The precursor solution was 10 mL titanium IV isopropoxide supplied by Sigma Aldrich mixed with 10 mL of anhydrous isopropanol (from Sigma) using a magnetic stirrer at 300 rpm. The titanium alkoxide reactivity was lowered by the use of acetylaceton. The spontaneous hydrolysis of the titanium isopropoxide was obtained by the quick addition of 75 mL of acidified water. The reacting medium was then heated to 100˚C under reflux for almost 8 hours (peptidization process). After this step, the dispersion was cooled down to room temperature, approximately 20˚C. A clear, yellow, anatase crystallized nanoparticles suspension was obtained, and stored in the dark.

The morphology and the particle sizes were characterized using a Philips CM 20 transmission electron microscope (TEM). The accelerating voltage was 200 kV. The samples were dispersed in methanol by ultrasonication. A drop of the suspension was then laid on a carbon-coated grid and dried. Selected Area Electron Diffraction (SAED) was performed to determine the crystalinity of the structure. The interplanar spacings were evaluated from the SAED patterns using the following formula:

λL = Rd (1)

where λL is the constant of the microscope, R is the ring radius, and d is the interplanar spacing. The constant of the microscope was calculated by measuring the radius of a gold standard pattern whose interplanar spacings were well documented in scientific publications [30] .

Four micro-organisms were used for photokilling experiments: Escherichia coli LE392, Enterotoxigenic Escherichia coli H10407, Lactobacillus casei rhamnosus Lcr35^® and Staphylococcus aureus (SA51). These bacteria have a size comprised between 0.5 and 5 µm. E. coli cells were cultured at 37˚C for 24 h in Nutrient Broth medium at pH 7.2 (Biokar diagnostics) containing Tryptone (10 g・L⁻¹), Meat extract (5 g・L⁻¹) and Sodium Chloride (5 g・L⁻¹) after 12 h of pre-culture in the same conditions. Lactobacillus casei rhamnosus Lcr35^® was cultured in De Man, Rogosa, Sharpe (MRS) medium (Bio-Rad, Mitry Mory, France) and S. aureus in Brain Heart Broth (Brain Heart Infusion 17.5 g・L⁻¹, Pancreatic digest of gelatin 10 g・L⁻¹, Sodium Chloride 5 g・L⁻¹, Disodium phosphate 2.5 g・L⁻¹, Glucose 2 g・L⁻¹, Biokar diagnostics) under the same conditions than E. coli strains. Cells were centrifuged at 2500 g for 15 min at 4˚C and the pellet was re-suspended in de-ionized water to prevent unintentional increase in cell numbers. The initial population of bacteria was determined by enumeration with a Petroff-Hausser Counting Chamber.

The Microtox^® Procedure employs the bioluminescent marine gram-negative bacterium Vibrio fischeri as test organism. The bacteria are exposed to a range of concentration of the TiO₂ in suspension being tested. The reduction in intensity of light emitted from the bacteria is measured along with standard solutions and control samples. Toxicity is, then, inversely proportional to the intensity of the light emitted after contact with the toxic substances. The change in light output and concentration of the toxicant produce a dose/response relationship. The results are normalized and the EC50 (concentration producing a 50% reduction in light) is calculated.

The basic test protocol (consisting of four test dilutions) was carried out to evaluate the ecotoxicity of the medium containing TiO₂ nanoparticles. All tests were performed using the Microtox 500 Analyser, and bioluminescence measurements were monitored at 0, 5 and 15 min of exposure. The effective concentrations causing 50% of bioluminescence inhibition were computed using the software for Microtox Omni Azur (AZUR environmental, 1998). Toxicity tests were performed in triplicate each week during a two months period and the results are expressed in mg・L⁻¹.

