E. coli CVCC2081. A total of 32 SPF level KunMing mice, 20 g weight each, were randomly divided into 4 groups (drug control, challenge control, AgNPs treatment and AgNPs-ebselen combination treatment) with 8 mice each group. All mice, except drug control, were injected intraperitoneally with 2 LD 50 doses of STEC CQ10, 1 h later, mice in AgNPs treatment group were injected subcutaneously with 2.4 mg AgNPs/Kg weight, in 0.5 ml solution. The mice in the combination group were injected subcutaneously with 2.4 mg AgNPs/Kg weight AgNPs and 30 mg ebselen/Kg weight in two sides of rear feet, respectively. Mice in drug control were injected with AgNPs and ebselen in two separate sites. Ebselen was dispersed in 0.5% CMC-Na solution with a final concentration of 5% DMSO.

Statistical analysis
All data were expressed as mean ± standard deviation (SD.). The ANOVA was used to assess the difference between treatments by using OriginPro 2017 C (Northampton, MA, USA.). A p-value less than .05 was considered as the significance level.

Characterization of AgNPs
AgNPs was synthesized with hypha lysate of Gibberella with the solution turning yellow and an absorption peak at 400 nm (Figure 1(A)). The size of AgNPs was in the range of 2-24 nm (Figure 1(B)) with irregular shapes of spheroidal, polygon, and flake observed by scanning electron microscopy (SEM) (Figure 1(C)).The results of mass spectroscopy indicated that the protein corona of AgNPs was composed of 38 proteins, among which there were 18 acidic proteins, 16 basic proteins, and 4 neutral proteins (Appendix Table A1). The abundance of protein indexed by isoelectric point (pI) indicated that S0D40 accounted for 25.12% (pI ¼ 10) and that W7MPB2, S0DUJ0 and S0DYW4 accounted for 41, 10.5, and 7% (pI ¼ 5.3, 6.5 and 5.8), respectively. Four neutral proteins accounted for 9.1% and the rest of the proteins were extremely low (Figure 1(D)).

Ascorbic acid quenched ROS induced by AgNPs on HeLa cells
It showed that the cytotoxicity of AgNPs was concentrationdependent. Lower than 2.5 lg/mL AgNPs exhibited less cytotoxicity (<10% death) and higher than 20 lg/mL caused 100% death. The cytotoxicity induced by interval concentration AgNPs correlated with cell growth stage, in particular, for 60-h-grown HeLa cells, 10 lg/mL AgNPs caused half cells dead, taken as IC50 value (Figure 2(A)).
Stimulated by 10 lg/mL AgNPs for 30 min, the cellular ROS of HeLa cells increased by 58.0%, which was significantly higher than that of control cell (p ¼ .0014) (Figure 2(B)). The addition of 800 lmol/L ascorbic acid quenched significantly ROS induced by AgNPs. It also reduced the intracellular ROS level of a normal cell, both to the same level, which decreased by 35% compared with the control.
The influence of ascorbic acid on cell apoptosis caused by AgNPs was detected by the same experiment. The total amount of apoptosis and death was 58.2% (30.1 and 28.1%, respectively) when HeLa cells were treated with 10 lg/mL AgNPs for 12 h (Figure 2(C)). When ascorbic acid was added in the same treatment, the amount slightly decreased to 54% (20.5 and 33.5%, respectively). There was no significant difference between these two groups but they were both significantly different from the control group. The result indicated that ascorbic acid did not reduce the apoptosis caused by AgNPs and they may rely on another mechanism to cause apoptosis, instead of the ROS pathway.

Ascorbic acid eliminated the lethal effect of AgNPs on bacteria
The MBC values of AgNPs toward a series of concentration of bacteria were measured (10 4 , 10 5 , 10 6 and 10 7 CFU/mL) in 12 h treatment. The results showed that the MBC values of AgNPs for E. coli were 2.5, 2.5, 10, and 20 lg/mL (Figure 3(A)). The MBC value for 10 4 -10 6 CFU/mL S. aureus was 5 lg/mL and for 10 7 CFU/mL was 10 lg/mL (Figure 3(B)), indicating that MBC values were changed in species-specific and concentration-dependent manners.
To characterize the kill effect of AgNPs, ascorbic acid was used to quench ROS.
The result showed that after 20 lg/mL AgNPs treatment for 3 h, the concentrations of E. coli and S. aureus were reduced from 10 7 CFU/mL to 10 3 CFU/mL and 10 CFU/mL, respectively. However, both E. coli and S. aureus maintained alive in the same order of magnitudes when 800 lmol/L ascorbic acid was added. The biomass of bacteria in the control groups was increased by 10 folds (Figure 3(C,D)). This suggested that the killing effect of AgNPs on bacteria is mostly dependent on the induction of ROS.

