Here, we show that resistance of Escherichia coli to TiO 2 photocatalysis involves defenses against reactive oxygen species. Results support the idea that TiO 2 photocatalysis generates damage which later becomes deleterious during recovery. We found this to be partly due to DNA attack via hydroxyl radicals generated by the Fenton reaction during recovery.Studies in the past few years have revealed that classical disinfection by chlorine or ozonation can generate carcinogenic and mutagenic by-products, thereby boosting research into alternative methods, such as photocatalysis (2,26,28). This process is based on the ability of a semiconductive catalyst (TiO 2 ) to kill bacteria upon illumination in aqueous solution (1,12,15,16,30). However, the basis for the bactericidal effect of photocatalysis is not well established.Active TiO 2 in anatase crystalline form behaves as a classical semiconductor. The bactericidal effect of photocatalysis with TiO 2 could be due to the presence of reactive oxygen species (ROS), such as superoxide (O 2 · Ϫ ), hydrogen peroxide (H 2 O 2 ), and hydroxyl radical (HO·), generated either by illuminated TiO 2 or by the illumination (mainly UV) of the cells. Most studies have concluded that HO·, directly generated by this process, is the main cause of the bactericidal effect of photocatalysis (5, 19).To prevent the harmful effects of ROS generated during the normal course of aerobic metabolism, especially that of the extremely reactive HO· able to damage DNA (18), bacteria like Escherichia coli are equipped with defenses, including catalases (KatG and KatE) and superoxide dismutases (SodA, SodB, and SodC) (4, 17). These defenses decrease H 2 O 2 and O 2 · Ϫ steady states and consequently limit the formation of HO·, for which no defense exists (22). Previous reports have shown that HO· is generated via a Fenton reaction (H 2 O 2 ϩ Fe 2ϩ 3 HO· ϩ HO Ϫ ϩ Fe 3ϩ ) and that regulating iron uptake by the transcriptional repressor Fur (3, 7) permits maintenance of low-level HO· production.Here, we aimed to investigate the resistance of E. coli to TiO 2 photocatalysis. Cells were grown in Luria-Bertani broth at 37°C on a rotary shaker (160 rpm) to an absorbance at 600 nm of 0.5. We then washed the cells twice with sodium phosphate (0.05 mol/liter, pH 7, 4°C) and resuspended them in sodium phosphate (0.05 mol/liter, pH 7) solution to a concentration of 2 ϫ 10 7 CFU/ml. As previously described (14), culture plates were illuminated from 310 nm to 800 nm with a xenon lamp in a Hanau Suntest system (AM1) at 550 W/m 2 light intensity with a filter cutting off wavelengths below 310 nm. Stopped bacterial growth in the exponential phase followed by incubation in phosphate buffer causes starvation and induction of the RpoS regulon, involved in resistance towards many environmental stresses (7,13). With this in mind, we used a mutant strain of E. coli to test whether the induction of the RpoS regulon protects cells against photocatalysis (Degussa P25, 20% rutile 80% anatase crystalline form; Degussa AG, Switz...