For inactivation kinetics measurements, an amount of 20 mL of de-ionized water was inoculated with Escherichia coli LE392 or Enterotoxigenic Escherichia coli H10407 suspension in order to achieve a concentration of 10⁶ CFU・mL⁻¹ (Colony-Forming Unit by mL). This suspension was placed in a Petri plate with TiO₂ nanoparticles to achieve a final concentration in TiO₂ of 1 g・L⁻¹. The slurries were continuously mixed and irradiated with UV (polychromatic fluorescent UV lamps (©Philips TLD 8 W) providing a total power of 48 W, in a configuration delivering 1.5 mW・cm⁻² at the liquid surface). A complete mixing was done with a sterilized Teflon magnetic stir bar placed in the Petri dish with a speed of 200 rpm.

Sampling of the solutions was done at requisite time intervals (from 0 to 30 min) by pipetting 1 mL from the suspension and serially diluted in 9 mL of Ringer’s solution. After sufficient mixing, 100 µL aliquots of each dilution were plated onto solid Nutrient Gelose medium (Biokar diagnostics) with agar 15 g・L⁻¹. Colony-Forming Units were counted after overnight incubation at 37˚C. All experiments were made in aseptic conditions to prevent any contamination in the media. The counts from three independent experiments corresponding to a particular sample were averaged. The method used for the LC₅₀ tests was similar to that used for inactivation kinetics and was performed on all bacteria strains with nanoparticles concentrations from 50 to 1200 mg・L⁻¹ TiO₂ under 30 min UV irradiation at 1.5 mW・cm⁻².

Generation of hydrogen peroxide by TiO₂ nanoparticules in an aqueous liquid suspension was determined as described by Batdorj et al. [31] with slight modifications. Aqueous solutions of TiO₂ particles concentrations ranging from 0 to 1200 mg・L⁻¹ were placed in Petri plates and continuously mixed in the dark or irradiated with UV for 30 minutes. H₂O₂ concentrations were measured after eliminating the nanoparticles by centrifugation for 15 min at 200000 g. One mL of supernatant was added to a solution containing 100 µl of 4-aminoantipyrine (4 mg・mL⁻¹ solution of 4-amino-2, 3-dimethyl-1-phenyl-3-pyrazolin-5-one, Sigma), 40 µl of water-satured phenol, 60 µl of horseradish peroxidase type VI-A (Sigma, 500 U・mL⁻¹ solution in sodium phosphate buffer pH 6) and 800 µl of phosphate buffer Na₂HPO₄/NaH₂PO₄ (0.1 M, pH 7). The reaction was allowed to proceed for 5 min and the absorbance was measured at 505 nm. The hydrogen peroxide was quantified using a standard curve performed with concentrations ranging from 10 to 200 µM.

Stable titanium dioxide nanoparticles in suspension are fabricated using a derivate sol gel process. Figure 1 shows the main steps of the synthesis protocol. This soft chemistry process is a one-pot, low temperature and efficient method to obtain highly dispersed colloids in a carrying liquid. The first step consists in the chelation of the titanium isopropoxide with an organic ligand, acetylacetone. This reaction is a substitution of alkoxy group of the titanium alcoxyd molecular species by beta diketone ligands. In consequence, the hydrolysis kinetic of the titanium precursor is lowered and undesirable precipitation avoided. Hydrolysis-condensation reactions were carried out by dropping acidified water in the homogeneous medium previously diluted in some isopropylic alcohol. The reacting mixture was heated under reflux for almost 8 hours to obtain a stable dispersion of TiO₂

nanoparticles in an aqueous liquid. Taking into account that the pH of liquid carrier is low, the TiO₂ mineral oxide nanoparticles have a positive surface charge.

A TEM picture and the associated SAED pattern of our as-synthesized sample are presented in Figure 2. The TEM image (Figure 2(a)) shows that most of the particles are elongated, some of them are spherical. From these TEM pictures, the mean crystallite diameter is approximately 8 nanometers.

The SAED patterns of the most intense spots are shown in Figure 2(b). The comparison, in Table 1, between the interplanar distances calculated from the SAED patterns and the tabulated ones obtained for the anatase crystallographic structure exhibits a good agreement and confirms the anatase crystalline structure of our synthesized sample.