AgNPs inhibits the activity of TrxR
To see whether AgNPs interrupt bacterial thiol-dependent antioxidant system to result in the elevation of ROS, we investigated the effects of AgNPs on the bacterial thioredoxin reductase activity. AgNPs exhibited to be an efficient inhibitor of E. coli TrxR. AgNPs was observed to have concentration-dependent inhibitory effects on TrxR, which decreased by 80% in the treatment of 0.54 lg/mL AgNPs. TrxR decreased by about 20% in the treatment of 0.017 lg/mL AgNPs (Figure 4).

Fractional inhibition concentration of AgNPs and ebselen
Given both AgNPs and ebselen have the inhibitory activities against bacterial TrxR [21][22][23][24], on the other hand, the similar level of IC 50 and MBC values of AgNPs for cells and bacteria, as well as their distinct response to ROS induced by AgNPs, we tried to combine with ebselen to decrease the dose of AgNPs on behalf of biosafety. The fractional inhibitory concentration (FIC) was measured with a standard method. The results showed that MIC of AgNPs for E. coli and S. aureus were 5 lg/mL and 6.67 ± 2.4 lg/mL, respectively, MIC of ebselen was 40 lmol/L and 2.5 lmol/L, respectively. However, by screening in series of combination, MIC of a combination of AgNPs and ebselen for E. coli was 0.3125 lg/mL AgNPs and 2.5 lmol/L ebselen, for S. aureus, 2.5 lg/mL AgNPs and 0.156 lmol/L ebselen. The FIC index for E. coli and S. aureus was 0.125 and 0.479 ± 0.14, respectively (Table 1). Obviously, AgNPs and ebselen showed synergistic antibacterial effect both on E. coli and S. aureus.

Synergistic killing effect of AgNPs and ebselen in vitro
The growth curve revealed that 5 lg/mL AgNPs or 20 lM ebselen had no inhibitory effect on 10 7 CFU/mL E. coli. However, the combination of AgNPs and ebselen showed synergistic antibacterial effect, in particular, 0.625 lg/mL AgNPs and 20 lM ebselen could completely kill 10 7 CFU/mL E. coli in 9 h treatment ( Figure 5(A)). Since 20 lmol/L could kill completely 10 7 CFU/mL S. aureus (data not shown here). It was observed that the growth of S. aureus was completely inhibited by 10 lM ebselen, but did not kill them, meanwhile, 5 lg/mL AgNPs only partially inhibited the growth of 10 7 CFU/mL S. aureus. However, the combination of 0.625 lg/mL AgNPs and 10 lM ebselen completely killed 10 7 CFU/mL S. aureus in 9 h treatment ( Figure 5(B)).

Ebselen mitigate cytotoxicity of AgNPs on HeLa cells
We also investigated the effect of AgNPs-ebselen combination on HeLa cells. The 60 h-grown HeLa cells were treated with a series of AgNPs-ebselen combinations for 12 h. The living cells were detected with intracellular content of ATP. The results indicated that the viability of HeLa cell decreased with increasing concentration of AgNPs. IC 50 value was about 10 lg/mL (consistent with the result of Figure 2(A)) and at this condition, 20 lM ebselen significantly increased the alive ratio of HeLa cells to 71.6%. At the treatments of 5 lg/mL AgNPs, 10 lM ebselen significantly mitigate the cytotoxicity of AgNPs, reach up to 100% alive. When the concentration of AgNPs over 20 lg/mL, it was hard to mitigate the cytotoxicity with the addition of ebselen ( Figure 6).

Synergistic antibacterial effect in cellular condition
Based on the above results, we subsequently investigated the synergistic antibacterial effects of AgNPs and ebselen in  the condition of coexistence of HeLa cell. Even though 0.625 lg/mL AgNPs and 20 lM ebselen (or 10 lM) combination was observed as a kill effect on E. coli and S. aureus in vitro, in cellular condition, the best antibacterial combination was 5 lg/mL AgNPs and 20 lM ebselen according to a couple of pilot experiments (Figure 7). In 12 h treatment, the timesurvival curves showed the AgNPs and ebselen combination possess bactericidal effect for S. aureus, but not for E. coli. The alive S. aureus decreased by 2.5 lgC after being treated with AgNPs-ebselen combination, while in treatment of AgNPs alone decreased by 1.5 lgC. The treatment of ebselen killed more than 99% S. aureus in 8 h, after that S. aureus grew 10-fold in 4 h. As for E. coli, the combination was observed better inhibitory role than AgNPs or ebselen treatment in 4 h and then gradually reached to similar inhibition condition (Figure 7).