The Microtox^® test has been routinely applied to treated waste waters or single compounds and mixtures of inorganic and organic compounds [32] . Furthermore, bioluminescence test becomes a recognized tool to investigate ecotoxicity of nanoparticles [33] . No visible precipitate was observed during the test over the two months period, which confirmed nanoparticles suspension stability. Our results showed EC50 values of respectively 43.75 ± 23.38 mg・L⁻¹ and 36.51 ± 20.55 mg・L⁻¹ at 5 min and 15 min. The calculated EC50 after 5 and 15 minutes exposure time are quite similar. The slight decrease could mean that the nanoparticles need a short time to diffuse into the cells and interact with lipids, carbohydrates, proteins and DNA [34] . Obtained EC50 values for TiO₂ particles are much higher, relative to the literature [35] [36] . This may be due to our particular and original way of synthesis of nanoparticles with the use of acetylacetone which is known as a toxic molecule [37] . Our TiO₂ nanoparticles with EC50 ranging from 36 to 44 mg・L⁻¹, can be classified as harmful to aquatic micro-or- ganisms (EC50 in the range of 10 - 100 mg・L⁻¹) according to the Commission Directive 93/67/EEC from the European Union for the assessment of risk to man and the environment of substances.

We have demonstrated the toxicity of our nanoparticle suspension in the dark on a very sensitive bacterium, Vibrio fischeri.

As in previous studies on Escherichia coli LE392 [38] where we clearly observed the total destruction of bacteria

after only 1 hour of treatment with 1 g・L⁻¹ TiO₂ suspension under UV irradiation, we could wonder what happens during this time duration. Figure 3 shows that after 10 min of treatment, approximately 40% of bacteria tested (E. coli LE392 and ETEC H10407) died. Ten minutes later, we can observe a drastic diminution of the population with around 80% of mortality. Finally, under these particular conditions, we clearly observed the total destruction of both strains of bacteria after only 30 minutes.

Freshly grown bacterial cultures (10⁶ CFU・mL⁻¹) were treated with 1 g・L⁻¹ of TiO₂ and irradiated with UV (1.5 mW・cm⁻²). This experiment was carried out in triplicate. Wang et al. [39] found quite similar results showing that a lower TiO₂ nanoparticles concentration (0.4 g・L⁻¹) had a similar inactivation effect on E. coli but after 2 h UVA irradiation.

The 30 minutes-LC50 tests were then performed on all strains in order to make a comparison between bacteria differing in cell wall structure and detoxification system implicating the catalase enzyme (Table 2).

Concentration-dependent mortality in E. coli exposed to TiO₂ suspension under 30 minutes UV irradiation (1.5 mW・cm⁻²) showed a linear profile for both strains at a concentration ranging from 100 to 600 mg・L⁻¹ (Figure 4). The LC50 calculated by linear regression were 340 mg・L⁻¹ for LE392 and 281 mg・L⁻¹ for ETEC H10407 (Table 2). The TiO₂ nanoparticles suspension had differing inactivation efficiency regarding Lactobacillus casei rhamnosus and Staphylococcus aureus. LC50 was calculated to be 195 mg・L⁻¹ for L. casei rhamnosus 35^® whereas the value of 585 mg・L⁻¹ was determinated for S. aureus. We can observe that concentration-de- pendent survival is higher for the S. aureus gram-positive catalase-positive bacteria compared to the other ones.

The fact that TiO₂ nanoparticles showed a lower effect on S. aureus than on the other ones, under the same conditions, indicates that the resistance of bacteria to TiO₂ nanoparticles is species-dependent. These differences might be due to different structural properties of cell wall and/or a higher self-defense property [40] or self-repair ability of S. aureus than the other ones. Only focusing on the cell wall property of bacteria tested, we can see that the more resistant one is S. aureus which has a gram-positive cell wall. This is in accordance with previous

results [41] [42] respectively using ZnO and Ag nanoparticles, which exhibited a much stronger antibacterial effect on gram-negative bacteria. This difference in antimicrobial activity between gram-positive and gram- negative micro-organisms is often attributed to the structure of their perspective cell walls [43] . On the other hand, our results are not similar with another report using ZnO nanoparticles that showed a much stronger antibacterial effect on gram-positive bacteria than on gram-negative ones [44] - [46] . In addition, van Grieken et al. [47] observed no significant differences between the photocatalytic inactivation of gram-negative and gram- positive bacteria for all experiments and concluded that despite their differences in cell wall structure, both E. coli and E. faecalis showed similar reaction to the treatment. Moreover, in our study, the most sensitive bacterium is Lcr35^® even if this micro-organism belongs to the gram-positive bacteria class. All these results confirm that the cell wall structure is not the primary factor involved in resistance to nanoparticles.