Synergistic antibacterial effect on MDR clinical isolates in vitro and in vivo
To elicit the synergistic effect of AgNPs and ebselen combination against MDR clinical isolate, STEC CQ10 was finally used  14 Synergistic a Data collected from three independent measurements are presented as mean ± SD. b Values of FIC index above 2.0 indicate antagonistic effects, values between 0.5 and 2.0 indicate additive effects, and values of less than 0.5 indicate synergistic effects. FIC: fractional inhibitory concentration; MIC: minimum inhibitory concentration; AgNPs: silver nanoparticles; SD: standard deviation.
to assess the antibacterial effect both in vitro and in vivo. The result indicated that the antibacterial efficiency against MDR clinical isolate in vitro were exactly same with China Veterinary Culture Collection Center (CVCC) model strain. Specifically, MBC for STEC CQ10 was 0.625 lg/mL AgNPs and 20 lM ebselen (Figures 5(A) and 8(A)).
To ensure the incompetence of AgNPs and ebselen combination to E. coli in biological condition, we carried out a mice infection model with STEC CQ10 strain. The results showed that the survival ratio of mice in treatment of AgNPs alone was slightly higher than the combination group (p ¼ .516), all of the mice in the challenge control group died in 36 h after challenge with 2 LD50 CQ10 strain (Figure 8(B)). The mice in the drug control group did not show any abnormal symptom, judging by its activities and food consumption.