Major constituents of the cell wall are each specific strains and the surface charge of the bacteria is associated with the presence of the ionized groups of the macromolecules [48] . Generally, the cell wall of gram-positive bacteria has a stronger negative charge than gram-negative bacteria. This negative charge is due to the presence of teichoic acid in gram-positive bacteria and lipophosphate in gram-negative ones [49] .

There are reports in the literature that show that electrostatic attraction between negatively charged bacterial cells and positively charged nanoparticles is crucial for the activity of nanoparticles as bactericidal materials. Nanoparticles are capable of penetrating bacterial cells and act as a catalyst, to inactivate enzymes that micro- organisms need for their metabolism by interacting with thiol groups of proteins, disrupt bacterial membranes and also affect DNA replication [50] [51] . In the case of nanoparticles of TiO₂ illuminated with UV, the produced hydrogen peroxide will contribute to this phenomenon [52] . Marugán et al. [40] found that bacteria by themselves had self-protection ability and could grow again after being injured. Therefore, the inactivation of bacteria requires a certain amount of cumulative damage.

In our study, we have to take into account the presence or absence in cells of an enzyme responsible for catalyzing the breakdown of hydrogen peroxide into water and molecular oxygen: catalase [53] [54] . The highest resistance of S. aureus encountered here could be explained by the combination of its cell wall gram+ and the diminution of extracellular and/or intracellular H₂O₂ concentration by catalase.

TiO₂ is a semiconductor [55] which can be excited by UV light. In these conditions, an electron of TiO₂ receives photon energy and is excited [56] . It then reacts with H₂O and/or O₂ and produces hydroxyl radicals and/or active oxygen species [57] . The active species further react with bacteria and inactivate them. The damage cannot be completed in a short time, even though there are enough radicals produced by photocatalytic nano-TiO₂ [39] .

With the aim of evaluating the H₂O₂ production capacity by TiO₂ nanoparticles in the dark or under UV irradiation after 30 minutes, we measured concentration of this molecule with regard with different nanoparticles concentrations (0 from 1200 mg・L⁻¹, Figure 5). The influence of the presence of bacteria on this parameter was also evaluated.

For this experiment, we chose S. aureus because it was the most resistant bacterium among the four tested. Its resistance may be due to a detoxification capacity of the external environment by a catalase activity. Effectively, in order to counteract excess ROS, various antioxidant mechanisms are activated in the organisms. The initial mechanisms that act to adjust antioxidant levels to protect the cells include changes in antioxidant gene expression [58] .

We observed (Figure 5) that in the dark the presence of H₂O₂ was proportional to TiO₂ concentration. The maximum value obtained was 13.23 ± 4.49 µmol・L⁻¹ with 1200 mg・L⁻¹ of nanoparticles in presence of S. aureus. This result shows that even in the dark, the TiO₂ nanoparticles cause the synthesis of hydrogen peroxide. Several studies indicate that certain nanomaterials, including metal oxide nanoparticles, have the potential to exhibit spontaneous ROS production based on material composition and surface characteristics [59] -[61] . The presence of S. aureus did not significantly affect this content. Under these conditions, it was not possible to show a detoxifying activity, by the micro-organism, in its environment.

Under UV irradiation, H₂O₂ concentration obtained was significantly greater than in dark condition. The maximum concentration (35 ± 1.66 µM hydrogen peroxide) was achieved with 1200 mg・L⁻¹ of nanoparticles without S. aureus. As in the dark condition, the bacteria did not change the content of H₂O₂ in their extracellular environment.

The greatest resistance of S. aureus to TiO₂ nanoparticles under UV irradiation is probably due to an intracellular detoxification process and wall thickness properties.