Discussion
Antibiotic generally targets a conserved bacterial domain which is absent or quite different from eukaryotes [27]. The clinical antibiotics usually attack either bacterial ribosome, cell wall synthesis and lipid membrane integrity, singlecarbon metabolic pathway or DNA maintenance. It has proved that bacteria had evolved pristine resistant genes in ancient time or develop a certain mechanism to the resistance of these antibiotics through mutation under antibiotic selection pressure [28][29][30]. At the perspective of this harsh condition, ROS-based therapeutic strategies drew increasingly interesting for biologist and medicinal chemistry community [3]. However, ROS exhibits nonspecific rapid oxidative damage to biomacromolecules, such as protein, lipid and DNA of both bacteria and eukaryotes. A problem raised to those of whom is concerned is how to regulate the level of ROS so as to meet the therapeutic role in biological conditions.
It is known that nanoparticle treatments could induce ROS both in vitro and in vivo. As for AgNPs, there were two purpose mechanisms for the production of ROS, including the dissolved silver ion and fenton reaction by chelatable iron ions [31][32][33]. In this study, 12 h-IC 50 of AgNPs for HeLa cells was 10 lg/mL (Figure 2(A)). After 3 h treatment of 10 lg/mL  . The time-survival curve of E. coli and S. aureus treated with AgNPs and ebselen coexistence with HeLa cells. HeLa cell has incubated in 96-well plates for 60 h before the addition of 10 7 CFU/mL bacteria. Final concentration of 5 lg/mL AgNPs, 20 lM ebselen, and 5 lg/mL AgNPs þ 20 lM ebselen were used in this experiment, respectively. The data collected from three independent experiments were expressed as mean ± SD. *p < .05, **p < .01, and ***p < .001.
AgNPs, ROS level of HeLa cell increased by 58% (Figure 2(B)). At the same time, the MBC of AgNPs for E. coli was 2.5-20 lg/mL and for S. aureus was 5-10 lg/mL, depending on the biomass of bacteria (Figure 3(A,B)). Given antioxidant ascorbic acid rescued E. coli and S. aureus and scavenged ROS in HeLa cells (Figures 2(B) and 3(C,D)), AgNPs displayed identical oxidative stress upon both bacteria and mammalian cells. The results were consistent with previous reports which addressed that the effective concentrations of AgNPs on viruses, bacteria, fungi, algae and mammal cells are in the same order of magnitude [34]. Therefore, a combination strategy should take into concern to reduce the dose of AgNPs while increase in bacterial sensitivity to ROS for a therapeutic purpose in biological condition.
In theory, ebselen would be one of the best candidate to combine with AgNPs since it selectively inhibits bacterial key antioxidative enzyme [24,35], while it displays antioxidant impact in mammalian cells [36]. In the present study, AgNPs dose-dependent inhibited E. coli TrxR in vitro and displayed synergistic antibacterial effect combining with ebselen for E. coli and S. aureus with FIC index less than 0.5 ( Table 1). The synergistic bactericidal effects were also observed in vitro. When combined with ebselen, the MBC of AgNPs for E. coli  (Figure 8(A)). The silver ion and ebselen combination caused a rapid depletion of glutathione and inhibition of thioredoxin system resulting in bactericidal effect [21]. Indeed, AgNPs combining with general antibiotics enhanced the antibacterial effects by induction of ROS [17]. The data demonstrated that the universal redox system could be a universal antibacterial target.
Ebselen and other selenium compounds have glutathione peroxidase and peroxiredoxin activities to protect mammalian cells from oxidative damage without down-regulation of endogenous antioxidant response [37]. The ROS scavenging activities of selenium compounds were measured in HEK293T and HeLa cells treated with H 2 O 2 [37]. In this study, 20 lM ebselen mitigated significantly cytotoxicity of HeLa cells induced by 10 lg/mL AgNPs, though it did not work when the concentration of AgNPs was over 20 lg/mL ( Figure 6). Given the MBC of AgNPs for E. coli and S. aureus less than 1 lg/mL combining with ebselen, this combination strategy can be effective to attack bacterial infection in biological condition. Unexpectedly, in bacteria-cell coexistence condition, the synergistic antibacterial effects were not as effective as in vitro, especially for E. coli. We chose the combination of 5 lg/mL AgNPs and 20 lM ebselen which was relatively safe for HeLa cells and displayed bactericidal effect to both E. coli and S. aureus according to the above results in vitro. The kill ratio of S. aureus was more than 99.9% in the presence of HeLa cell and the bactericidal effect of the combination group was significantly higher than that of other two groups (p<.05). However, under the identical biological condition and treatment, the growth of E. coli was inhibited but the biomass still increased more than 10-fold in 8 h of treatment (Figure 7). We further tested the antibacterial effects against MDR clinical E.coli strain, STEC CQ10. Though the efficiencies were the same as CVCC model strains in vitro, the AgNPs and ebselen combination did not increase the survival ratio of mice challenge with STEC CQ10 (Figure 8). It was easy to understand that S. aureus showed more sensitive to the treatment of AgNPs-ebselen combination than E. coli as S. aureus lacking of glutathione-dependent antioxidant system. However, we did not know why E. coli displayed resistance to the enhanced oxidative stress by AgNPs-ebselen combination in biological conditions. For most bacteria including E. coli, the master regulon OxyR mediate about 30 genes in response to oxidative stress to maintain ready-state of redox condition [38]. Recently, a novel mechanism of H 2 S-mediated protection against oxidative stress in E. coli was unveiled [39]. The mstA gene and L-cysteine are involved in the protection by the generation of H 2 S [39]. To our knowledge, it was the first report to describe E. coli resistance to oxidative stress in biological conditions. The answer to the underlying mechanism would be important to the coming global war against MDR bacterial infection.
In mice model, the survival ratio in AgNPs treatment was slightly higher than that of the control group (p ¼ 0.516). It is possible to improve the survival ratio of mice if we properly modify the protein corona of AgNPs. In this study, the AgNPs coated with 38 proteins of Gibberella sp. in the course of synthesis. We did not know whether or not it absorbed serum protein after administration and how these protein corona influences its antibacterial nature in vivo so far. It has been known that the formation of protein corona in a physiological environment caused nano-medicines loss of efficiency or missing of targets [40][41][42]. Miclaus et al. addressed that the weakly attached protein reduce nanocrystal formation in a serum-concentration-dependent manner, the strongly attached corona act as a site for sulphidation which decreases the toxicity of AgNPs [43]. On the other hand, Barbalinardo inferred that absorption of serum protein to AgNPs surface is essential for internalization and toxicity to cell [44]. Protein corona affects not only the pharmacokinetics of AgNPs in experimental animals but also the antibacterial efficiency in vitro. MBC values of bacteria and IC 50 values of cells measured in a culture medium with fetal bovine serum (FBS) were higher than the values measured in the medium without FBS [45,46].
Altogether, the development of oxidative stress-based non-antibiotics strategies would shed light on the fight against the increasingly MDR bacterial threats. Without a doubt, it requires interdisciplinary study and interlaboratory collaboration.